Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Biomarkers for acute exacerbation of chronic obstructive pulmonary disease Chen, Roy Yu-Wei 2016

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

Item Metadata

Download

Media
24-ubc_2016_may_chen_royyuwei.pdf [ 1.39MB ]
Metadata
JSON: 24-1.0229571.json
JSON-LD: 24-1.0229571-ld.json
RDF/XML (Pretty): 24-1.0229571-rdf.xml
RDF/JSON: 24-1.0229571-rdf.json
Turtle: 24-1.0229571-turtle.txt
N-Triples: 24-1.0229571-rdf-ntriples.txt
Original Record: 24-1.0229571-source.json
Full Text
24-1.0229571-fulltext.txt
Citation
24-1.0229571.ris

Full Text

  Biomarkers for Acute Exacerbation of Chronic Obstructive Pulmonary Disease  by  ROY YU-WEI CHEN B.Sc., The University of British Columbia, 2007 Dip.T, British Columbia Institute of Technology, 2012   A THESIS SUBMITTED IN PARTIAL FULLFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE   in   THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES  (Experimental Medicine)   THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)   April 2016   © Roy Yu-Wei Chen, 2016 ii  Abstract Rationale: There are currently no generally accepted and validated blood tests available for diagnosing acute exacerbations of chronic obstructive pulmonary disease (AECOPD). There is an urgent need of biomarkers that can guide therapeutic management in AECOPD. Based on literature review, systemic inflammation and mild cardiac dysfunction are often associated with AECOPD. We hypothesized that certain protein markers can indeed be useful in tracking and diagnosing AECOPD progression.   Methods: The study cohort consisted of 368 patients recruited in the chronic obstructive pulmonary disease (COPD) Rapid Transition Program who were hospitalized with a primary diagnosis of AECOPD, and 76 stable COPD patients who served as controls. We first determined the relationship of AECOPD of C-reactive protein (CRP) and the N-terminal of the prohormone brain natriuretic peptide (NT-proBNP). We then performed a discriminatory analysis using receiver-operating characteristics (ROC) curve in a logistic regression model. We compared the area under the curve (AUC) of 4 different combinations of CRP and NT-proBNP models. Lastly, we examined several potential biomarkers that were implicated in AECOPD.  Results: The demographic data of the cohort and the controls were well matched, with an average age of 68 versus 65 years old, 64% versus 77% male, and a forced expiratory volume in 1 second (FEV1) % predicted of 52% versus 58%. The CRP and NT-proBNP levels at exacerbation onset were found to be the highest and progressively decreased over time. Of the 4 models of ROC curves, the leave-one-out cross-validated model including both CRP and NT-proBNP had an AUC of 0.80. This model replicated well in an external LEUKO dataset. On the iii  other hand, D-Dimer, pulmonary and activation-regulated chemokine (PARC) and troponin I, showed minimal or no temporal changes during hospitalization and were no different than those with stable COPD.  Conclusions: In summary, this thesis demonstrated that biomarkers such as CRP and NT-proBNP are significantly elevated during AECOPD and decreased with recovery. Secondly, a combination of CRP and NT-proBNP could discriminate patients who were hospitalized for their AECOPD from stable patients. Together, these two biomarkers show promise in diagnosing and tracking AECOPD. iv  Preface  This dissertation is original, unpublished, independent work by the author Yu-Wei Roy Chen.  The work reported in Chapters 3-5 was covered by UBC-Providence Health Care Research Ethics Board Certificate number H11-00786.  v  Table of Contents  Abstract .......................................................................................................................................... iiPreface ........................................................................................................................................... ivTable of Contents ...........................................................................................................................vList of Tables .............................................................................................................................. viiiList of Figures ............................................................................................................................... ixList of Abbreviations .....................................................................................................................xAcknowledgements ......................................................................................................................xvDedication .................................................................................................................................. xviiChapter 1: Introduction ................................................................................................................1Chapter 2: A systematic review of diagnostic biomarkers of COPD exacerbation .................22.1 Methods........................................................................................................................... 32.1.1 Study population ......................................................................................................... 32.1.2 Literature search and article selection ......................................................................... 42.1.3 Data extraction and study quality assessment ............................................................. 42.2 Results ............................................................................................................................. 52.2.1 Search results .............................................................................................................. 52.2.2 Study characteristics ................................................................................................... 62.2.3 Biomarkers for the diagnosis of AECOPD ................................................................. 72.2.4 Study quality assessment ............................................................................................ 92.3 Discussion ..................................................................................................................... 29Chapter 3: Temporal relationship of CRP and NT-proBNP in COPD exacerbation ...........33vi  3.1 Methods......................................................................................................................... 343.1.1 Study subjects ........................................................................................................... 343.1.2 Specimens and measurement technique.................................................................... 353.1.3 Statistical analysis ..................................................................................................... 363.2 Results ........................................................................................................................... 383.2.1 Time course ............................................................................................................... 403.3 Discussion ..................................................................................................................... 41Chapter 4: Diagnostic performance of CRP and NT-proBNP in AECOPD ..........................464.1 Methods......................................................................................................................... 464.1.1 Study subjects ........................................................................................................... 464.1.2 Specimens and measurement technique.................................................................... 474.1.3 Statistical analysis ..................................................................................................... 484.2 Results ........................................................................................................................... 504.2.1 ROC curve analysis................................................................................................... 504.2.2 Kaplan-Meier survival analysis ................................................................................ 534.2.3 Multiple linear regression analysis ........................................................................... 544.2.4 Replication on external dataset ................................................................................. 564.3 Discussion ..................................................................................................................... 61Chapter 5: Relationship of other biomarkers in COPD exacerbation....................................675.1 Methods......................................................................................................................... 695.1.1 Study subjects ........................................................................................................... 695.1.2 Specimens and measurement technique.................................................................... 695.1.3 Statistical analysis ..................................................................................................... 70vii  5.2 Results ........................................................................................................................... 735.2.1 Time course ............................................................................................................... 735.2.2 ROC curve analysis................................................................................................... 765.3 Discussion ..................................................................................................................... 78Chapter 6: Conclusion .................................................................................................................83References .....................................................................................................................................85Appendices ....................................................................................................................................92Appendix A List of MeSH terms used for the systematic review in Chapter 2. ....................... 92Appendix B Guidelines for REMARK scores .......................................................................... 93Appendix C COPD definitions of 59 publications included in the review arranged by the latest published year ........................................................................................................................... 96Appendix D Biomarkers investigated in one studies .............................................................. 100Appendix E Modified REMARK (mREMARK) scores breakdown for the 14 studies listed in Table 2.4 ................................................................................................................................. 103 viii  List of Tables  Table 2.1 Study characteristics of 59 publications included in the review arranged by the latest published year ............................................................................................................................... 10Table 2.2 Patient characteristics of 59 publications included in the review arranged by the latest published year. .............................................................................................................................. 15Table 2.3 Top three most studied AECOPD biomarkers .............................................................. 22Table 2.4 Selected publications from the review with biomarker ROC analysis performance .... 24Table 3.1 Patient characteristics of the COPD Rapid Transition Cohort. .................................... 37Table 4.1 ROC curve characteristics of the four models. ............................................................. 52Table 4.2 ROC curve AUC comparisons across the four models................................................. 53Table 4.3 Survival curve analysis at 1-year time point of the four models. ................................. 54Table 4.4 Survival curve analysis of LOOCV model at various time points ............................... 55Table 4.5 Multiple linear regression analysis in modeling length of hospitalization ................... 59Table 4.6 Performance of the 4 models on the LEUKO dataset. .................................................. 60Table 4.7 ROC curve comparisons from the performance on LEUKO cohort. ........................... 60Table 5.1 ROC curve characteristics of the four biomarkers. ....................................................... 78Table 5.2 ROC curve AUC comparisons amongst the four biomarkers....................................... 78 ix  List of Figures  Figure 2.1 PRISMA flow diagram of study screening and selection. ............................................ 6Figure 3.1 CRP time course box-plots .......................................................................................... 39Figure 3.2 NT-proBNP time course box-plots. ............................................................................. 40Figure 4.1 Plot of ROC curves on the four models....................................................................... 51Figure 4.2 1-year Survival curve analysis for LOOCV model. .................................................... 55Figure 4.3 1-year Survival curve analysis for the linear combination model. .............................. 56Figure 4.4 1-year Survival curve analysis for NT-proBNP model. .............................................. 57Figure 4.5 1-year Survival curve analysis for CRP model. .......................................................... 58Figure 4.6 Length of hospital stay versus NT-proBNP concentration.......................................... 59Figure 5.1 Troponin I time course box-plots. ............................................................................... 72Figure 5.2 PARC time course box-plots. ...................................................................................... 73Figure 5.3 D-Dimer time course box-plots. .................................................................................. 74Figure 5.4 MPO time course box-plots. ........................................................................................ 75Figure 5.5 Plot of ROC curves for the four biomarkers. .............................................................. 