Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Use of autofluorescence in the detection of oral premalignant lesions in high-risk populations Ng, Samson Pak Yan 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2004-0579.pdf [ 7.68MB ]
Metadata
JSON: 831-1.0099780.json
JSON-LD: 831-1.0099780-ld.json
RDF/XML (Pretty): 831-1.0099780-rdf.xml
RDF/JSON: 831-1.0099780-rdf.json
Turtle: 831-1.0099780-turtle.txt
N-Triples: 831-1.0099780-rdf-ntriples.txt
Original Record: 831-1.0099780-source.json
Full Text
831-1.0099780-fulltext.txt
Citation
831-1.0099780.ris

Full Text

ABSTRACT Objectives: The current gold standard and more recent molecular markers in prediction of cancer risk for oral premalignant lesions (OPLs) rely upon biopsy samples of the OPLs. ' Clinicians' decision of when and where to biopsy would therefore heavily influence the results of histology and molecular markers. Unfortunately current clinical parameters used to make such decisions are far from adequate. The, objective of this thesis was to establish a multi-spectral fluorescent visualization (FV) technique and examine the potential role of FV as a visual aid to identify high-risk OPLs by examining the relationship between FV positivity and the conventional clinicopathological risk factors, including (1) cancer history, (2) clinical risk factors, (3) Toluidine blue (TB) staining, and (4) pathology. Materials and Method: 143 patients with a history of oral dysplasia (61) or oral cancer (82) were recruited from an ongoing prospective study (Oral Health Study) in the Oral Dysplasia Clinic at British Columbia Cancer Agency. Information collected included: FV results, demographics (age, gender, ethnicity), tobacco habit, history of oral cancer, TB staining, and clinicopathological findings. The FV examination was performed in a dark room. The oral mucosa was illuminated by an external light source with excitation wavelengths of 400 - 460 nm (violet/blue). The autofluorescence of oral tissue (normally pale green, defined as F V negative) was visualized by the examiner wearing glasses with longpass and notch filters that would only allow passage of green and red/orange fluorescence. Loss of the normal pale green colour or change of the colour to orange red are defined as FV +. Results: Of the 143 patients, 83 showed 1 or more FV positive oral sites and 60 showed no F V positive sites. There were no differences in the demographics and habits between FV positive and negative patients. A total of 319 oral sites from the 143 patients were examined with FV and 138 (43%) of these were FV positive. Cancer history of patients: There was no difference in either the % of patients with positive FV or the % of positive FV sites between patients with or without a history of oral cancer. Clinical risk factors (clinical diagnosis of OPLs): FV positive lesions showed a higher cancer risk as judged by clinical risk factors or diagnosis. Compared to FV negative lesions, FV positive lesions showed a significantly lower % of lesions diagnosed clinically as normal or equivocal (P < 0.0001), higher % of lesions diagnosed clinically as OPLs (P < 0.0001), were larger (P = 0.018) and more likely to be nonhomogeneous in appearance (P= 0.017). TB staining: FV positive lesions showed a higher cancer risk as judged by TB staining. A significantly higher % of FV positive lesions were TB positive compared to the FV negative lesions (P < 0.0001). Pathology: FV positive lesions showed a higher cancer risk as judged by pathology. A significantly higher % of FV positive lesions were dysplastic as compared to the FV negative lesions (P < 0.0001), and 91% of the 31 high-grade dysplasias were FV positive. Of 10 paired biopsies of oral lesions that were partly FV positive and negative (1 from FV + area, and another from FV - area but the same clinically similar-looking lesion), 6 pairs showed that the FV positive biopsy has the same histological diagnosis as the matching FV negative biopsy, and 4 showed that the FV positive biopsy with higher histological severity than the FV negative matching biopsies. Conclusion: This is the first study to investigate direct FV as a tool to identify high-risk OPLs in a large number of patients. As judged by clinical risk factors, TB staining and pathological findings, F V positivity seemed to identify high-risk OPLs. FV, a non-invasive procedure, could be a valuable visual aid to facilitate the decision of when and where to biopsy oral lesions. n TABLE OF CONTENTS Abstract i i Table of Contents i i i List of Tables vi List of Figures vii List of Abbreviations viii I. Introduction 1 1.1. Overview 1 1.2. Histology of Oral Mucosa 1 1.3. Etiologies of Oral Cancer 2 1.4. Oral Premalignant Lesions (OPLs) and Cancer Risk Factors 3 1.4.1. Clinical risk factors 3 1.4.2. Pathological risk factors 8 1.4.3. Need for visual aid 10 1.5. Mechanism of Fluorescent Visualization (FV) 11 1.5.1. Tissue optics 11 1.5.2. Interaction of oral tissue with light 14 1.6. Autofluorescence Examination as an Adjunct Tool in the Identification of Malignant and Premalignant Lesions 19 1.6.1. Direct fluorescent visualization as a visual tool 20 1.6.2. A F E as a tool for the identification of precancerous and cancerous lesions in other organs 21 1.6.3. A F E in oral premalignant lesions and cancer 23 II. Statement of the Problem 31 III. Objectives 32 IV. Null Hypothesis 33 V. Materials and Methods 34 5.1. Subject Selection 34 5.2. Devices 34 i i i 5.2.1. Direct fluorescent visualization (FV) examination device 34 5.2.2. Digital camera 35 5.3. Clinical Examination 35 5.3.1. Sequence of the examination 35 5.3.2. Questionnaire 36 5.3.3. White light examination 36 5.3.4. Autofluorescent examination 36 5.3.5. Toluidine blue examination 37 5.4. Histopathological Evaluation 37 5.5. Statistical Analysis 37 VI. Results 39 6.1. Demographics/habits 39 6.1.1. FV and demographics/habit information 39 6.1.2. Cancer history and demographics/habit information....) 40 6.1.3. Cancer history and FV 41 6.1.4. Cancer history, FV and demographics 41 6.2. Clinical attributes : 42 6.2.1. FV and clinical attributes 42 6.2.2. Cancer history and clinical attributes 44 6.2.4. Cancer history, FV and clinical attributes 45 6.3. TB staining 47 6.3.1. F V a n d T B 47 6.3.2. Cancer history and TB 48 6.3.3. Clinical attributes of OPLs and TB 48 6.3.4. Cancer history, clinical attributes and TB 49 6.4. Histological features 52 6.4.1. FV and histological features excluding dysplasia 52 6.4.2. FV, clinical diagnosis, TB staining and dysplasia 54 6.5. Combined diagnostic tools (clinical diagnosis, FV and TB staining) in identifying dysplasia 61 6.5.1. One diagnostic tool 61 6.5.2. Two diagnostic tools 61 6.5. Correlation of FV status with dysplasia in paired biopsy 64 VII. Discussion 66 7.1. Cancer history and diagnostic tools 66 7.2. FV positive lesions have increased cancer risk: Clinical prospective 67 7.3. FV positive lesions have increased cancer risk: TB staining 68 7.4. FV positive lesions have increased cancer risk: Histo-pathological prospective 69 7.4.1. FV and dysplasia 69 7.4.2. Clinical diagnosis and dysplasia 70 7.4.3. TB staining and dysplasia 71 iv 7.4.4. Use of FV to locate biopsy sites 71 7.4.5. Use of FV to screen subclinical lesions 72 7.4.6. Use of multiple diagnostic tools in identifying and monitoring dysplastic lesions.... 72 7.5. FV predicts histological features of biopsy specimens 74 7.6. Mechanism for selective staining by FV of OPLs 75 7.7. Future investigation 76 7.7.1. FV and outcome 76 7.7.2. FV and screening of the oral cavity 76 7.7.3. Limitations of current FV findings 76 7.9. Summary and Conclusion 76 Appendix 1: Oral Study Questionnaire 78 Appendix 2: Clinical Characteristics of OPLs 82 Appendix 3: Fluorescent visualization device 84 Appendix 4: FV positive lesion 84 Appendix 5: Published Abstract 85 References 86 v LIST OF TABLES Table 1. Classification of oral premalignant lesions 6 Table 2. Use of A F E to identify oral SCC and OPLs 27 Table 3. Sequence of steps for data collection of new patients 35 Table 4. FV and demographic/habit information 40 Table 5. Cancer history and demographic/habit information 41 Table 6. Cancer history and FV 41 Table 7. Demographic features of patients with or without a history of Head & Neck cancer42 Table 8. Association of FV with clinical attributes 44 Table 9. Association of cancer history with clinical attributes 45 Table 10. Association of FV with clinical attributes of primary lesions 46 Table 11. Association of FV with clinical attributes of mucosal sites in patients with a history o f H & N cancer 47 Table 12. Association of FV with TB staining 48 Table 13. Cancer history and TB 48 Table 14. Association of TB with clinical attributes 49 Table 15. Association of TB with clinical attributes of primary lesions (patients with no history of H & N cancer) 51 Table 16. Association of TB with clinical attributes of mucosal sites in patients with a history of H & N cancer 52 Table 17. FV and histological features (other than dysplasia) 54 Table 18. Sensitivity and specificity of FV (judging by gold standard histology) 55 Table 19. FV and degree of dysplasia 56 Table 20. Sensitivity and specificity of clinical diagnosis (judging by gold standard histology) 57 Table 21. Clinical diagnosis and degree of dysplasia 58 Table 22. Sensitivity and specificity of TB staining (judging by gold standard histology) 59 Table 23. TB staining and degree of dysplasia 60 Table 24. Combined diagnostic tool (clinical diagnosis, FV and TB staining) in identifying dysplasia 63 Table 25. . Correlation of F V status with dysplasia in paired biopsy 65 Table 26. Correlation of FV status with dysplasia in paired biopsies (only samples from clinically visible lesions) 65 v i LIST OF F I G U R E S Figure 1. Different degrees of dysplasia 9 Figure 2. Electromagnetic spectrum 12 Figure 3. Chemical structure of oxyhaemoglobin showing conjugated double bonds 13 Figure 4. Absorbance spectrum of haemoglobin 14 Figure 5. Structure of nicotinamide adenine dinucleotide (NAD + ) 15 Figure 6. Chemical structure of flavin adenine dinucleotide (FAD) 16 / vii LIST OF A B B R E V I A T I O N S AFE: Autofluorescent examination B C C A : British Columbia Cancer Agency CCD: Charge-coupled device CIS: Carcinoma in situ EMR: Electromagnetic radiation FAD: Flavin adenine dinucleotide F O M : Floor of mouth FS: Fluorescent signal FV: Fluorescent visualization FV +ve: Fluorescent visualization positive FV -ve: Fluorescent visualization negative LP Lichen planus N A D : Nicotinamide adenine dinucleotide OPLs: Oral premalignant lesions PAP: Papanicolaou smear SILS: Squamous intraepithelial lesion site TB: Toluidine blue UBC: University of British Columbia U V : Ultraviolet ray WHO: World Health Organization ViU I. INTRODUCTION 1.1. Overview Oral cancer is the 6 t h most common malignancy in the world. In Western countries it accounts for up to 6% of cancers (Johnson et al. 1998). In Canada, about 3,100 new cases of oral cancer arise every year, and 1,050 of these result in death (National Cancer Institute of Canada 2004). For the United States, nearly 19,400 new cases (1.5% of all new cancer cases) were diagnosed in a year and 5,000 people die of previously diagnosed oral cancer (American Society of Cancer 2003). In comparison with the West, the rates of oral cancer are much higher in Southeast Asia, accounting for up to 40% of all incidences. This high incidence is believed to be associated with the widespread practice of chewing tobacco in this region. In contrast, in Western countries, 80-90% of oral carcinomas are believed to be associated with a combination of smoking and alcohol habits (Marshall et al. 1992). Despite improvements in surgical, chemotherapeutic and radiation therapies, the 5-year survival rate of oral cancer patients has not improved in the last decade. It has remained at about 50% (Khuri et al. 2001). This is one of the lowest survival rates among the major human cancers. This poor prognosis is believed to result from the detection of tumours at a late stage and is associated with a high recurrence rate and an increased incidence of second primary cancer (Shiboski 2000; Khuri et al. 2001; Casiglia et al. 2001). Current strategies to improve this poor prognosis rely heavily on the identification and appropriate management of high-risk oral premalignant lesions (OPLs) with the goal of preventing progression of OPLs into invasive cancer. Although a number of clinicopathological characteristics are believed to have predictive value for the risk of malignant transformation of OPLs and they are currently used to guide clinicians in the decision of biopsies and management, these criteria are far from adequate. There is a need to develop other methods to facilitate the identification of high-risk OPLs. This study investigated direct fluorescent visualization (FV), a method of detecting autofluorescence in the tissue, as a visual aid in the detection of high-risk OPLs. It is theorized that during early carcinogenesis, the biochemical and morphological alterations that occur could result in changes in the normal autofluorescence, hence a means of early detection. This is the first study of direct FV in the oral cavity to focus on the OPLs with a large number of patients (JV = 143). The following introduction has two sections. The first introduces the current problems in the identification of high-risk OPLs by briefly reviewing the histology of the oral mucosa, cancer etiologies, clinicopathological changes during oral carcinogenesis, and their usage and limitations as risk factors in identifying high-risk lesions. The second introduces direct FV as a potential visual tool for high-risk OPLs by reviewing the concept and mechanism of autofluorescence, techniques of detection, and early study results of FV. 1.2. Histology of Oral Mucosa The oral cavity is lined with a moist mucosa, which is composed of stratified squamous epithelium and a lamina propria commonly known as the connective tissue. The physical barrier that separates the overlying epithelium from the underlying connective tissue is called the 1 basement membrane. It is an extracellular matrix, which provides mechanical support for epithelial cells. This membrane consists of two layers: the basal lamina and the lamina reticularis. The former is produced by the epithelium, and the latter is produced by the connective tissue (Burkitt 1993). The overlying stratified squamous epithelium is composed of four cell layer types: the stratum germinativum (stratum basale), the stratum spinosum (prickle cell layer), the stratum granulosum (granular layer), and the stratum corneum (cornified layer). The different layers of the oral epithelium represent a progressive maturation process. The cuboidal-shaped basal cells form a single-cell layer that exists between the epithelium and connective tissue. The cells within this basal cell layer contain progenitor cells that have the capacity to divide and give rise to more new basal cells or differentiate into prickle cells. As the cells mature, they migrate toward the surface, changing their shapes into more elongated and flattened forms. Once reaching the surface, they are eventually desquamated (Burkitt 1993; Berkovitz et al. 2002). Like many other parts of the body with mucosa lining such as esophagus and cervix, the majority of the oral cavity lining is not keratinized. The mucosa of the cheeks, lips, alveolus, floor of mouth, ventral surface of tongue, and soft palate is of this kind; it has a loose lamina propria, and is referred to as lining mucosa. However, some oral mucosal regions susceptible to mechanical forces, such as gingiva and hard palate, are characterized by a keratinized or parakeratinized epithelium and a thick lamina propria, and are referred to as the masticatory mucosa. This mucosa serves as an effective mechanical and permeability barrier. The dorsum of the tongue is composed of a specialized epithelium that is a mixture of both nonkeratinized and keratinized tissues, which are attached tightly to the underlying tongue muscle. It is referred to as the gustatory mucosa (Burkitt 1993; Berkovitz et al. 2002). The underlying connective tissue holds blood and lymphatic vessels as well as nerves and muscle fibres. This basic pattern of an epithelium and supporting connective tissue is analogous to the pattern of the epidermis and dermis of the skin. The bulk of the connective tissue is comprised of fine and coarse collagen. In normal tissue, collagen fibres are thin and loosely arranged in the superficial papillary layer while the deep reticular layer is dominated by thick, parallel bundles of collagen fibres. Elastin is another constituent that forms a fine interlacing network of fibres which follow the path of the interlacing collagen bundles. Over 90% of oral malignancies, excluding salivary gland tumors, develop from this stratified squamous epithelial tissue; hence the name squamous cell carcinoma (SCC) (Shiboski 2000). 1.3. Etiologies of Oral Cancer Many etiological factors are believed to be associated with carcinogenesis. This section discusses the two main factors: tobacco and alcohol. Tobacco is the most well-known and influential risk factor involved in cancer development. The correlation between tobacco use and oral cancer, as well as lung cancer, has been investigated in great detail, and most findings confirm the etiological role of tobacco in this disease. In Southeast Asia, particularly in India, the prevalence of chewing tobacco is believed to be responsible for the high incidence of oral cancer. In contrast, cigarette smoking and drinking are 2 thought to be the cause of this disease in the Western world. This combination of personal habits can increase the cancer risk by 3 to 15 times in the general population (Marshall et al. 1992). At least 2,000 chemicals have been identified in processed tobacco (Pershagen et al. 1996). Some of those chemicals function as initiators and some as promoters in carcinogenesis. Thus, a long duration of smoking is more likely to increase the risk of malignant transformation than a short exposure. The long duration gives initiating and promoting chemicals enough time to induce damage to D N A in cells. Another important risk factor related to oral cancer is alcohol consumption. Numerous studies have shown an increased oral cancer risk with drinking, and the risk tends to increase in a dose-response fashion (Marshall et al. 1992). Most studies focused on its combined effect with smoking rather than the effect of alcohol alone. When these two habits are combined, the cancer-initiating and promoting effects appear to multiply (Marshall et al. 1992). In summary, the most significant known etiological factors for oral cancer are tobacco and alcohol. Evidence of the relationship of such habits to oral cancer risk is extensive. Epidemiological evidence indicates that it generally takes decades of exposure to tobacco and alcohol for the development of precancerous and cancerous lesions. The length of the process allows early intervention i f carcinogenesis can be identified. 1.4. Oral Premalignant Lesions (OPLs) and Cancer Risk Factors In 1978, the World Health Organization proposed the universal definition of a premalignant lesion as "a morphologically altered tissue in which cancer is more likely to occur than in its apparently normal counterpart." As implied in the definition of the premalignancy, these lesions have an increased likelihood of cancer; however, the majority of premalignant lesions do not become cancerous. A number of clinicopathological features including morphological alterations of OPLs are known to have predictive value of the malignant risk of oral premalignant lesions. However, these risk factors or parameters are far from adequate in risk prediction. Our current ability to predict which OPLs will undergo malignant transformation is limited. The following sections discuss these risk factors and their limitations. 1.4.1. Clinical risk factors Current diagnoses of OPLs involve clinical identification of OPLs based on clinical risk factors, biopsies of the lesions and pathological assessment of the lesions. There are four main clinical risk factors: clinical appearance, location, lesion size, and history of head and neck cancer. These are discussed below along with some other risk factors. 1.4.1.1. Clinical appearance of OPLs and cancer risk In the oral cavity, most premalignancy presents as a white patch, also known as leukoplakia. Leukoplakia is defined as "a white patch or plaque that cannot be characterized clinically or pathologically as any other disease" (WHO, 1978). The less common clinical presentation is as a red patch, also known as erythroplakia. By definition this lesion is "a bright red, velvety plaque which cannot be characterized clinically or pathologically as any other definable lesion" (WHO 3 1978). Since the majority of OPLs are leukoplakia, the term leukoplakia has been used interchangeably with the term OPL, which includes both leukoplakia and erythroplakia. Leukoplakia can be classified into two major categories according to clinical appearance. Homogeneous leukoplakias are most common and refer to leukoplakias that are uniform in color and texture. They are predominantly white and have a smooth or slightly corrugated in texture. The other type, referred to as nonhomogeneous leukoplakias, includes lesions that are nonhomogeneous in color and/or texture. The surface may be nodular or verrucous and the lesion may appear speckled (red/white) There is strong evidence that nonhomogeneous leukoplakias pose a higher cancer risk than homogeneous leukoplakias. It is generally believed that nonhomogeneous leukoplakias have only a small cancer risk, speckled or nodular/verrucous leukoplakias have a high cancer risk and erythroplakias have the highest cancer risk with many of them already being cancer (Axell et al. 1996; Rajendran et al. 1989; Pindborg et al. 1968; Shafer and Waldron 1961). For example, Rajendran et al. (1989) reported that 48% of all carcinomas had a verrucous characteristic (numerous pointed projections), which is one of the features of a nonhomogeneous lesion. Pindborg et al. (1968) reported that 7 of 11 leukoplakia patients who developed oral cancer between 1955 and 1964 had speckled leukoplakias. A similar correlation between speckled leukoplakia and carcinoma was reported by Shafer and Waldron (1961). Because of the strong correlation between clinical appearance (homogenous vs. nonhomogenous) and the cancer risk of OPLs, many of the proposed staging systems include the clinical appearance criteria (Schepman and Van der Waal, 1995; Axell et al. 1996). It has been proposed that homogeneous leukoplakias represent early stage lesions that may progress inte to more worrisome nonhomogeneous leukoplakias (Bouquot and Whitaker 1994). Moreover, cancer is currently viewed as an acquired genetic disease driven by the accumulation of specific genetic alterations. It can be further hypothesized that as a lesion progresses, it acquires more genetic alterations, which induce phenotypical changes. In other words, with genetic change, it is more likely for lesions to develop more fissures, projections and erythema, which define the characteristics of late-stage leukoplakias. This implies that the clinical features seen in the late phases of the disease are likely to possess more genetic alterations, thus they tend to have a higher cancer risk than those in the early phases. More research is required in this area, as no longitudinal study has been done to investigate the correlation between clinical appearance and cancer progression. While the clinical appearance of OPLs has cancer risk prediction value, particularly in those cases of obvious nodular/verrucous leukoplakia or erythroplakia, clinical appearance is frequently inadequate in judging the cancer risk of OPLs for the following reasons: (1) OPLs are not always clinically visible, hence one could not use clinical appearance to judge the risk of the OPLs in these cases; (2) clinical appearance is not always an adequate predictor since even the most innocuous homogeneous leukoplakia could turn out to be carcinoma in situ (CIS); (3) lesions with less striking nonhomogeneous appearance could be easily misdiagnosed as homogeneous leukoplakia. Likewise, homogeneous leukoplakia with a slightly rough surface would be easily misdiagnosed as nonhomogeneous leukoplakia; and (4) most important of all, many reactive lesions with no precancerous predisposition can mimic leukoplakia (e.g., reactive hyperplasia and frictional hyperkeratosis) or erythroplakia (e.g., inflammation). Reactive lesions are common and are frequently misdiagnosed as leukoplakia and visa versa. These problems are reflected in a recent proposal for staging of OPLs, in which the clinical appearance was not used as a risk factor (Schepman and Van der Waal 1995). 4 1.4.1.2. Site of OPLs and cancer risk The location of an OPL has a high cancer risk predictive value. The oral cavity could be divided into high and low-risk sites. For countries in the Western world the high-risk oral sites are the floor of mouth (FOM), soft palate complex (Shiboski et al. 2000). The rest of the oral cavity is designated as low-risk. There is strong evidence to support the notion that the location of a lesion is a high risk predicting factor, which has been included as one of the risk predictors in 3 proposed staging system for OPLs (Axell et al. 1996; Pindborg et al. 1968; Schepman and Van der Waal 1995). Epidemiological investigations in the Western world have consistently shown that cancerous and pre-cancerous lesions reside in the high-risk sites. Studies have found that most oral cancer occurs in the horseshoe shaped high-risk regions. Mashberg and Meyers (1976) observed that 97.5% of 222 cancer cases developed at high-risk sites. This finding was confirmed by the same research group in 1986. The latter investigation showed that 84% of cancers were present on the F O M , the ventral tongue, and the soft palate complex. Waldron and Shafer in 1975, with 3,256 cases of leukoplakia, showed that 43% of biopsies from F O M and 24% from tongue were dysplasia and carcinoma, compared to a much lower rate of dysplasias and cancer in other sites of the oral cavity. Dysplastic lesions in high-risk sites are more likely to progress to cancer than those located in the low-risk sites despite-similar degrees of dysplasia. In a follow-up study by Kramer et al. (1978), 24 of 46 patients (52.2%) had carcinoma in the initial biopsies or developed carcinoma in biopsies attained from the F O M or the ventrolateral tongue during a follow-up period ranging from 1 to 19 years. Recent molecular studies support this notion. A study from Zhang et al. (2001) demonstrated that OPLs from high-risk sites are more likely to contain high-risk molecular patterns than morphologically similar OPLs from the low-risk sites. Though the lesion site is known to be one of the major factors affecting cancer risk, no one has provided a satisfactory explanation for the divergence of malignant potential in different regions of the mouth. A number of theories have been proposed. One popular theory is that high-risk sites are generally located in the lower part of the oral cavity (floor of mouth and ventrolateral tongue). These areas are generally bathed in saliva allowing for prolonged exposure to dissolved carcinogens. Another theory is that keratinization in many of the low risk sites (e.g., hard palate, gingiva, and cheek) may prevent carcinogen penetration to the dividing basal cells. In contrast, high-risk sites are usually covered by a fragile, thin squamous epithelium with very little or no keratin. An important factor for the high-risk predictive value of the site is that lesions located in the high-risk sites are much less likely to be reactive since floor of the mouth, ventral tongue and soft palate are much less likely to be traumatized than the cheek, gingiva and hard palate. The latter factor becomes less helpful, however, in the cases of patients with a prior history and treatment of oral cancer since reactive lesions frequently occur in high-risk areas after aggressive cancer treatment. Despite the high rate of malignant transformation for OPLs in the high-risk sites, many OPLs in the high-risk sites do not progress. Likewise, a small percentage of lesions at low-risk sites do progress. Tools to assist clinicians to further triage cancer risk of OPLs are needed. 5 1.4.1.3. Size of OPLs and cancer risk Another clinical factor that predicts cancer risk of OPLs is the size of the lesion. The bigger the premalignant lesion, the higher the cancer risk. Size refers to the combined size of all leukoplakias in the oral cavity. The cutoff size for different risks remains speculative. In 4 proposed staging systems for OPLs, it has been proposed that judging by size alone, an OPLs with a dimension less than 2 cm has the lowest cancer risk. A lesion measuring 2—4 cm has an intermediate risk and lesion larger than 4 cm high cancer risk (Schepman and van der Waal, 1995; van der Waal et al. 2000; Axell et al. 1996; Pindborg et al., 1968). Most OPLs measure less than 2 cm and have a low cancer risk. No one is clear what the actual risk value of the size of OPLs is. This is reflected in the fact that while all 4 proposed staging systems for OPLs included size as a risk factor, only 1 actually used size to categorize the risk of OPLs (see table below, Schepman and van der Waal, 2000). The other 3 staging systems gave no consideration to the risk value of size in the staging of OPLs (Schepman and van der Waal 1995; Axell et al. 1996; Pindborg et al.1968). Table 1. Classification of oral premalignant lesions Schepman and van der Waal 1995 Axell et al. 1996; Pindborg et al. 1968 van der Waal 2000 Stage 1 Any L, S b C,, P, or P2 Any L ,C , , P, or P 2 Lb Po Stage 2 Any L, S,, C 2 , P, or P2; any L, S2, C b P, or P2 Any L, C 2 , P, or P 2 L 2 5 Po Stage 3 Any L, S2, C 2 , P, or P2 Any L, any C, P 3 or P 4 L3, Po or L], L 2 , Pi Stage 4 Any L, any S, any C, P3 or P 4 U, Pi L: Leukoplakia size (Li < 2 cm; L 2 = 2-4 cm; L 3 > 4 cm) C: Clinical appearance of leukoplakia (Ci: homogeneous; C 2: non-homogeneous) P: Pathological features (P,: no dysplasia; P2: mild dysplasia; P3: moderate dysplasia; P4: severe dysplasia) S: Site of leukoplakia (S^ all oral sites except the floor of mouth and the tongue; S2: floor of the mouth and/or tongue) 1.4.1.4. Duration of OPLs and cancer risk It is not possible to discuss size without discussing the duration of the lesion. Not only is size important as a risk factor, but the change in size over time is also important. When a lesion is followed over time, it can increase or decrease in size, persist or completely disappear. Pindborg et al. (1968) followed 248 leukoplakia patients for 9 years. They reported that 20.1% of the leukoplakias disappeared completely and 17.8% showed a reduction in size. Only 3.3% of all leukoplakia increased in size, and 3.7%) developed into carcinoma. Unfortunately, this study failed to report whether leukoplakias that later became carcinomas had increased in size. Also, there was no report on whether the initial size of a lesion was associated with clinical outcome (complete regression, partial regression, or progression to cancer). Such additional information would be needed to confirm the correlation between size and cancer risk. Generally, a lesion that decreases in size or disappears completely over time or over the withdrawal of carcinogen (e.g., 6 quitting smoking) has a much lower risk than a similar sized lesion that increases in size or persists. Another general view is that the longer a lesion persists (remains the same size or increases in size), the higher the cancer risk. Conceptually, this makes sense since older lesions have a longer time to acquire genetic and histological changes than younger lesions. However, the literature varies in the reported time between initial identification of leukoplakia and diagnosis of cancer (duration of OPLs). Some reported that most carcinomas with leukoplakia originally developed 2 to 4 years after the onset of a white plaque, while others occurred decades later. Rarely did cancer occur quickly (Bouquot et al. 1988; Silverman et al. 1984). Others reported a shorter transition time. In one study (Leonardelli and Talamazzi 1950), 49% of the patients developed cancer within 3 months of lesion identification and 62% within 1 year. In another study, Pindborg et al. (1968) found that 36% of subjects developed malignancy within 1 year and 64% after 1 year. Of the 64%, half developed cancer after 3 or 4 years. Again the risk factor of duration has limited value in cancer prediction. 1.4.1.5. History of head and neck cancer and cancer risk Individuals with a history of aerodigestive tract cancers have an increased risk (10%-30% higher) of a second primary cancer or locoregional recurrence (Grant et al. 1993). This poor prognosis is generally attributed to undetected microinvasion of the primary tumour leading to an outgrowth of malignant cells after treatment and, eventually, recurrence at the former cancer site. The other possibility is multifocal carcinogenesis as a result of field cancerization. Because of this high cancer risk, patients with a history of head and neck in its cancer are followed up long term. A history of head and neck limited value as a risk factor. Patients with a history of oral cancer generally have a tendency to develop reactive lesions even to minor trauma. This is attributed to the effects of cancer that leaves the oral tissue fragile and slow to heal following trauma (Epstein and Scully 1997). For example, radiation therapy alters tissue appearance and decreases cellularity and vascularity, which predisposes the mucosa to breakdown that could be mistaken for an erythroplakia. If a biopsy is taken from such a lesion, the healing process is further delayed. Therefore, physicians tend to be reluctant to perform biopsies in these patients, and this frequently results in a delay of the diagnosis (Epstein and Scully 1997). Hence, any visual aid that can assist in detecting carcinomas and malignant margins would be in great demand in the medical field, as it would improve the survival rate and prognosis for oral cancer patients. 1.4.1.6. Other clinical risk factors for oral cancer Habits and genetic predisposition: Heavy smoking and drinking are strongly associated with oral cancer in the Western world. In this sense, heavy smokers and drinkers, particularly those of older age, are at risk for oral cancer, although the majority of these people do not develop cancer. Ironically, OPLs found in people without apparent etiologies (nonsmoker and social drinker) have higher cancer risk than those OPLs in heavy smokers and drinkers. It is generally presumed that these individuals are genetically predisposed to oral cancer. 7 Gender: Although oral cancer is more common in males than in females (Silverman 1994; van der Waal et al. 1997), OPLs in females have higher cancer risk than OPLs in males. The exact mechanism for this is unclear. Candidiasis: Candidiasis is a fungal infection. Studies have suggested that oral leukoplakias with associated candidiasis are at a higher risk of developing into cancer than those without Candida infection, thus raising the possibility that candidiasis of oral leukoplakia is a risk predictor (Krogh et al. 1987). However, it is not clear whether candidiasis is an independent risk factor, or it only reflects nonhomogeneous leukoplakia and dysplasia, since most of leukoplakia with candidiasis fall into these categories. 1.4.2. Pathological risk factors 1.4.2.1. Dysplasia and cancer risk Clinical risk factors help clinicians to decide whether a lesion needs to be biopsied and where to biopsy. The biopsy sample will then be sent for histopathological evaluation. Currently, histological diagnosis is the gold standard for determining the cancer risk for premalignant lesions. The gold standard or pathological risk factor is based mainly on the presence and degree of histological changes called "dysplasia." Dysplasia, as Lumerman describes, "is the diagnostic term used to describe the histopathological changes seen in a chronic, progressive, and premalignant disorder of the oral mucosa" (Lumerman et al. 1995). The microscopic changes used to diagnose dysplasia are the following (WHO 1978): • Loss of polarity of the basal cell • Presence of more than one layer of cells having a basaloid appearance • Increased nuclear-cytoplasmic ratio • Drop-shaped rete processes • Irregular epithelial stratification • Increased number of mitotic figures • Presence of mitotic figures in the superficial half of the epithelium • Cellular pleomorphism • Nuclear hyperchromatism • Enlarged nucleoli • Reduction of cellular cohesion • Keratinization of single cells or cell groups in the prickle layers Lesions with dysplasia are further classified into the categories of mild, moderate and severe. As shown in figure 1, the degree of dysplasia is determined by the extent of spread of dysplastic cells in the epithelial layers. In mild dysplasia, the cytological and architectural changes are seen in the lower third of the epithelium; in moderate dysplasia, such changes are seen in the lower half of the epithelium; in severe dysplasia, the dysplasia involves the lower two third of the epithelial layers. 8 Figure 1. Different degrees of dysplasia. The presence of dysplasia is a cancer risk predicting factor. Leukoplakia showing dysplasia is considered to have a higher risk for malignant transformation than those without dysplasia (Waldron and Shafer 1975; Banoczy et al. 1976; Lumerman et al. 1995). Epstein and Scully (1997) reported that 43% of dysplastic leukoplakia progressed to malignancy. Another study conducted by Silverman et al. in 1984 found that 36%> of leukoplakia with dysplasia became cancerous after 7.2 years, compared with only 15% of leukoplakia without dysplasia. The degree of dysplasia is an important factor of cancer risk. The higher the degree of dysplasia an OPL has, the higher the cancer risk. Literature supporting this comes from studies of the oral cavity and other organs such as uterine cervix, lung, esophagus, skin, pharynx, and larynx. There is a well-accepted histological progression model. As shown in figure 1, oral SCC is believed to progress from hyperplasia to an increasing degree of dysplasia, to CIS, and finally to an invasive carcinoma. CIS, a lesion regarded as cancer, is in fact closer to severe dysplasia in both pathology and clinical behaviour than to invasive SCC. Like severe dysplasia, changes of CIS are confined within the epithelium without destruction of the basement membrane. A CIS generally has dysplastic changes involving all epithelial layers (bottom-to-top). When cell changes are pronounced a diagnosis of CIS can also be made when the dysplastic changes only involve the lower two-thirds of the epithelium. Severe dysplasia and CIS have a high chance of progressing to an invasive cancer. A diagnosis of invasive SCC is based on the invasion of malignant cells through the basement membrane, the destruction of underlying connective tissue, and growth of tumour islands in the connective tissue. The presence and degree of dysplasia are the time tested gold standard in the diagnosis of cancer risk for many organs. These findings provide clinicians with excellent guidance in the management of severe dysplasia and CIS since these lesions have a high chance of progression into invasive lesions. However, even the gold standard has its limitations, since it is not good for cancer risk prediction of OPLs without dysplasia or with low-grade dysplasia. As mentioned previously, only a small percentage of such lesions become cancerous, and the model has no way of predicting which of these low-grade lesions have malignant potential. 9 A further limitation of dysplasia as a diagnostic criterion is that the identification of dysplasia require biopsy of a clinically visible lesion. Unfortunately, dysplasia per se may not be visible clinically. It is only visible when it is accompanied by an increase in epithelial or keratin thickness or both (leukoplakia) or a decrease in epithelial thickness (erythroplakia). Visual aids that do not depend upon clinical appearance of leukoplakia to identify dysplasia are needed. 1.4.3. Need for visual aid In summary, current clinicopathological risk factors are limited in value in the risk prediction of OPLs. This is strongly supported by the fact that there are huge variations in the reported rate of" cancer transformation of OPLs, varying from 0.13% to over 50% (Silverman et al. 1984; Rosati 1994; Bouquot et al. 1994; Roz et al. 1996; Papadimitrakopoulou et al. 1997; Lee et al. 2000; Soukos 2001; Amagasa et al. 1985), reflecting the huge differences in the diagnosis of OPLs. Other risk predictors are needed to further evaluate and categorize the malignant risk of OPLs. The development of molecular techniques has prompted scientists and clinicians to incorporate molecular techniques into such evaluations. One such technique, microsatellite analysis, has already been shown to produce genetic profiles for OPLs that are strong predictors of outcome (Rosin et al. 2000, 2002). Unfortunately, the value of histologic and molecular analyses are limited by the ability of the clinician a representative tissue sample. Since localization of biopsies can be difficult, based on clinical appearance alone, the development of new approaches, such as a visual aid that can facilitate the selection of high cancer risk sites and biopsy sites, is crucial. Toluidine blue (TB) is a vital stain that has been used as a visual aid for years. Many clinicians believed that its merit lies mainly in detecting early cancer or high-grade preinvasive lesions such as carcinoma in situ (CIS) and severe dysplasia. Its value in the diagnosis of low-grade (mild/moderate) dysplasia (representing the majority of oral premalignant lesions, OPLs) has been questioned since a significant number of these lesions cannot be stained with TB (Epstein et al. 2003; Epstein and Scully 1997). Recent studies, however, indicate that TB-positive low-grade dysplasia has higher cancer risk than morphologically similar TB-negative lesions, judging by molecular risk patterns (Epstein et al. 2003; Guo.et al. 2001) and outcome (unpublished study from this laboratory). Such studies have resulted in a revitalization of the studies of TB as a visual aid in clinical triage of high-risk OPLs from the low-risk ones, and in the selection of biopsy sites. While TB may be a promising visual aid, it has its limitations and weaknesses. Toluidine blue is messy to apply, tastes bad and may permanently stain the clothes of the patient or caregiver. Also, TB is commonly applied topically, which relies on the identification of a clinically visible lesion. These limitations support the development of other visual aids. This study investigated the possibility of the assessment of autofluorescence in tissue through fluorescent visualization (FV) as a means to detect high risk OPLs and early SCC. 10 1.5. Mechanism of Fluorescent Visualization (FV) 1.5.1. Tissue optics 1.5.1.1. Overview With the advent of proteomics, it becomes increasingly apparent that biochemical changes to cells occur early in carcinogenesis and can precede microscopic and macroscopic manifestations of cancer and precancer. Such changes can serve as valuable prognostic indicators for early disease. Among these early alterations are changes to molecules with fluorescent capacity such as nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), porphyrin compounds and elastin-collagen cross-links. Such changes can give rise to alterations in the visible fluorescence spectrum and may be used to delineate cancers and precancers from nonmalignant lesions clinically. Tissue fluorescence, therefore, can provide opportunities for the in situ evaluation of the biochemistry of tissue. In addition, microscopic and morphological changes in tissue undergoing malignant transformation may also affect the overall optical presentation of a lesion clinically by scattering light, and have an impact on the fluorescence emitted by tissue undergoing carcinogenesis, because it attenuates both the trigger of tissue fluorescence by the initiating, or excitation, light (less light hits the target fluorophores) and the amount of fluorescence reemitted (some fluorescent light is trapped in the tissue and does not reach the surface). These principles and how they impact the use of fluorescent visualization of epithelial tissue are presented in the following sections. In general, biomedical optics examination can provide opportunities for the evaluation of biochemical, architectural and physiological changes within intact oral mucosal tissues. Hence, opportunities to intervene might be enhanced i f earlier visualization of those precancerous changes can be achieved. Light interaction based upon tissue absorption, scattering properties, and fluorescence are explored in the following section. 1.5.1.2. Principles of light Electromagnetic radiation Light is a form of electromagnetic radiation (EMR), involving the movement of energy through space as a combination of electric and magnetic fields (Goaz and White 1994). Other examples of E M R are X-rays, ultraviolet (UV) rays, infrared radiation, microwaves and radio waves. The characteristics of visible light are described by using a dualistic theory that combines wave theory and quantum theory. The wave theory of E M R maintains that radiation is propagated in the form of a wave, similar to the waves from a disturbance of water. A l l E M R travels at the velocity of light, 3 x 108 m/s in vacuum, and exhibits in terms of wavelength (X) and frequency (v). Both properties are governed by the equation X * v = 3 * 108 m/s. Wave theory best accounts for the reflection, scattering, and refraction properties of tissue optics. Quantum theory depicts E M R as small, finite bundles of energy called photons. Each photon travels at the speed of light and contains a specific amount of energy expressed as units of 11 electron volts (eV). The relationship between wavelength and photon energy is described by the equation E = he/ X, in which E is the energy in kilo-electron volts, h is Planck's constant, and c is the speed of light. Based on this relationship, the shorter the wavelength, the more energy the photons contain. Quantum theory best describes the fluorescence properties of tissue optics. Figure 2 is a schema showing the spectrum for visible and invisible light. As the wavelength for the different types of E M R decreases from the left to the right side of the figure, energy increases. Thus, within the visible light spectrum, red/orange light has a longer wavelength than blue/violet. Infrared, which is invisible to humans, has a longer wavelength than visible red. On the other side of the visible light spectrum, ultraviolet light has an even shorter wavelength than visible violet, thus a higher energy level. Excitation photons used in this study include wavelengths from 400 to 460 nm (violet to blue). (waves per second) Wavelength [_ (in meters) -j gs Figure 2. Electromagnetic spectrum. The figure shows wavelengths for different types of radiation, with emphasis on visible light. 1.5.1.3. Impact of epidermal tissue architecture on optical properties In summary of Richards-Kortum and Sevick-Muraca's (1996) review article, optical properties of epidermal tissue reflect its structure and chemical composition. When a beam of light reaches the skin surface, a part of it will be reflected by the surface, while the rest will be refracted and transmitted into the skin. The light transmitted into the epidermal tissue will then be scattered and absorbed. After multiple scattering, some of the transmitted light will reemerge through the tissue surface. This reemergence is called diffuse reflection. The amount of diffuse reflection is determined by both the scattering and absorption properties of the tissue. The stronger the absorption, the less diffuse the reflection. Conversely, the stronger the scattering, the more diffuse the reflection. After the epidermal tissue absorbs photons, the electrically excited absorbing molecules may rapidly return to a more stable energy state by reemission of a photon with less energy, i.e. fluorescence emission. Since native fluorophores inside the skin are responsible for this fluorescence, it is termed autofluorescence. Fluorescence spectra are very sensitive to the chemical environment changes of the fluorophore molecules. 12 Tissue excitation The electronic configuration of a molecule is known as its ground state. Upon absorption of E M R (e.g., ultraviolet or short-wavelength visible light), an electron can be promoted from its ground state to an excited state through a process called electronic transition. Tissue absorption The absorption of energy as a result of an electronic transition causes an absorption band in the UV/visible light spectrum. A chromophore is the part of a molecule responsible for a U V or visible light spectrum. If a molecule absorbs wavelengths ranging from 180 to 400 nm, an ultraviolet light spectrum is obtained; i f it absorbs wavelengths ranging from 400 to 780 nm, a visible spectrum is obtained. Whether a molecule can be excited by ultraviolet or visible light depends on the availability of nonbonded electrons within it. Molecules that are excited by light usually contain conjugated double bonds. Haemoglobin is a good example of an organic molecule that contains conjugated double bonds, which result in multiple absorption peaks when exposed to the visible light spectrum (540-577nm). Figure 3 shows the chemical structure of haemoglobin after it combines with oxygen (oxyhaemoglobin). Figure 4 shows the absorbance spectrum for the molecule. H Figure 3. Chemical structure of oxyhaemoglobin showing conjugated double bonds. 13 400 500 600 700 800 Wavelength [nm] 900 1000 Figure 4. Absorbance spectrum of haemoglobin. The figure demonstrates that minimal absorption occurs between 650nm and 700 nm with peak absorptions at 540-780 nm (green-to-yellow region). 1.5.1.4. Principle of fluorescence Following the absorption of excitation photons from the UV-to-visible light region, electrons from chromophores are elevated to their excited states. The activated chromophores can undergo spontaneous relaxation of the excited electron via a radiative process (i.e. fluorescence). The relaxation of electrons to ground state is accompanied by the release of reemitted photons in the UV-to-visible light range at wavelengths lower than those absorbed. The light that emits from the molecules will produce an emission light spectrum (Richards-Kortum and Sevick-Muraca 1996). Reemission of fluorescence from chromophores, such as NAD(P)H, flavins, and porphyrins are in the U V (300^100 nm) and near-UV (400^160 nm) ranges. 1.5.2. Interaction of oral tissue with light The amount of auto fluorescence present on the tissue surface depends upon three main factors: 1. The quantity and quality of the fluorophores in the tissue: An increase in fluorophores results in an increase in fluorescent signals (FS). 2. The ability of the tissue to absorb light: An increase in the absorption of light results in a decrease in FS. 3. The ability of the tissue to scatter light: An increase in light scattering will result in a decrease in FS. The following sections discuss these three factors in the oral cavity. 1.5.2.1. Fluorophores in oral tissue There are numerous fluorophores in the human stratified squamous epithelium and the underlying stroma that contribute to fluorescence signals. There are three major fluorophores that are of importance to this study (Nelson and Cox 1993; Svistun et al. 2004): 14 1. Metabolic co-factors: nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD + ) 2. Collagen and elastin fibre cross-links 3. Porphyrin NADH and FAD+ N A D H and F A D + are water-soluble cofactors that undergo reversible oxidation and reduction in many of the electron transfer reactions of metabolism. Indeed, N A D H and F A D + fluorescence signals may be closely related to the cellular metabolic activity (Pavlova et al. 2003; Benavides et al. 2003). Figure 5 shows the chemical structure of nicotinamide adenine dinucleotide (NADH). N A D H is a coenzyme for oxidation reactions (Chance 1989). It is composed of two nucleotides joined through their phosphate groups by a phosphoric acid anhydride bond. One of the nucleotides is linked with nicotinamide (derived from the vitamin nicotinic acid) and the other nucleotide is linked with adenine. The positively charged molecule results from the positively charged nitrogen of the pyridine ring. N A D + and its reduced form N A D H are used as a coenzyme for catabolic reactions. The reduced pyridine nucleotide N A D H emits fluorescence peak at 450 nm upon U V excitation. O Figure 5 . Structure of nicotinamide adenine dinucleotide (NAD ). Figure 6 shows the chemical structure of the second pyrimidine nucleotide, FAD. It is derived from the vitamin riboflavin and is characterized by the presence of an isoalloxazine ring (flavin nucleotides). F A D serves as a coenzyme for several enzymes including fatty acyl-CoA dehydrogenase, dihydrolipoyl dehydrogenase, succinate dehydrogenase, and alpha-glycerophosphate dehydrogenase. The reduced form of F A D does not have fluorescence properties. However, oxidized F A D has an emission peak at 520-530 nm upon excitation at 440 nm (Benavides et al. 2003). 15 Figure 6. Chemical structure of flavin adenine dinucleotide (FAD). FAD containing an isoalloxazine ring (indicated with a rectangle) is classified as the chromophore. Support for the involvement of N A D H and F A D + in cellular autofluorescence comes from a recent study that used confocal microscopy to localize the source of autofluorescence in epithelial tissue. Pavlova et al. (2003) examined 10 pairs of colposcopically normal and abnormal cervical biopsies after excitation with U V (351-364 nm) and visible light (488 nm). They found that the fluorescence was localized to N A D H and F A D + in cellular mitochondria. Strikingly, the distribution of this cellular fluorescence altered with degree of dysplasia. It was restricted to basal cells in normal tissue but visible in the bottom one-third of the epithelium of low-grade dysplasias, and extended throughout the lower two-thirds of the epithelium in high-grade dysplasia. These findings have yet to be reviewed in oral tissue. In addition, the contribution of these fluorophores towards the overall clinically visible fluorescence requires further evaluation. In contrast to increased FS from N A D H and F A D + , clinical observations suggested an overall lack of FS in oral precancerous lesions and cancer (Svistum et al. 2004; Onizawa et al. 2002). Collagen fibres Although N A D H and F A D + contribute to the autofluorescence of normal oral and cervical mucosa as well as skin, the main source of autofluorescence was found from the connective tissue collagen fibres and the cross-links of the collagen fibres (Pavlova et al. 2003). Collagen and their cross-links fluoresce upon excitation with either ultraviolet light or visible light of short-wavelength (e.g., around 488 nm). The changes in collagen fibres and their cross-links hence play a very important role in changes of autofluorescence of mucosa tissue. Pavlova et al. (2003) looked at stromal autofluorescence in association with cervical dysplasia and showed that the fluorescence density of the stromal matrix immediately beneath the epithelium (the matrix density located 0.25-0.30 mm beneath the basement membrane) decreased as the tissue progressed from normal- to high-grade dysplasias. However, there were 16 no microscopically visible changes in the connective tissue between the dysplasia and normal mucosa. This finding suggested that the reduction of collagen fibres concentration was part of the tumorigenesis process, in which enzymes secreted by tumour cells or even precancerous cells degrade stromal collagen fibres. Of interest, a previous study (Mitrani and Marks 1982) showed a 10-fold increase in the level of procollagen in the dermis immediately beneath the SCC of the skin by using immunofluorescence titers of anti-procollagen antibody. The level of procollagen is considered to reflect the activity of collagen synthesis. This result is interesting because it supports the theory of dynamic changes (high turn-over rate) of collagen-dominant stroma immediately underneath the cancerous epidermis. Porphyrin A final fluorophore that has been associated with autofluorescence in epithelial tissue is porphyrin. Porphyrin is a core component of haemoglobin, myoglobin, and cytochromes (Nelson and Cox 1993). It is formed by the incorporation of iron atoms into protoporphyrin IX. Protoporphyrin IX is constructed from four molecules of the monopyrrole derivative porphobilinogen. Biosynthesis of porphyrin is regulated by the concentration of heme protein products, such as haemoglobin, which can serve as a feedback inhibitor of early steps in porphyrin synthesis. This biosynthesis pathway occurs in mammals, bacteria, and plants. Porphyrin compounds fluoresce in red upon excitation. Excitation-emission wavelengths for porphyrin compounds are 490 nm / 630 nm (Benavides et al. 2003) although 410 nm / 605-700 nm wavelengths may also be involved (Ingrams et al. 1997). Red fluorescence can be seen normally on the dorsum tongue and dental plaques. During carcinogenesis, some tumours also emit red fluorescence, suggesting the presence of porphyrin. The origin of the porphyrin, however, remains speculative. It is possible that with tumorigenesis, there are accumulations of porphyrinogen intermediates that are excreted as porphyrins or porphyrin-related compounds (Onizawa et al. 2002; Harris and Werkhaven 1987; further described in sections 1.6.3.1-2). An alternative explanation is that some of the red fluorescence are products of bacterial infection (Inaguma et al. 1999; Onizawa et al. 2003). 