77 x  List of Abbreviations  AECOPD: acute exacerbation of chronic obstructive pulmonary disease Aα-Val360: fibrinogen cleavage product ANOVA: analysis of variance Anti-VP1 IgG1: immunoglobulin G1 antibody against viral protein 1 AUC: area under the curve BMI: body mass index BNP: brain natriuretic peptide BPI: bactericidal permeability increasing protein CD: cluster of differentiation CCL: chemokine C-C motif ligand CI: confidence interval COPD: chronic obstructive pulmonary disease CRP: c - reactive protein CT: computed tomography CV: coefficient of variation CXCL: chemokine C-X-C motif ligand ECLIPSE: evaluation of COPD longitudinally to Iidentify predictive surrogate endpoints ECG: electrocardiogram ECP: eosinophil cationic protein EDTA: ethylenediaminetetraacetic acid ELISA: enzyme-linked immunosorbent assay xi  EPO: erythropoietin ESR: erythrocyte sedimentation rate FE: frequent exacerbators/exacerbations FEV1: forced expiratory volume in 1 second FEU: fibrinogen equivalent unit FSH: follicle stimulating hormone FVC:  forced vital capacity  GFR: glomerular filtration rate GOLD: global initiative for chronic obstructive lung disease GPx: erythrocytic glutathione peroxidase Hgb: hemoglobin HMGB: high mobility group box HRV: human rhinovirus ICU: intensive care unit IE: infrequent exacerbators/exacerbations IFNγ: interferon gamma IG: immunoglobulin IGF: insulin-like growth factor IL: interleukin IP: interferon-γ inducible protein IQR: interquartile range K-M: Kaplan-Meier L: litre xii  LH: luteinizing hormone LOD: limit of detection LOOCV: leave-one-out cross validation LPS: lipopolysaccharide LTB4: leukotriene B4 MCP: monocyte chemoattractant protein MeSH: medical subject headings MFAP: microfibrillar associated protein µg: microgram mg: milligram mL: millilitre MMP: matrix metallopeptidase MPIF = myeloid progenitor inhibitory factor MPO: myeloperoxidase MR-proANP: mid-regional prohormone of atrial natriuretic peptide N/A: not available NE: non-exacerbators/exacerbations ng: nano gram NO: nitric oxide NPPV: non-invasive positive pressure ventilation NS: not significant NT-proBNP: amino-terminal of the prohormone brain natriuretic peptide PARC: pulmonary and activation-regulated chemokine xiii  PCT: procalcitonin PE: pulmonary embolism PRISMA: preferred reporting items for systemic reviews and meta-analysis proBNP: prohormone of brain natriuretic peptide QC: quality control RBP: retinol-binding proteins REMARK:  recommendations for tumor marker prognostic studies ROC: receiver-operating characteristics SEM: standard error of the mean SD: standard deviation sICAM: soluble intercellular adhesion molecule sIL-1R: soluble interleukin 1 receptor SLPI: secretory leukocyte protease inhibitor SP: surfactant protein sTNFR: soluble tumor necrosis factor  receptors sRAGE: soluble receptor for advanced glycation end-products sTREM: soluble triggering receptor expressed on myeloid cells suPAR: soluble urokinase-type plasminogen activator receptor TEAC: trolox equivalent antioxidant capacity TGF: transforming growth factor TIMP: tissue inhibitors of metalloproteinase TNF: tumor necrosis factor T3: tri-iodothyronine xiv  T4: thyroxine TSH: thyroid stimulating hormone WBC: white blood cellsxv  Acknowledgements  I am not a man of many words; this will be brief, but by no means unimportant nor insincere.  First I would like to sincerely thank my supervisors Drs Don Sin and Paul Man for having me in the laboratory. I have minimal research experience when I started, and it definitely was a life-changing moment when they took me in as a graduate student. I am really grateful for this opportunity to experience research to the fullest. Additionally, I would also like to thank Dr Chris Carlsten, who is part of the supervisory committee, and Dr Mustafa Toma, who is the external examiner, for their helpful feedback on my project.  Secondly, I would like to thank all members of Sin laboratory. At one point or another, I have received tremendous mental and physical support from these extraordinary individuals. They are the real stars. Best of luck to everyone wherever life takes you.  Thirdly, I would like to thank everyone from the Centre for Heart Lung Innovation mainly for the science, but as well for food, and drinks. In extension, I would like to thank the knowledgeable folks at the PROOF Centre for their comprehensive computational support.  Fourthly, I would like to thank all the patients and the rest of the staffs that are involved with the COPD Rapid Transition Program. Without them, my thesis work would not have existed in the first place.  xvi  Last, but definitely not least, I would like to thank my family for their support morally and financially. xvii  Dedication        THIS PAGE IS INTENTIONALLY LEFT BLANK.     Not!     To: Family (Bun, Mom, Dad, Debbie-toe), Friends, and Pets (Cats and Dogs = Mao-mao, Coco, Kiki, Little-Jade, and Little-Strong).  LONG LIFE AND PROSPERITY!!  1  Chapter 1: Introduction In this thesis, we explored the complex symptomatic occurrence called exacerbations, or “flare-ups”, in chronic obstructive pulmonary disease (COPD) patients. Currently there are no accepted blood-based biomarkers in place to diagnose and treat acute exacerbation of COPD (AECOPD). We hypothesized that certain protein biomarkers can be useful in diagnosing and tracking AECOPD. More details will be provided in later chapters but here are brief descriptions of chapters to follow. In chapter 2, we summarize the current knowledge of AECOPD biomarker research in a systematic review, and provide potential future directions. Stemming from this review, we focus on two biomarkers: C-reactive protein (CRP) and the amino-terminal of the prohormone brain natriuretic peptide (NT-proBNP) in chapter 3, which have shown potential in guiding therapeutic treatment in AECOPD. We then determine the temporal relationship of these two biomarkers to AECOPD. In chapter 4, we determine the discriminatory power of various combinations of these two biomarkers in the context of AECOPD. Then, we assess the relationship of these two biomarkers to clinical outcomes including mortality and length of hospitalization. Lastly, we validate our model in an external cohort to replicate our findings. In chapter 5, we examine four additional biomarkers and determine their temporal relationship as well as their discriminatory power. In the final chapter, we summarize the main findings from each chapter and discuss how this thesis as a whole adds to the management of COPD exacerbations.  2  Chapter 2: A systematic review of diagnostic biomarkers of COPD exacerbation Chronic obstructive pulmonary disease (COPD) is a debilitating disease that is characterized by reduced lung function, breathlessness, decreased productivity, and poor quality of life (1). Currently, COPD is the only major cause of mortality with a rising death rate and it is estimated that by 2030 COPD will become the fourth leading cause of death worldwide (2, 3). The natural history of COPD is often marked by periodic exacerbations in which symptoms of breathlessness and sputum production worsen acutely, resulting in emergency hospital admissions and hospitalizations (1, 4, 5). In Canada, acute exacerbations of COPD (AECOPD) account for the highest rate of hospital admissions and repeat hospitalizations (6), with an estimated economic burden amounting to $4.5 billion dollars each year in direct and indirect costs (7).    Owing to their heterogeneity and the lack of available diagnostic laboratory tests, AECOPD are often diagnosed based on clinical gestalt, which is subjective and variable within and across physicians. Forced expiratory volume in the first second of expiration (FEV1) has conventionally been used to guide therapy in stable COPD; however, it is a poor indicator of a patient’s exacerbation status (1). Biological molecules or biomarkers that can reflect disease activity and fluctuate in accordance with disease state could theoretically provide a more objective determination of a patient’s health status before, during, and after an AECOPD event (8, 9). Biomarkers could further allow physicians to provide personalized care for each patient by tailoring targeted therapies based on biomarker levels, thus avoiding unnecessary side effects of prolonged exposure to drugs or conversely incompletely treating an AECOPD.   3  There have been numerous articles published over the past decade, which have focused on the discovery and assessment of biomarkers in relation to AECOPD (10). Similarly, there have been a wide variety of sample types that have been collected for this purpose including exhaled breath condensate, sputum, nasal wash, blood, bronchoalveolar lavage, and lung biopsies. In this review, we have focused our attention on blood-based biomarkers to diagnose exacerbations. This type of sample has clear advantages that make clinical translation facile including non-invasiveness, ease of collection, widespread availability of laboratories that can procure and process these samples, and the ability to standardize measurements for most assays. The aims of this systematic review were to determine which plasma or serum molecules have been evaluated (and published) as possible biomarkers to diagnose AECOPD and to ascertain the quality of these publications with the view of determining which molecules, if any, have the greatest potential for clinical translation.    2.1 Methods 2.1.1 Study population Our population of interest was defined as COPD patients of any age and gender, who had experienced exacerbations, and as a consequence, required medical attention and admission to a hospital for treatment. We included studies that examined patients longitudinally (i.e. onset of AECOPD versus convalescence), and also those that were cross-sectional (i.e. AECOPD versus stable COPD). Studies that focused on exacerbation biomarkers to guide therapeutic treatments were excluded. The biomarkers, which we subsequently reviewed after study selection were categorized and described in terms of their use in the diagnosis of AECOPD onset.  4  2.1.2 Literature search and article selection We employed our search strategy in accordance with the Preferred Reporting Items for Systemic Reviews and Meta-Analysis (PRISMA) guidelines (11). We searched articles in MEDLINE (1966-2015), EMBASE (1980-2015), CINAHL (1982-2015), and Cochrane databases by using specific Medical Subject Headings (MeSH) terms. We used Elsevier ScienceDirect as an additional source.  The MeSH terms included a combination of: chronic obstructive pulmonary disease or COPD, exacerbations, acute exacerbation, or AECOPD, biomarkers or biological markers, diagnosis or diagnostic, and blood or serum or plasma (a detailed list of MeSH terms is provided in Appendix A). Two authors (YWRC and JML) independently screened the titles and abstracts based on the articles’ relevance to our MeSH terms, with disagreements resolved by iteration and consensus. Primary articles that were published in English, focused on human subjects, and performed analysis on blood specimens were retained. Entries of review articles, conference abstracts, book chapters, editorials, or original articles that performed biomarker assessments on sputum samples, or breath condensate were excluded. References from selected articles were also reviewed to ensure the inclusion of all relevant articles.  2.1.3 Data extraction and study quality assessment To assess the quality of publications reporting biomarkers, we first screened the 59 relevant original papers for the two important analytic components of biomarker studies, as recommended by Sin and colleagues (12). These components included: 1) the use of receiver operating characteristics (ROC) area-under-the curve (AUC) statistics or equivalent in reporting the performance of the biomarker as a diagnostic tool for AECOPD, and 2) the reporting of biomarker findings in a replication cohort or in a sub-cohort of the parent study. We additionally 5  used the Guidelines for the REporting of Tumor MARKer Studies (REMARK) to rank these selected publications according to the quality of reporting of biomarker data (See Appendix B for a detailed REMARK checklist) (13). Because REMARK was created for oncology studies, we modified the criteria to enable its use for AECOPD (resulting in mREMARK). For instance, where the guidelines mentioned recommendations on tumor biomarkers (points 13 and 15 of Appendix B), we simply replaced these terms with "AECOPD biomarkers". In cases where the guidelines referred to standard prognostic variables (in points 14 and 17 of Appendix B), we replaced this concept with lung function measurements, which have significant prognostic value in COPD. A collection of mREMARK scores were then collated and ranked. Higher mREMARK scores are considered reflective of a higher quality study.  2.2 Results 2.2.1 Search results The initial search revealed a total of 2,732 studies, of which 111 were duplicate articles (see Figure 2.1). We also excluded 2,362 other articles by screening titles and abstracts of these articles because they were not relevant to this analysis, leaving 270 articles eligible for full-text review. Among these articles, 211 articles were excluded because of several other factors, which are outlined in Figure 2.1.  In total, 59 studies were included in the qualitative analysis. A flow diagram of the study screening and selection is shown in Figure 2.1.       6   Figure 2.1 PRISMA flow diagram of study screening and selection. 59 Studies were included for review whereas the rest of the studies were excluded.  2.2.2 Study characteristics The study characteristics are displayed in Table 2.1 and the patient characteristics are listed in Table 2.2. The definitions used by all of the studies pertaining to COPD diagnosis, AECOPD, and stable COPD are listed in Appendix C. The majority of the studies (53 out of 59) defined COPD diagnosis based on the GOLD criteria, in which the patients had FEV1% < 80% predicted 7  and FEV1/FVC <70%, and a bronchodilator response of <12% with respect to FEV1% (1). 33 out of 59 studies defined an AECOPD based on worsening of symptoms including dyspnea, cough, or sputum production, which led to the intensification in the use of maintenance medications and/or institution of “rescue” medications (1, 5). Definitions of stable COPD were highly variable among the 59 studies, with duration free of exacerbation ranging from 3 weeks to 3 months (see Appendix D). The majority of the studies (12 out of 59) defined stable COPD as free of exacerbation in the preceding 4 weeks.  