1.5.2.2. Tissue absorption of light In the oral cavity, the main endogeneous tissue factor that absorbs light, called chromophore, is haemoglobin. Oxy- and deoxyhaemoglobin is responsible for light absorption at wavelengths below 600 nm (wavelength range used in this study) and abovelOOO nm (Jacques et al. 2001). For oxygenated haemoglobin absorption, peaks are below 420 nm, at 540 nm, and at 575 nm (Sciubba 1999; Inaguma and Hashimoto 1999). Due to the large amount of light absorption from haemoglobin, optical penetration beyond 1 mm from the perfused tissue layer is impractical in wavelengths below 600 nm. The focus of this study is the oral mucosal epithelium and the underlying lamina propria; these are both superficial to the perfused tissue layer, namely the subconnective tissue. The light energy propagates through these two layers before being absorbed by haemoglobin under nonhaemorrhagic conditions. However, in the event of increased vascularity in the superficial connective tissue layer, increased light absorption by haemoglobin occurs. Increased vascularity in the superficial connective tissue layer can occur both during carcinogenesis (increased angiogenesis) and during inflammatory processes. 17 1.5.2.3. Tissue scattering of light Tissue scattering arises from the microscopic heterogeneity of refractive indices between extracellular, cellular, and subcelluar components. The scattering property as a result of changes in cell morphology, size, and shape has been critically evaluated in several kinds of tissue. The following tissue scattering factor are discussed: 1. Mitochondria 2. Nuclei 3. Cell size 4. Surface/stromal structure Mitochondria Beauvoit et al. (1994) determined that mitochondria, which occupy 22% of hepatocyte volume, are the predominate component for light scattering in the liver. Studies may apply to carcinogenesis in general, including those of the oral cavity, which is accompanied by a marked increase in the amount of mitochondria in the epithelial cells. It is possible that dysplasia shows a decrease in FV as a result of increased mitochondria and light scattering. Nuclei Nuclei scatter light and hence an increase in nuclei number, size and hyperchromatism, such as those seen during carcinogenesis, should result in an increase in light scattering and decrease in FS. Current literature supports this. Troy et al. (1996) examined tissue scattering properties of human mammary carcinoma and normal breasts and found that increased scattering was associated with the appearance of numerous enlarged nuclei in these tissues. Gurjar et al. (2001) also quantified the tissue scattering properties of colon tissue ex vivo in relation to nucleus size, cell pleomorphism, and hyperchromatism and reported that they were able to distinguish dysplasia from benign samples with this approach. These results were supported by a further study of in vivo scattering measurements in OPLs and cancer tissue that showed a correlation with the percentage of cells containing an enlarged nucleus (Sciubba 1999). Cell size Chance et al. (1995) also demonstrated the effect of cell size towards light scattering in perfused liver samples. The proposed effect could be demonstrated in a mathematical model, although in vivo study was required to validate the effect from tissue temperature and tissue osmotic balance. They suggested that small changes in solute concentrations impacted the scattering coefficient by altering the osmotic pressure and thereby cell sizes, or by changing the relationship of refractive index between scatterers. However, their view was challenged by Sloot et al. (1988), who suggested that the changes of scattering properties were due to the relative volumetric ratio of cytoplasm to nucleus rather than actual cell size. Tissue surface and stromal arrangement Increased roughness of the tissue topography results in increases in light scattering in both excitation and emission pathways. For example, both an ulcerated lesion and nonhomogenous lesion of the oral mucosa share the common feature of losing the smooth mucosa surface. The increased roughness of an ulcer and nonhomogenous lesion could attenuate the propagation of light. Although there is no in vivo experiment to demonstrate the extent and significance of this effect, the mathematical models (Richards-Kortum and Sevick-Muraca 1996) suggest that the direction of light travel within the tissue medium (i.e., the epithelium and the connective tissue) 18 depends upon the incidence angle of excitation light entering the tissue medium (air-tissue interface). In other words, instead of traveling in a constant optical path, excitation light travels differently i f it illuminates the tissue surface perpendicularly or obliquely. More importantly, the light scattering effect is different between macroscopically rough surfaces and adjacent smooth surfaces clinically. Similar scattering phenomena also arise at the interface between intercellular and extracelluar compartments due to the microscopic heterogeneities of refractive indices. Lynn et al. (1993) measured the skin reflectance in new born infants, and suggested that tissue scattering increased as the number and size of collagen fibres in the dermis increased. The latter suggested that structures in the stroma may also play a role in tissue scattering. Whether the scattering effect from collagen fibres significantly affects the availability of their FS emission present on the tissue surface has yet to be determined. 1.6. Autofluorescence Examination as an Adjunct Tool in the Identification of Malignant and Premalignant Lesions One of the most challenging questions faced by clinicians in the field of oral cancer prevention is how to identify premalignant lesions that have the potential to become carcinomas. A possible key is to examine the fluorescence properties of oral tissue, also known as autofluorescence examination (AFE). This examination has been recently introduced to the oral oncology community as a possible approach for identifying malignant tumours and premalignant lesions with malignant transformation potential (Svistun et al. 2004; Gillenwater et al. 1998; Betz et al. 1999; Onizawa et al. 2003). In addition, direct oral tissue fluorescent visualization (FV) may aid clinicians in selecting biopsy sites for lesions that are clinically difficult to identify .The data supporting this use for F V devices are extremely limited. This study examines the relationship between the positivity of FV and the clinicopathological features of OPLs, in order to generate the information necessary to evaluate this approach to the clinical assessment of high-risk patients. Currently, there are a few technical approaches to conducting an A F E in clinical and laboratory settings: 1) Different combinations of the excitation and emission wavelengths are selected in order to yield greater sensitivity and specificity in neoplastic lesion identification. This approach is mainly used in controlled laboratory settings as this investigation requires precise spectrophotometer readings (Wang et al. 1999; van Staveren et al. 2000; Muller et al. 2003) 2) The other approach relies on one predetermined excitation (blue) and two emission (green and red) wavelengths. A computing machine, LIFE®, will then compare the signals (intensity) from both green and red sensors with a predetermined threshold, and determine the presence or absence of neoplastic potential (Lam et al. 2000; Kulapaditharom and Boonkitticharoen 1998). Combined with conventional endoscopic examination, this approach is practiced worldwide for pulmonary neoplastic lesion detection. 3) The approach used in this study is called direct fluorescent visualization (FV). The technique relies on human eyes to visualize and evaluate the intensity of green fluorescent signals, and determine the presence or absence of neoplastic potential (Svistun et al. 2004; Onizawa et al. 1996). 19 1.6.1. Direct fluorescent visualization as a visual tool Before A F E became reintroduced as a facilitating tool in clinical diagnosis of oral cancer, it was used to evaluate dermatological lesions. FV in the diagnosis of oral tissue dates back to 1950, when clinicians used a Wood's Lamp (an ultraviolet flash light) to estimate prognosis of oral cancer lesions based on the presence of red fluorescence at the centre of the lesion (Ronches and Providence 1954). This investigative technique then fell out of favour. Over time, A F E technology has been continuously refined and applied to answer questions in diverse areas including forensic sciences, dermatology, and recently in gynecology, otolaryngology, and pulmonology (Kolli et al. 1995; Chang et al. 2002; Lam et al. 1998; Delank et al. 2000; Andersson-Engels et al. 2000). The direct fluorescent visualization (FV) method employed by this study is of recent development. In the following sections, this method is briefly reviewed and the advantages and potential applications of this technique are discussed. 1.6.1.1. Direct fluorescent visualization (FV) method This method relies heavily on the visual acuity, and integrity of colour perception (absence of colour-blindness) of the examiner. The oral tissue is examined after reducing ambient light as much as possible. An excitation light source that consists of blue and violet wavelengths is directed at the tissue of interest. The examiner then observes the autofluorescence emitted from the oral mucosa using a pair of goggles with filters that block the reflection of excitation light from the illuminated tissue. FV examination starts at least 5 seconds after ambient light is reduced and the excitation light is turned on. This lag time allows the examiner to reestablish visual acuity in an overall dark working environment. Normally, most of the normal oral tissue fluoresces in green that can be seen by humans. The present of fluorescent green is classified as FV negative (FV -ve). In contrast, compared to adjacent tissue or tissue from the control contralateral side, oral tissue is classified FV positive (FV +ve) if there is a reduction of green fluorescence so that it is no longer visible to humans. Oral re-evaluation of FV-positive regions in 3 months is recommended, to allow traumatic related inflammation or ulcer to heal before a second autofluorescence examination. This provision is aimed at reducing the possibility of a false positive that could result from such a change. If the lesion is still positive, a biopsy of the FV-positive site is recommended for histological evaluation. 1.6.1.2. Advantage of direct FV as a tool in identifying OPLs and early SCC FV is an exciting novel technology in oral oncology. Compared with TB, FV has some distinct advantages. FV technique is noninvasive. It is non-toxic, clean and can be used to scan the entire oral cavity without relying on the clinical identification of a oral lesion. One of the invariable characteristics of visual aids or markers for high-risk lesions is that the ones high in specificity are low in sensitivity, and vice versa. The availability of visual aids that have different sensitivities and specificities would be advantageous in the clinical diagnosis since the visual aid can complement each other. It is therefore possible that FV can complement TB as a visual aid in the diagnosis of high-risk oral lesions. 20 1.6.1.3. FV as a potential tool in secondary prevention of OPLs and early SCC 0 There are two forms of cancer prevention: primary prevention and secondary prevention. Primary prevention involves the identification and elimination of exposure to cancer-causing agents and habits. This involves altering alcohol and tobacco use for oral cancers. Secondary prevention is accomplished through screening programs that identify patients at increased risk to allow time to intervene early in the progression of the disease. Over the last couple of decades, there has been an increasing emphasis on screening programs, largely due to early success with the PAP (Papanicolaou) smear screening program in decreasing cervical cancer incidence and morality. Medical and technological advances have created many new approaches that have yet to be evaluated as screening tools. In order for such procedures to be effective and accepted by the public, they must be cost-effective, noninvasive and easy to perform. Two national cancer-screening programs are currently in place in British Columbia. These are mammography for breast cancer detection and the PAP smear for cervix cancer detection. The fecal occult blood test for colon cancer detection has been proposed. As yet, there is no validated screening tool for precancer and early oral cancer anywhere in the world. There are no available technologies for the use in the screening of high-risk populations. Exfoliative cytology has been evaluated as a minimally invasive tool for assessing oral mucosal lesions (Sciubba 1999). This technique allows sampling of the full thickness of oral mucosal epithelium. The sample is then evaluated by pathologists with the assistance of a computerized scanner for abnormal cytology features. A comprehensive data analysis supporting this screening method is still ongoing (Rick 2003; Poate et al. 2004). OPLs tend to be more keratinized than lesions at uterine cervix, hence smears from an oral lesion tend to show squamous cells without nucleated cells for diagnosis, leaving a false negative result. More importantly, as a screening tool, the whole oral cavity should be screened. Taking smears of the whole oral cavity is impractical. FV can potentially be a screening tool for the whole oral cavity and a tool for secondary prevention. 1.6.2. A F E as a tool for the identification of precancerous and cancerous lesions in other organs AFE in cervical and esophageal lesions Benavides et al. (2003) published a study that focused on the development of a cost-effective in vivo imaging tool to detect cervical precancer lesions. The device was used to ascertain fluorescence emission patterns in a range of tissue types, including samples from 126 histologically normal sites, 31 high-grade squamous intraepithelial lesion site (SILS), 63 low-grade SILS, and 107 inflammatory and/or metaplasia sites. They used excitation wavelengths of 345 nm (UV) and 445 nm (visible light, similar to the wavelengths used this study). The data suggested a significant correlation between the reduced autofluorescent green region and the pathology results. A classification algorithm was determined based on the fluorescence properties in three colour channels: red, blue, and green. This algorithm performed well for the discrimination of normal vs. high-grade SILS, and columnar normal-grade vs. low-grade SILS. Georgakoudi et al. (2002) reported that NAD(P)H and collagen as in vivo quantitative fluorescent biomarkers of epithelial precancerous changes. They identified and adjusted for distortions introduced in measured tissue fluorescence spectra by tissue scattering and absorption. Fluorescence and reflectance spectra were simultaneously measured at sites from, 35 21 patients with suspected cervical lesions and 7 patients with Barrett's esophagus. Using multivariate curve resolution analysis, the contribution of N A D H and collagen to the overall reduction of fluorescent signal could be isolated. In both tissue cases, low collagen and high NAD(P)H fluorescence characterized high-grade dysplastic lesions when compared with nondysplastic tissues. Biopsy samples of squamous metaplasias and high-grade squamous intraepithelial lesions (precancer) exhibited a 5-fold decrease in collagen fluorescence relative to colposcopically normal squamous epithelium, and a 2-fold increase in the N A D P H fluorescent contribution for the HSILS (precancerous lesion) compared with the benign squamous metaplasia site. The authors suggested that tissue deoxygenation, an increased number of cells, and increased levels of metabolic activity in epithelial cells could be responsible for the increase in N A D P H content in precancerous tissue. The decrease in collagen might be attributed to the presence of collagenases such as matrix metalloproteinases, which are involved in the breakdown of collagen. AFE in lung and upper airway lesions Lung imaging fluorescence endoscopy, or LIFE® (developed by Xi l l ix Technologies Corp.), is the most well researched autofluorescence imaging system for bronchoscopic evaluation of clinical abnormalities in the lung and upper airways. This system employs a helium-cadmium laser light (442 nm) for tissue excitation and two image-intensified CCD cameras to record the tissue fluorescence signals. The ratio of red (enhanced) to green (reduced) signals are computerized and compared with a predetermined value. A pseudocolour image that corresponded to the suspected cancer sites appears on a video monitor, to assist the clinician in determining which tissue requires further follow-up. This device has been used by numerous clinics worldwide as a resource to guide biopsy collection (Lam et al. 2000; Delank et al. 2000; Kulapaditharom and Boonkitticharoen 1998). It is one of the primary devices used to follow chemopreventive trials. The later trials are used to evaluate the ability of drugs to halt or reverse early histological change in tissue and hence prevent cancer development. Among drugs studied to date are vitamin A and several of the synthetic retinoids, and more recently, anti-inflammatory agents (Lam et al. 2002, 2003) The device has also been used, although less frequently to complement microlaryngoscopy and assist clinicians in detecting and delineating laryngeal malignancies (Zargi et al. 2000; Delank et al. 2000). AFE in brain lesions Chung et al. (1997) surveyed the excitation-emission matrix of a few human brain tissue samples using ex vivo techniques in order to determine the feasibility of identifying gliomas with this approach. Three excitation-emission wavelength pairs were chosen to compare normal and tumour tissues: 360 nm / 470 nm (to detect N A D H alterations), 440 nm/ 520 nm (flavin), and 490 nm / 630 nm (porphyrin). Fluorescent emissions of N A D H were lower in human brain tumours (including metastatic tumours, astrocytoma, and medulloblastoma) than in normal brain tissues. In contrast, flavin and porphyrin emissions varied in intensity according to tumour types. AFE in skin lesions Several studies evaluated autofluorescence as a noninvasive aid to skin cancer detection (Andersson-Engels et al. 2000; Na et al. 2001; Panjehpour et al. 2002; Wang et al. 2003). Among them, the Panjehpour article is one of the more extensive studies. It measured the overall intensity of skin fluorescence emission using short-wavelength (410 nm) visible light in 279 skin sites from 49 patients. The results suggested that basal cell carcinoma and squamous cell carcinoma exhibited a fluorescence emission that was weaker than that of normal and premalignant sites (reduction of FV, or FV +ve). Retrospective data was used to establish a 22 fluorescence intensity threshold and this was used to assess the prospectively collected data. As a result, 90% of the cancer sites and 82 % of the normal sites could be correctly discriminated using this threshold. The author attributed the reduction of fluorescence intensity in skin cancer sites to the absorption of fluorescent emission from increased blood supply and from melanin in tumour sites and to the destruction of collagen and elastin cross-links at tumour sites. Na et al. (2001) attempted to use this approach to demarcate the clinically invisible margin of skin basal cell carcinoma using autofluorescence intensity. Complete removal of this tumour is important, since recurrence was reported for 30-67% of cases (Sigurdsson and Agnarsson 1998; Sussman and Liggins 1996; Rippey and Rippey 1997). Twenty-one lesion sites were measured with fluorescence intensity from the lesion centre outwards. A loss of fluorescence intensity could be found up to 10 mm away from the clinical margin. Although these results were inconclusive as there is no indication of outcome for cases with different margins, they did support the possibility of using A F E for such a purpose. 1.6.3. A F E in oral premalignant lesions and cancer 1.6.3.1. A F E as a tool for identification of OPLs and oral SCC in animal and cell culture models FV-positive oral lesions from hamster oral cancer model One of the most commonly used animal models for the study of oral cancer development is the 7,12-dimethylbenz[a]anthracene (DMBA)-induced hamster buccal pouch carcinogenesis model. The association of alterations with autofluorescence and the progression through premalignancy to cancer have been studied extensively in this model, using both ex vivo and in vivo analysis (Balasubramanian 1995; Wang et al. 1999, 2003). In general, these studies all supported the potential value of using autofluorescence spectroscopy to detect pre-cancerous and cancerous lesions, with data showing that the approach is capable of discriminating between specimens of normal mucosa / hyperkeratosis, dysplasia, and carcinoma with a high degree of accuracy. Of these studies, one by Wang et al. (2003) clearly supported a potential role for FV as a diagnostic tool. This study used a fibre optics-based fluorescence spectroscopy system with 330 nm excitation light to measure autofluorescence spectra of 75 buccal pouch sites of D M B A -treated hamsters, including normal mucosa and lesions at different stages of carcinogenesis. Peak emission was found at 380 nm (hypothesized to be collagen) and 460 nm. Compared to the normal mucosa, there was an increase in the intensity ratio of 380 nm / 460 nm in hyperkeratosis, then a decrease in the intensity ratio with increasing degree of dysplasia and to SCC. By choosing a proper threshold, accurate identification rates of 86% for hyperkeratosis, 87 % for normal, and 89 % for dysplasia were found. In addition, a sensitivity of 92% and a specificity of 95 % were observed for differentiating benign (normal or hyperkeratosis) from premalignant (dysplasia) or malignant sites. AFE revealing increased porphyrin red fluorescence in oral lesions from hamster oral cancer model A study by Onizawa et al. (2002) is of specific interest because this study was conducted to analyze the spectral properties of the red fluorescence observed in human and experimental animals (see section 1.6.2) and to spectrometrically establish the involvement of porphyrin production in that fluorescence. The study used an animal model that used 9,10-dimethyl-l, 2-benzanthracene application to the lateral border of the hamster tongue to induce tongue carcinomas. Fluorescent samples were obtained from 18 hamsters after excitation at 404 nm and 23 compared with samples from 5 patients with SCCs (all emitting red/orange fluorescence after excitation), and with cell media collected from 3 human oral SCC. The contralateral, untreated border of the tongue of six hamsters served as controls. Ten mild dysplasia, five severe dysplasia and three SCC cases were analyzed from the hamster samples. Red/orange fluorescence was observed in all 3 experimentally induced SCCs, in 1 of 10 mild dysplasia, and 1 of 5 severe dysplasia. A l l controls were negative for fluorescence. The spectral profile of the experimentally induced cancers changed with malignancy, with increases in intensity at 634 nm and 672 nm and decreases at 582 nm. Porphyrin peak intensity at 634 nm and 672 nm increased as the severity of pathology progressed from control, to mild dysplasia, to severe dysplasia, and to SCCs; however, the difference was significant only for intensity changes from control, hyperplasia, or early cancer to invasive cancer. In contrast, the Zn-protoporphyrin IX peak intensity of 582 nm showed a decrease as malignancy progressed; however, again the results were significant only for comparisons between control and SCC samples. Of specific importance was the observation that the ratios of the intensity values of 582 nm over 634 nm might be of diagnostic utility in differentiating between nonmalignant and malignant tissues. These values were consistently altered not only in experimental carcinomas but also in the clinical cancer samples, and cultured cancer cells. Furthermore, the study showed that the oral cancer cell itself can produce porphyrin and that its accumulation in tumour tissue produces red fluorescence under Ultra-violet (UV) excitation. 