41 studies evaluated biomarkers longitudinally in the same patients at onset of and recovery from AECOPD. 42 studies evaluated biomarkers cross-sectionally between patients with AECOPD and stable COPD patients, and/or healthy controls. Approximately half of the studies included in the review were performed in Europe (47%), with the United Kingdom being the most prevalent location. Most of the studies were single-centered (86%) and all were prospective in design. 81% of the studies had a relatively small study size (defined as less than 100 exacerbating patients). The total number of patients included in the 59 studies was 5431, with a range of 9 to 333 COPD patients per study. Patients were mostly males, and the mean age of all COPD participants were 64 years with a mean FEV1% of approximately 47%.  2.2.3 Biomarkers for the diagnosis of AECOPD Biomarkers evaluated covered a wide range of molecules: acute phase reactants such as            C-reactive protein (CRP), erythrocyte sedimentation rate (ESR) and fibrinogen; cytokines such as interleukin (IL)-6, IL-8, and tumor necrosis factor-alpha (TNF-α); molecules of cardiac origin such as brain natriuretic peptide (BNP); molecules involved in collagen formation such as matrix 8  metalloproteinase (MMP)-9; and molecules involved in fatty acid processing such as adiponectin. The most commonly studied biomarker was CRP, followed by IL-6 and TNF-α (see Table 2.3). The use of CRP as a biomarker was investigated in 28 studies, in which 26 of these reported a statistically significant increase in concentration during AECOPD versus stable COPD and/or healthy controls. The CRP assays that used were highly variable; five studies used immunonephelometric assay, four studies used immunoturbidimetric assay, and three studies used immune latex agglutination assay. Despite these differences in techniques, the reported CRP results were congruent with each other. IL-6 was investigated in 18 studies, in which 13 showed significant increases during AECOPD versus stable COPD. Three studies also reported increased IL-6 levels during AECOPD, but the statistical analysis was either non-significant or not reported. TNF-α was investigated in ten studies, but with variable statistical significance. Seven out of ten studies reported significant increases in TNF-α concentrations during AECOPD compared to stable COPD, whereas two did not show statistical significance and one showed borderline statistical significance. Nevertheless, TNF-α concentrations were numerically higher in AECOPD compared with stable COPD or healthy controls. The measurement methods used for both IL-6 and TNF-α were all immuno-based assays, such as enzyme-linked immunosorbent assay (ELISA).  51 biomarkers that were investigated only in single study were detailed in Appendix E along with statistical comparisons. Approximately half of these biomarkers were reported with an increased concentration during AECOPD, and the other half reported the opposite. 18 of these biomarkers were not statistically different between levels during AECOPD versus stable state.  9  2.2.4 Study quality assessment Of the 59 articles, only 12 articles (20%) reported an ROC analysis (see Table 2.4). The articles had a median mREMARK score of 13/20, with a range from 6 to 18/20; the detailed breakdowns of the scores are shown in Appendix E. Only one (2%) article, by Bafadhel et al, utilized a second, independent cohort to replicate biomarker performance (14). This study by Bafadhel et al, which had the highest mREMARK score at 18, included 145 AECOPD patients, and used a combination of serum biomarkers and sputum biomarkers to classify patients into 4 distinct exacerbation phenotype clusters (14). CRP, CXCL10 and eosinophil % peripheral count were useful in distinguishing bacterial-, virus-, and eosinophil-associated exacerbations, respectively. The AUC results were 0.65 (CI: 0.57-0.74) for CRP, 0.76 (CI: 0.67-0.86) for CXCL10, and 0.85 (CI: 0.78-0.93) for peripheral eosinophil count. Findings of bacterial-associated exacerbations via CRP were replicated with a comparable AUC of 0.70 (CI: 0.59-0.82). In addition, findings of virus-associated exacerbations via CXCL10 were also replicated with a comparable AUC of 0.65 (CI: 0.52-0.78).10  Table 2.1 Study characteristics of 59 publications included in the review arranged by the latest published year Reference  Year Country Single or Multi-centre Study Design Biomarkers in serum or plasma  Group Comparisons (Longitudinal / Cross-sectional) Time course Andelid, K., et al. (15) 2015 Sweden S P CRP, MPO, Neutrophil Elastase, WBC L & CS                             (A vs Smoking Cont vs Non-Smoking Cont) Yes Gumus, A., et al. (16) 2015 Turkey S P CRP, fibrinogen, SuPAR L & CS (A vs Cont) Yes Chang, C. , Yao, W. (17) 2014 China S P CRP, IL-6  L Yes Chang, C., et al. (18) 2014 China S P CRP, IL-6, WBC  L Yes Fattouh, M. Alkady, O. (19) 2014 Egypt S P CRP, Fibrinogen, WBC  L & CS (A vs C vs Cont) Yes Johansson, S.L., et al. (20) 2014 Denmark M P CRP, MFAP4, SP-D, WBC  L & CS (A vs C vs Cont) Yes Labib, S., et al. (21) 2014 Egypt S P Desmosine  L & CS (C vs Cont) Yes Lee, S.J., et al. (22) 2014 Korea S P Osteopontin  L & CS (A vs C vs Cont) Yes Liu, H.C., et al. (23) 2014 Taiwan S P IL-8, IL-17 CS (C vs Cont) No Liu, Y., et al. (24) 2014 China S P CD34+ cells, CRP, MMP-9, NT-proBNP  CS (A vs C vs Cont) No Meng, D.Q., et al. (25) 2014 China S P Adrenomedullin, CRP, WBC  L & CS (A vs C vs Cont) Yes Nikolakopoulou, S., et al. (26) 2014 Greece S P Angiopoietin-2, CRP  L Yes Nishimura, K., et al. (27) 2014 Japan S P BNP  L & CS (A vs C) Yes Omar, M.M., et al. (28) 2014 Egypt S P Adiponectin  L & CS (A vs C vs Cont) Yes Oraby, S.S., et al. (29) 2014 Egypt S P Adiponectin  L & CS (A vs C vs Cont) No Urban, M.H., et al. (30) 2014 Austria S P sRAGE  L Yes Zhang, Y., et al. (31) 2014 China S P CRP, Fibrinogen, HMGB1, sRAGE  L Yes Zhao, Y.F., et al. (32) 2014 China S P Copeptin, CRP, Procalcitonin  L Yes Adnan, A.M., et al. (33) 2013 Syria S P ECP, Eotaxin/CCL11, IL-8  CS (A vs C vs Cont) No Carter, R.I., et al. (34) 2013 UK S P Aα-Val360  L Yes   11  Reference  Year Country Single or Multi-centre Study Design Biomarkers in serum or plasma  Group Comparisons (Longitudinal / Cross-sectional) Time course Gao, P., et al. (35) 2013 China S P CRP, IL-6, MMP-9, Serum Amyloid-A CS (A vs Cont) No Jin, Q., et al. (36) 2013 China S P RBP4  CS (A vs C vs Cont) No Mohamed, N.A., et al. (37) 2013 Egypt S P Adiponectin, CRP, IL-6, IL-8, TNF-α  CS (A vs C vs Cont) No Scherr, A., et al. (38) 2013 Switzerland S P Pancreatic stone protein/regenerating protein CS (A vs C vs Cont) No Shoukry, A., et al. (39) 2013 Egypt S P IL-6, TNF-α, Thyroid hormone T3, T4, TSH  CS (A vs C vs Cont) No Stanojkovic, I., et al. (40) 2013 Serbia S P Beta-crosslaps, CRP, MMP-9, TIMP-1  L & CS (A vs C vs Cont) Yes Chen, H., et al. (41) 2012 China S P 507 inflammatory mediators (including CCL22, CCL28, Cerberus 1, Growth Hormone R, IL-1β, IL-17, IL-19, Lymphotoxin beta, MMP-10, Thrombopoietin, Toll-like receptor 4) CS (A vs C vs Cont) No Falsey, A.R., et al. (42) 2012 USA S P Procalcitonin  L & CS                               (A vs Pnemonia Cont) Yes Huang, J., et al. (43) 2012 UK M P Desmosine CS (A vs C vs Cont) No Ju, C.R., et al. (44) 2012 China S P CRP, SP-D  L & CS (A vs C vs Cont) Yes Koczulla, A.R., et al. (45) 2012 Germany S P Alpha-1 antitrypsin, CRP, Procalcitonin, WBC  CS (A vs C vs Smoking Cont) No Kwiatkowska, S., et al. (46) 2012 Poland S P MMP-9, TIMP-1 CS (A vs C vs Cont) Yes Marcun, R., et al. (47) 2012 Slovenia S P NT-proBNP, Troponin T  L Yes Mohamed, K.H., et al. (48) 2012 Egypt S P CRP, ESR, Procalcitonin, WBC  L & CS (A vs Cont) No Pazarli, A.C., et al. (49) 2012 Turkey S P CRP, ESR, Procalcitonin, WBC  CS (A vs C) No Rohde, G., et al. (50) 2012 Germany S P sTREM-1  L & CS (A vs C vs Cont) Yes Shaker, A., et al. (51) 2012 Egypt S P FSH, IGF-1, LH, Testosterone  L & CS (A vs Cont) Yes Yerkovich, S.T., et al. (52) 2012 Australia S P Anti-VP1 IgG1 , IL-21 CS (A vs C) No 12  Reference  Year Country Single or Multi-centre Study Design Biomarkers in serum or plasma  Group Comparisons (Longitudinal / Cross-sectional) Time course Bafadhel, M., et al. (14) 2011 UK S P 24 biomarkers (including CCL4, CCL17, CRP, CXCL11, ECP, eosinophil % count, IFNγ, IL-5, IL-6, IP-10, Neopterin, Procalcitonin, Serum Amyloid-A, SP-D, TNFR1, TNFR2) L No Chen, H., et al. (53) 2011 China S P 40 inflammatory mediators (including betacellulin, CCL17, CCL22, CCL23/MPIF-1, CCL25, CCL27, CCL28, CXCL11, IL-9, MCP-3, MCP-4, osteopontin) L & CS (A vs C vs Cont) Yes Lacoma, A., et al. (54) 2011 Spain M P MR-proANP  L & CS (A vs C) Yes Lacoma, A., et al. (55) 2011 Spain M P CRP, Neopterin, Procalcitonin  L & CS (A vs C) Yes Lim, S.C., et al. (56) 2011 Korea S P IL-6, IL-8, TNF-α, T-Lymphocyte Apoptosis  CS (A vs C vs Cont) No Markoulaki, D., et al. (57) 2011 Greece M P CRP, EPO, Fibrinogen, Hgb, IL-6, TNF-α  L Yes Krommidas, G., et al. (58) 2010 Greece M P Adiponectin, CRP, IL-6, Leptin, TNF-α  L Yes Quint, J.K., et al. (59) 2010 UK M P CRP, IL-6, IP-10  L & CS (A vs C) Yes Koutsokera, A., et al. (60) 2009 Greece S P CRP, Fibrinogen, IL-6, Serum Amyloid-A, TNF-α  L Yes Kythreotis, P., et al. (61) 2009 Greece S P IGF-1, IL-1β, IL-6, IL-8, Leptin, TNF-α  L & CS (A vs Cont) Yes Shakoori, T.A., et al. (62) 2009 Parkistan S P SP-D  CS (A vs C vs Cont) No Karadag, F., et al. (63) 2008 Turkey S P IL-6, NO, TNF-α  L & CS (A vs C vs Cont) Yes Stolz, D., et al. (64) 2008 Switzerland S P CRP, BNP, Procalcitonin  L Yes Groenewegen, K.H., et al. (65) 2007 Netherlands S P BPI, IL-6, sIL-1RII, sTNFR55, sTNFR75, TEAC  L & CS (A vs Cont) Yes Perera, W. R., et al. (66) 2007 UK S P CRP, IL-6 L Yes   13  Reference  Year Country Single or Multi-centre Study Design Biomarkers in serum or plasma  Group Comparisons (Longitudinal / Cross-sectional) Time course Pinto-Plata, V.M., et al. (67) 2007 USA S P IL-6, IL-8, LTB4, SLPI, TNF-α  L Yes Hurst, J.R., et al. (68) 2006 UK Multi Centre P 36 biomarkers (including Adiponectin, CRP, CCL4, CCL5, CCL23/MPIF-1, Eotaxin-2, IFNγ, IL-1Ra, IL-6, IL-8, IL-17, IP-10, MCP-1, MMP-9, MPO, PARC/CCL18, sICAM-1, TGF-α, TIMP-1, TNFR1, TNFR2) L Yes Phua, J., et al. (69) 2006 Spain S P sTREM-1  CS (A vs C vs Cont) No Roland, M., et al. (70) 2001 UK S P Endothelin-1  L Yes Fiorini, G., et al. (71) 2000 Italy S P ECP, IgE, MPO  CS (A  vs  C  vs  Cont) No Note: For studies that included more than 10 biomarkers, not all markers are listed. Abbreviations: S = S, M = multi centre, P = prospective, CS = cross-sectional, A = acute exacerbation group, C = stable COPD group, Cont = healthy control group, FE = frequent exacerbators, NE = non-frequent exacerbators. Biomarker abbreviations: Aα-Val360 = fibrinogen cleavage product, Anti-VP1 IgG1 = immunoglobulin G1 antibody against viral protein 1, BNP = brain natriuretic peptide, BPI = bactericidal permeability increasing protein, CD = cluster of differentiation, CCL = chemokine C-C motif ligand, CXCL = chemokine C-X-C motif  ligand, ECP = eosinophil cationic protein, EPO = erythropoietin, ESR = erythrocyte sedimentation rate, FE = frequent exacerbators, FSH = follicle stimulating hormone, GPx = erythrocytic glutathione peroxidase, Hgb, = hemoglobin, HMGB = high mobility group box, IFN = interferon , IG = immunoglobulin, IGF =  insulin-like growth factor, IL = interleukin, IP =  interferon-γ inducible protein, LH = luteinizing hormone, LTB4 = leukotriene B4, MCP = monocyte chemoattractant protein,  MFAP = microfibrillar associated protein, MMP = matrix metallopeptidase, MPIF = myeloid progenitor inhibitory factor, MPO = myeloperoxidase, MR-proANP = Mid-regional prohormone of atrial natriuretic peptide, NE = non-exacerbators, NO = nitric oxide, NT-proBNP = amino-terminal of the prohormone of brain natriuretic peptide, PARC = pulmonary and activation-regulated chemokine , RBP = retinol-binding proteins, sICAM = soluble intercellular adhesion molecule, sIL-1R = soluble interleukin 1 receptor, SLPI = secretory leukocyte protease inhibitor, SP = surfactant protein, sTNFR = soluble tumor necrosis factor  receptors, sRAGE = soluble receptor for advanced glycation end-products, sTREM = soluble triggering receptor expressed on myeloid cells, suPAR = soluble urokinase-type plasminogen activator receptor, TEAC = 14  Trolox equivalent antioxidant capacity, TGF = transforming growth factor, TIMP = tissue inhibitors of metalloproteinase, TNF = tumor necrosis factor, T3 = Triiodothyronine, T4 = thyroxine, TSH = thyroid stimulating hormone, and WBC = white blood cells.15  Table 2.2 Patient characteristics of 59 publications included in the review arranged by the latest published year. Reference  Year Number of patients Age  (Mean ± SD) Sex (M:F) FEV1% Predicted (Mean ± SD) BMI in kg/m2  (Mean ± SD) Smoking Status Smoking History in Pack-years (Mean ± SD) Andelid, K., et al. (15) 2015 60 AECOPD 62 (45-76)* 26:34 60 (29-97)* ^ N/A All Smokers 40 (14-156)* 10 Smoker Controls 50 (26-64)* 2:8 106 (83-119)* ^ N/A All Current Smokers 27 (12-44)* 10 Non-Smoker Controls 68 (47-70)* 3:7 120 (97-137)* ^ N/A All Non-Smokers 0 Gumus, A., et al. (16) 2015 43 AECOPD 68 ± 8 40:3 37 (25-49)* 27.4 (4.6) 10 Current Smokers  53 (33-73)* 30 Controls 64 ± 7 25:5 N/A 28.9 (4.8) 10 Current Smokers 40 (23-57)* Chang, C. , Yao, W. (17) 2014 57 FE COPD 67 (63-74)* 50:7 45 (36-54)* 23.8 (20.4-26.0)* 12 Current; 40 Ex-Smokers 30 (20-40)* 78 NE COPD 66 (60-74)* 69:9 52 (43-55)* 21.3 (18.8-25.5)* 17 Current; 50 Ex-Smokers 20 (15-30)* Chang, C., et al. (18) 2014 93 AECOPD 67 (61-74)* 83:10 47 (43-55)* 21.6 (19.0-25.7)* 21 Current Smokers 25 (22-34)* Fattouh, M. Alkady, O. (19) 2014 98 AECOPD 62.29±7.03 82:16 53.41 ± 7.469 N/A 24 Current; 58 Ex; 16 Non-Smokers N/A 30 Controls N/A 23:7 88.4 ± 5.137 N/A 5 Current; 7 Ex; 17 Non-Smokers N/A Johansson, S.L., et al. (20) 2014 14 AECOPD 66 ± 8 N/A 30 ± 11 N/A 5 Current Smokers 60 (34)* 69 Stable COPD 62 ± 7 32:37 50 ± 16 26.2 ± 4.7 21 Current; 48 Ex-Smokers 40 (20)* 54 Smoker Controls 46 ± 12 N/A N/A N/A All Smokers N/A 52 Non-Smoker Controls 41 ± 14 N/A N/A N/A All Non-Smokers N/A Labib, S., et al. (21) 2014 20 AECOPD 66.4 ± 11.2 17:3 N/A N/A 9 Current; 11 Ex-Smokers 55.6 ± 23.7 20 Controls 59.3 ± 13.6 15:5 N/A N/A N/A 10.1 ± 4.2   16  Reference  Year Number of patients Age  (Mean ± SD) Sex (M:F) FEV1% Predicted (Mean ± SD) BMI in kg/m2  (Mean ± SD) Smoking Status Smoking History in Pack-years (Mean ± SD) Lee, S.J., et al. (22) 2014 64 AECOPD 73.1 ± 8 58:6 52.2 ± 25.3 20.5 ± 3.4 16 Current Smokers; 48 Ex-Smokers 44.4 ± 21.0 68 Stable COPD 70.3 ± 6.3 64:4 58.8 ± 20.7 21.9 ± 3.5 17 Current Smokers; 51 Ex-Smokers 38.1 ± 13.3 30 Controls 70.8 ± 4.1 26:4 98.1 ± 19.5 23.6 ± 3.3 3 Current Smokers; 5 Ex-Smokers N/A Liu, H.C., et al. (23) 2014 9 AECOPD 72 ± 4.7 9:0 N/A 22.3 All Smokers N/A 12 Asthma 48 ± 8.5 5:7 N/A 24.0 5 Current; 7 Non-Smokers N/A 10 Controls 44 ± 8.5 5:5 N/A 24.2 All Non-Smokers N/A Liu, Y., et al. (24) 2014 27 AECOPD 70 ± 9 24:3 34 ± 13 19.3 ± 3.6 All Ex-Smokers 38.5 26 Stable COPD 65 ± 8 24:2 37 ± 17 20.2 ± 2.4 All Ex-Smokers 40.9 24 Controls 66 ± 9 20:4 101 ± 11 20.9 ± 2.0 All Ex-Smokers 39.2 Meng, D.Q., et al. (25) 2014 79 AECOPD 67.0 ± 9.4 67:12 35.05 ± 17.98 20.9 ± 3.5 N/A 30 (0–45)* 29 Stable COPD 64.9 ± 8.3 25:4 34.94 ± 13.18 22.3 ± 3.2 N/A 42 (30–60)* 20 Controls 61.6 ± 9.0 15:5 103.3 ± 8.42 21.9 ± 1.8 N/A 30 (6.5–34.5)* Nikolakopoulou, S., et al. (26)   2014 90 AECOPD 69.64 ± 9.93 67:23 N/A N/A Current or Ex-Smokers N/A Nishimura, K., et al. (27) 2014 61 AECOPD 75.4 ± 7.6 49:12 56.0 ± 23.8 20.1 ± 3.3 N/A 76 ± 41 190 Stable COPD 71.7 ± 8.7 166:24 66.6 ± 27.4 21.9 ± 3.2 32 Current; 158 Ex-Smokers 71 ± 42 Omar, M.M., et al. (28) 2014 15 Obese AECOPD 53.13 ± 5.08 15:0 N/A 32.9 ± 1.9 All Smokers N/A 25 Non-Obese AECOPD 54.88 ± 5.25 25:0 N/A 22 ± 1.7 All Smokers N/A 7 Obese Controls 48.75 ± 5.4 7:0 N/A 34 ± 1.3 All Smokers N/A 8 Non-Obese Controls 47.75 ± 6.43 8:0 N/A 23.5 ± 2 All Smokers N/A Oraby, S.S., et al. (29) 2014 68AECOPD 53.43 ± 5.26 68:0 43.75 ± 0.75 29.33 ± 6.08 All Smokers 30.97 ± 13.15 20 controls 52.24 ± 4.86 20:0 79.46 ± 0.27 20.22 ± 3.57 Non Smokers N/A 17  Reference  Year Number of patients Age  (Mean ± SD) Sex (M:F) FEV1% Predicted (Mean ± SD) BMI in kg/m2  (Mean ± SD) Smoking Status Smoking History in Pack-years (Mean ± SD) Urban, M.H., et al. (30) 2014 29 AECOPD 64 ± 7.6 8:21 37 ± 12 25 ± 5.8 N/A 69 ± 46 Zhang, Y., et al. (31) 2014 44 AECOPD 68.2 ± 8.5 32:12 47.5 ± 13.5 24.4 ± 4.2 19 Current; 11 Ex; 14 Non-Smokers 39.5 ± 17.6 Zhao, Y.F., et al. (32) 2014 78 AECOPD 74 ± 12 53:25 40.87 ± 14.16 N/A N/A 47 (35–68)* 81 Stable COPD 70 ± 19 59:22 57.04 ± 19.74 N/A N/A 45 (32–60)* Adnan, A.M., et al. (33) 2013 35 AECOPD 65.37 74:14 N/A N/A  N/A 30 Stable COPD 63.36 N/A N/A  N/A 23 Controls 62.78 N/A N/A All Non-Smokers N/A Carter, R.I., et al. (34) 2013 81 AECOPD 65.75 ± 0.92# 45:36 73.28 ± 2.86# N/A All Smokers or ex-smokers N/A Gao, P., et al.  2013 83 AECOPD 63.23 ± 11.42 61:22 39.8 ± 14.7 21.6 ± 4.8 40 Current; 43 Non-Smokers 19.11 ± 11.92 26 Controls 60.44 ± 13.42 25:1 93.0 ± 14.7 24.6 ± 3.7 9 Current; 17 Non-Smokers 15.32 ± 13.85 Jin, Q., et al. (36) 2013 100 AECOPD 76 (66-91)$ 72:28 40 (24–66)$ 20.7 (17.2-26.4)$ N/A N/A 46 Stable COPD 75 (65–89)$ 34:12 N/A 21.6 (17.8-27.5)$ N/A N/A 50 Controls 75 (65–86)$ 34:16 N/A 23.4 (19.5-26.8)$ N/A N/A Mohamed, N.A., et al. (37) 2013 40 AECOPD 68.32 ± 6.6 N/A 55.28 ± 24 20.99 ± 5.2 N/A N/A 20 Stable COPD 64.4 ± 5.6 N/A 71.52 ± 22 24.33 ± 3.7 N/A N/A 20 Controls 62.72 ± 4.2 N/A N/A 26.4 ± 4.0 N/A N/A Patel, A.R.C., et al. (72) 2013 55 AECOPD 72.1 ± 8.4 32:23 46.7 ± 18.5 27.1 ± 5.5 11 Current Smokers 44 (21-74)* 98 Stable COPD 72.1 ± 8.9 60:38 52.0 ± 18.9 26.8 ± 5.6 20 Current Smokers 45 (25-79)* Scherr, A., et al. (38) 2013 200 AECOPD 70 (42-91)$ 114:86 40 ± 18.3 N/A N/A 45 ± 28 133 Stable COPD 64.9 ± 12.3 99:34 47 ± 17 N/A N/A N/A 40 Controls 59.2 ± 7 22:18 N/A N/A N/A N/A Shoukry, A., et al. (39) 2013 20 AECOPD 56.8 ± 8.3 18:2 44.6 ± 11.9 25.2 ± 2.5 All male ex-smokers; females non-smokers N/A 40 Stable COPD 54.5 ± 7.6 35:5 54.8 ± 14.5 24.8 ± 3.4 All male ex-smokers; females non-smokers N/A 20 Controls 53.6 ± 8.2 17:3 94.7 ± 12.6 24.3 ± 1.58 All Non-Smokers N/A 18  Reference  Year Number of patients Age  (Mean ± SD) Sex (M:F) FEV1% Predicted (Mean ± SD) BMI in kg/m2  (Mean ± SD) Smoking Status Smoking History in Pack-years (Mean ± SD) Stanojkovic, I., et al. (40) 2013 85 AECOPD 61.9 ± 7.9 34:51 41.9 ± 17.4 AECOPD 24.4 ± 5.2 28 Current; 41 Ex; 16 Non-Smokers N/A 57.9 ± 25.3 Stable COPD 47 Controls 59.1 ± 4.4 22:25 N/A 24.7 ± 2.8 14 Current; 21 Ex; 12 Non-Smokers N/A Chen, H., et al. (41) 2012 6 AECOPD 69 ± 10.35# 3:3 55.72 ± 12.57# 18.64 ± 3.42# 3 Current; 3 Non-Smokers N/A 6 Stable COPD 61.50 ± 8.43# 3:3 64.72 ± 15.23# 19.76 ± 4.42# 3 Current; 3 Non-Smokers N/A 6 Controls 61.17 ± 9.16# 3:3 84.72 ± 10.57# 22.28 ± 1.25# 3 Current; 3 Non-Smokers N/A Falsey, A.R., et al. (42) 2012 184 AECOPD 66.7 ± 13.3 97:87 N/A N/A N/A N/A Huang, J., et al. (43) 2012 102 AECOPD 72 (66-79)* 43:59 47 ± 18 26 ± 7 55 Current; 33 Ex-Smokers ≥ 10 53 Stable COPD 65 ± 7 23:30 62 (44-77)* 27 (23-30)* 28 Current: 25 Ex-Smokers ≥ 10 26 Controls (Non-Smokers) 51 ± 13 17:29 106 (95-115)* 27 (22-28)* All Non-Smokers N/A 20 Controls  (Smokers) 52 ± 9 9:11 93 (88-100)* 27 (21-30)* All Smokers N/A 19 Controls (group 2) 68 (65-73)* 18:1 N/A N/A 10 Current; 9 Non-Smokers N/A Ju, C.R., et al. (44) 2012 40 AECOPD 66.05 ± 5.67 38:2 33.78 ± 7.39 18.07 ± 3.39 All Ex-Smokers N/A 71 Stable COPD 65.17 ± 6.80 54:17 37.76 ± 14.93 19.15 ± 3.19 All Ex-Smokers N/A 60 Controls 63.98 ± 5.77 21:39 97.1 ± 8.90 22.64 ±2.24 All Non-Smokers N/A   19  Reference  Year Number of patients Age  (Mean ± SD) Sex (M:F) FEV1% Predicted (Mean ± SD) BMI in kg/m2  (Mean ± SD) Smoking Status Smoking History in Pack-years (Mean ± SD) Koczulla, A.R., et al. (45) 2012 18 AECOPD 72.2 ± 11.8 13:5 59.2 ± 14.5 N/A 1 Current; 17 Ex-Smokers 43.9 ± 26.4 17 Stable COPD 66.6 ± 7.8 11:6 56.3 ± 17.5 N/A 3 Current; 14 Ex-Smokers 38.8 ± 21.5 10 Controls 36.1 ± 11.5 6:4 103.5 ± 6.8 N/A 3 Ex-Smokers 4 ± 7.4 Kwiatkowska, S., et al. (46) 2012 17 AECOPD 68 ± 10 10:7 57 ± 15 N/A All Current Smokers 33 ± 19 22 Controls 57 ± 11 14:8 95 ± 12 N/A All Current Smokers 28 ± 11 Marcun, R., et al. (47) 2012 127 AECOPD 70 ± 10 89:38 34 ± 15 N/A 28 Current; 99 Ex-Smokers 48 ± 30 Mohamed, K.H., et al. (48) 2012 50 AECOPD 64.6 ± 8.0 37:13 53.2 ± 9.5 N/A 42 Current Smokers 40.4 ± 8.9 10 Controls 56.3 ± 11.5 8:2 88 ± 10.4 N/A 6 Current Smokers 28 ± 6 Pazarli, A.C., et al. (49) 2012 68 AECOPD 65.9 ± 0.97# 58:10 N/A 26.6 ± 0.58# N/A 46.1 ± 2.89# 50 Stable COPD 64.1 ± 1.22# 46:4 N/A 26.9 ± 0.78# N/A 45 ± 2.88# Rohde, G., et al. (50) 2012 118 AECOPD 66 (13)* 95:23 35.1 (20.9)* 26.9 (6.3)* N/A 30 (32)* 64 Stable COPD 67 (19)* 56:8 45.4 (29.6)* 27.5 (7.7)* N/A 30 (47)* 13 Smoker Controls 47.5 (4)* 7:6 98.4 (13.5)* 26.5 (6.6)* All Current Smokers 40.5 (26.8)* Shaker, A., et al. (51) 2012 20 AECOPD 57.34 ± 12.65 20:0 46.39 ± 12.15 23.61 ± 4.86 N/A 48.57 ± 15.37 10 Controls 55.58 ± 13.21 10:0 91.23 ± 14.46 25.32 ± 5.72 N/A 46.39 ± 13.46 Yerkovich, S.T., et al.  2012 32 AECOPD 69.6 (62.8-74.2)* 21:11 31 (21-45) N/A 9 Current Smokers 45 (35-72)* 28 Stable COPD 69.8 (62.9-73.4)* 15:13 46 (35-60) N/A 8 Current Smokers 53 (37-72)* Bafadhel, M., et al. (14) 2011 145 AECOPD 69 (43-88)$ 101:44 52 ± 2# N/A 42 Current; 100 Ex-Smokers 49 (10-153)$ Chen, H., et al. (53) 2011 7 AECOPD N/A N/A N/A N/A N/A N/A 5 Stable COPD N/A N/A N/A N/A N/A N/A 5 Controls N/A N/A N/A N/A N/A N/A   20  Reference  Year Number of patients Age  (Mean ± SD) Sex (M:F) FEV1% Predicted (Mean ± SD) BMI in kg/m2  (Mean ± SD) Smoking Status Smoking History in Pack-years (Mean ± SD) Lacoma, A., et al. (54) 2011 217 AECOPD 71.38 ± 9.97 204:13 102 <40%; 51 40%-59%; 17 60%-80% 47 N/A N/A 38 Current; 150 Ex; 23 Non-Smokers; 6 N/A N/A 46 Stable COPD 70.93 ± 10.37 45:1 11 <40%; 22 40%-59%; 3 60%-80%; 10 N/A N/A 15 Current; 23 Ex-Smokers; 8 N/A N/A Lacoma, A., et al. (55) 2011 217 AECOPD 71.4 ± 9.9 204:13 102 <40%; 51 40%-59%; 17 60%-80% 47 N/A N/A 38 Current; 150 Ex; 23 Non-Smokers; 6 N/A N/A 46 Stable COPD 70.9 ± 10.3 45:1 11 <40%; 22 40%-59%; 3 60%-80%; 10 N/A N/A 15 Current; 23 Ex-Smokers; 8 N/A N/A Lim, S.C., et al. (56) 2011 17 AECOPD 68.3 ± 7.7 13:4 36.7 ± 14.5 N/A 9 Current Smokers 35.2 ± 23.7 21 Stable COPD 64.9 ± 7.1 20:1 56.9 ± 15.6 N/A 11 Current Smokers 36.1 ± 19.2 12 Controls 62.5 ± 7.9 11:1 103.0 ± 11.5 N/A 7 Current Smokers 33.3 ± 17.2 Markoulaki, D., et al. (57) 2011 93 AECOPD 66 ± 9 64:29 45 (34–64)* 27 (25–33)* 52 Current; 41 Ex-Smokers 85 ± 45 Krommidas, G., et al. (58) 2010 63 AECOPD 67.4 ± 9.1 54:9 42.8 ± 13.4 27.7 ± 5.3 38 Current; 25 Ex-Smokers 92.5 ± 49.8 Quint, J.K., et al. (59) 2010 136 AECOPD 72.6 ± 8.4 83:53 53.9 ± 18.7 26.3 ± 5.8 41 Current Smokers 51.1 ± 38.6 70 Controls 67.4 ± 8.7 28:42 112.1 ± 28.3 26.0 ± 5.1 12 Current Smokers 18.4 ± 20.9 Koutsokera, A., et al. (60) 2009 30 AECOPD 69.3 ± 1.7# 28:2 38.6 ± 3.5# 25.7 ± 1.0# 14 Current; 16 Ex-Smokers 89.7 ± 10.0# Kythreotis, P., et al. (61) 2009 52 AECOPD 65.8 ± 8.3 43:9 44.1 ± 11.4 25.6 ± 4.1 All current or ex-smokers N/A 25 Controls 65.9 ± 9.6 19:6 89.9 ± 8.0 27.9 ± 4.1 N/A N/A Shakoori, T.A., et al. (62) 2009 13 AECOPD 60 ± 13 13:0 45 ± 21 22 ± 5 9 Current; 4 Ex-Smokers 53 ± 23 14 Stable COPD 62 ± 11 14:0 50 ± 24 22 ± 4 8 Current; 6 Ex-Smokers 75 ± 41 54 Controls 36 ± 11 54:0 98 ± 13 26 ± 6 35 Current; 5 Ex; 13 Non-Smokers 16 ± 20   21  Reference  Year Number of patients Age  (Mean ± SD) Sex (M:F) FEV1% Predicted (Mean ± SD) BMI in kg/m2  (Mean ± SD) Smoking Status Smoking History in Pack-years (Mean ± SD) Karadag, F., et al. (63) 2008 20 AECOPD 68.60 ± 5.87 20:0 36.00 ± 9.92 24.00 ± 5.41 N/A 61.50 ± 22.05 83 Stable COPD 65.54 ± 7.66 83:0 46.41 ± 14.43 25.25 ± 4.84 N/A 53.58 ± 25.63 30 Controls 64.10 ± 7.68 30:0 85.09 ± 10.24 26.32 ± 3.07 N/A 16.77 ± 15.78 Stolz, D., et al. (64) 2008 208 AECOPD 70.3 ± 9.9 94:114 41 ± 17 N/A 94 Current; 97 Ex-Smokers 45 ± 27.9 Groenewegen, K.H., et al. (65) 2007 21 AECOPD 66.7 ± 9.0 15:6 35.0 ± 14.4 23.5 ± 4.7 7 Current; 13 Ex; 1 Non-Smokers 40 ± 20 20 Controls 60.6 ± 3.4 14:6 108.2 ± 14.2 25.9 ± 2.7 1 Current; 11 Ex; 8 Non-Smokers 20 ± 15 Perera, W. R., et al. (66) 2007 73 AECOPD 69.3 ± 7.8 39:34 45 ± 18 26.2 ± 6.3 20 Current Smokers 48.1 ± 34.9 Pinto-Plata, V. M., et al. (67)  2007 20 AECOPD 72 ± 8 N/A 41 ± 13 25.8 ± 5.9 All Ex-Smokers 68 ± 27 Hurst, J.R., et al. (68) 2006 90 AECOPD 70.1 ± 8.2 54:36 43.9 (27.5–56.8)* 24.7 (20.8–29.4)* 25 Current Smokers 45 (29–59)* Phua, J., et al. (69) 2006 43 Stable COPD 72 ± 14 37:6 N/A N/A N/A 50 (0-180) 63 Controls 33 ± 11 43:20 N/A N/A N/A N/A Roland, M., et al. (70) 2001 71 AECOPD 68.2 ± 7.8 N/A 39.8 ± 17.0 N/A 26 Current Smokers 44.6 ± 34.1 Fiorini, G., et al. (71) 2000 17 AECOPD 69.5 ± 2 13:4 N/A N/A 10 Current Smokers 1 pack per day 11 Stable COPD N/A N/A N/A N/A N/A N/A 11 Controls 70.6 ± 4.4 4:7 N/A N/A 1 Smoker 1 pack per day Numerical results displayed as Mean ± SD unless otherwise indicated by symbols. Symbols: N/A: Not available, *: Median (IQR), #: Mean ± SEM, and $: Mean (Range), ^: Without the use of bronchodilation prior to lung measurement. Abbreviations: AECOPD = acute exacerbation of COPD group, FE = frequent exacerbators, and NE = non-frequent exacerbators.    22  Table 2.3 Top three most studied AECOPD biomarkers Reference Year Number of patients CRP IL-6 TNF-α Assay methodology Andelid, K., et al. (15) 2015 60 AECOPD ^ +   N/A Gumus, A., et al. (16) 2015 43 AECOPD ↑ +, ^ +   N/A Chang, C. , Yao, W. (17) 2014 135 AECOPD ↑ + ↑  Latex agglutination assay, ELISA Chang, C., et al. (18) 2014 93 AECOPD ↑ + ↑ +  Latex agglutination assay, ELISA Fattouh, M. Alkady, O. (19) 2014 98 AECOPD ↑ +, ^ +   Latex agglutination assay Johansson, S.L., et al. (20) 2014 14 AECOPD : 69 COPD ↑   N/A Liu, Y., et al. (24) 2014 27 AECOPD : 26 COPD ↑ +, ^ +   Immunoturbidimetric assay Meng, D.Q., et al. (25) 2014 79 AECOPD : 29 COPD NP   N/A Nikolakopoulou, S., et al. (26) 2014 90 AECOPD ↑ +   Immunoturbidimetric assay Zhang, Y., et al. (31) 2014 44 AECOPD ↑ +   Hospital analyzer Zhao, Y.F., et al. (32) 2014 78 AECOPD : 81 COPD ↑ +   Immunonephelometric assay Gao, P., et al. (35) 2013 83 AECOPD ^ + ^ +  ELISA Mohamed, N.A., et al. (37) 2013 40 AECOPD : 20 COPD ↑ + ↑ +, ^ + ↑ +, ^ + Immunonephelometric assay, ELISA Patel, A.R.C., et al. (72) 2013 55 AECOPD : 98 COPD ↑ +   Electro-chemiluminescence        immune assay Shoukry, A., et al. (39) 2013 20 AECOPD : 40 COPD  ↑ −, ^ + ↑ +, ^ + ELISA Stanojkovic, I., et al. (40) 2013 85 AECOPD ↑ +   Immunoturbidimetric assay Ju, C.R., et al. (44) 2012 40 AECOPD : 71 COPD ↑ +, ^ +   ELISA Koczulla, A.R., et al. (45) 2012 18 AECOPD : 17 COPD ↑ +, ^ +   N/A Mohamed, K.H., et al. (48) 2012 50 AECOPD ↑ +   Hospital analyzer Pazarli, A.C., et al. (49) 2012 68 AECOPD : 50 COPD ^ +   N/A Bafadhel, M., et al. (14) 2011 145 COPD ↑ + ↑ +  Immunoturbidimetric assay,     Multiplex immuno assay Chen, H., et al. (53) 2011 7 AECOPD : 5 COPD NP   N/A Lacoma, A., et al. (55) 2011 217 AECOPD : 46 COPD ↑ +   Immunofluorescent assay Lim, S.C., et al. (56) 2011 17 AECOPD : 21 COPD  ↑ +, ^ + ↑ +, ^ + ELISA Markoulaki, D., et al. (57) 2011 93 AECOPD ↑ + ↑ + ↑ + Immunonephelometric assay, ELISA Krommidas, G., et al. (58) 2010 63 AECOPD ↑ + ↑ + ↑ + Immunonephelometric assay, ELISA Quint, J.K., et al. (59) 2010 136 AECOPD ↑ + ↑ +  Luminometric assay, ELISA Koutsokera, A., et al. (60) 2009 30 AECOPD ↑ + ↑ + ↑ + Immunonephelometric assay, ELISA 23  Reference Year Number of patients CRP IL-6 TNF-α Assay methodology Kythreotis, P., et al. (61) 2009 52 AECOPD  ↑ + ↑ + ELISA Karadag, F., et al. (63) 2008 20 AECOPD : 83 COPD  ↑ −, ^ + ↑ ±, ^ + ELISA Stolz, D., et al. (64) 2008 208 AECOPD  NP  EMIT Groenewegen, K.H., et al. (65) 2007 21 AECOPD  ↑ +  ELISA Perera, W. R., et al. (66) 2007 73 AECOPD ↑ + ↑ +  ELISA Pinto-Plata, V. M., et al. (67) 2007 20 AECOPD  ↑ + ↑ − ELISA Hurst, J.R., et al. (68) 2006 90 AECOPD ↑ + ↑ + ↑ − Chemiluminescent proteome array Abbreviations: ↑ = biomarker increased during AECOPD, ^ = biomarker increased during AECOPD compared to healthy controls, + = Statistically significant ( P-value < 0.05), − = Not statistically significant (P-value > 0.20), ± = Borderline statistically significant ( P-value = 0.05-0.20), NP = studied biomarker but statistically analysis not performed or not published, : = independent group comparison between exacerbating COPD patients versus stable COPD patients, ELISA = Enzyme-linked immunosorbent assay, EMIT = Enzyme multiplied immunoassay technique, N/A = Not available or not specified.    