1.6.3.2. A F E as a tool for the identification of human oral SCCs The delineation of oral carcinomas is usually not an easy task because 1) tumours usually start out as surface lesions which spread over the surface of mucosa and sometimes merge with surrounding inflammation or leukoplakia (Strong et al. 1968, 1984); 2) some lesions are very small, asymptomatic, or morphologically similar to benign lesions and are easily missed clinically; 3) at the early stage of cancer, colour changes may not be very different from the changes seen in the surrounding normal mucosa; 4) sometimes, the only clue to the detection of such lesions is the area of erosion, where the thin covering layer of keratin is missing (Strong et al. 1968, 1984); and 5) due to the unique characteristic of multicentric origin of oral cancer, it is not unexpected for multiple or satellite tumours to exist adjacent to the main tumour mass which are often missed during clinical examination (Strong et al. 1968, 1984). The following is a summary of several of the key publications that describe the use of A F E to detect or delineate oral SCCs (table 2). FV positive site associated with oral SCC Kulapaditharom and Boonkitticharoen (1998) used the LIFE® system, a helium-cadmium laser (excitation at 442 nm) to examine the head and neck cancer region in 25 patients suspected for malignancy in order to determine whether potential primary, recurrent, or residual cancer sites could be detected. Of the 32 examination sites, only 11 were in the oral cavity with the majority of the remaining sites (21) in the tracheobronchial tree or larynx. In all cases, the diseased tissue fluoresced with much weaker signals in the green region. Using pathological diagnosis as a gold standard, the system correctly identified all 11 oral cancer cases, including 2(1 tonsil and 1 lateral tongue) that were not visible with conventional light examination. The sensitivity was 100% and specificity 87.5%> for this small number of cases compared to a sensitivity of 87.5%) 24 and specificity of 50% with conventional white light. The authors attributed false positives to the presence of an inflammatory process. Betz et al. (1999) performed spectrophotometric analysis on 49 histology-proven SCCs using excitation wavelengths of 375 nm to 440 nm. Spectrophotometric measurements were taken in 36 of these patients from tumours, their surrounding host tissue, and from several clinically normal standard screening points (palate, buccal mucosa, dorsum of tongue, inner lip). A blocking filter was used to restrict the light transmission to wavelengths greater than 525 nm, thus blocking the excitation light completely. In 20 of the 30 patients, malignant lesions were demarcated from normal tissue by their appearance as a darker shade of green compared to light green at normal sites. Cancers located on the floor of the mouth or palate were easier to differentiate from innocuous tissue than from the rim of the tongue. Ten of the 30 tumours also demonstrated red fluorescence. Spectral analysis was able to differentiate tumours and normal tissue in 34 of 36 (94.4%>) patients so examined, suggesting that FV might be a promising tool for the diagnosis of oral malignancy. FV positive sites associated increased cell thickness One of the characteristic features of intraepithelial transformation and subsequent disease progression are changes in mucosal thickness. Koll i et al. (1995) explored the changes in autofluorescence intensity with respect to the epithelial thickness in cancerous epithelium to determine whether native autofluorescence could be used to distinguish such changes. The study used a handheld fibreoptic probe attached to a fluorescent spectrometer. Fluorescence analysis was performed on untreated oral cancer sites and normal mucosa from the corresponding contralateral site in 19 patients before surgical excision. Ratios of intensities at various emission wavelengths were compared with epithelial thickness as measured in the biopsies. A n additional 12 biopsies from patients with cancer of the upper aerodigestive tract were analyzed for fluorescence in vitro within 2 hours of surgical excision. Florescence measurements, were made with an excitation at 340 nm with an emission at 360-660 nm, and with an excitation at 200-360 nm and an emission at 380 nm. The results showed a significant difference in the spectra obtained in neoplastic and normal tissue with both of these excitation/emission regimes. Although a significant correlation between reduced emission intensities and the increased epithelial thickness was observed in study samples (P = 0.01), the correlation was only 40%), suggesting that other factors are involved in the altered fluorescence. FV positive sites associated with decreased cell differentiation Schantz et al. (1998) conducted an exploratory study with 35 oral SCC patients to look for an association between cancer differentiation status and the probability of disease recurrence with florescence emission spectrums. Twenty-nine of the patients had primary cancer and six had a disease recurrence or a second oral malignant tumour. The analysis was done with a handheld fibreoptic probe attached to a fluorescent spectrometer and used an excitation at 290^-30 nm and an emission at 450 nm. This range would include excitation and emission peaks for both N A D H and flavins. In contrast to the previous discussion of the increased fluorescent signal of N A D H and FAD in metabolically active cells (Georgakoudi et al. 2002), emission intensities in this study were significantly higher in the normal mucosa compared to cancers from the same patients. In addition, poorly differentiated tumours showed lower excitation maximums than well-differentiated or moderately differentiated lesions. Of interest, a comparison was also made between patients who remained free of disease with longitudinal follow-up and those who had recurrence found upon analysis or developed recurrence of the index cancer during follow-up. Recurrence included local, regional, or distant disease. Twenty-nine of these patients were 25 included in the latter analysis with a median follow-up of 29 months. Ten patients either had recurrence at initial analysis or developed it during follow-up. Patients whose tumours or whose normal mucosae had low excitation maxima were more likely to develop recurrence. However, this data was not significant (P = 0.11). Although the sample size was small, these data suggested that A F E might be used to predict disease recurrence. AFE reveals increased red autofluorescence in oral SCC Onizawa et al. (2003) specifically focused on wavelengths that allowed assessment of protoporphyrins and designed a study to evaluate the relationship between autofluorescence in these wavelengths and clinical features of oral SCCs. Fifty-five oral SCC sites were assessed, of which 50 (91%) emitted orange and red fluorescence under an ultraviolet flash lamp (360 nm). The size of the fluorescent site and the number of fluorescence-positive cases increased with stage, present in 70% of T l , 90% of T2, and 100% of T3 and T4 cancers. No relationship was found between the clinical appearance (ulcerative, protrusive, superficial) and the fluorescence characteristics. Porphyrin content in tissue swabs of the tumours was assessed biochemically using high-performance liquid chromatography. Protoporphyrin IX and coproporphyrin were identified in the tumour samples, but were also shown to be present in control samples from dental plaque and healthy tongues of these patients. However, in 4 out of eight cancer sites, the elution patterns were different from the false positive sources, with two additional peaks for protoporphyrin IX. These data suggested that porphyrins might be associated with the cancer tissue itself, although the presence of porphyrins in dental plaque may indicate the involvement of microorganisms in the generation of these compounds. A similar but earlier study by Inaguma and Hashimoto (1999) supports the above findings for porphyrin-like red fluorescence in oral carcinoma and normal tissue. This study evaluated in vivo fluorescence spectral characteristics of oral carcinoma after excitation at near-UV (410 nm) with emission measured in the red wavelength (605-700 nm). Extracts from carcinomas were evaluated with capillary electrophoresis. The study involved 78 patients with suspected oral cancers and 30 nondysplastic lesion sites. A major red fluorescent band at 630 nm and smaller bands at 665 nm and 690 nm were noted in 85% of oral carcinomas, but not in the normal mucosa. Ulcerated areas of tumour sites showed stronger red fluorescence than non-ulcerated tumour sites. Inaguma argued that the presence of porphyrin-like fluorescence was the result of cancer cell metabolites in addition to bacterial contamination. In support of this argument, they noted that porphyrin fluorescence was present in non-ulcerated lesion sites, cross-sections of SCCs converted from normal mucosa and in OPLs with intact mucosa. These areas were free of bacterial colonies in theory. Based on the presence of porphyrin-like fluorescence, this technique could achieve a diagnostic sensitivity of 86% in oral SCCs and a specificity of 96%. 26 4> > La o u s u SB V i . o s t= o s < o O OO T 3 o o 00 u. 3 c CJ o cd cd s-cd n. o o CJ CJ oo oo u. —> CJ O oo £ .2 "O o o oo UH -o o o co o cn .T^  > u. <U UH > cu Q cu fc O Cu ca s 5 8 CU o K ca s ca ft. c ca ca ca ft. 13 o u O SC i-e SC ed u CO! T3 » tiat one O .2 feren T3 ed '-^ c <D feren rre rre Idif 03 n. a iffe Idif rec rec cor >> £ »les non SCC poorl ed to samp ed to _c _c C O _s t-cd oo oo mp oo mp u. U. mp u- mp —> —> co - > co (N o u. -o c cd o E 2 3 T3 X <u .ti cd O CO td 2 o o cd 0) o CO CD 60 e '5b C d S -> -> S3 O0 T3 •O 3 U X i co i- CJ g oo O o cn O cd <u CO cd O o 3 E oo UH o 3 O da O0 <» o o o o j d '5 > cd -w O Z X cd o "E. D . cd O Z x cd o 15. a. cd +-» o Z ic .£ - • "3 •§ o <a ,E co o Z ON O O u ss +N s O O 00 O CJ oo c".S cd .3 < o u, a. cn cn a> c 00 j£ 15 •-O u T3 C cd c cn O cd o c cd Q ^ < ^ Z o u oo -o c cd o a. O U oo T3 c cd "a. Urn o .2 g S W)| o s •- « ft a* -a 3 o W) CN c 4) o T i n O o o O o CN O o" ON m o c o CO o i n r^ o c n I o o c<1 T3 Hi o vo CO o o vo o o s a. o c o ON i n c o m i n o o u s 2 o Pi E o 3 CD O •5 o ON ON ON o N i n ON ON cd "o oo ON ON o N C cd X o oo cd "S cd cd N co 'e ° O CN ON ON ON T3 — § 2 cd O E E 3 IS 2P « 2 cd = X 21 s s-> o V u e a u «> 4> i-© s c o R to <*> b- S _ Jo o s 5^  60 _03 "o o s o 03 o —> -o ^ X ~ < ? a A-S o 5 .S 8 I J - TO 00 o o M « o * -5 o u. 7; * es 3 <Z> u < ><r-00 T3 <a 0) 03 O 03 3 -o O c/> c/} <Z5 O 3 T3 •a c o 03 CL) c^ > o e o C 03 c s 03 5 8 cn 00 cn 00 O o C =3 60 . 2 o 60 N 03 2 a s c£ £ « a~ 03 j_ •a 73 ^ > 03 c u 60 03 60 o .S <-> o < T3 C 03 o c 03 CJ •2 S cn S O S ••B « o IT) 00 i n m cn cn w cn o s a cn o o (N "3 4> 03 03 C <u 00 = ON 5 2 1.6.3.3. A F E as a tool for the identification of human OPLs In contrast to studies on oral SCC, there has been far less research conducted on the efficiency of A F E to detection human premalignant lesions. Ingrams et al. (1997) performed a small study with 11 patients attending an otolaryngology-head and neck surgery clinic to determine whether fluorescence spectroscopy could differentiate normal and dysplastic tissues in vitro and to obtain information on the optimal excitation wavelength for identifying dysplasias in future in vivo studies. Patients were identified as having clinically suspicious oral mucosal lesions. Biopsies were taken from abnormal and normal regions and examined spectroscopically. On histological examination, 10 specimens were classified as abnormal (this included dysplastic or malignant diagnoses, unfortunately, with no indication of relative numbers of each) and 12 samples were classified as normal (healthy or with benign changes). Significant differences were seen between the 2 groups with the most striking observed at the excitation wavelength of 410 nm, where 20 or 22 samples were correctly identified as abnormal by fluorescence red spectrum. The increased fluorescence red for dysplastic and cancer tissues occurred with emission peaks above 600 nm with the most prominent peaks at 635 nm and 690 nm, which were not seen in the corresponding normal tissue. The ratio of prominent peak intensity (640 nm / 615 nm) and the area under the cure from 600 nm to 650 nm for normal and dysplastic (or cancer) lesions was used to establish a cutoff line. Using this algorithm, the sensitivity and specificity to differentiate histologically normal from dysplastic and cancerous tissues were 90% and 91%. Muller et al. (2003) used trimodal spectroscopy to assess reflectance, light scattering, and fluorescence in 15 patients with known upper aerodigestive tract cancers, including 12 patients with cancer in the oral cavity. The objective was to determine how well such spectral features correlated with early biochemical and histological change. Spectral data were collected from 91 tissue sites in these patients with 53 biopsies of varying histology (16 biopsies were normal, 11 mild/moderate dysplasia, 8 severe dysplasia, 12 SCC, 2 inflammation, and 1 keratosis). These spectral data were compared to the data obtained from 38 tissue sites, none of which were biopsied, in 8 healthy volunteers. Diagnostically significant information was found to lie in the emission wavelength range of 350-600 nm, where intrinsic fluorescence of oral cavity tissue comprised collagen and N A D H . Based on the spectral data, N A D H appeared to increase with disease progression and collagen appeared to decrease. Reflectance spectra suggested that the scattering phenomena became prominent with the extension of dysplasia and the enlargement of epithelial cell nuclei. Scattering increased with keratinization, higher in normal keratinized mucosa than nonkeratinized mucosa. The researchers concluded that this combination of spectroscopy could distinguish malignant and precancerous tissues from normal tissue with a sensitivity and specificity of 96%. The diagnostic sensitivity for fluorescence is calculated to be 83% for cancer and 53% for OPLs using the raw data supplied in the article. Finally, Gillenwater et al. (1998) used fluorescent spectroscopy to collect data from the oral cavities of 8 healthy controls and 15 patients with premalignant or malignant oral lesions. Patients attended a tertiary care head and neck cancer clinic. Histological diagnosis included 15 normal, 5 reactive atypia, 12 dysplasia, 4 CIS, and 9 cancers (45 in total). There was no indication of whether patients with dysplasia had a prior history of cancer or of prior treatment for any of the cases. Data from 3 patients were later excluded due to instrument error, leaving 12 patients with 33 sites for analysis. Spectral data was collected with 3 excitation wavelengths (337 29 nm, 365 nm, and 410 nm) and an emission range of 350-700 nm and used to develop diagnostic algorithms. Peak fluorescence intensity for the spectra showed both inter-individual and intra-individual variations. Of interest, there were even some variations in the same individual for contralateral sites. The person-to-person variation at each site was greater than the contralateral variation within a particular site in the same person and the variation among different anatomical sites within the same person. For the 3 excitation wavelengths assessed, the fluorescent emission intensity ratio of red to blue showed less variation between healthy sites and individuals compared with the absolute peak intensity. In addition, the excitation wavelength of 410 nm showed the least variation in intensity ratio. Compared with the fluorescence spectrum on the contralateral sites, the intensity of the blue peak was higher in a normal site, and that of the red peak was higher in an abnormal site. Red-to-blue (635 nm / 490 nm) peak ratios were greater in abnormal tissue than at a contralateral normal site within the same patient. By establishing a cutoff line (+ 1 SD) in the peak intensity and the red-to-blue ratio of healthy sites, the sensitivity and specificity of differentiating healthy from dysplastic or cancer sites was determined to be 94.1% and 100% respectively. 1.6.3.4. FV character in other human oral lesions In addition to exploring the autofluorescence of cancers, Inaguma and Hashimoto (1999) also evaluated autofluorescence in 43 other patients with oral lesions diagnosed as benign based on clinical appearance and history. The assessment again was for porphyrin-like fluorescence. The set included 15 leukoplakia (no histological diagnosis given), 3 lichen planus, 17 inflammatory disease, 2 pigmentation cases, and 6 benign tumours. Only 1 case of acute necrotizing ulcerative gingivitis (ANUG) and 2 of the leukoplakia showed porphyrin-like fluorescent emission. Chen et al. (2003) observed the fluorescence spectrum of oral submucous fibrosis upon excitation at 330 nm. Oral submucous fibrosis is a chronic scarring disease characterized by epithelial atrophy and progressive deposition of collagen in the lamina propria and submucosa of the oral mucosa. This mucosal condition usually results from the chewing of the areca quid. An increase in emission at 380 nm (which corresponds to collagen) was observed for the fibrotic tissue compared to normal tissue. Emission at 460 nm (corresponds to NADH) was decreased. These profiles are in contrast to those reported for cancer, where emission is decreased at 380 nm and increased at 460 nm. 30 II. S T A T E M E N T O F T H E P R O B L E M Oral SCC is believed to progress through sequential stages from hyperplasia to increasing degrees of dysplasia (mild, moderate and severe) to CIS and finally to invasive SCC. Once invasive cancer is formed, the prognosis is one of the worst among major human cancers (< 50% five year survival rate). Early diagnosis and management of OPLs before they become invasive lesions is one key in improving the dismal prognosis. Most of OPLs do not progress into cancer. Differentiation of high cancer risk OPLs from low-risk OPLs is critical for the management of these lesions. Histology the current gold standard for cancer risk assessment of OPLs and more recent molecular markers both rely upon biopsy samples of the OPLs. If clinicians do not take a sample from a high-risk lesion, or take a sample but not from the highest cancer risk area of an OPL, the gold standard and molecular cancer risk assessment will be inadequate. The decision of when and where to biopsy is currently based on clinical risk factors of the OPLs, which mostly present clinically as leukoplakia and sometimes as erythroplakia. However the clinical risk factors are limited in their ability to identify high-risk lesions (see Section 1.4.1). The identification process is further complicated by the fact that a proportion of OPLs are not clinically visible or apparent. Development of visual aid that could facilitate the clinical identification of high-risk OPLs is highly needed. TB, a visual aid in experienced hands, has long been shown to facilitate the identification of early oral SCC and some dysplastic OPLs. Recent molecular studies have shown that TB positive dysplasias have a greater risk profile than TB negative lesions with similar degrees of dysplasia, suggesting TB could be a useful tool in triage cancer risk of low-grade dysplastic lesions. However, TB has its weakness of being a chemical (associated with potential side effects of chemicals), a dye (messy stains for both patients and the caregivers), and frequently relies upon the clinical" identification of OPLs since most clinicians resort to the topical application method (the rinse method is more irritating and less tolerated). The development of new visual aid that could compliment TB is desired. Recently, FV has attracted intense interest as a visual aid in the diagnosis of cancer because it is non-invasive in nature and does not involve chemicals. Studies have found FV to be effective in the diagnosis of cancer in a number of organs including lung, cervix and skin. A limited number of studies of FV have been done on premalignant lesions with little information regarding its value as a visual tool in human OPLs. Clinical evaluation of FV in the detection of OPLs and cancer has been limited to retrospective computer analysis of spectrometer readings in a laboratory setting, e.g., animal models and cell culture. There is inadequate data to show the effectiveness of using human eyes to detect and differentiate autofluorescence of OPLs in a pure clinical setting. This thesis is the first study to investigate the direct FV as a potential visual aid to identify high-risk OPLs. 31 III. O B J E C T I V E S 1. To determine autofluorescence properties (FV +ve vs. FV -ve) of oral lesions or sites of previous lesions in patients with a prior diagnosis of oral dysplasia or SCC. 2. To determine the following clinical information: a. Demographics: Age, gender, and ethnicity b. Habit: Tobacco c. OPLs: Site, size, clinical appearance, TB staining, and others 3. To determine the following pathological information: a. Dysplasia: Presence and degree b. Thickness of the epithelium and keratin c. Surface topography: Smooth and rough d. Inflammation: Presence and degree 4. To compare the cancer history and above clinicopathological features of FV positive lesions with those of negative lesions. 32 IV. N U L L H Y P O T H E S I S FV-positive OPLs show no difference in cancer risk compared with FV-negative OPLs, judging by both clinical and pathological risk factors. If data show FV-positive OPLs to have more clinicopathological high-risk features (e.g., more common in cancer patients and in high-risk sites, larger and nonhomogeneous, and more being dysplastic and with higher degree of dysplasia), it suggests that FV positive OPLs have a higher cancer risk than FV-negative OPLs, thus supporting the alternative hypothesis. 33 V. MATERIALS AND METHODS 5.1. Subject Selection Patients with a history of histologically confirmed dysplastic OPLs or oral SCC were recruited from the "Oral Health Study", an ongoing longitudinal study on oral leukoplakia. These patients were primarily referred by community dentists or the Head and Neck Oncology Group at the British Columbia Cancer Agency (BCCA) for assessment and care. A l l clinical examinations were conducted in the division of Oral Oncology / Dentistry at the Agency during the study period from January 2003 to March 2004. Subsequent histopathological reviews were conducted in the Oral Pathology section at the Vancouver Hospital and Health Sciences Centre. The subjects were informed of the study procedures, and their informed consent was received. This study was approved by the Institutional Review Board of the University of British Columbia as part of an ongoing longitudinal study of patients with histories of either oral cancer or dysplasia. A preliminary set of results collected from the first 5 months of the study was presented previously (Appendix 5, Zhang et al. 2004). This aim of that preliminary trial was to identify the feasibility of continuing this direct fluorescent visualization project for this thesis. The data was also used to calculate the required sample size for this thesis. As there was no amendment made in the study design between the pilot study and this thesis, the data from the pilot study was incorporated as part of the results of this thesis. 5.2. Devices 5.2.1. Direct fluorescent visualization (FV) examination device For autofluorescence examination, an external light source is needed. This light is used to illuminate the target tissue and is called the excitation light. When target tissue fluoresces under the excitation light and the light produced by the tissue is called autofluorescence (a colour different from the excitation light). In order for the examiner to see the autoflurescent light only, a filter is needed to block out the colour of the excitation light by the examiner wearing a pair of glasses or goggles with the filter (Appendix 3). In this study, the excitation light (extraoral light source) was blue and violet in colour and was produced by a halogen arc lamp in a device called SCD-1 from Biomax to examine the oral mucosa of all test subjects. An integrated parabolic reflector collects the light from the light bulb and produces a homogenous output beam through a diffuser. The wavelength of the excitation light was in the range of 400^160 nm by using 460 nm bandpass filters centred at 425 nm (Chroma Technology Corp., USA). The measured illumination intensity at the usual working distance (3-6 inches) was approximately 1-4 mW/cm 2 at 440 nm excitation. The oral mucosa autofluorescence induced by the extraoral excitation light was green and red in colour. The examiner viewed the autofluorescence by wearing a pair of goggles or glasses with a long pass filter and a notch filter to allow only the passage of green (475-560nm) and red (620-740nm) fluorescent emission (Schott Glass Technologies, USA). 34 The FV examination was conducted in the absence of ambient light, and at least five seconds after ambient light was turned off and excitation light was turned on. 5.2.2. Digital camera Clinical images of the oral lesions were obtained at each patient review. This allowed for comparison of lesions at re-evaluation appointments and clinical record audits, and were essential for case presentation. A l l clinical pictures were taken with a highly sensitive colour CCD camera (FujiFilm FinePix S2 Pro) equipped with Nikon Macro Speedlight SB-29s and Nikkor AF Micro 105 mm 1:2.8D lens for recording intraoral pictures. For autofluorescence pictures, an additional long pass filter was attached to the front of the camera lens to prevent the blue excitation light from reaching the camera sensor. Normal oral mucosa appeared light green on the acquired images. However, diseased tissue showed lower autofluorescence intensity in general, allowing suspicious lesions to be demarcated from the adjacent normal tissue as a weaker shade of green (Appendix 4). 5.3. Clinical Examination 5.3.1. Sequence of the examination Table 3 summarizes the steps involved in the study. A questionnaire was used to collect demographics (age, gender, ethnicity) and smoking history. The patients' medical history and pathology reports were reviewed to identify sites and diagnoses for all previous oral biopsies and to obtain a record of past and current treatments. Any missing pathology reports that were identified by the clinician at this time were requested from the patients' personal physicians. The review was followed by a clinical examination (discussed below) and a biopsy if required. Table 3. Sequence of steps for data collection of new patients Step Action 1. Discuss study 2. Acquire informed consent 3. Collect contact information 4. Complete questionnaire 5. Review medical history 6. Review pathology reports 7. Perform oral examination under standard white light 8. Take photographs of clinical lesions under white light 9. Perform autofluorescent examination 10. Take photographs of FV-positive lesions 11. Perform toluidine blue staining 12. Take post-toluidine blue photographs 13. Perform biopsy i f warranted A l l examinations described below were performed under the direct supervision of Dr. M . Williams, an oral medicine specialist. / 35 5.3.2. Questionnaire For each patient visit, the following demographics, habits and other information were collected in a questionnaire (see appendix 1): • Demographics: Date of birth, gender, and ethnicity • Habit: Tobacco • Prior history of cancer 5.3.3. White light examination A l l subjects underwent a comprehensive oral mucosal examination to identify all clinical lesions. Particular attention was paid to previously identified lesion sites (e.g., sites of prior oral SCC). A l l examinations were conducted in the Oral Dysplasia Clinic at BC Cancer Agency (Vancouver Cancer Centre) in one of three operatories. The ambient light condition in the three operatories was illuminated by ceiling-mount fluorescent tube with a color temperature below 50000 degrees Kelvin. Each dental chair was also equipped with ceiling-mount adjustable overhead light with a color temperature approximately at 5000 degrees Kelvin. The examination was made in the following sequence, with photographs taken: • Lesion present or not: 0 = no lesion, normal looking, scar, graft, or equivocal; 1 = lesion present. • Clinical diagnosis when there was a lesion: H = hyperplastic lesion or polyp or inflammatory reactive lesions; LP = lichen planus; OPLs = leukoplakia • Clinical characteristics of lesions were recorded if a diagnosis of OPL was made (see appendix 2): o History o Site (marked on a grid map) o Size o Clinical appearance (homogeneous vs. nonhomogeneous) o Margin: discrete (well-defined) or diffuse (not well-defined). 5.3.4. Autofluorescent examination • FV negative: Most of normal mucosa appeared as various shades of pale green, and this was classified as FV negative. • FV positive: Tissue showed a distinct reduction in the normal pale green colour and appeared as dark green to black to orange, and this was classified as FV positive. This designation involved a comparison of a lesion site with both adjacent tissue and, as an anatomic control, with tissue on the contralateral site. At least 2 of 3 experienced examiners (Poh, Williams, and I) had to agree for the FV status to be designated either negative or positive. A l l FV positive lesions were measured in size and photographed, if clinically allowed (Appendix 4). 36 5.3.5. Toluidine blue examination The TB staining (OraTest) involved topical application of 1% TB and destaining with acetic acid (1%) as previously described (Epstein et a l . 1997). Lesions were classified as TB negative or positive (strong or weak). Since the designation of some weakly stained lesions was challenging, consensus meetings were held weekly to clarify ongoing complex cases. The clinical findings were categorized by me and confirmed by Dr. M Williams. This information was transferred to the tracking sheet (Appendix 2). Further photos were taken of all TB-positive lesions. 