24  Table 2.4 Selected publications from the review with biomarker ROC analysis performance Reference  Sample Size  Blood-based Biomarker AUC (95% CI) Sens  Spec  Replica-tion ROC Notes  mREMARK Score (/20) Bafadhel, M., et al. (14) 145 COPD CRP  1) 0.65 (0.57-0.74)  4) 0.70 (0.59-0.82) 1) 0.60  4) 0.65 1) 0.70  4) 0.65 External 1) Discriminate bacterial AECOPD cluster  2) Discriminate viral AECOPD cluster  3) ROC for eosinophil associated exacerbations  4) Validation performance  18 IL-6 1) 0.67 (0.58-0.76) N/A N/A Serum Amyloid-A 1) 0.67 (0.58-0.76) N/A N/A TNFR1 1) 0.62 (0.53-0.71) N/A N/A TNFR2 1) 0.60 (0.50-0.70) N/A N/A CXCL10 2) 0.76 (0.67-0.86)  4) 0.65 (0.52-0.78) 2) 0.75  4) 0.70 2) 0.65  4) 0.60 CXCL11 2) 0.67 (0.56-0.78) N/A N/A IFNγ 2) 0.65 (0.54-0.75) N/A N/A Peripheral Eosinophil % 3) 0.85 (0.78-0.93) 3) 0.90 3) 0.60 IL-5 3) 0.65 (0.55-0.76) N/A N/A CCL17 3) 0.63 (0.53-0.73) N/A N/A   25  Reference  Sample Size  Blood-based Biomarker AUC (95% CI) Sens  Spec  Replica-tion ROC Notes  mREMARK Score (/20) Lacoma, A., et al. (55) 217 AECOPD 46 Stable COPD CRP 1) 0.527  (0.721-1.61)  2) 0.683 (0.6530-1.76) N/A   N/A N/A   N/A None 1) Discriminate bacterial AECOPD   2) Discriminate bacterial AECOPD with clinical symptoms 16 Procalcitonin 1) 0.523  (0.712-1.38)  2) 0.663 (0.652-1.42) N/A   N/A N/A   N/A  Neopterin 1) 0.609 (0.389-0.904)  2) 0.699 (0.305-0.867) N/A  N/A   Stolz, D., et al. (64) 208 AECOPD BNP 1) 0.55 (0.41-0.68)  2) 0.56 (0.45-0.66) N/A  N/A N/A  N/A None 1) Discriminate 6-month mortality 2) Discriminate 2-year mortality 15 Jin, Q., et al. (36) 64 AECOPD RBP4 0.88 (0.78-0.94) N/A N/A None Discriminate AECOPD patient mortality 15 Hurst, J.R., et al. (68) 90 AECOPD CRP only 1) 0.73 (0.66-0.80)  2) 0.88 (0.82-0.93) N/A  N/A N/A  N/A None 1) Discriminate AECOPD versus Stable COPD  2) Discriminate AECOPD with CRP and one major symptom versus Stable COPD 14 CRP, MMP-9, and MPIF-1 1) 0.75 (0.67-0.82) N/A N/A  All 36 biomarkers 1) 0.79 (0.73-0.86) N/A N/A  26  Reference  Sample Size  Blood-based Biomarker AUC (95% CI) Sens  Spec  Replica-tion ROC Notes  mREMARK Score (/20) Gumus, A., et al. (16) 43 AECOPD suPAR  Fibrinogen  CRP 0.807 (0.715-0.900)  0.763 (0.663-0.862)  0.695 (0.583-0.807) N/A  N/A  N/A N/A  N/A  N/A None Discriminate AECOPD on day 1 versus day 7 13 Shakoori, T.A., et al. (62) 13 AECOPD 14 Stable COPD 54 Controls SP-D 1) 0.76  (0.597-0.916)   2) 0.68 (0.48-0.89) 1) 0.77    2) 0.77 1) 0.74    2) 0.63 None 1) Discriminate AECOPD versus stable COPD and Controls  2) Discriminate AECOPD versus stable COPD 13 Falsey, A.R., et al. (42) 184 AECOPD 56 Pneumonia 16 Bacterial and viral AECOPD 25 viral AECOPD Procalcitonin 1) 0.76 (0.68-0.84) 0.75 (0.67-0.82)   2) 0.7013  (0.53-0.87) N/A    2) 0.31 N/A    2) 0.96 None 1) Discriminate AECOPD versus pneumonia patients at day 1 or 2  2) Discriminate bacterial and viral AECOPD versus viral AECOPD alone 12 Quint, J.K., et al. (59) 72 AECOPD IP-10 (CXCL10)    1) 0.78 (0.65-0.91)  2) 0.82 (0.74-0.90) N/A N/A None 1) Discriminate HRV-positive AECOPD versus HRV-negative AECOPD  2) Discriminate HRV-positive AECOPD with coryzal symptoms versus HRV-negative AECOPD 11   27  Reference  Sample Size  Blood-based Biomarker AUC (95% CI) Sens  Spec  Replica-tion ROC Notes  mREMARK Score (/20) Pazarli, A.C., et al. (49) 68 AECOPD 50 Stable COPD Procalcitonin 1) 0.887  (0.804-0.970)  2) 0.894 (0.821-0.967) 1) 0.82   2) 0.95 1) 0.91   2) 0.78 None 1) Discriminate mild versus moderate/severe exacerbation  2) Discriminate patients with NPPV versus without 10 Phua, J., et al. (69) 43 COPD 72 Pneumonia 35 Asthma 63 Controls sTREM-1 0.77 (0.70-0.84) 0.81 0.65 None Discriminate  patients requiring antibiotics versus those that do not 10 Adnan, A.M., et al. (33) 35 AECOPD 30 Stable COPD 23 Controls Eotaxin   Eotaxin and  IL-8  Eotaxin and ECP 1) 0.703 2) 0.872  N/A   N/A 1) 0.382 2) 0.738  1) 0.771 2) 0.923  1) 0.657 2) 0.985 1) 0.968 2) 0.783  1) 0.900 2) 0.609  1) 0.833 2) 0.652 None 1) Discriminate AECOPD versus stable COPD  2) Discriminate Stable COPD versus Controls  6 Selected studies from the review that contained biomarker characteristics and ROC performances. Only Studies that performed ROC analysis were included in this table and those without biomarker ROC analysis were not included. REMARK scores are assigned based on whether the studies met the 20 reporting recommendations. The table is arranged in descending order based on the REMARK scores. Abbreviations: AUC = area under the curve, CCL = chemokine C-C motif ligand, CI = confidence interval, CXCL = chemokine C-X-C motif ligand, ECP = eosinophil cationic protein, FE = frequent exacerbators, HRV = human rhinovirus, ICU = intensive care unit, IFN = interferon, IL = interleukin, IP = interferon-γ inducible protein, LVD = left ventricular dysfunction, MMP-9 = matrix metallopeptidase-9, MPIF-1 = myeloid progenitor inhibitory factor-1, N/A = not available, NPPV = non-invasive positive pressure ventilation, RBP = retinol-28  binding protein, REMARK =  recommendations for tumor marker prognostic studies, ROC = receiver-operator characteristics, Sens = sensitivity, Spec = specificity, SP-D = surfactant protein-D, TNFR = tumor necrosis factor receptor.29  2.3 Discussion In this systematic review, we have identified 59 studies that have prospectively evaluated a wide range of blood-based biomarkers in the diagnosis of AECOPD; however, we found a number of deficiencies in the literature that have likely impeded the translation of these biomarkers into widespread clinical use. Very few of these studies reported performance of their biomarkers using an ROC analysis and even fewer replicated their findings in an external cohort. Along these lines, many biomarkers have only been tested in single centres, again raising the necessity for validation of these results. Moreover, the definitions employed for AECOPD and stable states were inconsistent across studies, making it difficult to assess overall biomarker performance. Until these gaps in the literature are addressed, a biomarker that can accurately and consistently diagnose AECOPD may be challenging to achieve. The best studied biomarkers to date have reflected inflammatory and cytokine pathways (CRP, IL-6, and TNF-α); however, of these, only CRP concentrations appeared to be consistently elevated in the AECOPD state compared to convalescence. Still, only four studies evaluating CRP used an ROC analysis and only one study employed a second validation cohort.  Given these discrepancies in the literature, minimum standard criteria implemented for every biomarker study may help to strengthen the quality of biomarker research. Currently, there is no standardized method of assessing the quality of COPD biomarker performance (12). We propose here the use of the modified REMARK (mREMARK) score, derived from the oncology literature to better serve COPD. In the field of oncology, REMARK guidelines for biomarker studies have been in place since 2005 to facilitate the translation of biomarkers from discovery to clinical trials (13). These detailed and rigorous criteria help to provide a realistic and 30  reproducible performance assessment and importantly include the requirement that studies report estimated effects of biomarkers, validation, comparisons to standard prognostic variables, and transparent statistical methods. Although the REMARK checklist was originally developed to assess the quality of biomarker studies in oncology, we believe that with minimal modifications, it may serve as an assessment tool in COPD exacerbation biomarker discovery. In accordance with mREMARK guidelines, very few among the selected studies in this systematic review would be deemed of “good” quality.   In addition to the mREMARK criteria, we recommend the use of ROC analyses and AUC statistics to objectively evaluate biomarker performance (73). Such analyses confer distinct advantages over individual measures of sensitivity and specificity, and certainly over simple t-test statistics calculating significance between case and control biomarker levels. For one, ROC analyses allow for an unbiased assessment of the best cut-off point for biomarker levels (74, 75). Second, the AUC statistic considers both sensitivity and specificity in reporting a test’s discriminative power, without regard to the prevalence of disease in specific populations (74, 76). It further allows an objective comparison of biomarker performance across studies and platforms whereby a biomarker with an AUC >0.85 is considered to have high accuracy while a biomarker with an AUC between 0.7 and 0.85 has borderline potential for clinical translation and warrants further refinement and validation. (12, 77).   Furthermore, even in the 12 studies that reported an ROC analysis (some with remarkable AUC values), caution must be exercised when interpreting the results. The vast majority of these studies performed statistical analyses on discovery cohorts only without reproducing the 31  performance in an external cohort. These studies run the risk of over-fitting their statistical model to the initial discovery cohort, particularly if the discovery cohort is small in size, and often times these initially optimistic findings may perform poorly in a new cohort. In the absence of external cohorts, methods such as cross-validation may provide an additional degree of validation and thus confidence that the results could be replicated in wider use. In this approach, the statistical model is applied to successive data sets in which one or more samples have been removed, and tested on the left out sample(s); the cross-validation AUC is therefore an estimate of the model’s discriminative power in the samples used. Types of cross-validation include leave-one-out cross-validation, k-fold cross-validation, and repeated random-split cross-validation. In leave-pair-out cross validation, a case and its corresponding control are left out in each iteration of the cross-validation (78, 79). In k-fold cross-validation, data are split into k subsets of equal size, with each one serving as the test set and the remaining k-1 subsets serving as the training set in successive iterations. In repeated random-split cross-validation, data are split into test and training sets randomly and repeatedly (79). Given the risk of overly optimistic results with standard AUC statistics, we suggest that all future biomarker studies employ cross-validation techniques, even in studies with available external validation cohorts.  There were a few limitations with our systematic review. First, the decision to use the mREMARK checklist to rank the selected studies in our review was arbitrary, as alluded to previously, but one made based on the fact that there are no alternative ranking methods for biomarker studies in COPD exacerbations. Nevertheless, the ranking scores via the mREMARK guidelines were objective measures and provided some guidance as to how to judge the quality of biomarker studies. Second, given the heterogeneity of biomarkers studied, we were not able to 32  perform a meta-analysis on the results. However, we summarized the top three most studied biomarkers and their respective statistical significance to provide an overview. Last, we recognize there could be some potential publication bias, due to the fact that negative studies tend to be less published than positive studies.  In summary, while we found a number of studies that have evaluated potential candidate biomarkers for AECOPD diagnosis, we also identified a number of deficiencies in the COPD biomarker literature that make it difficult to fully evaluate the performance of these biomarkers. Standardized guidelines such as the mREMARK score and the use of ROC curves may help to streamline biomarker performance reporting, while external validation or at least internal cross-validation techniques may help instil confidence that biomarkers can be translated successfully into the real-world clinical realm.   33  Chapter 3: Temporal relationship of CRP and NT-proBNP in COPD exacerbation Chronic obstructive pulmonary disease (COPD) is a heterogeneous and debilitating disease that affects millions worldwide, and is projected to rank fourth in terms of mortality in 2020 (1). Exacerbations are empirically defined as the worsening of a patient’s symptoms from daily baseline in terms of dyspnea, coughing and sputum production (1). This definition is purely subjective, making it difficult for healthcare providers and researchers to objectively determine exacerbations from patients’ day-to-day variation in symptoms. Hence, there is a need of biomarkers that can guide therapy treatment in COPD exacerbations.  C-reactive protein (CRP) is an acute-phase systemic inflammatory biomarker that is known to be associated with COPD exacerbation (68, 80). Their circulation levels are generally higher in COPD patients compared to healthy controls (81, 82), and can rise to even higher levels during acute exacerbation of chronic obstructive pulmonary disease (AECOPD) (68). CRP levels are associated with all-cause, cardiovascular, and cancer mortality (83).   Brain natriuretic peptide (BNP) has been used clinically to screen and diagnose acute decompensated heart failure (84). The main stimulus for BNP release is cardiac stress in the cardiomyocytes related to volume overload (85). In the context of COPD, BNP levels have been shown to be elevated compared to healthy controls (86) and in AECOPD versus stable COPD patients (27). Amino-terminal of the prohormone brain natriuretic peptide (NT-proBNP) is the inactive fragment that is released in conjunction with BNP (87). NT-proBNP has been 34  investigated in its utility in AECOPD with or without left ventricular dysfunction (88), in AECOPD with respiratory failure (89), and in AECOPD with ischemic heart disease (72).   To our knowledge, CRP and NT-proBNP have not been studied concurrently in the same patient over the full time course of AECOPD. As inflammatory events associated with exacerbation can impact both the pulmonary and cardiac systems, our aim was to investigate the temporal relationship between CRP and NT-proBNP during AECOPD. We hypothesized that CRP and NT-proBNP levels are elevated in AECOPD and can be used to distinguish between exacerbating versus stable COPD patients. Part of this study has been presented in abstract form previously (90).  3.1 Methods 3.1.1 Study subjects  This observational study includes patients recruited into the COPD Rapid Transition Program between July 2012 and April 2015. At the time of writing, the cohort consisted of 368 AECOPD patients who were hospitalized at St Paul’s Hospital and Vancouver General Hospital in Vancouver British Columbia. All patients included in the analysis have a confirmed primary diagnosis of AECOPD as deemed by general internists and pulmonologists who cared for these patients during their hospitalization.  