5.4. Histopathological Evaluation The determination whether a lesion was in need of a biopsy was made mainly according to the patient's history, clinical characteristics of the lesion, and TB staining results (routine clinical procedure and not based on this study), regardless of their FV status. Consequently not all of the FV positive lesions were biopsied. In some cases, however, paired biopsies (FV +ve vs. FV -ve) from a patient were taken. Sometimes, histological features within the same biopsy specimen sample varied. The most representative section was used to assess the following histological parameters and data recording: • Pathology diagnosis of the sample, and presence and degree of dysplasia: N (normal, scar, graft, or equivocal), H (reactive, inflammatory, polyp, or hyperplastic lesions without dysplasia), LP (lichen planus), D l (mild dysplasia), D2 (moderate dysplasia), D3 (severe dysplasia), CIS, and SCC. Histological diagnoses of the study samples were confirmed by two pathologists, Dr. L. Zhang and Dr. R. Priddy, the Provincial Oral Biopsy Service, UBC, using criteria from the World Health Organization (WHO, 1978). • Thickness of the epithelium: (measured by Ng and Zhang) o Epithelial and keratin thickness through pathology assessment (only done for obvious cases), e.g., acanthosis (0 = no or mild, 1 = yes), atrophy (0 = no or mild, 1 = yes), hyperkeratosis (0 = no or mild, 1 = yes; type: hyperparakeratosis or hyperorthokeratosis) o Epithelial and keratin thickness were ascertained in 50um and 12.5 um intervals respectively using a microscope eyepiece with an internal grid calibrated with a 0.01 mm stage microscopic ruler. • Surface topography: smooth/normal, rough, and verrucous • Degree of inflammation: 0 = absent or scanty, 1 = mild and moderate, and 2 = severe inflammation 5.5. Statistical Analysis Each lesion or site of previous lesion (e.g., site of previous SCC and site of previous dysplasia) was given a unique identification code that linked to the patient number. Differences and associations between different study groups (e.g., FV positive vs. FV negative lesions) were examined using either Fisher's exact test for categorical variables (gender, smoking habit, ethnicity, presence of OPLs, site, and clinical appearance of OPLs, TB staining, 37 dysplasia, cancer history, histological diagnosis, increase or decrease in epithelial and keratin thickness, and inflammation) or t-test for continuous variables (age, OPL size, measurement of epithelial thickness). The T-test and the Fisher's exact test were performed using GraphPad InStat® version 3.00 for Windows 95/NT, GraphPad Software, San Diego, California, USA. A l l tests were 2-sided. P < 0.05 was considered to be statistically significant. The pilot study indicates that a sample size of approximately 40 is required (at least 20 patients with FV positive and 20 patients with FV negative) to detect a relative risk of 2 with 80% power at an alpha level of 5% (2-sided). The pilot study has a power of approximately 95%> and thus an adequate sample size for detecting the difference in dysplasia proportions. This full study has a power of approximately 99% to detect the different rates of dysplasia seen. This calculation is performed by a web-based statistics program (Lenth 2004) which runs on the Windows® X P professional platform. 38 VI. R E S U L T S The following results were presented as a whole, and then compared between FV positive and negative groups, and between patients with and without a history of cancer. 6.1. Demographics/habits This section presents the information from questionnaires, including demographics, tobacco habit, history of oral cancer, number of lesions examined with FV and number of lesions biopsied. 6.1.1. FV and demographics/habit information Table 4 summarizes the demographic data of all patients in this study. A total of 143 patients were examined under fluorescent visualization (FV). Patient age ranged from 21 to 89 years, with a mean of 60. Fifty-six percent were male, and sixty-four percent had a smoking habit (ever-smoker). Fifty-seven percent of the cases had a history of head and neck cancer. Forty-seven percent of patients had a biopsy taken as part of their clinical evaluation. A total of 319 sites had undergone FV. Of these, 142 (45%) sites were biopsied. FV positive patients were defined as patients having one or more FV positive lesions with or without FV negative lesions. FV negative patients were those with only F V negative lesion(s). When FV positive and negative patients were compared, there were no statistical difference in the age, gender, tobacco habit and history of cancer (all P > 0.05). However, patients with FV positive lesions were more likely to have their lesions biopsied as compared to patients with only negative lesions: 57%> of FV positive patients had biopsies vs 33% FV negative patients, (P = 0.007); and 56%) of all FV examined sites were biopsied in FV positive patients vs. 25%> FV examined sites in FV negative patients (P < 0.0001). 39 Table 4. FV and demographic/habit information All patients FV-positive patients (%) FV-negative patients (%) P value Total 143 83 60 — Mean age (yrs) ± S.D. 60 ± 14 62 ±14 58 ± 14 0.13 Male sex— # (%) 80 (56%) 43/83 (52%) 37/60 (62%) 0.306 Tobacco use ever— # (%) 91 (64%) 52/83 (63%) 39/60 (65%) 0.861 History ofH & N cancer— # (%) 82 (57%) 51/83 (63%) 31/60 (52%) 0.30 % patients with biopsies 67 (47%) 47/83 (57%) 20/60 (33%) 0.007 # of sites examined with FV 319 205 114 % of sites with positive FV 138 (43%) 138/205 (67%) 0/114(0%) % of sites biopsied 142 (45%) 114/205 (56%) 28/114(25%) <0.0001 6.1.2. Cancer history and demographics/habit information Table 5 compared the demographic/habit information between patients with and without a history of oral SCC. There was no difference in the demographic/habit information (all P > 0.05). However, more patients without a history of oral SCC had their lesions biopsied than patients with a history of oral SCC (53% vs. 37%, P =0.007). 40 Table 5. Cancer history and demographic/habit information Patients with Patients without All patients history of SCC history of SCC P value Total 143 82 61 — Mean age (yrs) ± S.D. 60 ± 14 62 ± 15 58± 13 0.154 Male sex— # (%) 80 (56%) 48/82 (59%) 32/61 (53%) 0.499 Tobacco use ever— # (%) 91 (64%) 52/82 (63%) 39/61 (64%) 1 % patients with biopsies 67 (47%) 32/82 (39%) 35/61 (57%) 0.042 # of sites examined with FV 319 165 154 % of sites biopsied 142 (45%) 61/165 (37%) 81/154 (53%) 0.007 6.1.3. Cancer history and FV Table 6 examined FV results in patients with or without a history of oral SCC. There was no difference in either the % of patients with positive FV or the % of FV examined sites positive for FV (both P > 0.05). Table 6. Cancer history and FV With a history of oral cancer Without a history of oral cancer P value # of patients 82 61 % of patients + for FV 51/82 (62%) 32/61 (52%) 0.304 # of sites examined with FV 165 154 % of sites + for FV 1 02/165 (62%) 97/154(63%) 0.908 6.1.4. Cancer history, FV and demographics Table 7 examined patients with and without cancer separately regarding their demographics and FV results. There were no demographic/habit differences between FV positive and negative patients in both the patients with and without a history of oral cancer (all P > 0.05). F V positive patients in both the cancer and non-cancer groups had more biopsies compared to the FV negative patients: in patients with a history of oral SCC, 49%> of FV positive patients had biopsies compared to 14%> of FV negative patients (P < 0.0001); in patients without a history of oral SCC (P = 0.004). 41 Table 7. Demographic features of patients with or without a history of Head & Neck cancer Patients with a history of S C C All patients FV-positive patients (%) FV-negative patients (%) P value1 Total 82 51 31 — Mean age (yr) ± SD 62 63 ± 15 60 ± 16 0.50 Male sex— # (%) 48 (82%) 28/51 (55%) 20/31 (65%) 0.49 Tobacco use ever— # (%) 52 (63%) 31/51 (61%) 21/31 (68%) 1.00 # of sites examined with FV 165 108 57 # (%) of sites +for FV 70 (42%) 70/108 (65%) 0 # (%) of sites biopsied 61 (37%) 53/108 (49%) 8/57(14%) <0.0001 Patients without a history of S C C Total 61 31 29 — Mean age (yr) ± S.D. 58 61 ± 13 56 ±12 0.15 Male sex— # (%) 32 (52%) 15/32 (47%) 17/29(59%) 0.44 Tobacco use ever— # (%) 39 (64%) 21/32 (66%) 18/29 (62%) 0.80 # of sites examined with FV 154 97 57 # (%) of sites +for FV 65 (42%) 65/97 (67%) 0 # (%) of sites biopsied 81 (53%) 60/97 (62%) 21/57(37%) 0.004 FV-positive vs. FV-negative cases 6.2. Clinical attributes 6.2.1. FV and clinical attributes A total of 319 oral sites were examined with FV in the study patients. Tables 8 showed clinical characteristics of the lesions. Of the 319 FV examined lesions, the clinical diagnoses were clinically normal or equivocal including scar and graft area as (n = 96), reactive lesions such as fibroepithelial hyperplasia (9), lichen planus (15) and OPLs/leukoplakia (199). Of the 199 OPLs, 50% were located at the high-risk regions, 64% were nonhomogeneous, and 28% were TB positive (6.3.1.) with a mean size of 254 mm2. The majority of the lesions have a diffuse/indistinct margin (70%) compared to a discrete/distinct margin, and 62% showed a rough surface. 42 When the FV positive sites were compared with F V negative sites, a significantly lower proportion of FV positive lesions were clinically diagnosed as normal/equivocal (12% vs. 49%>, P < 0.0001), and higher proportion as OPLs (78% vs. 51%, P < 0.0001). No differences were noted in the inflammatory reactive lesions between the groups. There was no difference in the %> of OPLs located at high-risk sites and %> of OPLs with diffuse margins between FV positive and negative lesions. A marginal increase in the percentage of lesions with rough surface was noted in the FV positive OPLs (69%>) as compared to the F V negative OPLs (55%) (P = 0.053). Compared to the FV negative lesions, FV positive OPLs were larger (297 mm 2 vs. 192, P = 0.018), and more likely to be nonhomogeneous (72%> vs. 55%o, P = 0.017). These results indicate that judging by clinical characteristics of OPLs, FV staining could identify high-risk OPLs. 43 Table 8. Association of FV with clinical attributes A l l FV+ (%) FV - (%) P value Clinical diagnosis of sites examined 319 \ 3 3 j g j # of normal/equivocal1 106/319(37%) 16/138(12%) 90/183 (49%) <0.0001 % of reactive lesions 9/319(3%) 5/138(4%) 4/183 (2%) 0.508 # of lichen planus 15/319(5%) 9/138(6%) 6/183 (3%) 0.194 # with clinical dx of OPL 201/319(63%) 108/138 (78%) 91/183 (51%) <0.0001 Clinical risk factors of OPLs % of high risk sites Mean area (mm2 ± SD) % of nonhomogenous lesions 199 108 91 100/199(50%) 57/108(53%) 254 ±331 297 ±375 43/91 (47%) 0.478 192 ±225 0.018 127/198(64%) 77/107(72%) 50/91 (55%) 0.017 Other features of OPLs % with diffuse margin % of lesions with rough surface 135/195 (70%) 75/106(71%) 60/89(67%) 0.643 119/191(62%) 72/105 (69%) 47/86(55%) 0.053 Include scar and graft. 6.2.2. Cancer history and clinical attributes Table 9 examined the clinical attributes between patients with and without a history of oral SCC. Compared to the non-cancer group, a significantly higher percentage of lesions examined with FV in the cancer group were normal/equivocal (38% vs. 21%, P - 0.001), and a significantly lower proportion of lesions in the cancer group were clinically diagnosed as OPLs (56% vs. 70%, P = 0.015). OPLs in the cancer and non-cancer groups showed no difference in the location, size, % of nonhomogeneous lesions and diffuse margins. OPLs in the cancer group were more likely to be rough surfaced as compared to the non-cancer group (73% vs. 53%, P = 0.005). 44 Table 9. Association of cancer history with clinical attributes AH With SCC history Without SCC history P value Clinical diagnosis of sites examined 319 165 154 with FV # of normal/equivocal1 96 63/165 (38%) 33/154 (21%) 0.001 % of reactive lesions 9 6/165 (4%) 3/154 (2%) 0.504 # of lichen planus 15 4/165 (2%) 11/154(7%) 0.063 # with clinical dx of OPL 199 92/165 (56%) 107/154 (70%) 0.015 Clinical risk factors of OPLs 199 92 107 % of high risk sites 100/199 (50%) 42/92 (45%) 58/107(54%) 0.257 Mean area (mm2 ± SD) 254 ±331 270 ±337 242 ±327 0.565 % of nonhomogenous lesions 127/198 (64%) 58/91 (64%) 69/107 (65%) 1 Other features of OPLs % with diffuse margin 135/195 (70%) 60/90 (71%) 71/105 (67%) 1 % of lesions with rough surface 119/191 (62%) 65/89 (73%) 54/102 (53%) 0.005 Include scar and graft. 6.2.4. Cancer history, FV and clinical attributes Tables 10 and 11 examined patients with and without cancer separately regarding their clinical attributes and FV results. In both cancer and non-cancer groups, when the FV positive sites were compared with FV negative sites, a significantly lower proportion of FV positive lesions were clinically diagnosed as normal/equivocal: 6%> vs. 36%) for non-cancer group (P < 0.0001,Table 10), and 16%o vs. 55%> in cancer group (P < 0.0001, Table 11); and higher proportion of FV positive lesions were clinically diagnosed as OPLs: 80%> vs. 62%> in non-cancer group (P = 0.021, Table 10), and 77%> vs. 40%> in cancer group (P < 0.0001, Table 11). In non-cancer group, the number of reactive lesions was too small to be compared; there was a significantly increased percentage of FV positivity in clinically diagnosed lichen planus (14%> vs. 2%>, P < 0.009). For the cancer group, the number of lichen planus was too small to be compared; there was marginal increase in FV positivity noted in clinically diagnosed reactive lesions (7%> vs. 1%>, P = 0.088). When the clinical risk factors for OPLs were examined, neither group of lesions showed differences in the %> of lesions with diffuse margins between FV positive and negative lesions. 45 For other OPL clinical risk factors, the two groups (lesions from patients with and without a history of oral SCC) illustrated different results. For the lesions from patients without a history of cancer, compared to the F V negative lesions, the FV positive lesions were larger (308 vs. 175, P = 0.036), and more likely to be nonhomogeneous (75% vs. 55%, P = 0.043), but no differences were noticed for the location and margin of the OPLs, and % of OPLs with rough surface (all P > 0.05). For OPLs from patients with a history of oral SCC, compared to the FV negative lesions, FV showed an increase in % of lesions with rough surfaces (81% vs. 60%, P = 0.03), a marginal increase in the percentage of lesions located in the high-risk sites (54% vs. 32%, P = 0.0857). There was no difference in other clinical parameters (all P > 0.05). These results indicate that judging by clinical characteristics of OPLs, FV could identify high-risk OPLs from patients without a history of oral cancer. Since clinical risk factors did not seem to be effective in differentiate high and low-risk OPLs clinically in patients with a history of oral cancer, it was not surprising that FV could not identify high-risk OPLs from patients with a history of oral cancer. Table 10. Association of FV with clinical attributes of primary lesions All FV+ (%) FV - (%) P value Clinical diagnosis of sites examined with FV 154 65 89 # of normal/equivocal1 36/154(23%) 4/65 (6%) 32/89 (36%) <0.0001 % of reactive lesions 3/154(2%) 0/65 (0%) 3/89 (3%) 0.234 # of lichen planus 11/154 (7%) 9/65 (14%) 2/89 (2%) 0.009 # with clinical dx of OPL 107/154 (70%) 52/65 (80%) 55/89 (62%) 0.021 Clinical risk factors of OPLs 107 52 55 % of high risk sites 58/107 (54%) 27/52 (52%) 31/55 (56%) 0.700 Mean area (mm2 ± s) 242± 327 308 ±390 175 ±235 0.036 % of nonhomogenous lesions .69/107(65%) 39/52 (75%) 30/55 (55%) 0.043 Other features of OPLs % with diffuse margin 71/105 (67%) 36/52 (69%) 35/53 (66%) 0.835 % of lesions with rough surface 54/102 (53%) 28/51 (55%) 26/51 (51%) 0.843 Include scar and graft. 46 Table 11. Association of FV with clinical attributes of mucosal sites in patients with a history of H & N cancer A l l FV+ (%) FV - (%) P value Clinical diagnosis of sites examined with FV 165 73 92 % of normal/equivocal1 63/165 (38%) 12/73 (16%) 51/92 (55%) < 0.0001 % of reactive lesions 6/165 (4%) 5/73 (7%) 1/92(1%) 0.088 % of lichen planus 4/165 (2%) 0/73 (0%) 4/92 (4%) 0.131 % with clinical dx of OPL 92/165 (56%) 56/73 (77%) 36/92 (40%) <0.0001 Clinical risk factors of OPLs 92 56 36 % located at high risk sites 42/92 (45%) 30/56 (54%) 12/36 (32%) 0.0857 Mean area (mm2 ± s) 270±337 307 ±393 215 ±204 0.215 % of nonhomogenous lesions 58/91 (64%) 38/55 (69%) 20/36 (56%) 0.265 Other features of OPLs % with diffuse margin 64/90 (71%) 39/54 (72%) 25/36 (69%) 0.815 % of lesions with rough surface 65/89 (73%) 44/54 (81%) 21/35 (60%) 0.031 Include scar and graft. 6.3. TB staining 6.3.1. FVandTB As shown in Table 12, of the 199 OPLs, 28%> were TB positive in this study. For OPLs from patients without a history of oral SCC, 22%> of the 107 OPLs were TB positive. A higher proportion (34%>) of OPLs from patients with a history of oral SCC were TB positive. When compared to FV negative lesions, FV positive lesions showed significantly higher proportion of TB positive OPLs (41%> vs. 12%>, P < 0.0001). This significant increase remained when the lesions were divided into those from patients with and without a history of cancer (P = 0.0009 and P = 0.0067 respectively). These results suggested that FV positivity and TB positivity were strongly associated. 47 Table 12. Association of FV with TB staining All FV+ (%) FV - (%) P value Total number of OPLs 199 108 91 % with TB positive staining 55/199 (28%) 44/108 (41%) 11/91 (12%) <0.0001 Patients without a history of oral SCC # of OPLs 107 52 55 % with positive TB staining 24/107 (22%) 19/52 (37%) 5/55 (9%) 0.0009 Patients without a history of oral SCC # of OPLs 92 56 36 % with positive TB staining 31/92 (34%) 25/56 (45%) 6/36(17%) 0.007 6.3.2. Cancer history and TB Table 13 examined TB results in patients with and without a history of oral SCC. There was no difference in the % of patients with positive TB in patients with and without a history of oral cancer (29% vs. 25%, P = 0.574). OPLs in patients with a history of oral cancer showed a marginal increase in the proportion of TB positive lesions as compared to the lesions from patients without a history of oral SCC (34% vs. 22%, P = 0.082). Table 13. Cancer history and TB With a history of oral cancer Without a history of oral cancer P value # of patients 82 61 % of patients + for TB 24/82 (29%) 15/61 (25%) 0.574 # of sites examined with TB 92 107 % of sites + for TB 31/92 (34%) 24/107 (22%) 0.083 6.3.3. Clinical attributes of OPLs and TB Table 14 compared the clinical characteristics of TB positive and negative lesions. Compared to the TB negative lesions, TB positive lesions showed a significantly lower proportion of lesions clinically diagnosed as normal/equivocal (4% vs. 36%, P < 0.0001) but higher proportion of OPLs (92% vs. 56%, P < 0.0001). 48 Regarding to the clinical risk factors of OPLs, there were no differences in the proportion of lesions located at the high-risk sites and proportion of lesions with diffuse margins. However, compared to the TB negative lesions, the size of TB positive lesions was marginally increased 2 2 (327 mm vs. 225 mm , P - 0.054), TB positive lesions were more likely to be nonhomogeneous in clinical appearance (85% vs. 57%, P = 0.0001), and more likely to have a rough surface (85% vs. 54%, P< 0.0001). These results indicate that judging by clinical characteristics of OPLs, TB staining could identify high-risk OPLs. Table 14. Association of TB with clinical attributes All TB positive TB negative P value i (%) (%) Clinical diagnosis of sites examined 319 60 259 with FV # of normal/equivocal1 96 2/60 (4%) 94/259 (36%) < 0.0001 % of reactive lesions 9 2/60 (3%) 7/259 (3%) 0.679 # of lichen planus 15 1/60 (2%) 14/259(5%) 0.319 # with clinical dx of OPL 199 55/60 (92%) 144/259(56%) < 0.0001 Clinical risk factors of OPLs 199 % of high risk sites 100/199 (50%) 33/55 (60%) 67/144 (47%) 0.113 Mean area (mm2 ± SD) 254 ±331 327 ±436 225 ± 273 0.054 % of nonhomogenous lesions 128/198 (64%) 47/55 (85%) 81/143 (57%) 0.0001 Other features of OPLs % with diffuse margin 137/195 (70%) 42/54 (78%) 95/141 (67%) 0.167 % of lesions with rough surface 119/191 (62%). 45/53 (85%) 74/138 (54% <0.0001 Include scar and graft. 6.3.4. Cancer history, clinical attributes and TB Tables 15 and 16 examined patients with and without cancer separately regarding their clinical attributes and TB results. In both cancer and non-cancer groups, when the TB positive sites were compared with TB negative sites, a significantly lower proportion of TB positive lesions were clinically diagnosed as normal/equivocal: 4%> vs. 25% for non-cancer group (P = 0.017,Table 15), and 3% vs. 47% in cancer group (P < 0.0001, Table 15); and higher proportion of TB positive lesions were clinically diagnosed as OPLs: 92% vs. 65% in non-cancer group (P = 0.005, Table 15), and 91%) vs. 47% in cancer group (P < 0.001, Table 16). In non-cancer group, the number of reactive lesions was too small to be compared; there was no difference in TB positive rate in clinically diagnosed lichen planus (P = 0.692). For the cancer group, the number of lichen planus was too small to be compared; there was no difference in the TB positive rate in the reactive lesions (P = 0.604). When the clinical risk factors for OPLs were examined, neither groups of lesions showed differences in the % of lesions with diffuse margins between TB positive and negative lesions. For other OPL clinical risk factors, the two groups (lesions from patients with and without a history of oral SCC) illustrated different results. For the lesions from patients without a history of cancer, compared to the TB negative lesions, the TB positive lesions were larger (359 vs. 207, P = 0.045), and more likely to be nonhomogeneous (96% vs. 55%, P = 0.0002), but no differences was noted for the location and margin of the OPLs, and % of OPLs with rough surface (all P > 0.05). For OPLs from patients with a history of oral SCC, compared to the TB negative lesions, TB positive lesions showed an increase in % of lesions with rough surfaces (90% vs. 64%, P = 0.011), and in the percentage of lesions located in the high-risk sites (61% vs. 38%, P = 0.046). These results indicate that judging by clinical characteristics of OPLs, TB could identify high-risk OPLs from patients without a history of oral cancer. For patients with a history of oral cancer, TB did not seem to be able to markedly differentiate the high from low risk lesions, particularly in view of the fact that most of oral SCCs were in high-risk sites, hence the differences in the risk sites would not help very much in the differentiation. Since clinical risk factors did not seem to be effective in differentiate high and low-risk OPLs clinically in patients with a history of oral cancer, it was not surprising that TB could not identify high-risk OPLs from patients with a history of oral cancer. 50 Table 15. Association of TB with clinical attributes of primary lesions (patients with no history of H & N cancer) All TB positive (%) TB negative (%) P value Clinical diagnosis of sites examined with FV 154 26 128 # of normal/equivocal1 33 1/26(4%) 32/128(25%) 0.017 % of reactive lesions 3 / o 3/128(2%) 1 # of lichen planus 11 1/26 (4%) 10/128(8%) 0.692 # with clinical dx of OPL 107 24/26(92%) 83/128(65%) 0.005 Clinical risk factors of OPLs 107 24 83 % of high risk sites 58/107 (54%) 14/24 (58%) 44/83 (53%) 0.817 Mean area (mm2 ± SD) 242 ± 327 359 ±472 207 ±261 0.045 % of nonhomogenous lesions 69/107 (65%) 23/24 (96%) 46/83 (55%) 0.0002 Other features of OPLs % with diffuse margin 71/105 (67%) 18/24(75%) 53/81 (65%) 0.462 % of lesions with rough surface 54/102 (53%) 18/23 (78%) 36/79 (46%) 0.008 Include scar and graft. 51 Table 16. Association of TB with clinical attributes of mucosal sites in patients with a history of H & N cancer All TB positive TB negative (%) (%) P value Clinical diagnosis of sites examined with FV % of normal/equivocal1 % of reactive lesions % of lichen planus % with clinical dx of O P L 165 63 6 4 92 34 1/34(3%) 2/34 (6%) 0 31/34(91%) 131 62/131(47%) 4/131 (3%) 4/131 (3%) 61/131 (47%) <0.0001 0.604 0.588 <0.001 Clinical risk factors of OPLs % located at high risk sites 2 42/92(45%) 19/31(61%) 23/61 (38%) 0.046 Mean area ( m m ' ± S D ) 270 ±337 302 ±413 252 ± 289 0.506 % of nonhomogenous lesions 58/91(64%) 23/31 (74%) 35/60(58%) 0.170 Other features of OPLs % with diffuse margin 6 4 / 90 (71%) 23/30 (77%) 41/60 (68%) % of lesions with rough surface 65/89(73%) 27/30(90%) 38/59(64%) 0.468 0.011 Include scar and graft 6.4. Histological features 6.4.1. FV and histological features excluding dysplasia Table 17 examined the influence of histological features (excluding dysplasia) on FV. Of the 131 biopsies, 36 (28%) showed normal thickness of the epithelium, while the 17 (13%) were atrophic and 78 (59%) were acanthotic, as assessed pathologically. When the thickness of the epithelium was measured, the mean thickness was 131 um. When keratin layer was assessed, 77 (59%) demonstrated hyperkeratosis, and most of the keratin layer was parakeratin (65%). When the inflammation was examined, 56% showed pronounced inflammation in the lamina propria. When the surface topography was examined, 38% of the lesion showed a rough surface whereas 62% showed a smooth surface. The histological features were compared between FV positive and negative biopsies. There were no differences in measured epithelial thickness (388 vs. 326 um, P = 0.199), in the proportion of 52 atrophic lesions (16% vs. 9%, P = 0.303), and in the proportion of the lesions with rough surface (41%o vs. 34.5%o, P = 0.585) between F V positive and negative groups. Compared to the F V negative biopsies, F V positive biopsies showed a significantly lower proportion of lesions with normal thickness of epithelium (17%> vs. 42%, P = 0.003), higher proportion of lesions with increased epithelial thickness (acanthosis) (67%> vs. 49%>, P = 0.048) as assessed pathologically, lower proportion of lesions with hyperkeratosis (41% vs. 65%, P - 0.008), and higher proportion of lesion with pronounced inflammation (76% vs. 30%, P < 0.0001). 53 Table 17. FV and histological features (other than dysplasia) #of cases FV positive (%) FV negative (%) P value Thickness of epithelium 131 76 55 By path assessment Normal 36 13/76(17%) 23/55 (42%) 0.003 Atrophy 17 12/76(16%) 5/55 (9%) , 0.303 Acanthosis 78 51/76 (67%) 27/55 (49%) 0.048 By measurement (um) 131 388 ±303 3261211 0.199 Keratin By path assessment 76 55. No hyperkeratosis (normal) With hyperkeratosis 64 77 45/76 (59%) 31/76 (41%) 19/55 (35%) 36/55 (65%) 0.008 By type of keratin 56 42 Orthokeratosis Parakeratosis 34 64 18/56 (32%) 38/56 (68%) 16/42 (38%) 26/42 (62%) 0.669 Inflammation 74 54 No/mild Moderate/severe 56 72 18/74 (24%) 56/74 (76%) 38/54 (70%) 16/54 (30%) < 0.0001 Surface topography 76 55 Smooth Not smooth 81 50 45/76 (59%) 31/76 (41%) 36/55 (65.5%) 19/55 (34.5%) 0.585 6.4.2. FV, clinical diagnosis, TB staining and dysplasia 6.4.2.1. FV and dysplasia Table 18 examined the ability of FV as a tool to identify dysplasia. A significantly higher percentage of dysplastic lesions were FV positive as compared to the hyperplastic lesions (70%> vs. 31%o, P < 0.0001). The sensitivity and specificity of FV in detecting dysplasia were 70%» and 69%> respectively. The significant differences remained even when the lesions were divided into cancer and non-cancer group (both P < 0.05). These results indicated that judging by dysplasia, FV could identify high-risk OPLs. 54 Table 18. Sensitivity and specificity of FV (judging by gold standard histology) Hyperplasia Dysplasia P value All biopsies FV positive FV negative 39 12/39(31%) 27/39 (69%)2 92 64/92 (70%)' 28/92 (30%) <0.0001 Patients with no history of oral cancer FV positive FV negative 22 5/22 (23%) 17/22 (77%)2 54 35/54 (65%)' 19/54(35%) 0.001 Patients with a history of oral cancer (#) FV positive FV negative 17 7/17(41%) 10/17 (59%)2 38 29/38 (76%)' 9/38 (24%) 0.016 Sensitivity rate. Specificity rate. Table 19 examined the association between FV and the presence and degree of dysplasia. With increasing degree of dysplasia, there was an increase in the F V positivity: 31% hyperplasia, 59% low-grade dysplasia (mild/moderate), and 91% high-grade dysplasia were FV positive (P < 0.0001). The difference was also significant between different histological groups (between hyperplasia and low-grade dysplasia, P = 0.008; between low-grade and high-grade dysplasia, P = 0.002). The trend remained for both lesions from patients without a history of oral SCC: 23% of hyperplasia, 56% of low-grade dysplasia and 87% of high-grade dysplasia were FV positive (P = 0.005); and for patients with a history of cancer: 41% of hyperplasia, 64% low-grade dysplasia and 94% high-grade dysplasia (P = 0.006). For lesions from patients without a history of cancer, there was still difference between hyperplasia and low-grade dysplasia (P = 0.016), and approaching difference between low and high-grade dysplasia (P = 0.056). For lesions from patients with a history of cancer, approaching difference was noted between low and high grade dysplasias (P = 0.053); however, a higher percentage of hyperplasia showed FV positivity and the difference between the low-grade and hyperplasia in these patients were no longer significant (P = 0.206). These results suggested that FV positivity increased with degree of dysplasia. For OPLs from patients without a history of oral SCC, there were a significant increase in proportion of FV positive lesions from hyperplasia to low-grade dysplasia and another significant increase from low to high-grade dysplasia. However, for OPLs from patients with a history of oral cancer, there was no difference between hyperplasia and low-grade dysplasia. Only high-grade dysplasia showed approaching significant increase in FV positive lesions. 55 Table 19. FV and degree of dysplasia Hyperplasia Low-grade dysplasia P value' High-grade dysplasia P value2 AH biopsies 39 61 31 FV positive FV negative 12/39(31%) 27/39 (69%) 36/61 (59%) 25/61 (41%) 0.008 28/31 (91%) 3/31 (9%) 0.002 Patients with no history of 22 39 15 oral cancer (#) FV positive FV negative 5/22 (23%) 17/22 (77%) 22/39 (56%) 17/39(44%) 0.016 13/15 (87%) 2/15(13%) 0.056 Patients with a history of 17 22 16 oral cancer(#) FV positive FV negative 7/17(41%) 10/17(59%) 14/22 (64%) 8/22 (36%) 0.206 15/16(94%) 1/16(6%) 0.053 _ 2 Comparison between hyperplasia and low-grade dysplasia. Comparison between low- and high-grade dysplasia. 6.4.2.2. Clinical diagnosis and dysplasia Table 20 examined the ability of clinical diagnosis of OPLs as a tool to identify dysplasia. A significantly higher percentage of dysplastic lesions were clinically diagnosed as OPLs as compared to the hyperplastic lesions (85%> vs. 67%>, P = 0.032). When the biopsies were divided into those from patients with and without a history of oral SCC, the significant difference observed earlier was only seen in patients without a history of cancer (85%o of dysplasia were clinically diagnosed as OPLs as compared to 55%> hyperplastic lesions, P = 0.007). Clinical diagnosis could not differentiate dysplasia from hyperplasia in those biopsies from patients with a history of cancer. This is in contrast to that seen using FV (Table 20). These results indicated that judging by dysplasia, clinical diagnosis based on clinical risk factors could identify high-risk OPLs. However, when the lesions were divided into lesions from patients with and without a history of oral SCC, it showed that clinical risk factors only worked in lesions from patients without a history of oral SCC and not the lesions from patients with a history of oral SCC. 56 Table 20. Sensitivity and specificity of clinical diagnosis (judging by gold standard histology) Hyperplasia Dysplasia P value All biopsies OPL Not OPL 39 26/39 (67%) 13/39 (33%)2 92 78/92 (85%)' 14/92(15%) 0.032 Patients with no history of oral cancer (#) OPL Not OPL 22 12/22 (55%) 10/22 (45%)2 54 46/54 (85%)' 8/54(15%) 0.007 Patients with a history of oral cancer (#) OPL Not OPL 17 14/17(82%) 3/17 (18%)2 38 32/38 (84%)' 6/17(16%) Sensitivity rate. Specificity rate. Table 21 examined the association between clinical diagnosis and the presence and degree of dysplasia. With increasing degree of dysplasia, there was a marginal increase in the proportion of lesions with a diagnosis of OPLs: 67% hyperplasia, 85% low-grade dysplasia, and 84% high-grade dysplasia were clinically diagnosed as OPLs (P = 0.063). The difference resulted from a jump from hyperplasia (67%) to low-grade dysplasia (85%) (P = 0.046), but clinical diagnosis did not differentiate low from high-grade dysplasias (85% vs. 84%, P = 1). When the biopsies were divided into those from patients with and without a history of oral SCC, only biopsies from patients without a history of cancer showed an increase between hyperplasia and low-grade dysplasia (85%) (P - 0.016), but again the difference could not be discerned by clinical diagnosis between low and high-grade dysplasia (85% vs. 87%, P = 1). In patients with a history of oral cancer, clinical diagnosis could not differentiate hyperplasia and dysplasia: 82% hyperplasia, 86% low-grade dysplasia and 81% high-grade dysplasia were diagnosed clinically as OPLs (P = 0.902). These results suggested that while clinical diagnosis based on clinical risk factors could identify more dysplastic lesions, these risk factors do not differentiate low-grade dysplasias from high-grade ones in lesions from patients without a history of oral SCC. Clinical risk factors were ineffective in differentiating lesions from different histological groups in patients with a history of oral SCC. 57 Table 21. Clinical diagnosis and degree of dysplasia Hyperplasia Low-grade dysplasia P value High-grade dysplasia P value # of biopsies Clinical diagnosis OPL Not OPL 39 26/39 (67%) 13/39 (33%) 61 52/61 (85%) 9/61 (15%) 31 0.046 26/31 (84%) 5/31 (16%) Patients with no history of oral cancer(#) Clinical diagnosis OPL Not OPL 22 12/22 (55%) 10/22 (45%) 39 33/39 (85%) 6/39(15%) 0.016 15 13/15 (87%) 2/15 (13%) Patients with a history of oral cancer (#) Clinical diagnosis OPL Not OPL 17 14/17(82%) 3/17(18%) 22 19/22(86%) •3/22(14%) 16 13/16(81%) 3/16(19%) 0.682 'Comparison between hyperplasia and low-grade dysplasia. Comparison between low- and high-grade dysplasia. 6.4.2.2. TB staining and dysplasia Table 22 examined the ability of TB staining as a tool to identify dysplasia. A significantly higher percentage of dysplastic lesions were TB positive as compared to the hyperplastic lesions (41% vs. 13%, P = 0.0002). These results indicated that judging by dysplasia, TB could identify high-risk OPLs in both patients with and without a history of oral SCC. 58 Table 22. Sensitivity and specificity of TB staining (judging by gold standard histology) Hyperplasia Dysplasia P value All biopsies 39 92 TB positive 5/39(13%) 38/92 (41%)' 0.0002 TB negative 34/39 (87%)2 54/92 (59%) Patients with no history of oral cancer (#) 22 54 TB positive 2/22 (9%) 18/54 (33%)' 0.043 TB negative 20/22 (91%)2 36/54 (67%) Patients with a history of oral cancer (#) 17 38 TB positive 3/17(18%) 20/38(53%)' 0.019 TB negative 14/17 (82%)2 18/38 (27%) a _ Sensitivity rate. Specificity rate. Table 23 examined the association between TB staining and the presence and degree of dysplasia. With increasing degree of dysplasia, there was an increase in the TB positivity: 13% hyperplasia, 26% low-grade dysplasia (mild/moderate), and 71% high-grade dysplasia were TB positive (P < 0.0001). When the individual histological groups were examined, the difference was mainly from the difference between low and high-grade dysplasias (71% vs. 26%, P < 0.0001) as opposed to between low-grade dysplasia and hyperplasia (26% vs. 13%, P = 0.135). The trend of significant increase with increasing dysplasia remained for both biopsies from patients with no history of cancer: 9% of hyperplasia, 18% of low-grade dysplasia and 73% of high-grade dysplasia were TB positive (P < 0.0001); and for biopsies from patients with a history of cancer: 18% of hyperplasia, 41% low-grade dysplasia and 69% high-grade dysplasia (P = 0.012). When the individual histological groups were compared, the lesions from patients without a history of cancer showed a significant difference between high and low-grade lesions (73% vs. 18%, P < 0.0001) but not between low-grade dysplasia and hyperplasia (18% vs. 9 % , P = 0.467). There were no differences between the histological groups in lesions from patients with a history of oral cancer (both P > 0.05). When presence and degree of dysplasia were examined, the differences (for all samples together and for samples from patients with no history of cancer) were only between low-grade dysplasia and high-grade dysplasia (both P < 0.0001), contrary to FV, where the difference lied between both hyperplasia and low-grade dysplasia (P = 0.008) and between low-grade dysplasia and high-grade dysplasia (P = 0.002). These results suggested that TB could identify more dysplastic lesions but these risk factors do not differentiate low from high-grade dysplasias in lesions from patients without a history of oral SCC. Clinical risk factors were ineffective in differentiating lesions among different histological groups in patients with a history of oral SCC. These results suggested that TB could differentiate high from low-grade dysplasia and hyperplasia in patients without a history of oral SCC. For lesions from patients with a history of oral SCC, there was an increase of TB positive staining with increasing degree of dysplasia but the differences between different histological groups were not significant with exception of high-59 grade dysplasia to hyperplasia (69% vs. 18%>, P = 0.005) or to hyperplasia and low-grade dysplasia grouped together (69%> vs. 31%, P - 0.016). Table 23. TB staining and degree of dysplasia Hyperplasia Low-grade dysplasia P value1 High-grade dysplasia P value All biopsies TB positive TB negative 39 61 5/39(13%) 16/61(26%) 34/39 (87%) 45/61 (74%) 0.135 31 22/31 (71%) 9/31 (29%) <0.0001 Patients with no history 22 39 of oral cancer(#) TB positive 2/22(9%) 7/39(18%) TB negative 20/22(91%) 32/39(82%) 0.467 15 13/15(73%) 2/15 (13%) <0.0001 Patients with a history of oral cancer(#) TB positive TB negative 17 3/17(18%) 14/17(82%) 22 9/22 (41%) 13/22 (59%) 0.169 16, 11/16(69%) 5/16(31%) 0.112 Comparison between hyperplasia and low-grade dysplasia. "Comparison between low- and high-grade dysplasia. c 60 6.5. Combined diagnostic tools (clinical diagnosis, FV and TB staining) in identifying dysplasia Table 24 examined different combination of the three diagnostic tools (FV, clinical diagnosis and TB) in detecting dysplasia. 6.5.1. One diagnostic tool When only one tool (clinical diagnosis based on clinical risk factors, FV or TB) was assessed by its ability to detect dysplastic lesions, clinical diagnosis showed the highest false positive rate (67%) while the TB showed the lowest false positive rate (13%o). In other words, clinical diagnosis had the lowest specificity rate (33%) while the TB had the highest specificity rate (87%>). On the other hand, clinical diagnosis showed the highest sensitivity rate or true positive rate (85%>) and the TB showed the lowest sensitivity rate (41%>). FV was in between. When different degree of dysplasia was looked at, FV had the highest sensitivity in detecting high-grade dysplasia (91%>) as compared to clinical diagnosis (84%>) and TB staining (71%>). 6.5.2. Two diagnostic tools Using two diagnostic tools with either or both positive: When the clinical diagnosis was used in conjunction of either FV or TB (either clinical &/or FV, either clinical &/or TB), a high false positive rate was noted (74%> and 67%> respectively). On the other hand, i f only the two visual aids (FV &/or TB) were used, a lower false positive rate (36%>) (or higher specificity) was noted than the clinical diagnosis combined with a visual aid. The sensitivity was highest for clinical diagnosis in conjunction with FV: 92%> of all dysplasia were identified and 100%) (31/31) high-grade dysplasia were identified. Clinical diagnosis in conjunction with TB identified 87%> of dysplasia including 90% of high-grade dysplasia. When the two visual aids were used, 75% of dysplasia were identified including 97% of high-grade dysplasia. Two diagnostic tools positive: Positive for two diagnostic tools resulted in higher specificity or lower false positive rate. Lesions positive for 2 diagnostic tools (both clinical & FV +, both clinical & TB +, both FV & TB positive) showed a false positive rate of 8-23%>. Although a much smaller number of cases were positive for two diagnostic tools, their sensitivity for high-grade dysplasia remained high (65%> to 74%>). If the proportion of the high-grade dysplasia with two markers positive was calculated over all the cases positive for both markers instead of all high-grade dysplasia, a high proportion of the lesions positive for two diagnostic tools or markers were high-grade dysplasia: 35%> for both clinical & FV +, 49%> for clinical & FV +, and 56% for FV & TB +. In summary, a significant proportion of lesions positive for two diagnostic tools were high-grade dysplasia, varying from 35% to 56%>. 61 Clinical + & FV-: Lesions belong to this category had a low sensitivity for dysplasia (23%), particularly the high-grade dysplasia (10%>). Nonetheless, some high-grade dysplasia were still missed by FV but diagnosed by clinical criteria. Clinical + & TB -: The results were very similar to clinical + & FV -. Again, even high-grade dysplasia could be missed by TB while diagnosed by clinical criteria (4/16, 25%>). Clinical - & FV+: The FV examination in this study was centred on clinically visible lesions or site of previous lesions with the exception of one patient. As shown in Table 8, 30 of the 138 FV positive sites were not diagnosed as OPLs. Of these 30 FV positive but clinically negative sites, 16 were clinically regarded as normal/equivocal (no obvious lesions), 5 were reactive lesions and 9 lichen planus. Ten biopsies were taken from the 30 FV positive sites that were not diagnosed as OPLs clinically including 4 from clinically normal/equivocal sites. Of these 10 biopsies, 70% turned out to be dysplastic, and 50%> (5/10) were high-grade dysplasias, which account for 16%> of all high-grade dysplasia in this study (Table 23). The result indicated that FV could identify dysplastic lesions that were clinically innocuous. Furthermore, of the 4 biopsies from clinically normal/equivocal areas (3 at previous lesion sites and 1 at a new site), 3 lesions were high-grade dysplasias. This suggested that even in the absence of obvious clinical lesions, FV could identify high-risk OPLs. A l l + (clinical, FV & TB): low chance to be nondysplastic (8%), about a third dysplasia positive for all three markers, and more than half (58%>) high-grade dysplasia were such lesions. However, there was no marked improvement in the sensitivity and specificity compared to using two tools. Only clinical + (FV -, TB -): 38% nondysplastic lesions belong to this group, a low percentage of dysplasia belong to this (17%>) and very low high-grade dysplasia belong to this (3%>). 62 at gra lasi i •SS sp a >, *> -"O 03 °t» 6X1 •— • c. £ 2. as .2 "S. SO Q • c« a « os vo CN O N O N CO 5 I I* 5 I V in ro O N o o -~< C5 o o 3 o s 0s- 0s-Tt- 7-^ i 1 OO O N s—^ ^-^ •^ -^  -^ .— CO co co VO oo CN CN CN CN 0 s N ? 6 S -in O N VO 00 m CN , , , , VO vo VO CN vo VO in CO NT 0 s O O NT 0 s O — 9 N vo oi N: © s o s in vo co co XT o s O N 00 CO <Y> oo ( N CO CN o CN V NT in vo o CN > «a *•> 0 s 0s- 0 s O in VO CN s—^  N—^ .—> vo , 1 co — co co in 0s-m oo 3 i 1 8 2 I * ? « ©\ o s O N CO VO CN O N CO 0s-O N co + O o s DJD 03 u + > + 05 H ^2 VO in Os VO 0 s 0s- 0s- 0 s 0s-in O CN oo r~ O N oo s—s s— CN CN CN CN CN O N O N O N O N O N OO OO m o VO CO 00 00 NT 0s-O N co O N CN CN CK in iv-, 2 «s o ^ o V CN ct Os VO VO Os ro N^  0 s r~ vo O N co vo CN 2 °° + + > « H u •-. o o "S O O e e OX) BJj pa .2 .2 H -3 •S •-. _^ o CS CS u > <w <u JS co — s ° <a o V NT CN VO CN O N m 0 s CO CN 0 s O N co CN O N VO co 0s-ro O N O N CO CO O N m VO ~ VO VO 5 S NT 0 s VO co CN Os co co 0s-oo O N co VO co + + > 5 Vi yi "5 'ee • O O C S .2 .2 •3 "3 W "3 "« ^ •s -s > ' u o fa. -M o o o 03 CQ CQ CN °0 «5 NT NT 0 s 6 S co t~-CN CO CN O N 00 CO NT 0 s 00 CN O N CN — > fa. + '1 oa e H es °0 + fa. «8 o s + '-3 u u V u •St Os 0 s 00 in ro oo Nf 0s-ro CN Os (44%) (65%) (8%) (8%) O N CO O N co O N co CO co 00 CO in CN o co 03 H > fa. C3 u Os N: 0 s co (26%) 0 s CN (30%) (45%) 0 s CO (21%) (25%) VO jo VO CN CN VO VO VO vo ro oo O CN ro in 0 s CN O N — VO 0 s 00 co O N co O c WD S3 03 H > fan v—' + 'Ji = e < o -a c cs ^ cd cj •S 1 •-a £ t j fen °- « OiJ co cj ' >• c I -a co • "° C g S Cd cj i cd 1.-S)1 6 0 - c a. J= ^ >> •SP « -a ^ •- -S co T; oo CJ P CCJ « <u <N iS i M ttc^ .S £ o -a s — c O cd •-C o cd . o •-" "> > 23 > .C- cd SS -5? — -S "° 52 E = Q g 5 >;= 2 " H O > .a to S .E x S o vo to fJ -S C * *S CJ ^ u ?5 - co g CO W m V O co — P. C E 3 3 _ C O O ^ ° o « « c = f. a o N ^ fJ cj ^ • 'r-; -= ° cd cd CJ CJ 1 CO -o o CJ c 1 /1 i l l 0--5 x cj cd CO o ,C3 c c -3 5 _o .3 -3 (L> ^ CJ CJ VO •a c CO .3 CJ S o CO ' •3 CJ •S 73 E <+-• o o C3 I " t i -es _y" « 2 c " O ^ c 1 CO C '—' co - T3 cd 6.5. Correlation of FV status with dysplasia in paired biopsy Twenty patients had paired biopsies (40 biopsy specimens) from a FV positive area and a FV negative. The FV positive and negative samples from each patient were at least 1 cm apart. A l l FV positive biopsies were from clinical lesion area (FV + and clinical +). For FV negative lesions, some were from the same clinically visible lesion (FV - and clinical +), and others were taken from clinically normal-looking area adjacent to the clinical lesions (FV - and clinical -). Tables 25 examined the relationship between FV and dysplasia in paired biopsies using all the 20 pairs of samples (40) including those FV negative lesions from apparently clinically normal area. Of these, 7 pairs showed similar histological diagnosis between F V positive and negative biopsies: 5 pairs nondysplastic, 1 pair low-grade dysplasia, and 1 pair high-grade dysplasia. The remaining 13 pairs all showed a more severe histology in the FV positive biopsies compared to the negative biopsies: 7 pairs with FV negative lesions non-dysplastic but FV positive lesions low-grade dysplasia; 3 pairs with FV negative lesions non-dysplastic but F V positive lesions high-grade dysplastic; and 3 pairs with FV negative lesions low-grade dysplasia but FV positive lesions high-grade dysplasia. No significant difference, however, was noted between the two groups (P = 0.167). Table 26 only used paired samples that were both taken from clinically visible lesions (matching), and included 10 pairs of the samples. Of these, 6 pairs had similar diagnoses between the FV positive and matching negative biopsies: 4 pairs both showed non-dysplasia, 1 pair both showed low-grade dysplasia, and 1 pair both showed high-grade dysplasia. The remaining 4 pairs showed a more severe histology in FV positive biopsies compared to the matching negative samples: 1 pair with FV negative lesions non-dysplastic but matching FV-positive biopsy low-grade dysplasia; 3 pairs with FV negative lesions low-grade dysplasia but matching FV positive biopsies high-grade dysplasia. There was still no statistical significance between the two groups. 64 Table 25. Correlation of FV status with dysplasia in paired biopsy FV negative Non- Low-grade High-grade dysplastic/normal dysplasia dysplasia Total P value FV positive Non-dysplastic/normal Low-grade dysplasia 5 0 0 7 1 0 5 8 0.767 High-grade dysplasia 3 3 1 7 Total 75 Table 26. Correlation of FV status with dysplasia in paired biopsies (only samples from clinically visible lesions) Matching FV negative FV positive Non-dysplastic/normal Low-grade dysplasia High-grade dysplasia Total P value Non-dysplastic/normal 4 0 0 4 Low-grade dysplasia 1 1 0 2 0.079 High-grade dysplasia 0 3 1 4 Total 5 4 7 65 Vll. DISCUSSION The prognosis of oral SCC remains one of the worst among major human cancers, with 5-year survival rate of approximately 50%. Diagnosis and management of oral premalignant lesions is one key to improving the dismal prognosis. The majority of premalignant lesions, however, do not progress into cancer. Improving our ability to identify the high-risk premalignant lesions is crucial for improving the prognosis of oral SCC. Pathological assessment of a biopsy sample for the presence and degree of dysplasia is the current gold standard in judging cancer risk of OPLs. Recently researchers have also shown that molecular analysis of biopsy samples could markedly improve our ability in predicting the cancer risk of OPLs. However, histology and molecular analysis are largely dependent upon the decision of a clinician to biopsy an OPL and the choice of the biopsy site. Currently the decision regarding when and where to biopsy is based on the clinical risk factors including site, size and clinical appearance of the lesions, habit and cancer history of the patients. The clinical criteria, however, are far from adequate in the identification, and this has prompted intense interest in the research and development of new visual aid. This thesis is the first study to investigate fluorescent visualization (FV) as a tool in the identification of high-risk OPLs using a large number of patients. The 143 patients used in this study were part of an ongoing longitudinal study in the Dysplasia Clinic at B C C A . These patients had either a history of primary oral dysplasia or a history of oral SCC. The oral lesions were followed and studied for risk factors that were associated with cancer progression or development of second oral malignancy. The ability of FV as a tool in identifying high-risk OPLs were examined and compared by (1) patients' history of head and neck cancer; (2) clinical diagnosis based on clinical characteristics of the oral lesions; (3) TB staining results, and (4) dysplasia (gold standard). The preliminary results suggested that FV-positive sites had a higher cancer risk than FV-negative sites. 7.1. Cancer history and diagnostic tools As discussed before, one of the important factors in the determination of the cancer risk of OPLs is patient risk. The most important patient risk factor is a history of head and neck cancer. Leukoplakia in patients with a history of oral cancer has a much higher risk of progressing to cancer than leukoplakia in patients without such a history (Grant et al., 1993). A risk-predicting tool should therefore identify more high-risk lesions in patients with a history of cancer compared to patients without a history of cancer. In this study, no difference was noted in the proportion of FV positive patients or FV positive sites between patients with and without a history of oral SCC (Table 6). In comparison, clinical diagnosis (of OPLs based on clinical characteristics of lesions) did not differentiate the risk of lesions between patients with and without a history of cancer either. In fact, it was worse than FV. Based on the clinical characteristics, compared to lesions from patient 66 without a history of oral cancer, lesions from patients with a history of oral cancer actually showed a lower proportion of lesions diagnosed as OPLs (56% vs. 70%, P - 0.015, Table 9), even though there were no differences in the site, size and clinical appearance of lesions in the two groups of patients (most important factors in the diagnosis of OPLs). Such results are contrary to conventional wisdom that more OPLs should be present in patients with a history of oral cancer as compared to those without. Part of the problems resides in the fact that the establishment of clinical risk factors of OPLs (size, site and appearance) were generally based on studies of primary oral lesions, not from oral lesions of patients with a history of oral cancer, hence these factors may not be applicable to the cancer patients as suggested by the results of this study. The anatomy and physiology of oral mucosa are altered after aggressive treatment of cancer, such as surgery and radiation. Such mucosal changes include scar formation, post-radiation mucosal atrophy/erythema, and fragility of the tissue that is hard to heal in response to trauma. A l l of these non-specific reactive changes could make it hard to differentiate from true leukoplakia and erythroleukoplakia. Interestingly, the study results showed that a higher proportion (73%) of lesions with a rough surface were noted in patients with a history of oral SCC than those without a history of cancer (53%, P = 0.005, Table 9), suggesting roughness of the leukoplakia instead of erythematous component may be more diagnostic of OPLs in patients with a history of oral SCC. Of the three clinical risk-predicting tools studied in this thesis (FV, clinical diagnosis and TB staining), TB staining was the only one showed an approaching significant difference between the proportion of TB positive lesions in patients with a history of oral cancer (34%>) as compared to lesions from patients without a history of oral cancer (22%, P = 0.083, Table 13). However, no difference was noted in the proportion of TB positive patients or TB negative patients between patients with and without a history of oral SCC (29% vs. 25%o, P = 0.574). 7.2. FV positive lesions have increased cancer risk: Clinical prospective The most important 3 clinical risk characteristics for OPLs are size, site and appearance of the lesions. Literature has shown that these clinical parameters are associated with histological findings of dysplasia, the gold standard forjudging malignant risk. Furthermore, there are also indications that these parameters may be cancer risk predictors independent of dysplasia. For example, a study from this research team has shown low-grade dysplasia from high-risk sites contain increased high-risk molecular pattern than those morphologically similar lesions located at the low-risk sites (Zhang and Rosin, 2001). The study results showed that FV facilitated the identification of high-risk lesions as judged by these clinical parameters (size, site and clinical appearance of the OPLs, Table 8). While there was no difference in the proportion of lesions in the high-risk regions (53% of FV positive vs. 47%o negative, P = 0.478), the FV positive lesions were larger (297 mm vs. 192 mm in negative lesions, P = 0.018) and more likely to be nonhomogeneous (72% in FV positive vs. 55% in negative, P = 0.017). In addition, the difference in the proportion of lesions with rough surfaces was also approaching significant: 69% in FV positive vs. 55% in FV negative (P = 0.053). A significantly higher number of FV positive lesions were judged to be OPLs according to the clinical criteria (78%) as compared to the FV negative lesions (61%>, P < 0.0001), and a significantly lower number of FV positive lesions were regarded as normal/equivocal (12%) as 67 compared to the FV negative lesions (49%, P < 0.0001). Similar results were obtained when the association between FV and clinical characteristics was examined in patients with and without a history of oral SCC separately (Tables 10 & 11), in both cases, a significantly higher number of FV positive lesions were judged to be OPLs according to the clinical criteria as compared to the FV negative lesions and a significantly lower number of FV positive lesions were regarded as normal/equivocal as compared to the FV negative lesions. These findings are important as they support the hypothesis that FV identifies high-risk oral lesions. Interestingly, while an increased proportion of FV positive lesions in the cancer patients were judged to be OPLs, there were no differences in the general criteria used to make the diagnosis of OPLs including site, size and non-homogeneous appearance (all P > 0.05). The reasons for this remain unclear although it is speculated that TB staining probably had influenced the clinicians' decision in the diagnosis of OPLs (see discussion below). In comparison, TB staining showed similar results as FV and also identified more lesions clinically diagnosed as OPLs based on the clinical risk characteristics. While there was no difference in the proportion of lesions in the high-risk regions (60%> of TB positive vs. 47%> negative, P = 0.113), the TB positive lesions showed marginal increase in the size (327 mm 2 vs. 225 mm 2 in negative lesions, P = 0.054) and more likely to be nonhomogeneous (85%> in TB positive vs. 57% in negative, P = 0.0001). In addition, the difference in the proportion of lesions with rough surfaces was also significant: 85%> in TB positive vs. 54%> in TB negative (P < 0.0001). Compared to TB negative lesions, TB positive lesions had a significantly lower proportion of lesions diagnosed as normal/equivocal (4%> vs. 36%>, P < 0.0001, Table 14) and higher proportion of lesions diagnosed as OPLs (92% vs. 56%, P < 0.0001). Similar results were obtained when the association between TB and clinical characteristics was examined in patients with and without a history of oral SCC separately (Tables 15 & 16), in both cases, a significantly higher number of TB positive lesions were judged to be OPLs according to the clinical criteria as compared to the TB negative lesions and a significantly lower number of TB positive lesions were regarded as normal/equivocal as compared to the TB negative lesions. Similar to FV, while an increased proportion of TB positive lesions in the cancer patients were judged to be OPLs, there were no differences in the general criteria used to make the diagnosis of OPLs including site, size and non-homogeneous appearance (all P > 0.05). 