The diagnosis of each patient was then reviewed carefully by two physicians who were not involved in the case and the diagnosis was validated based on chart review. In the cases where the primary diagnoses were not deemed to be AECOPD in the opinion of both external reviewers, the patients were excluded in our analysis. All the patients included in this analysis received standard anti-exacerbation care during their hospitalization, 35  including short-acting bronchodilators, prednisone and antibiotics as necessary. Upon discharge, a team of healthcare professionals subsequently followed the status of patients. In addition to the hospitalized AECOPD cohort, 76 stable COPD patients (different than the patients in the AECOPD cohort) were recruited from the St. Paul’s Hospital COPD clinic and they served as non-exacerbating COPD controls. Demographic data included age, sex, body mass index (BMI), ethnicity, current smoking status, and smoking duration. The study is registered in publicly available ClinicalTrials.gov website with Identifier: NCT02050022 (registered January 28, 2014).  3.1.2 Specimens and measurement technique  Following informed consent, blood samples were collected from patients on day 1 and 3 of hospitalization, once at discharge, and as well as on day 30 and day 90 post-admission date. Blood components were processed as per standardized protocol, and stored in barcode-labelled aliquot tubes in the -80°C freezer until analysis.   Serum CRP was measured via a high-sensitivity assay on the Advia® 1800 Chemistry System analyzer (Siemens Healthcare GmbH, Erlangen, Germany), in the Clinical Laboratory of St Paul’s Hospital (Department of Pathology and Laboratory Medicine, Vancouver, BC). The analytical range of the assay is 0.2 to 200.0 mg/L including an auto-dilution capability on board the analyzer. In cases where the values were greater than the analytical range, the samples were manually diluted by a laboratory technician. The samples were analyzed in an unbiased fashion through blinding of technicians to the patient characteristics.   36  Plasma NT-proBNP was measured from ethylenediaminetetraacetic acid (EDTA) whole blood specimens on the RAMP® diagnostic rapid kit (Response Biomedical Corp, Vancouver, BC, Canada). RAMP assay is based on the principle of quantitative immunochromatography, with a measurement range of 18 to 35,000 ng/L.  Baseline lung function measurements were performed at the time of convalescence (i.e. at day 30 or day 90) for AECOPD patients, and at COPD clinic visits for stable COPD control patients. Spirometry was used to obtain lung function parameters after bronchodilator administration.   3.1.3 Statistical analysis  The study population consisted of an AECOPD group and a COPD stable control group as shown in Table 3.1. Continuous variables that were not normally distributed were natural-logarithmically transformed prior to a Student’s t-test analysis. Categorical dichotomous variables are displayed as counts and were compared using a chi-square test. Analysis of variance (ANOVA) was performed for the time course analysis for the AECOPD patients followed by Bonferroni’s multiple comparison post-hoc analysis for CRP and NT-proBNP. Pair-wise t-test was also used to determine significant changes in the two biomarkers at each time point versus the control group, with p-values adjusted by the Bonferroni method. All statistical analysis was performed using R version 3.1.3 with R Studio version 0.99.441, as well as GraphPad Prism version 5.04 (GraphPad Software Inc., San Diego, CA, USA). P-values of less than 0.05 (on a two-tailed test) were considered significant.   37      Table 3.1 Patient characteristics of the COPD Rapid Transition Cohort.    AECOPD  (n = 386) COPD Stable Controls  (n = 51) ( P-value  Age (years)  67.9±11.8  64.6±10.3  0.0574  Sex (Male %)  63.7%  76.5%  0.1012  BMI (kg/m2)  26.7±7.7  27.2±7.4  0.6158  Height (m)  1.67±0.10  1.70±0.11  0.7216  Ethnicity (White %)  82.6%  92.2%  0.4556  Current Smoker (%)  60.9%  51.0%  0.2534  Smoking Duration (pack-years)  51.7±39.3  46.8±20.4  0.6034  FEV1 (Litres)  1.43±0.57  1.84±0.77  0.0232  FEV1 % predicted  51.8±17.5  58.1±20.8  0.2534  FVC (Litres)  2.74±0.74  3.29±1.10  0.0474  FVC % predicted  76.4±15.2  82.0±20.2  0.3319  FEV1 / FVC ratio  0.53±0.16  0.54±0.14  0.6544  FEV1 / FVC (%)  67.9±21.0  68.4±17.4  0.3382  Continuous variables are presented as mean ± SD. Comparisons made via t-test after natural-log transformation. Dichotomous variable are presented as counts (% total). Comparisons are made via chi-square test. Lung Function measurements were taken from post-bronchodilator stable/convalescent samples (i.e. at Day 30 or Day 90).  38  3.2 Results Demographic data and lung function measurements are shown in Table 3.1. The mean age and sex of the AECOPD group were 67.9±11.8 (mean±SD) and 63.7% men, whereas in COPD control they were 64.6±10.3 and 76.5%, respectively. The majority of the subjects were white Caucasians (82.6% and 92.2%) and over half were current smokers (60.9% and 51.0%). The p-value for age approached statistical significance (0.0574), but otherwise, the rest of the demographic data showed no statistical significance between the two groups.  As for lung functions data, both FEV1 and FVC in litres showed significant difference when compared between AECOPD and COPD control groups (p-value of 0.0232 and 0.0474, respectively). The rest of the lung function comparisons showed no significant difference.  Figure 3.1 CRP time course box-plots. controls. The data are expressed as Tukey boxplots, in which the box represents the 25percentile. The whiskers extend to 1.5 times of the interquartile range on either side of the box, and the outliers plotted separately. The y-axis is displayed on a naturalBonferroni correction.  CRP values for AECOPD patients at five time-points and stable COPD th, the median, and the 75-log scale. The p-values represent the pair39   th -wise t-test with Figure 3.2 NT-proBNP time course boxstable COPD controls. The data are expressed as Tukey boxplots, in which the box represents the 25and the 75th percentile. The whiskers extend to 1.5 times of the interquartile range on either side of the box, and the outliers plotted separately. The y-axis is displayed on a naturalwith Bonferroni correction.   3.2.1 Time course  Box-plots of CRP results are shown in Figure 27.7 [5.9-78.9] mg/L (median [IQR]) and progressively decreased over time. CRP levels are 8.5 [2.2-22.4] mg/L at discharge, and 5.8 [2.13.71e-15, respectively. Within the AECOPD group time points, there -plots. NT-proBNP values for AECOPD patients at five -log scale. The p-values represent the pair3.1. Median CRP levels were highest on day 1 at -14.0] mg/L at day 90 with p-values of 1.00ewas a significant 40   time-points and th, the median, -wise t-test -05 and 41  decreasing trend (p < 2.2e-16). In addition, the day 1 level was significantly higher than the level of the stable controls of 2.8 [1.3-7.1] mg/L (p = 7.91e-12). There was no significant difference between day 30 CRP levels compared to the control group, however, there was a significant difference between day 90 and the control (p = 0.021).  Box-plots of NT-proBNP results are displayed in Figure 3.2. NT-proBNP median levels were highest on day 1 with 367 [168-1189] ng/L and significantly decreased over time (187 [69-544] ng/L at discharge and 115 [46-309] at day 90; p = 0.043 and 5.81e-13, respectively). Within the AECOPD group time points, there was a significant decreasing trend (p < 2.2e-16), although there was no difference between day 1 and day 3 (p = 1.000). The day 1 level was also significantly greater than that of the control group (85 [28-230] ng/L; p = 7.91e-12). There was no significant difference between day 30 and day 90 compared to the control group NT-proBNP levels.  3.3 Discussion In this study, we demonstrated the following key findings: 1) CRP and NT-proBNP were both elevated during AECOPD and decreased during treatment and recovery. 2) Levels of CRP and NT-proBNP of AECOPD patients were significantly higher than those in our stable COPD controls. 3) By day 30 post-hospitalization when patients have returned home, there were no significant differences in CRP and NT-proBNP levels between the exacerbators and stable COPD controls. To our knowledge, this is the first study to have examined CRP and NT-proBNP congruently at five time points for AECOPD patients.   Our finding of elevated levels of CRP during exacerbation is consistent with published literature (16, 68, 80, 82, 91). Given the role of CRP as an acute-phase protein, it tracks well with 42  treatment progression. As shown in Figure 3.1, the box-plots indicating temporal response demonstrated a dramatic decline in plasma CRP levels from day 1 to discharge followed by a relatively horizontal slope. The decline in CRP levels represents approximately 4.5 fold difference between exacerbation onset versus recovery. The exact mechanism for this observation is not fully known. Corticosteroid use in AECOPD treatment has been shown to reduce CRP levels (92). The CRP reduction may also reflect removal of the inciting event (e.g. viral or bacterial infection). Interestingly, CRP levels at day 90 were significantly higher than those of stable COPD controls. One possible explanation could be due to the cessation of systemic corticosteroids, which may have led to a small “rebound” effect on CRP (82, 92).   The elevation of NT-proBNP in AECOPD is also consistent with previous published literature (89). The box-plots in Figure 3.2 showed a significant decreasing trend from day 1 to discharge (approximately 3-fold change), while the median levels between day 1 and 3 were not significantly different. This could be attributed to the fact that NT-proBNP has a longer half-life in blood compared with BNP (half-life of 120 versus 20 minutes) (87). BNP and NT-proBNP are released into circulation in a 1:1 ratio (87). BNP concentrations have been reported to be elevated in patients with renal failure, independent of heart failure status (93). Part of the clearance method is through passive excretion, in which the glomerular filtration rate (GFR) is inversely related to BNP/NT-proBNP concentrations (94). In regards to our cohort, COPD patients with kidney disease were excluded from our analysis, and the elevated values seen on Figure 3.2 were unlikely due to renal impairment.  43  Interestingly, a portion of our patients have NT-proBNP levels had mildly elevated levels (i.e. 367-1189 ng/L) during day 1 and day 3. Our finding of mild elevation in NT-proBNP in our cohort may suggest that AECOPD patients experienced mild cardiac decompensation. Studies have shown that a small proportion of AECOPD patients with high BNP levels have diastolic and systolic dysfunction (27). Another study reported elevations in both NT-proBNP and troponin-T levels in AECOPD patients admitted to the intensive care unit (ICU) with left ventricular dysfunction (88). It has been reported that the prevalence of COPD in chronic heart failure patients is approximately 20-26% (95, 96). In our cohort, the prevalence of heart failure is 5.2% (18 patients), with a median NT-proBNP value of 3603 [1092 – 8235] ng/L (median [IQR]). These patients had NT-proBNP concentrations above the 75th percentile of the boxplot during exacerbation onset in Figure 3.2. Although the prevalence of 5% is lower than 20% as reported in larger cohort studies, it is plausible that some patients might have been previously undiagnosed for heart failure. In addition, there was little evidence of these patients having cardiac injury, as the measurement for cardiac troponin I, which was mostly negative (see Figure 5.1). One possible mechanism is that our AECOPD patients developed cardiac distress secondary to exacerbation. It has been known that exacerbations in COPD patients can lead to lung hyperinflation (97), which in turn, could impact the heart as a response and thus resulting in the NT-proBNP release. Another possibility is that lung inflammation during AECOPD may spill over into the systemic circulation, causing myocardial stress and dysfunction, which in turn may lead to elevated NT-proBNP levels. In a murine model, acute instillation of lipopolysaccharide (LPS) directly into lungs induced a transient state of cardiac dysfunction (with reduced cardiac output). Interestingly, inhibition of interleukin-6 (IL-6) prevented cardiac dysfunction in this model, suggesting that lung and systemic inflammation plays a critical role in the pathogenesis 44  of myocardial dysfunction related to acute lung injury. While there is a scarcity of human data, it is conceivable that the inflammatory response during AECOPD may be an important driver of (subclinical) myocardial dysfunction (leading to mild elevations in NT-proBNP).   The strength of this study is in the time course of CRP and NT-proBNP blood levels, spanning a total duration of approximately 90 days (or 3 months) with a moderate sample size of 368 AECOPD patients. Although our time course contains 5 time points similar to studies published from other groups (72, 98), our design covered a wider duration of time points beyond 8 weeks. Our data showed that these two biomarkers are modifiable and responded well to the progression of AECOPD treatment, and therefore, they can be used potentially as end-points in future AECOPD biomarker studies.  Our study has a few limitations. First, we lacked “true” baseline values of the two biomarkers prior to exacerbation for our cohort. This was due to the nature of our recruitment design, in which we enrolled patients when they experienced a full-blown COPD exacerbation that required hospitalization. We used convalescent samples from day 30 or day 90 as an alternate representation of baseline for our patients. Second, we made the assumption that the day 1 samples from our cohort were taken at the peak of exacerbation. However, this may not have been the case of all patients and it was unclear how long the patients waited prior to seeking medical attention. This could have contributed to the overall variability in our biomarker measurements.   45  In summary, both CRP and NT-proBNP blood levels are significantly elevated during AECOPD and decreased with treatment and in recovery. The change in the magnitude of both biomarkers could offer a more objective measure of AECOPD progression. Elevated levels in NT-proBNP may suggest transient decompensation of left and/or right ventricles. These two proteins are promising biomarkers for diagnosing and tracking AECOPD progression. 