7.3. FV positive lesions have increased cancer risk: TB staining TB has been used to identify both oral SCC and dysplasia, although the sensitivity of the latter varies in different studies. Recent studies using molecular techniques have suggested that TB staining could be a cancer risk-predicting parameter independent of dysplasia since TB positive lesions were shown to contain increased high-risk molecular pattern as compared to those TB negative lesions of similar degree of dysplasia (Epstein et al., 2003; Zhang et al., submitted). The study results here showed that a significantly higher number of FV positive lesions were TB positive as compared to the FV negative lesions (41% vs. 12%, P < 0.0001, Table 12). The results remained the same when the association between FV and TB staining was examined in patients with and without a history of oral SCC separately. In both cases, FV positive lesions 68 contained a markedly higher proportion of TB positive lesions. In patients without a history of cancer, 37% of FV positive lesions were TB positive as opposed to 9%> in F V negative lesions (P = 0.0009). In patients with a history of oral cancer, 45% of FV positive lesions were TB positive as opposed to 17% FV negative lesions (P - 0.0067). These findings again support the hypothesis that FV identifies high-risk oral lesions. 7.4. F V positive lesions have increased cancer risk: Histo-pathological prospective Histopathology is the gold standard in judging the malignant risk of OPLs and this gold standard is time tested and has been used for decades in different organs and tissues such as upper aerodigestive tracts, lung, breast, uterine cervix and skin. It is therefore very important that any new risk markers, such as visual tools, be examined for its ability in identifying dysplastic lesions. In the following sections, the ability of FV as a tool in identifying dysplastic lesions is discussed. As a comparison, the clinical diagnosis and TB were also compared for their ability to identify dysplastic lesions. 7.4.1. FV and dysplasia The study results showed that FV positive lesions contained a markedly higher percentage of dysplastic lesions than the FV negative lesions (70% vs. 30%, P < 0.0001, Table 18). The sensitivity of FV to identify dysplastic lesions was 70% and the specificity 69%>. Sensitivity of a marker represents the true positive rate, i.e., FV positive dysplasia cases/total dysplasia cases, and a 70% sensitivity means that FV could diagnose 70% of all dysplasia with a 30% false negative rate. The specificity of a marker represents the true negative rate, i.e., FV negative hyperplasia cases/total hyperplasia cases, and a 69% specificity rate means that FV could identify 69% of non-dysplastic (hyperplastic) lesions specifically with a false positive rate of 31%. When the association of FV with dysplasia was examined in patients with and without a history of oral cancer separately, both cases showed that FV positive lesions had increased proportion of dysplasia as compared to the FV negative lesions (both P < 0.05, Table 18). As compared to those without a history of oral SCC, FV in patients with a history of oral SCC showed a higher sensitivity (76% vs. 65%> in patients without a history of cancer) but lower specificity (59%> vs. 77%> in patients without a history of cancer), indicating that this tool could detect more dysplastic lesions in patients with a history of cancer even though the false positive rate increased. Importantly, this thesis showed that FV examination facilitated the identification of high-grade dysplastic OPLs, which are known to have a very high-risk for malignant transformation (see Chapter one). When different histological groups were examined, with increasing degree of dysplasia, there was an increasing proportion of FV positive dysplasia: 31% hyperplasia, 59% of low-grade dysplasia and 91% high-grade dysplasia (Table 19). In fact, in patients with a history of oral cancer, 94% of all high-grade dysplasia were identified by FV. Such a high sensitivity of FV for high-grade dysplasia could guide the diagnosis and management of this particular group of lesions with very high cancer risk. 69 Of course, a marker with increased sensitivity is invariably accompanied by decreased specificity. While there was a significant difference between low grade dysplasia and hyperplasia (P = 0.016), and between high and low grade dysplasia (P = 0.056, approaching significant) in patients without a history of cancer, the patients with a history of cancer showed no difference between FV positive rate between low grade dysplasia and hyperplasia as a result of decreased specificity and increased false positive rate in the hyperplasia). However, importantly, the difference between high and low grade dysplasia still approaching significant (P = 0.053). In summary, our results showed a strong association between FV positivity and presence and degree of dysplasia, supporting the hypothesis that FV could be a useful tool to facilitate the identification of high-risk OPLs. 7.4.2. Clinical diagnosis and dysplasia As a comparison to FV, the ability of clinical diagnosis of OPLs based on clinical features of oral lesions to identify dysplasia was examined (Table 20). The sensitivity of clinical diagnosis in identifying dysplasia was slightly higher than FV: 85% for all lesions, 85%> for lesions from patients without a history of cancer, and 84%> for lesions from patients with a history of cancer. However, clinical diagnosis had a much lower specificity as compared to FV: 33%> for all lesions (compared to 69%> with FV), 45% for lesions from patients without a history of cancer (77% for FV); and as low as 18%> for lesions from patients with a history of cancer ( 59%> for FV). Such low specificity (18%>) or high false positive rate (82%>) of clinical diagnosis of lesions in patients with a history of cancer demonstrated the difficulty of clinical differentiation of reactive lesions in these patients from true premalignant lesions. The higher specificity rate of FV could be used to help clinicians to lower the false positive rate and avoid unnecessary biopsy of these patients when the previous cancer treatment already made the oral cavity fragile and hard to heal. A further problem for clinical diagnosis in identifying dysplastic lesions is that the sensitivity of clinical diagnosis of dysplasia actually resulted from its sensitivity in identifying low-grade dysplasia: 67%> of histologically diagnosed nondysplastic lesions were clinically diagnosed as OPLs as opposed to 85%> of low-grade dysplasia that were called OPLs clinically (P = 0.046). However, there was no difference between the proportion of low-grade dysplasia that were called OPLs clinically (85%) and the high-grade dysplasia (84%, P = 1, Table 20). When the lesions were divided into those from patients with and without a history of oral cancer, a similar result was observed for lesions from patients without a history of oral cancer: a difference was noted between the hyperplasia group and low-grade dysplasia (P = 0.016), but not between the low and high-grade dysplasia (P = 1). The problem of clinical diagnosis of OPLs was most pronounced in cancer patients since there was a similar percentage of all histological groups were clinically called OPLs: 82%> of nondysplastic lesions, 86%> low-grade dysplasia and 81%> of high grade dysplasia were called OPLs clinically. This again indicated that clinical criteria used for diagnosis of OPLs such as size and nonhomogeneous appearance might not apply to patients with a history of cancer and visual aid such as FV is needed. 70 7.4.3. TB staining and dysplasia As another comparison to FV, the ability of TB staining to identify dysplasia was examined (Table 22). Overall, the sensitivity of TB staining in identifying dysplasia was much lower than that of FV: 41% for all lesions (70%> for FV), 33% for lesions from patients without a history of cancer (65%> for FV); and 53%> for lesions from patients with a history of cancer (76%> for FV). This suggested that the FV as a visual tool would miss less dysplastic lesions than TB staining. However, TB staining had a higher specificity as compared to FV: 87%> for all lesions (69%> for FV), 91%) for lesions from patients without a history of oral cancer (77% for FV); and 82%> for lesions from patients with a history of cancer (59%> for FV). In another word, TB staining has lower false positive rate: 13%> vs. 31% for FV for all lesions; 9%> vs. 27% for FV lesions from patients without a history of oral cancer; and 18%> vs. 41%> for F V for lesions from patients with a history of oral cancer. When the individual histological group was looked at (Table 23), the sensitivity of TB actually came from high-grade dysplasia (71%> were TB positive) whereas only a low percentage of hyperplasia (13%>) and low-grade dysplasia (26%>) were TB positive (no difference was observed between hyperplasia and low-grade dysplasia, P = 0.135). As a comparison, FV was even more sensitive in identifying high-grade dysplasia (91% FV positive), and there was also a significant difference between low-grade dysplasia and hyperplasia in FV positive except the cancer patients. In summary, the results of this thesis showed that FV had higher sensitivity whereas TB had higher specificity and two visual aids could compliment each other clinically. 7.4.4. Use of FV to locate biopsy sites The ability of clinicians to identify the worst area of a lesion to biopsy is critical for the diagnosis and management of OPLs. As discussed above, FV positive lesions are more likely to be dysplastic, and this would support to position the site of biopsy to the FV positive area i f only part of the OPL is FV positive. A stronger support of this however would come from paired biopsies of the same lesions. To test the hypothesis that FV positive area would show at least the same or more severity of histology as the FV negative area in the same lesion of similar clinical appearance, paired biopsies were taken from a small number of patients with oral lesions that were partly FV positive and partly FV negative (Table 26). Of the 10 pairs of matched biopsies (both FV positive and negative biopsies were from similar looking clinical lesions), 6 FV positive biopsies showed the same severity of histology as the FV negative biopsies of the same oral lesions; whereas the remaining 4 pairs demonstrated increased histological severity of the lesions in FV positive biopsies as compared to the FV negative biopsies, e.g., high-grade dysplasia vs. low-grade dysplasia. There was an approaching statistical significance (P = 0.079) even with such a small number of paired biopsies. These findings support the notion that direct FV technique facilitates the choice of biopsy location with the highest risk. 71 7.4.5. Use of FV to screen subclinical lesions OPLs mostly present as leukoplakia and occasionally as erythroplakia clinically. Conventionally, these white or red lesions are detected by direct examination with naked eyes judging by their site, size, and clinical appearance. However, not all OPLs are clinically visible since dysplasia per se may not be clinically visible and it is the accompanying thickening of the epithelium and/or keratin that makes a dysplasia clinically white (leukoplakia) or atrophy of the epithelium that makes a dysplasia clinically red (erythroplakia). Such problem is demonstrated by the fact that clinically apparently normal looking biopsy margins from an OPL or cancer could be dysplastic histologically, and by the fact that second oral malignant tumours could appear right under our watchful eyes and their premalignant stages of lesions totally escape clinical detection even though patients with a history of oral cancer are carefully followed. A visual aid that could help clinician to detect OPLs that are not clinically visible could therefore not only help to determine the subclinical extension of a clinically visible lesions (help planning the treatment extension of the lesions), but also help to identify new lesions that are totally missed by naked eyes. A visual aid that could identify these changes could be useful as a screening tool to screen OPLs and early SCC in the high-risk populations, such as patients with a history of oral cancer or dysplasia or elderly heavy smokers. As discussed before, FV has the distinct advantage over TB as a visual tool since FV is not a chemical (which could have chemical associated side effects), not a dye (which has low acceptance of patients), and could be used for the whole oral cavity without reliance of clinical identification of a lesion (topical application of TB relies on clinical identification of lesions; whereas oral rinse of TB is used much less frequently because of a messier stain of the whole oral cavity and increased irritation). Such advantages suggest that FV could be used in the future for screening of the whole oral cavity for high-risk areas, particularly for the high-risk populations. This study was designed to develop and test the technique of direct FV and the FV examination was centred on the lesion or lesional sites of previous oral cancer or dysplasia. It is therefore not possible to comment on the ability of FV to screen the whole oral cavity or apparently lesion free mucosa areas. However, some of the previous lesion sites that showed no obvious clinical lesions (normal/equivocal) had demonstrated FV positivity (16 of the 138 FV positive sites). Of these 16 clinically normal/equivocal sites, 4 were biopsied, and 3 of the 4 showed high-grade dysplasia and 1 normal (see Figure 1). Such results suggest that FV could identify high-risk OPLs that are not clinically apparent or visible. 7.4.6. Use of multiple diagnostic tools in identifying and monitoring dysplastic lesions Currently the diagnosis of high-risk OPLs clinically is solely dependent upon clinical diagnosis based on the clinical risk characteristics of OPLs for the majority of clinicians. As discussed above, clinical diagnosis had a high false positive rate and it still missed a small proportion of high-grade dysplasias (16%) despite of its high sensitivity (84%). A further problem of the clinical diagnosis as shown in the results of this thesis was that it was not reliable in cancer patients. Addition of visual aid should help the diagnosis. 72 7.4.6.1. Lesions diagnosed clinically as OPLs but negative for FV or TB: low cancer risk The traditional clinical diagnosis without positive visual aid would generally identify lesions without dysplasia. The sensitivity of clinical diagnosis alone without positive visual aid was low: 23% for clinical + but FV -, and 37% for clinical + but TB positive, i.e., the majority of dysplasia did not belong to this group. Furthermore, an even lower sensitivity was present for high-grade dysplasias: 10%> for clinical + but F V -, and 25% for clinical + but TB -. * Such combination of tools could be used to identify the low-risk lesions. 7.4.6.2. Lesions diagnosed as OPLs clinically with an additional visual aid: Increased sensitivity in identifying dysplasia The study results had shown that clinical diagnosis in conjunction with either FV or TB increased the sensitivity of detecting dysplasia. Clinical diagnosis with the help of FV (either clinical &/or FV +) had the highest sensitivity and detected 100% (31/31) high-grade dysplasia, as compared to clinical diagnosis with TB (either clinical &/or TB +) detecting 90% high-grade dysplasia. As expected the increased sensitivity was accompanied by decreased specificity or increased false positive rate Interestingly when two visual aids (FV &/or TB +) were used without the clinical diagnosis, the sensitivity was only slightly lower than clinical diagnosis in conjunction with FV or TB, but the sensitivity of identifying high-grade dysplasia was very high (97%), and there was a much lower false positive rate (36%>) compared to clinical &/or FV (74%>) and clinical &/or TB (67%>). These results suggest that use of two visual aids together may be a good way in identifying high-risk OPLs. 7.4.6.3. Lesions positive for two diagnostic tools: high cancer risk. Using two diagnostic tools markedly increased the specificity of the diagnosis or lowered the false positive rate by looking at two markers both positive. When two diagnostic tools were positive, that is both clinical & FV + or both clinical & TB +, or both FV & TB +, the false positive rate was lowered (8%> to 23%>). In fact, when two markers are positive, a high-proportion of these were high-grade dysplasia: 35%> for both clinical & FV +, 49%o for clinical & FV +, and 56%o for FV & TB + (see Section 6.5.2, 'two diagnostic tool positive'). One of the major problems in monitoring high-risk OPLs is to detect change in the degree of dysplasia. For example, clinicians are reluctant to biopsy previous oral cancer sites repeatedly because the sites are generally fragile and hard to heal as a result of aggressive treatment. The use of clinical diagnostic tool in conjunction with visual tools could triage those lesions that are progressing as compared to those lesions remain low-risk or regress. For example, i f a lesion was clinical positive but negative for the visual aid, it might have a high chance of being low-risk; the appearance of positivity with two diagnostic tools, however, could indicate an increased cancer risk or progression to a high-grade of dysplasia. 73 7.5. FV predicts histological features of biopsy specimens Correlation of FV status with the tissue architectural and pathological changes would improve our understanding of the mechanism of FV positivity. Few studies were done in this field (see chapter one). In general, in addition to dysplasia, some literature has indicated that increase in the thickness of epithelium &/or keratin, destruction of the mature collagen fiber and increased blood flow in the immediate subepithelial region would result in reduction of normal fluorescence (FV +). Conflicting with the literature, the FV status was not associated with the surface topography of the suspected lesions. It was explained in the Chapter one that the surface irregularity of the lesion could results in light scattering and alteration of fluorescent signal available for clinical interpretation. It is not clear why the rough surface of mucosa was not associated with reduction in the fluorescence or increase in FV positivity. It is possible with more sensitive instrument, such potential reduction could be detected but this was not the case using direct FV. Also conflicting with the literature, the FV positive status was not associated with increased keratin. In fact, a lower percentage of FV positive lesions were hyperkeratotic compared to the FV negative lesions (41% vs. 65%). Instead of keratin blocking the fluorescence, the author observed using a fluorescent microscope (data not shown) that epithelial keratin did emit a relatively strong auto-fluorescent signal. This phenomenon would be clinically interpreted as FV negative (present of auto-fluorescent). This also consisted with the FV negative findings in clinical diagnosis of linea albia and verrucous-type lesion (by clinical observation only). However, the strong fluorescent signal from keratin would override the fluorescent signal underneath resulting potential false FV negative interpretation. Consistent with the literature (see Chapter one), the thesis showed that the thickening of epithelium resulting in a reduction of fluorescence or increased FV positivity. General hypothesis for this is that the main source of normal green fluorescence comes from the underlying connective tissue. Thickening of the epithelium resulted in blocking, scattering and absorption of the fluorescence emitted by the connective tissue (FV +). The most striking difference was noted in the presence of pronounced inflammation: FV positive lesions showed a markedly increased proportion of lesions with pronounced inflammation as compared to the negative lesions (76%> vs. 30%, P < 0.0001). It is possible that the intense inflammation resulted in destruction of the mature collagen fibers (lower fluorescence emission), or the presence of band like inflammatory cells (a physical barrier to deflect or scatter stromal fluorescent light become visible) and increase in local vascularity (increased blood cells increases absorption of fluorescence). This finding explains the fact that many inflammatory reactive lesions were FV positive. The clinical expertise in differentiating inflammatory lesions and in repeat observation in two weeks (allow inflammation to die down) should play an important role in differentiating these false positive lesions (reactive inflammatory lesions). For example, lichen planus showed a high rate of FV positivity. Clinically lichen planus in general looks different from leukoplakia. 74 7.6. Mechanism for selective staining by FV of OPLs Why is FV preferentially found in OPLs? The definitive mechanisms underlying fluorescent visualization process are unknown. Fluorescent visualization is meant to detect superficial lesions by exploiting its limited penetration power, on the order of several hundred microns of emitted fluorescence light as a measure of changes. Decrease of green fluorescence of pre-cancerous and cancerous tissue due to thickening of epithelium or overlying tumour tissue and a reduction in concentration of flavins (fluorescent emission at 520 nm) was purposed. Indeed, autofluorescence may not reflect differentiation status but, rather, may reflect some other tumour property associated with tumour differentiation, such as the metabolic activity of the tumour mass, the number and density of neoplastic cells, cellular proliferation or the tendency of cells within the tumour mass to undergo terminal differentiation and apoptosis (Schantz et al. 1998). In addition to the epithelial components of cancerous development, supporting stroma also plays an important role in tumour invasion. Matrix metalloproteinase (MMP) 2 and 9, also known as gelatinases, from oral mucosal epithelial cell may contribute to the pathogenesis of oral squamous cell carcinoma transformation. Tsai et al. (2003) showed that both M M P types were highly expressed in human oral SCC compared with normal keratinocytes. The degradation of basement membrane proteins laminin and type IV collagen were attributed to the high M M P activities during early-stage metastasis. Pavlova et al. (2003) also pointed out that stromal collagen fluorescence decreased by a factor of 5 in precancerous cervical tissue. When high-grade dysplastic cells committed to invade, M M P from committed cells would degrade epithelial base membrane and then supporting stroma before histological invasion became evident. Once the integrity of the base membrane and the supporting stroma was breached, the autofluorescent signal would reduce and eventually be lost. This could explain the fact that almost all high-grade dysplastic sites were FV positive. V 75 7.7. Future investigation 7.7.1. FV and outcome To avoid over-interpretation of the thesis data, it is critical to determine i f FV positivity would be an indicator for disease progression in the next study. Although several risk predictors (clinical features, TB staining, and dysplasia) have supported the hypothesis that FV may be a useful tool in identifying high-risk OPLs, the ultimate proof needs to come from the progression data. Follow up patients and collect of outcome data would be important in the determination of the role of FV as a diagnostic tool. 7.7.2. FV and screening of the oral cavity This study was centered on oral lesions and site of previous oral lesions. Studies are needed to see whether FV could be used as a noninvasive tool to screen the whole oral cavity to identify lesions from apparently clinically normal looking areas. This would be particularly important for potential usage of this technique for screening the high-risk population, such as screening individuals with a history of heavy tobacco and alcohol use. 7.7.3. Limitations of current FV findings This study has its own limitations in methodology. The selected populations in this study were all referred from the community and hospitals for management of oral dysplasia. The prevalence of having histological oral dysplasia is 70%, which does not reflect the prevalence in the community. As a result, the subsequent positive and negative predictive values from this data need to be interpreted with caution. It is anticipated that a much lower positive predictive value and a much higher negative predictive value would result i f this technique is being used in the general community. In addition, this study was not a randomized single- or double-blinded study. Both the clinicians and pathologists knew all the pertinent information of the lesions before clinical decisions or pathological diagnoses were made. Although the decisions of tissue biopsy were obviously often made in the FV-positive group, the bias could possibly be related to either the clinical appearance of the lesions other than the FV status, or the FV status alone. However, this bias can be controlled by performing the FV procedure independent to the attending clinician who made the decision (including biopsy) on the management of the patient. Finally, the FV status of the lesion was limited to either positive or negative. However, there were cases that the FV status was neither positive nor negative, requiring extensive consensus. This confusion could be clarified by introducing FV-equivocal status in the future study. The FV-equivocal status could represent a level of cancer risk different from FV positive does. 7.9. Summary and Conclusion In summary, a total of 143 patients with 319 lesions were evaluated using direct fluorescent visualization technique. Concurrent biopsy was performed on 142 lesions . The data showed that 76 FV-positive lesions identified high-risk oral lesions as judged by clinical risk features of oral premalignant lesions (size, site, and clinical appearance of OPLs), by TB staining, and by presence and degree of dysplasia, the gold standard in current assessment of cancer risk of OPLs. Not only did FV, when used alone, show a high sensitivity in identifying dysplasia (70%) and high-grade dysplasia (91%), but its usage together with clinical diagnosis and TB staining also markedly increased the sensitivity of identifying high-grade dysplasia (100%) and the specificity of the diagnosis. When used together with TB, the two markers complement each other with the high sensitivity of FV and the high specificity of TB. This data also showed that FV could identify high-risk lesions that were otherwise clinically not apparent. In conclusion, after a detailed evaluation and statistical analysis of the available data, the null hypothesis of no difference in cancer transformation risk of OPLs regardless of FV status is rejected. The alternative hypothesis that FV-positive OPLs show a higher cancer risk compared with FV-negative OPLs, judging by both clinical and pathological risk factors, is accepted. 