46  Chapter 4: Diagnostic performance of CRP and NT-proBNP in AECOPD Forced expiratory volume in the first second (FEV1) have been traditionally used in the diagnosis and management of COPD patients (99). However, FEV1 alone does not account for the variability and complexity seen in COPD patient phenotypes (1). There is a critical need of biomarkers that can guide therapeutic management especially during exacerbations.  In Chapter 3, we showed that clinically available laboratory tests such as CRP and NT-proBNP track well with AECOPD progression of successful treatment. In this chapter, we will extend these findings by addressing 3 specific aims: First, we will determine the discriminatory power of these two biomarkers singly and in combination in the context of AECOPD. Second, we will determine the relationship of these two biomarkers to outcomes including mortality and length of hospitalization. Third, we will replicate the findings of our models in an external cohort.  4.1 Methods 4.1.1 Study subjects We used data from the COPD Rapid Transition Program. The cohort has been described in detail in the previous chapter. In brief, the cohort consisted of 368 COPD patients who were hospitalized with a physician diagnosis of an exacerbation in the emergency department. The patients were given standard anti-exacerbation therapies including prednisone and/or antibiotics. In addition to this group, another group of 76 stable COPD patients were recruited from the St. Paul’s Hospital COPD clinic and they served as the non-exacerbating COPD controls for the exacerbation group. Demographics data were available for these patients included age, sex, BMI, 47  ethnicity, current smoking status, and smoking duration. The study is registered with ClinicalTrials.gov with an Identifier: NCT02050022 (registered January 28, 2014).  4.1.2 Specimens and measurement technique Specimen collection and storage condition have been described in detail in the previous chapter. All patients provided written informed consent to participate in the study.   Serum CRP was measured via a high-sensitivity assay on the Advia® 1800 Chemistry System analyzer (Siemens Healthcare GmbH, Erlangen, Germany), located in the Clinical Laboratory of St Paul’s Hospital (Department of Pathology and Laboratory Medicine, Vancouver, BC). The analytical range of the assay is 0.2 to 200.0 mg/L including an auto-dilution capability on board the analyzer. In cases where the samples were over the analytical range, they were manually diluted.  Plasma NT-proBNP was measured on whole blood specimens collected in EDTA tubes using the RAMP® diagnostic rapid kit (Response Biomedical Corp, Vancouver, BC, Canada). RAMP assay uses quantitative immunochromatography and has a measurement range of 18 to 35,000 ng/L.  Baseline lung function measurements were performed at the time of convalescence (i.e. at day 30 or day 90) for AECOPD patients, and at the COPD clinic visits for stable COPD control patients. Spirometry was used to obtain lung function parameters after bronchodilator administration.   48  4.1.3 Statistical analysis The study population consisted of AECOPD group and COPD stable control group as described in the previous chapter. Continuous variables that are not normally distributed were natural-logarithmically transformed prior to Student’s t-test analysis. Categorical dichotomous variables are displayed as counts and compared by chi-square test.   Receiver-operating characteristic (ROC) curves were generated based on logistic regression model for diagnosis, with exacerbation (i.e. day 1 or day 3) and convalescent samples (i.e. day 30 or 90) used as a continuous binary classifier ranging from 0 to 1. We compared the area under the curve (AUC) of 4 ROC models that included: 1) a leave-one-out cross-validation (LOOCV) model that included both CRP and NT-proBNP, 2) a linear combination of both biomarkers, 3) NT-proBNP alone, and 4) CRP alone. In the leave-one-out cross-validation model, the algorithm cycled through the complete dataset, and a sample was excluded systematically (total n – 1) with each iteration. The logistic regression were subsequently computed based on the total n – 1 samples, and allowed to predict on the one sample that was left out. For example, we have a total sample size of n = 444 (the cross-validation algorithm would loop over 444 iterations), and in each iteration, a logistic regression model was built on the remainder 443 samples. At the end, a collection of 444 binary probabilities ranging from 0 to 1 from each iteration was compiled and then used in ROC curve analysis computation. This provided a method of internal validation. Both LOOCV and the linear combination model of CRP and NT-proBNP employed the formula: z = a*x + b*y + c, where x and y represents CRP and NT-proBNP, respectively, a and b are the slopes, and c is the intercept. Sensitivity was computed according to the formula: Sensitivity = True Positive / (True Positive + False Negative). Specificity was calculated as follows: 49  Specificity = True Negative / (True Negative + False Positive). Binomial exact test was used to compute the 95% confidence interval for the AUC in the ROC analysis (see Table 4.1). To determine the “optimal” cut-off value for the biomarker analysis, we calculated the Youden’s index. Youden’s index was defined as J = maximum (sensitivity + specificity – 1) across the full range of all data points on the ROC analysis.  DeLong’s method was used to generate the p-values comparing the various ROC curves (see Table 4.2).  We then performed a Kaplan-Meier (K-M) survival analysis (n = 255) based on whether the patients died or survived. The mortality data were obtained by the clinical coordinators at both St Paul’s Hospital and Vancouver General Hospital. We arbitrarily had set our survival analysis at the 1-year time point. At the time of writing, only 255 of the 368 patients had reached the 1-year time point. Subsequently, the optimal cut-off point of the biomarkers based on the Youden’s index is shown in Figures 4.2 through Figure 4.5. The figures are displayed as cumulative mortality on the y-axis. The p-value for the K-M analysis was generated using a log-rank test (see Table 4.3). The multiple linear regression analysis was used to model length of hospitalization versus the two biomarkers, and was adjusted for age, sex and current smoking status. The values used in this portion of the analysis for both biomarkers were from the first onset sample (i.e. Day 1 or 3) for the exacerbating patients.  To replicate our models, we measured CRP and NT-proBNP in the Zileuton to treat adults with chronic obstructive pulmonary disease cohort (the LEUKO Study). The details of the LEUKO Study are published elsewhere; ClinicalTrials.gov identifier: NCT00493974 (100). The LEUKO Study was ended prematurely due to low recruitment rate. It consisted of a total of 119 patients 50  in which 59 were subjected to placebo and 60 were administered zileuton. Out of the 119 LEUKO patients that had blood work procured, we obtained a total of 88 samples, which we considered as our external cohort for replication. These 88 samples consisted of 52 day 1 exacerbation samples and 36 day 30 post-exacerbation samples. The day 30 samples were treated as convalescent controls for this cohort. Upon applying the 4 logistic regression models to predict on the LEUKO samples (total samples = 88), ROC curves and the associated parameters were generated (see Table 4.5). In Table 4.6, the performance characteristics of each model were compared using DeLong’s method as described above. Prediction accuracy for each of the models was calculated as follows: Accuracy = True Positive + True Negative / (True Positive + True Negative + False Positive + False Negative).  All statistical analysis was performed using R version 3.1.3 with R Studio version 0.99.441, as well as MedCalc version 14.8.1 (MedCalc Software, Ostend, Belgium). P-values less than 0.05 were considered statistically significant.  4.2 Results 4.2.1 ROC curve analysis ROC curves are displayed in Figure 4.1. ROC models were compared across 4 different analyses: 1) LOOCV model based on CRP and NT-proBNP, 2) CRP and NT-proBNP in linear combination, 3) NT-proBNP alone, and 4) CRP alone. AUC for LOOCV model was 0.796, linear combination was 0.844, NT-proBNP alone was 0.796, and CRP alone was 0.784. The corresponding test sensitivity, specificity, and Youden’s index for optimal cut-off for all 4 models are shown in Table 4.1. The optimal cut-off for NT-proBNP and CRP alone based on 51  these analyses were 164 ng/L and 12.3 mg/L. Determining the optimal cut-off for combination biomarker models is complicated. There are numerous combinations of CRP and NT-proBNP values that could exceed the cut-off of 0.544. For example, a combination of 4.9 mg/L in CRP and 1189 ng/L in NT-proBNP would produce an index value of 0.663 over the cut-off point of 0.544 for LOOCV model, and this would be categorized as having an AECOPD. On the other hand, a CRP level of 1.3 mg/L and NT-proBNP level of 70 ng/L would yield a value of 0.381 and therefore, would be considered as stable COPD by the model.    Figure 4.1 Plot of ROC curves on the four models. Receiver operating characteristic (ROC) curve for 1) LOOCV  CRP and NT-proBNP, 2) CRP and NT-proBNP, 3) NT-proBNP alone, and 4) CRP alone. The ROC curve is used in 52  discriminating patients with AECOPD. CRP = C-reactive protein, NT-proBNP = N-terminal of the prohormone brain natriuretic peptide, and LOOCV = Leave-one-out cross validation.  In direct comparisons, the ROC curve AUC differences are listed in Table 4.2. LOOCV model is not significantly different from the other 3 models (vs linear combination p = 0.0985, vs NT-proBNP p = 0.9935, vs CRP p = 0.7134, respectively). Linear combination of both biomarkers produced an AUC of 0.844, which was significantly different than each of the biomarkers alone (vs CRP p = 0.0001 and vs NT-proBNP p = 0.0235, respectively). There was no significant difference in the AUC between NT-proBNP alone and CRP alone curves.  Table 4.1 ROC curve characteristics of the four models.  Variable Sample size (cases : controls) AUC AUC   95% CI Sens Spec Youden’s index P-value of comparison to reference line  (AUC = 0.5) LOOCV  CRP +  NT-proBNP 368:76  0.796  0.761-0.831  0.6250  0.8498  0.544  <0.0001  CRP +  NT-proBNP 368:76 0.844 0.807-0.877 0.7636 0.7763 0.725 <0.0001  NT-proBNP  368:76 0.796 0.741-0.850  0.7799 0.7237 164 <0.0001  CRP 413:76 0.784 0.735-0.834  0.6077 0.8684 12.3 <0.0001  Variables in the ROC curve analysis with performance characteristics. AUC = Area under the curve, CI = confidence interval computed via the binomial exact method, CRP = C-reactive protein, LOOCV = Leave-one-out cross-validation, NT-proBNP = N-terminal prohormone of brain natriuretic peptide, Sens = sensitivity, Spec = specificity, and Youden’s index = Optimal cut-off.     53   Table 4.2 ROC curve AUC comparisons across the four models.  ROC curves AUC1 – AUC2  Absolute AUC difference P-value LOOCV CRP + NT-proBNP vs  CRP + NT-proBNP 0.7962 – 0.8444  0.0482 0.0985  LOOCV CRP + NT-proBNP vs NT-proBNP 0.7957 – 0.7959  0.0002 0.9935  LOOCV CRP + NT-proBNP vs  CRP  0.7957 –  0.7836 0.0121 0.7134  CRP + NT-proBNP vs NT-proBNP  0.8444 – 0.7959  0.0484  0.0235 CRP + NT-proBNP vs CRP 0.8444 – 0.7836  0.0608 0.0001 NT-proBNP vs CRP 0.7959 – 0.7836  0.0123 0.7100 ROC curve comparisons with AUC differences and pair-wise comparisons. AUC = Area under the curve, CRP = C-reactive protein, NT-proBNP = N-terminal prohormone of brain natriuretic peptide, LOOCV = Leave-one-out cross-validation. P-values are computed by pair-wise comparison via the DeLong method.   4.2.2 Kaplan-Meier survival analysis Table 4.3 summarizes the 1 year all-cause mortality rate based on the 4 models from the ROC analysis. When we categorize the cohort mortality rate based on the respective optimal cut-offs, only the NT-proBNP alone model had a significant difference in the curves (34.7% for > 164 ng/L cut-off vs 13.6% for ≤ 164 ng/L cut-off). The LOOCV, linear combination, and CRP alone models did not yield a significant difference in the mortality rate at 1-year (Approximately 33% for > cut-off and 23% for ≤ cut-off). Figures 4.2-4.5 displayed each of the cumulative mortality curves categorized by optimal cut-off point along with the log-rank test p values from Table 4.3. The LOOCV model is examined further with 90-day interval for survival analysis and the results are in Table 4.4. Only the 90-day mortality rate is significantly different between those above 0.544 cut-off value versus those below it.  54  4.2.3 Multiple linear regression analysis Table 4.5 summarizes the variables, β estimates, and p-values from the multiple linear regression models. It is computed with length of hospital stay as the dependent variable and the two biomarkers as the independent variable while adjusted for age, sex, and current smoking status. The formula of the regression is: Length of Stay ~ CRP + NT-proBNP + Age + Sex + Current Smoker. Of the two biomarkers, only NT-proBNP was statistically significant with p-value = 0.0203. The overall adjusted-R2 value for the linear regression model is 0.067. Nevertheless, the relationship between NT-proBNP and length of stay was significant (Figure 4.6).  Table 4.3 Survival curve analysis at 1-year time point of the four models.  Variable Number at risk Observed 1-year Mortality rate (%) 95%  CI P-value LOOCV CRP + NT-proBNP              Combined > 0.544 174 45  31.8% 22.7-39.8% 0.0672          Combined ≤ 0.544 97  15  21.1% 10.1-30.7%  CRP + NT-proBNP              Combined > 0.724 212 51 30.5% 22.2-37.9% 0.134          Combined ≤ 0.724 59 9 20.1% 6.5-31.8%  NT-proBNP              NT-proBNP > 164 211 55 32.9% 24.3-40.4% 0.00207          NT-proBNP ≤ 164 60 5 12.8%  1.1-23.1%  CRP              CRP > 12.3 182 49 27.2% 19.3-34.3% 0.981         CRP ≤ 12.3 111 31 29.2% 18.1-38.9%  Survival Curve Analysis of four models categorized based on the optimal cut-off point. CI = confidence interval, CRP = C-reactive protein, NT-proBNP = N-terminal prohormone of brain natriuretic peptide, LOOCV = Leave-one-out cross-validation. P-values are computed by log-rank tests.   55   Table 4.4 Survival curve analysis of LOOCV model at various time points   0-day  90-day  180-day  270-day  1-year  LOOCV CRP + NT-proBNP       Combined > 0.544 Number at risk  174  137  92  61  46                                 Number of events 0 29 32 41  45  Combined ≤ 0.544 Number at risk  97 88  50  32  23                                 Number of events  0  6  12  14  15  P-value  N/A  0.0152  0.174  0.0845  0.0672  Number at risk and events for five particular time points (0-day, 90-day, 180-day, 270-day, and 1-year) of the LOOCV CRP + NT-proBNP model. P-values are computed by log-rank tests.   