77 APPENDIX 1: ORAL STUDY QUESTIONNAIRE 1. In addition to being Canadian or a landed immigrant, what is your ethnic or cultural heritage? (Check one box only): • White • East or South-east Asian (eg. China, Japan, Indonesia, Philippines, Vietnam) • South Asian (eg. India Pakistan, Sri Lanka) • First Nations • Black • Other (Please Specify) 2. a) What is the highest grade (or year) of high school or elementary school that you have completed? Grade Never attended school b) How many years of post-secondary school have you completed (college, university)? Years None 3. a) Have you ever used chewing tobacco? Yes • No • b) Have you ever used betel nut? Yes • No • 4. Have you ever regularly smoked cigarettes, cigars or pipes more than once per week for one year or longer? Yes • No • If Yes, please specify: a) At what age did you begin smoking: Cigarettes? Cigars? Pipes? b) Do you currently smoke: Cigarettes? Yes • No • Cigars? Yes • No • Pipes? Yes • No • 78 c) If you have quit smoking, at what age did you permanently stop: Cigarettes? Cigars? Pipes? d) Looking back over your entire life, on average, how many did you usually smoke per day? Before Age In your In your In your In your 60's & 20 years 20's 30's 40's 50's older Cigarettes Cigars Pipes • 5. Looking back over the last year, please think about your exposure to the smoke of others, either at home, at work, and in public places (such as restaurants, recreational facilities). Are you regularly exposed to smoke of others: At home? Yes • No • At work? Yes • No • In public places? Yes • No • If Yes, to any of the above, please specify: How often are you regularly exposed to smoke of others: Never Less than once More than once a At least Daily a month month but less than once a once a week week At home? • • " • • • 79 At work? • • • • ' • In Public • • • • • Places? 6. Looking back over your entire life, please check the age periods in which you were daily exposed to the smoke of others. Before Age In your In your In your In your 60's & 20 years 20's 30's 40's 50's older t • • • • ' • • 80 7. Have you ever regularly consumed alcoholic beverages more than once per month for one year or longer? Yes • No • If Yes, please specify: a) At what age did you begin drinking: Beer? Wine? Spirits (liquor)? b) Do you currently drink: Beer? Yes • No • Wine? Yes • No • Spirits (liquor)? Yes • No • c) If you have quit drinking, at what age did you permanently stop: Beer? Wine? Spirits (liquor)? d) On average, how much did you usually drink per week: Beer bottles Wine glasses Spirits (liquor) (shots - 1 oz.) 8. Have any of your immediate family members (parents, brothers/sisters, daughters/sons, grandparents, aunts/uncles related by birth not marriage) had cancer in the head and neck region (excluding skin cancer)? Yes • No • If Yes, please specify all who had head and neck cancer: • Parents • Brothers/sisters • Daughters/sons • Grandparents • Aunts/uncles related by birth not marriage 81 CN oo Q T3 s C3 e =s c cj '3 CL •o •o >, ;>> >-> td < a <u OX I* u OS td 35 > Z td a. Z O 35 td ° <u —. • -a Z o O u _ CD td 3 _j ^ b. T o Z z ° o ° E- s ft, o 5 u J Cd £ ^ CJ CO U OJ cd i-0) c <L> oo a> 2 x w o 2 Z UJ CJ H Z z .2 2 £ 35 b [VI co J -a ."2 00 O-Vi Z o c-u o -1 S Si PS § £ ° o <= Ed ^ • J Z, -a S -a O OJ •a >-!_ ft o C N -• o •a CJ OJ CJ a. ^ -5 © O A .1 td c/5 — O CN td u z <; a: < c CL. OJ J 2 P -a ^ aj a. > a." CO cj > cd cd S - I o 3 «I *CN Ld * ca t— o x 5 r— — 3 o CJ 3 fc CJ > , o CL 00 2 C/J H J td CJ M S ir o O ii H CN 0 0 if o ' f t o o I? § > & r r 1 , 1 |tN ro la- II bo _H © I C/J Q. C O <D u ca J= «& 5 <D OH 1/3 < U o - J Q 5 a —i u o u odo H a z u - J - J O a o Q % - J a u o u z u u W o 3 U U z < at O u. ° l 8 » z ° </5 £ U J > " C£ II Cu — < ° II a s a > O II o -r- — Z C3 CJ O 5 U o O z CJ o C/! - > I H Z U J l - ~ O o ° U ~ 2 u <u 3 j E g U E -U O c S " . 2 a. a-O .2 ID a o a a J U J A P P E N D I X 3: F L U O R E S C E N T VISUALIZATION D E V I C E Schematic diagram and the actual devices (light source and eye goggles) for Fluorescent Visualization Examination A P P E N D I X 4: F V POSITIVE LES ION • Reduced fluorescent green as compared to adjacent tissue; or contra-lateral tissue is classified as FV positive. A P P E N D I X 5: P U B L I S H E D A B S T R A C T r International Association for Dental Research abstract number 1242, March 11, 2004, Hawaii Convention Center Exhibit Hall http://iadr.confex.com/iadr/2004Hawaii/techprogram/abstract_44189.htm 1242 Detection of High-risk Oral Premalignant Lesions by Multi-Spectral Fluorescent Visualization L. ZHANG1, S.P. N G 1 , M . WILLIAMS 2 , D. L A R O N D E 1 , C.F. POH 1 , C. M A C A U L A Y 2 , and M.P. ROSIN 3 , 1 University of British Columbia, Vancouver, Canada, 2 British Columbia Cancer Agency, Vancouver, Canada, 3 British Columbia Cancer Agency, Simon Fraser University, Vancouver, Canada Objectives: The prediction of cancer risk for oral premalignant lesions (OPLs) is currently based upon histological diagnosis and, more recently, molecular analysis. These approaches rely on the clinician's ability to identify lesions requiring biopsy based on clinical features, which is quite often a difficult decision. The objective of this study was to explore the feasibility of using multi-spectral fluorescent visualization (FV) for its ability to identify high-risk OPLs. Methods: This abstract describes early results from an ongoing prospective study of patients with OPLs. FV, demographics, OPL clinical features, histology and outcome were collected from 69 OPLs. Results: 42 of the 69 OPLs showed a loss of autofluorescence (FV-positive). There was no significant difference in age, gender or smoking habits for FV-positive and -negative cases. Clinical attributes were also similar with no significant difference in site, size, clinical appearance (homogeneous vs. nonhomogeneous) and toluidine staining of OPLs, although F V -positive lesions tended to be larger (352 ± 357 mm 2 vs. 224 ± 253, P = 0.15) and toluidine positive (55% vs. 33%>, P = 0.0913). Histologically, epithelial &/or keratin thickness was similar for the two groups. FV-positive lesions were more likely to be obviously inflamed (71% vs. 37%, P = 0.0063) or ulcerated (31% vs. 4%, P = 0.0059). Strikingly, a significantly higher proportion (81 %>) of FV-positive lesions were dysplastic compared to the FV-negative lesions (41%, P = 0.0009). Increasing degree of dysplasia was accompanied by increasing FV positivity: 33%> nondysplastic lesions, 67%> of mild/moderate dysplasias and 100%> of severe dysplasias were FV-positive (P = 0.0004). When the outcome was studied, all 5 OPLs that progressed into cancer were FV-positive (12%> vs. 0%>, P = 0.1487). Conclusion: FV could facilitate the identification of high-risk OPLs (Supported by grant R01DE13124, NIDCR). 85 R E F E R E N C E S Abe, S., K.Izuishi, H.Tajiri, T. Kinoshita, and T. Matsuoka. 2000. Correlation of in vitro autofluorescence ndoscopy images with histopathologic findings in stomach cancer. Endoscopy 32:281-86. Amagasa T., E. Yokoo, K. Sato, N . Tanaka, S. Shioda, M . Takagi. 1985. A study of the clinical characteristics and treatment of oral carcinoma in situ. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 60:50-55. Andersson-Engels, S., G. Canti, R. Cubeddu, et al. 2000. Preliminary evaluation of two fluorescence imaging methods for the detection and the delineation of basal cell carcinomas of the skin. Lasers Surg Med 26:76-82. Axell, T., J. J. Pindborg, C. J. Smith, and I. van der Waal. 1996. Oral white lesions with special reference to precancerous and tobacco-related lesions: Conclusions of an international symposium held in Uppsala, Sweden, May 18-21 1994. International Collaborative Group on Oral White Lesions. J Oral Pathol Med 25:49-54. Balasubramanian S., V . Elangovan, S. Govindasamy. Fluorescence spectroscopic identification of 7,12-dimethylbenz[a]anthracene-induced hamster buccal pouch carcinogenesis. Carcinogenesis 16:2461-5. Banoczy, J., and A. Csiba. 1976. Occurrence of epithelial dysplasia in oral leukoplakia. Oral Surg Oral Med Oral Pathol 42:766-74. Beauvoit, B., T. Kirai, B. Chance. 1994. Contribution of the mitochondrial compartment to the optical properties of the rat liver: A theoretical and practical approach. Biophys J 67:2501-10. Benavides, J. M . , S. Chang, S. Y . Park, et al. 2003. Multispectral digital colposcopy for in vivo detection of cervical cancer. Optics Express 11:1223-1233.Berkovitz, B. K. , G. R. Holland, and B. J. Moxham. 2002. Oral Mucosa. In Oral Anatomy, Histology and Embryology. 3rd ed. Mosby International Ltd,London, 220-229. Betz, C. S., M . Mehlmann, K. Rick, et al. 1999. Autofluorescence imaging and spectroscopy of normal and malignant mucosa in patients with head and neck cancer. Lasers Surg Med 25:323-34. Bouquot, J., L. Weiland, D. Ballard, and L. Kurland. 1988. Leukoplakia of the mouth and pharynx in Rochester, M N , 1935-1984; incidence, clinical features and follow-up of 463 patients from a relatively unbiased patient pool. J Oral Pathol 17:436. Bouquot, J. E., and S. B. Whitaker. 1994. Oral leukoplakia—rationale for diagnosis and prognosis of its clinical subtypes or "phases". Oral Med 25:133-140. Bouquot, J. E. 1994. Malignant transformation in precancers of head and neck. Available from: [http://www.maxillofacialcenter.com/Table21 .html]. 86 Bruice, P. Y . 1995. Spectroscopy and the electromagnetic spectrum. Organic Chemistry. Prentice Hall, New Jersey, 534-537. Burkitt, H. G., B. Young, and J. W. Heath. 1993. Oral tissue. In Wheather's Functional Histology. 3rd ed. Churchill Livingstone. Cancer Facts & Figures 2003, American Cancer Society, 2003. Available from: [http://www.cancer.org/downloads/STT/CAFF2003PWSecured.pdf] Casiglia, J., and S. B. Woo. 2001. A comprehensive review of oral cancer. Gen Dent 49: 72-82. Chance, B., H. Liu, T. Kitai, and Y. Zhang. 1995. Effects of solutes on optical properties of biological materials: Models, cells, and tissues. Anal Biochem 227:351-62. Chance, B. 1989. Metabolic heterogeneities in rapidly metabolizing tissues. J. Appl. Cardiol 4:207-221. Chang, S. K. , M . Follen, A. Malpica, et al. 2002. Optimal excitation wavelengths for discrimination of cervical neoplasia. IEEE Trans Biomed Eng 49:1102-1110. Chen, H . M . , C. Y . Wang, C. T. Chen, et al. 2003. Auto-fluorescence spectra of oral submucous fibrosis. J Oral Pathol Med 32:337-43. Chung, Y . G., J. A . Schwartz, C. M . Gardner, et al. 1997. Diagnostic potential of laser-induced autofluorescence emission in brain tissue. J Korean Med Sci 12:135^42. Delank W., B. Khanavkar, J. A . Nakhosteen, and W. Stoll. 2000. A pilot study of autofluorescent endoscopy for the in vivo detection of laryngeal cancer. Laryngoscope 110:368-73. De Veld, D. C , M . Skurichina, M . J. Witjes, et al. 2003. Autofluorescence characteristics of healthy oral mucosa at different anatomical sites. Lasers Surg Med 32:367-76. Electromagnetic Spectrum: The Joy of Visual Perception: A Web Book . Available from: http://www.yorku.ca/eye/ Epstein, J. B., C. Oakley, A. Millner, S. Emerton, E. van der Meij, and N . Le. 1997. The utility of toluidine blue application as a diagnostic aid in patients previously treated for upper oropharyngeal carcinoma. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 83:537-47. Epstein, J. B., and C. Scully. 1997. Assessing the patient at risk for oral squamous cell carcinoma Spec Care Dentist 17:120-8. Epstein, J. B., L. Zhang, C. Poh, et al. 2003. Increased allelic loss in toluidine blue-positive oral premalignant lesions. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 95:45-50. Feigelson, H. S., R. K. Ross, M . C. Yu, G. A . Coetzee, and B. E. Henderson. 1996. Genetic susceptibility to cancer from exogenous and endogenous exposures. J Cell Biochem. Suppl 25:15-22. 87 Fittkow, C. T., S. Q. Shi, E. Bytautiene, G. Olson, G. R. Saade, R.E. Garfield. 2001. Changes in light-induced fluorescence of cervical collagen in guinea pigs during gestation and after sodium nitroprusside treatment. JPerinat Med 29:535-43. Flavin adenine dinucleotide structure: available in www.gwu.edu/~mpb/c/fad.gif Georgakoudi, I., B. Jacobson, M . G. Muller, et al. 2002. NAD(P)H and collagen as in vivo quantitative fluorescent biomarkers of epithelial precancerous changes. Cancer Res 62:682-87. Gillenwater, A. , R. Jacob, R. Ganeshappa, et al. 1998. Noninvasive diagnosis of oral neoplasia based on fluorescence spectroscopy and native tissue autofluorescence. Arch Otolaryngol HeadNeckSurg 124:1251-58. Goaz, P. W., and S. C. White. 1994. Radiation Physics. Oral Radiology. 3rd ed. St. Louis: Mosby international Ltd, London, 4-6. Grant, W. E., C. Hopper, A . J. MacRobert, P. M . Speight, and S. G. Bown. 1993. Photodynamic therapy of oral cancer: Photosensitisation with systemic aminoclaevulinic acid. Lancet 342:147-8. Guo, Z., K. Yamaguchi, M . Sanchez-Cespedes, W. H. Westra, W. M . Koch. D. Sidransky. 2001. Allelic losses in OraTest-directed biopsies of patients with prior upper aerodigestive tract malignancy. Clin Cancer Res 7:1963-8. Gurjar, R. S., V . Backman, L. T. Perelman, et al. 2001. Imaging human epithelial properties with prolonged light scattering spectroscopy. Nat Med 7:1245-8. Haemoglobin structure: available from: www.chem.uwa.edu.au/enrolled_students/ BIC sectl/sectl.2.htm Harris, D. M . , and J. Werkhaven. 1987. Endogenous porphyrin fluorescence in tumors. Lasers Surg Med 7 A67-72. Hsu, T. C , L. R. Shirley, and H. Takanari. 1983. Cytogenetic assays for mitotic poisons: The diploid Chinese hamster cell system. Anticancer Res 3:155-9. Inaguma, M . , and K. Hashimoto. 1999. Porphyrin-like Fluorescence in Oral Cancer. In vivo fluorescence spectral characterization of lesions by use of a near-ultraviolet excited autofluorescence diagnosis system and separation of fluorescent extracts by capillary electrophoresis. Cancer 86:2201-2221. Ingrams, D. R., J. K. Dhingra, K. Roy, et al. 1997. Autofluorescence Characteristics of Oral Mucosa. Head Neck 19:27-32. Ishwad, C , R. Ferrell, K . Rossie, et al. 1996. Loss of heterozygosity of the short arm of chromosomes 3 and 9 in oral cancer. Int J Cancer 69:1-4. Jacques S etal. 2001. Optical measures for quality control during photodynamic therapy. Plenary Talk at Intl. Photodynamic Association meeting, June 7, 2001, Vancouver BC, Canada. 88 Johnson, N . 1998. Diagnosing oral cancer: Can toluidine blue mouthwash help? FDI World 2:22-26. Khuri, F. R., S. M . Lippman, M . R. Spitz, R. Lotan, and W. K . Hong. 1997. Molecular epidemiology and retinoid chemoprevention of head and neck cancer. JNatl Cancer Inst 89:199-211. Khuri, F. R., E. Kim, J. Lee, et al. 2001. The impact of smoking status, disease stage, and index tumor site on second primary tumor incidence and tumor recurrence in the head and neck retinoid chemoprevention trial. Cancer Epidemiol Biomarkers Prev 10:823-9. Kol l i , V . R., A . R. Shaha, H. E. Savage, P. G. Sacks, M . A. Casale, and S. P. Scantz. 1995. Native cellular fluorescence can identify changes in epithelial thickness in-vivo in the upper aerodigestive tract. Am J Surg 170:495-8. Kramer, I. R., R. B. Lucas, J. J. Pindborg, L . H. Sobin. 1978. Definition of leukoplakia and related lesions: an aid to studies on oral precancer. Surg Oral Med Oral Pathol Oral Radiol Endod 46:518-39! Krogh, P., B. Hald, and P. Holmstrup. 1987. Possible mycological etiology of oral mucosal cancer: Catalytic potential of infecting Candida albicans and other yeasts in production of N-nitrosobenzylmethylamine. Carcinogenesis 8:1543-8. Kulapaditharom, B., and V . Boonkitticharoen. 1998. Laser-induced fluorescence imaging in localization of head and neck cancers. Ann Otol Rhinol Laryngol 107:241-6. Lam , S., T. Kennedy, M . Unger, et al. 1998. Localization of bronchial intraepithelial neoplastic lesions by fluorescence bronchoscopy. Chest 113:696-702. Lam, S., C. MacAulay, J. C. Le Riche, and B. Palcic. 2000. Detection and localization of early lung cancer by fluorescence bronchoscopy. Cancer Supplement 89 (11): 2468-73. Lam, S., C. MacAulay, J. C. Le Riche, et al. 2002. A randomized phase lib trial of anethole dithiolethione in smokers with bronchial dysplasia. JNatl Cancer Inst 94:1001-9. Lam, S., C. MacAulay, J. C. Le Riche, A. F. Gazdar. 2003. Key issues in lung cancer chemoprevention trials of new agents. Recent Results Cancer Res 163:182-95. Discussion 264-6. Latimer, P., 1979. Light scattering vs. microscopy for measuring average cell size and shape. BiophysJ 27:117-126. Lee, J. J., W. K. Hong, W. N . Hittleman, et al. 2000. Predicting cancer development in oral leukoplakia: Ten years of transitional research. Clin Cancer Res 6:1702-10. Lenth, R.V. 2004. Java applets for power and sample size. Available from: http://www.stat.uiowa.edu/~rlenth/Power/index.html. Leonardelli, G. B., F. Talamazzi. 1950. Leukoplakias of the oral cavity and precancerous conditions. Arch Ital Otol Rinol Laringol 61:288-301. 89 Lumerman H. , Freedam P., S. Kerpel. 1995. Oral epithelial dysplasia and the development of invasive squamous cell carcinoma. Surg Oral Med Oral Pathol Oral Radiol Endod. 79:321-9. Lynn, C. J., I. S. Saidi, D. G. Oelberg, and S. L. Jacques. 1993. Gestational age correlates with skin reflectance in newborn infants of 24-A2 weeks gestation. Biol Neonate 64:69-75. Marshall, J. R., S. Graham, B. P. Haughey, et al. 1992. Smoking, alcohol, dentition and diet in the epidemiology of oral cancer. Eur J Cancer B Oral Oncol 28B:9-15. Mashberg, A. , and H. Meyers. 1976. Anatomical site and size of 222 early asymptomatic oral squamous cell carcinomas: A continuing prospective study of oral cancer. II. Cancer 37:2149-57. Mitrani, E., and R. Marks. 1982. Procollagen localisation in normal, premalignant and malignant lesions of the epidermis. Arch Dermatol Res 274:21-28. Mullef, M . G., T. A . Valdez, I. Georgakoudi, et al. 2003. Spectroscopic detection and evaluation of morphologic and biochemical changes in early human oral carcinoma. Cancer 97:1681-92. Na, R., I. Stender, and H. C. Wulf. 2001. Can autofluorescence demarcate basal cell carcinoma from normal skin? A comparison with protoporphyrin IX fluorescence. Acta Derm Venereol 81:246-9. National Cancer Institute. Surveillance, Epidemiology, and End Results Program public-use data,1973-1998. Rockville M D : National Cancer Institute, Division of Cancer Control and Population Sciences, Surveillance Research Program, Cancer Statistics Branch. Released April 2001, based on the August 2000 submission. National Cancer Institute of Canada: Canadian Cancer Statistic 2004, Toronto, Canada 2004. Nelson, D. L. , and M . M . Cox. 1993. Principles of Bioenergetics. In Principles of Biochemistry 2nd ed. 393-4. New York: Worth Publishers. . 1993. Bioenergetics and Metabolism. In Nelson and Cox ,711-2 Nicotinamide adenine dinucleotide (NAD + ) structure : available from www.gwu.edu/~mpb/ coenzymes.htm Nordstrom, R. J., L. Burke, J. M . Niloff, and J. F. Myrtle. 2001. Identification of cervical intraepithelial neoplasia (CIN) using UV-excited fluorescence and diffuse-reflectance tissue spectroscopy. Lasers Surg Med 29:118-27. Onizawa, K. , H. Saginoya, Y . Furuya, and H. Yoshida. 1996. Fluorescence photography as a diagnostic method for oral cancer. Cancer Lett 108:61-6. Onizawa, K., N . Okamura, H. Saginoya, H. Yusa, T. Yanagawa, and H. Yoshida. 2002. Analysis of fluorescence in oral squamous cell carcinoma. Oral Oncol 38:343-8. 90 Onizawa, K., N . Okamura, H. Saginoya, H. Yusa, and H. Yoshida. 2003. Characterization of autofluorescence in oral squamous cell carcinoma. Oral Oncol 39:150-6. Oxy-haemoglobin structure. Available from: www.chem.uwa.edu.au/enrolled_students/ BICsectl/sectl.2.htm Palcic, B., S. Lam, J. Hung, and C. MacAulay. 1991. Detection and localization of early lung cancer by imaging techniques. Chest 99:742-3. Panjehpour, M . , C. E. Julius, M . N . Phan, T. Vo-Dinh, and S. Overholt. 2002. Laser-induced fluorescence spectroscopy for in vivo diagnosis of non-melanoma skin cancers. Lasers Surg Med 31:367'-73. Papadimitrakopoulou, V. , J. Izzo, S. M . Lippman, et al. 1997. Frequent inactivation of pl6INK4a in oral premalignant lesions. Oncogene 14:1799-803. Pavlova, I. K. , K. Sokolov, R. Drezek, A. Malpica, M . Follen, and R. Richards-Kortum. 2003. Microanatomical and biochemical origins of normal and precancerous cervical autofluorescence using laser-scanning fluorescence confocal microscopy. Photochem Photobiol 77:550-5. Pershagen, G. 1996. Smokeless tobacco. Br Med Bull 52:50-7. Pindborg, J. J., O. Joist, G. Renstrup, and B. Roed-Petersen. 1968. Studies in oral leukoplakia: A preliminary report on the period prevalence of malignant transformation in leukoplakia based on a follow-up study of 248 patients. J Am Dent Assoc 76:767—71. Poate, T. W. J., J. A . G. Buchanan, T. A . Hodgson, P. M . Speight, A . W. Barrett, D. R. Moles, C. Scully, S. R. Porter. 2004. An audit of the efficacy of the oral brush biopsy technique in a specialist oral Medicine unit. Oral Oncol 40: 829-834. Ramaswamy, G., V . R. Rao, S. V . Kumaraswamy, and N . Anatha. 1996. Serum vitamins' status in oral leukoplakias—a preliminary study. Eur J Cancer B Oral Oncol 32B: 120-2. Rajendran, R., C. K. Sugathan, J. Augustine, D. M . Vasudevan, and T. Vijayakumar. 1989. Ackerman's tumour (Verrucous carcinoma) of the oral cavity: A histopathologic study of 426 cases. Singapore DentJ 14:48-53. Richards-Kortum, R., and E. Sevick-Muraca. 1996. Quantitative optical spectroscopy for tissue diagnosis. Annu Rev Phys Chem 47:555-606. Rick, G. M . 2003. Oral brush biopsy: The problem of false positives. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 96:252. Rippey, J. J., and E. Rippey. 1997. Characteristics of incompletely excised basal cell carcinomas of the skin. MedJAust 166:581-3. Ronches, F., and R. I. Providence. 1954.The fluorescence of cancer under the Wood Light. J Oral Surg (Chic) 7:967-971. 91 Rosati, C. 1994. Prevention of oral cancer. Canadian Task Force on the Periodic Health Examination, 826-36. Ottawa: Health Canada. Rosin, M . P., X . Cheng, C. Poh, et al. 2000. Use of allelic loss to predict malignant risk for low-grade oral epithelial dysplasia. Clin Cancer Res 6:357-62. Rosin, M . P., W. L. Lam, C. Poh, N . D. Le, R. J. L i , T. Zeng, R. Priddy, L.Zhang. 2002. 3pl4 and 9p21 loss is a simple tool for predicting second oral malignancy at previously treated oral cancer sites. Cancer Research 62:6447-50. Roz, L., C. Wu, S. Porter, C. Scully, et al. 1996. Allelic imbalance on chromosome 3p in oral dysplastic lesions: An early event in oral carcinogenesis. Cancer Res 56:1228-31. Schantz, S. P., V . Koll i , H . E. Savage, et al. 1998. In vivo native cellular fluorescence and histological characteristics of head and neck cancer. Clin Cancer Res 4:1177-82. Schepman, K. P., and I. van der Waal. 1995. A proposal for a classification and staging system for oral leukoplakia: A preliminary study. Eur J Cancer B Oral Oncol 31B:396-8. Sciubba, J. J. 1999. Improving detection of precancerous and cancerous oral lesions. Computer-assisted analysis of the oral brush biopsy. U.S. Collaborative OralCDx Study Group. J Am Dent Assoc 130:1445-57. Shafer, W. G., C. A. Waldron. 1961. A clinical and histopathologic study of oral leukoplakia. Surg Gynecol Obstet 112:411-20. Shibuya, H., T. Amagasa, K. Seto, K. Ishibashi, J. Horiochi, and S. Susuld. 1986. Leukoplakia-associated multiple carcinomas in patients with tongue carcinoma. Cancer 57:843-6. Shiboski, C. H. , S. C. Shiboski, and S. Silverman Jr. 2000. Trends in oral cancer rates in the United states, 1973-1996. Community Dentist Oral Epidemiol 28:249-56. Sigurdsson, H., and B. A . Agnarsson. 1998. Basal cell carcinoma of the eyelid. Risk of recurrence according to adequacy of surgical margins. Acta Opthalmol Scand 76:477-80. Silverman, S., C. Migliorati, and F. Lozada. 1984. Oral leukoplakia and malignant transformation. A follow-up study of 257 patients. Cancer 53:563-8 Silverman, S. 1994. Oral cancer. Semin Dermatol 13:132-7. Silverman, S., M . Gorsky, and G. Kaugars. 1996. Leukoplakia, dysplasia, and malignant transformation. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 82:117. Silverman Jr, S., P. B. Sugerman. 2000. Oral premalignancies and squamous cell carcinoma. Clin Dermatol 18:563-8. Sloot, P. M . , A . G. Hoekstra, and C. G. Figdor. 1988. Osmotic response of lymphocytes measured by means of forward light scattering: Theoretical considerations. Cytometry 9:636^11. 92 Smith, P.W. 2002. Fluorescence emission-based detection and diagnosis of malignancy. J Cell Biochem. Suppl 39:54-59. Soukos, N . 2001. e-Medicine Journal, 2, #7. Available from: www.emedicine.com Strong, M . S., C. W. Vaughan, J. S. Incze. 1968. Toluidine blue in the management of carcinoma of the oral cavity. Arch Otolaryngol 87:527-31. Strong, M . S., J. Incze, C. W. Vaughan. 1984. Field cancerization in the aerodigestive tract—its etiology, manifestation, and significance. Journal of Otolaryngology 13:1-6. Sussman, L. A. , D. F. Liggins. 1996. Incompletely excised basal cell carcinoma: A management dilemma? AustNZJSurg 66:276-8. Svistun, E., R. Alizadeh-Naderi, A . El-Naggar, R. Jacob, A . Gillenwater, R. Richards-Kortum. 2004. Vision enhancement system for detection of oral cavity neoplasia based on autofluorescence. Head Neck 26:205-15. Troy T. L., Page D. L. , E. M . Sevick-Muraca. 1996. Optical properties of normal and diseased breast tissues: prognosis for optical mammography. JBiomed Opt 1:342-355. Tsai, C. FL, Y . S. Hsieh, S. F. Yang, M . Y . Chou, and Y . C. Chang. 2003. Matrix Metalloproteinase 2 and matrix metalloproteinase 9 expression in human oral squamous cell carcinoma and the effect of protein kinase C inhibitors: Preliminary observations. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 95:710-6. van der Waal, I., K. P. Schepman, E. H. van der Meij, and L. E. Smeele. 1997. Oral leukoplakia: A clinicopathological review. Oral Oncology 33:291-301. van der Waal, I., K.P. Schepman, E.H. van der Meij. 2000. A modified classification and staging system of oral leukoplakia. Oral Oncol 36:264-266. van Staveren, H. J., van Veen R. L . , Speelman O. C , Witjes M . J., Star W. M . , and J. L. Roodenburg. 2000. Classification of clinical autofluorescence spectra of oral leukoplakia using an artificial neural network: A pilot study. Oral Oncol 36:286-93. Waldron, C. A. , and W. G. Shafer. 1975. Leukoplakia revisited: A clinicopathologic study of 3256 oral leukoplakias. Cancer 36:1386-92. Wang, C. Y. , H. K. Chiang, C. T. Chen, C. P. Chiang, Y . S. Kuo, S. N . Chow. 1999. Diagnosis of oral cancer by light-induced autofluorescence spectroscopy using double excitation wavelengths. Oral Oncol 35:144-50. Wang, C. Y. , T. Tsai, H. C. Chen, S. C. Chang, C. T. Chen, C P . Chiang. 2003. Autofluorescence spectroscopy for in vivo diagnosis of DMBA-induced hamster buccal pouch pre-cancers and cancers. J Oral Pathol Med 32:18-24. Wilson, B. C , and S. L. Jacques. 1990. Optical reflectance and transmittance of tissues: Principles and applications. IEEE J Quantum Electron 26:2187-40. WHO. 1978. See World Health Organization 1978. 93 World Health Organization. 1978. Definition of leukoplakia and related lesion: An aid to studies on oral precancer. Oral Surg Oral Med Oral Pathol 46:518-39 Zakrzewska, J. M . , V . Lopes, P. Speight, and C. Hopper. 1996. Proliferative verrucous leukoplakia: A report of ten cases. Oral Med Oral Pathol Oral Radiol Endod 82:396-401. Zargi, M . , I. Fajdiga, and L. Smid. 2000. Autofluorescence imaging in the diagnosis of laryngeal cancer. Eur Arch Otorhinolaryngol 257:17-23. Zhang, L., and MP. Rosin. 2001. Loss of heterozygosity: a potential tool in management of oral premalignant lesions? Journal of Oral Pathology & Medicine 30(9): 513-20. Zhang, L. , C. F. Poh, W. L. Lam, J. B. Epstein, X . Cheng, S. Zhang, R. Priddy, J. Lovas N . D. Le, and M . P. Rosin. 2001. Impact of localized treatment in reducing risk of progression of low-grade oral dysplasia: molecular evidence of incomplete resection. Oral Oncology. 37(6): 505-12. Zhang, L. , S. P. Ng, M . Williams, D. Laronde, C. F. Poh, C. Macaulay, M . P. Rosin. 2004. Detection of high-risk oral premalignant lesions by multi-spectral fluorescent visualization. Paper presented at the International Association for Dental Research 82 n d General Session, Honolulu, Hawaii. http://iadr.confex.com/iadr/2004Hawaii/ techprogram/abstract_44189.htm 94 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099780/manifest

Comment

Related Items