Figure 4.2 1-year Survival curve analysis for LOOCV model. Cumulative all-cause mortality rate over 1-year period stratified according to the optimal cut-off point for LOOCV model including both CRP and NT-proBNP 56  levels. The p value is based on the log-rank test on whether there is a significant difference between the curves (n = 271).   Figure 4.3 1-year Survival curve analysis for the linear combination model. Cumulative all-cause mortality rate over 1-year period stratified according to the optimal cut-off point for CRP and NT-proBNP levels. The p value is based on the log-rank test on whether there is a significant difference between the curves (n = 271).  4.2.4 Replication on external dataset  The ROC models from the above were validated in the LEUKO dataset, in which the models were used to distinguish between exacerbating patients versus those in stable states. The performance and associated parameters are shown in Table 4.6. The LOOCV model when 57  applied to the LEUKO dataset, gave the best accuracy at 0.614 (accuracy calculation described in methods 4.1.3) and with the highest AUC at 0.642. Table 4.7 consisted of all the pair-wise comparisons between the AUCs and all comparisons were not statistically significant.   Figure 4.4 1-year Survival curve analysis for NT-proBNP model. Cumulative all-cause mortality rate over 1-year period stratified according to the optimal cut-off point for NT-proBNP levels. The p value is based on the log-rank test on whether there is a significant difference between the curves (n = 271). NT-proBNP = N-terminal prohormone of brain natriuretic peptide. 58   Figure 4.5 1-year Survival curve analysis for CRP model. Cumulative all-cause mortality rate over 1-year period stratified according to the optimal cut-off point for CRP levels. The p value is based on the log-rank test on whether there is a significant difference between the curves (n = 293). CRP = C-reactive protein.   59  Table 4.5 Multiple linear regression analysis in modeling length of hospitalization  Variable β  Estimate  Standard Error 95% CI for β  P-value CRP 0.040560  0.028022 -0.014363-0.095482  0.1492 NT-proBNP 0.070437 0.030129 0.011385-0.129489  0.0203 Age 0.007820  0.004427 -0.000857-0.016497  0.0787 Sex 0.035934  0.094574 -0.149428-0.221296 0.7043 Current Smoker -0.114751  0.104638 -0.319837-0.090335  0.2740  The CRP and NT-proBNP values are natural-log transformed (n = 224).   Figure 4.6 Length of hospital stay versus NT-proBNP concentration. NT-proBNP and length of stay are both plotted on natural-log scale. The linear relationship is significant (p = 0.0203) from multiple linear regression analysis. The shaded area represents the 95% confidence interval for the regression line. The model has an adjusted R-squared value of 0.067 based on a sample size of 224 patients. The model was adjusted for age, gender and smoking status.   60  Table 4.6 Performance of the 4 models on the LEUKO dataset. Performance on LEUKO  (n = 52 cases: 36 controls) Accuracy Accuracy  95% CI AUC AUC  95% CI Sens  Spec LOOCV  CRP + NT-proBNP 0.614 0.504-0.716  0.642 0.523-0.761  0.5577  0.6944  CRP + NT-proBNP 0.568  0.458-0.673  0.640 0.519-0.761 0.4808 0.7778 NT-proBNP  0.580 0.470-0.684  0.625 0.503-0.748  0.4423  0.7778  CRP 0.602 0.492-0.705  0.607  0.488-0.726  0.4423  0.8333  Accuracy is calculated by True Positive + True Negative / (True Positive + True Negative + False Positive + False Negative). AUC = Area under the curve, CI = confidence interval computed via the binomial exact method, CRP = C-reactive protein, NT-proBNP = N-terminal prohormone of brain natriuretic peptide, LOOCV = Leave-one-out cross-validation, Sens = sensitivity, and Spec = specificity.  Table 4.7 ROC curve comparisons from the performance on LEUKO cohort. Performance comparisons of ROC curves on LEUKO cohort AUC1 – AUC2  Absolute AUC difference P-value LOOCV CRP + NT-proBNP vs  CRP + NT-proBNP 0.6416 – 0.6400  0.0016 0.9557  LOOCV CRP + NT-proBNP vs  NT-proBNP 0.6416 – 0.6253  0.0163 0.6552 LOOCV CRP + NT-proBNP vs  CRP  0.6416 – 0.6074  0.0342 0.3802  CRP + NT-proBNP vs  NT-proBNP  0.6400 – 0.6253  0.0147  0.2188 CRP + NT-proBNP vs  CRP 0.6400 – 0.6074  0.0326 0.5770  NT-proBNP vs  CRP 0.6253 – 0.6074   0.0179 0.7775 AUC = Area under the curve, CRP = C-reactive protein, NT-proBNP = N-terminal prohormone of brain natriuretic peptide, and LOOCV = Leave-one-out cross-validation. P-values are computed by pair-wise comparison via the DeLong method.  61   4.3 Discussion In this study, we demonstrated the following findings: 1) Using LOOCV model of CRP and NT-proBNP, we obtained comparable ROC AUCs to non-cross-validated, linear combination of the two biomarkers (LOOCV AUC 0.796 vs linear combination AUC 0.844, p = 0.0985). 2) Only the LOOCV at 90-days and NT-proBNP alone models produced survival curves that were statistically significant, when we categorized the cohort by their optimal cut-off point. 3) Higher circulating NT-proBNP levels at the exacerbation onset were associated with longer length of hospital stay. 4) Our logistic regression models were validated in an external cohort with reproducible AUCs that were comparable statistically, providing support for the combinatorial use of the two biomarkers in AECOPD diagnosis and treatment monitoring.  To our knowledge, this is the first study to examine the combinatorial utility of both CRP and NT-proBNP in differentiating patients who were experiencing acute hospitalization for COPD from stable patients. Our ROC curve analysis from the LOOCV model suggested that the combinatorial usage of the two biomarkers agreed with the physician diagnosis approximately 80% of the time. As suggested in previous studies, an AUC of 0.80 is considered “good” in terms of discriminatory performance (12, 68, 77). By utilizing LOOCV methodology, we mitigated the risk of over-fitting.   Our findings are consistent with smaller, published studies. Previous studies have evaluated CRP either alone or in conjunction with other biomarkers in the diagnosis of AECOPD (16, 68, 101). One of these studies by Gumus et al. examined 43 AECOPD patients admitted to the hospital 62  and reported an AUC of 0.695 using day 1 and day 7 CRP levels (16). The other study by Helmy et al. reported the usefulness of CRP in combination with IL-6 levels to diagnose AECOPD patients admitted into ICU, and on their 28-day mortality (101). CRP and IL-6 together in these patients produced an AUC of 0.851 for the 28 day mortality.  Finally, Hurst et al. reported an AUC of 0.73 for CRP alone, and with the addition of a major symptom, they improved the AUC to 0.88 (68). Here in our study, we obtained an AUC of 0.784 for CRP alone, and an AUC of 0.796 for the LOOCV model even without the addition of symptoms. These results are comparable to those reported from Hurst et al., with the advantage of being cross-validated and without the use of symptoms in AECOPD diagnosis.  Secondly, main findings from the survival curve analysis suggested that NT-proBNP level at exacerbation onset is associated with worse outcomes. We see a significant proportion of all-cause mortality when we stratified the cohort by the optimal cut-off on the patients’ first NT-proBNP level (Figure 4.4). This elevation might suggest that our patients have experienced cardiac stress in addition to pulmonary stress. Other studies have reported higher BNP or NT-proBNP levels correlate with mortality in COPD exacerbations (47, 102, 103). Many factors could cause an increase in BNP and/or NT-proBNP levels (104). For example, pulmonary hypertension, pulmonary embolism, liver cirrhosis, and renal failure could elicit a rise in BNP/NT-proBNP levels (104). It is speculated that the mechanism of BNP/NT-proBNP release seemed to involve more than simple cardiac stretch. The patients in our cohort might have underlying co-morbidities that are not yet diagnosed. Nonetheless, we demonstrated that NT-proBNP is useful in the context of diagnosing and tracking AECOPD treatment progression.  63  Interestingly, when we added CRP into the models, namely in LOOCV and linear combination, we did not detect a significant difference in all-cause mortality (Figure 4.2 & 4.3). CRP is a well characterized systemic, non-specific inflammatory biomarker, and is previously reported to be associated with mortality in mild to moderate COPD patients (83). In the cases of in-hospital treatment success versus failure for AECOPD patients, Crisafulli et al. reported that those whom had treatment failure had higher CRP values both at exacerbation onset and day 3, versus those that had been successfully treated (91). In regards to our finding, one interpretation of the survival curves seen in Figure 4.2 & 4.3 might be that we have indeed, treated these AECOPD patients adequately during their hospital stay. Hence, no statistical difference was seen when we dichotomized the cohort as per model cut-off at 1-year point. Based on the time course figures from the previous chapter, those that have initially high levels of both biomarkers eventually got better just as well as those that have initially low levels below the cut-off. Further note that the majority of our patients did not experience in-hospitalization death; the majority survived hospitalization and were discharged, then the mortality incidents arose within 1-year follow up. On the other hand, another interpretation could be that we simply did not have enough sample size to reach a significant statistical level, as the curves do show complete separation trend visually. This could also explain that at 90-day time-point the LOOCV model gave a significant difference, whereas the rest of the time points did not (See Table 4.4). Nevertheless, the specific causes of death in these patients warrant further investigation, and perhaps certain commonality amongst those that died could be elucidated.  Thirdly, higher initial NT-proBNP result at exacerbation onset is associated with longer length of hospital stay (Figure 4.6). This finding is consistent as the patients with higher values are most 64  likely sicker and requiring more extensive medical intervention and thus, prolonging their stay in the hospital. Our multiple linear regression model has an R2 value of 0.067, which indicated that our model explained only 6.7% of total variation seen in our data. Despite the wide variability, the linear relationship between the predictor NT-proBNP and length of stay remained significant. Interestingly, NT-proBNP cut-off level of 164 ng/L is considered only mild elevation. For patients of all ages, a NT-proBNP value of less than 300 ng/L is used to rule out acute congestive heart failure (105). A level surpassing 1000 ng/L would be considered more indicative of heart failure. Here from our data, we showed a lower threshold level that is approximately half of 300 ng/L, and suggested that the threshold to diagnose AECOPD is less than those typically used for heart failure guidelines. Interestingly, there have been reports of elevated levels of BNP without heart failure, indicating potential usage of subclinical BNP levels beyond the realm of cardiovascular diseases (106, 107). It might be applicable in the future to stratify the lower range of NT-proBNP for the context of AECOPD, in a similar approach as the high-sensitivity CRP was done for cardiovascular risks currently implemented for heart-related diseases (108).  Lastly, our logistic regression models were validated in LEUKO cohort with reproducible AUCs. The LEUKO cohort consists of patients that are less severe in symptoms and sickness as our own Rapid Transition cohort, which we recruited patients that were exacerbating severe enough for hospitalization. This could explain for the overall decrease in AUCs seen in Table 4.6. The LEUKO cohort consisted a total of 119 patients spanning two groups: 60 treatment with Zileuton versus the 59 placebo (100). Each patient have two paired samples (at day 1 and day 30) but have a smaller fold change than our Rapid Transition cohort (1.5 times fold change in LEUKO vs 3.3 fold change in Rapid Transition, p < 0.05 for both CRP and NT-proBNP). The median 65  values of biomarkers at the onset are relatively lower as well (LEUKO vs Rapid Transition: CRP 10.9 vs 27.7 mg/L and NT-proBNP 230 vs 367 ng/L, respectively). The LEUKO cohort also have relatively shorter length of stay (LEUKO: approximately 3-4 days vs Rapid transition: 6 days) and with substantially less death (LEUKO: n = 3) during one month follow-up. In addition, the LEUKO patients were recruited at twelve hospitals with diagnoses of AECOPD assessed by numerous physicians. There may have been variations between physicians in defining and determining a patient as having an AECOPD, and as well as whether the patient has recovered or not. This multi-centred nature of the LEUKO study might have influenced the variability on AECOPD diagnoses seen in the cohort, and subsequently affected the accuracy of our models indicated on Table 4.6. To truly replicate our models and findings, ideally we would be applying our models to an external cohort with similar recruitment criteria and severity of symptoms as our own. Despite this fact, the LOOCV model still produced the best accuracy and the highest AUC (see Table 4.6).   There are few limitations to our study. The ROC optimal cut-off for the combinatorial models cannot be easily interpreted as concentrations. One would need to compute the biomarker concentrations into the logistic formula and solve for the response variable to determine whether it is higher or lower than the cut-off. Although the calculations can be done computationally in a program (such as R), it is relatively less straight forward. Secondly, due to the nature of clinical studies, our demographical data for certain variables such as BMI are limited. Henceforth, these variables with high missing rate are not included in our multiple linear regression analysis. This decision perhaps could account for the small percentage of total variation (6.7%) seen in our linear regression model. Thirdly, we used either day 1 or day 3 samples in our analysis 66  collectively. Not all patients have samples obtained on day 1 of hospital admission due to the nature of the enrolment process. Although samples were collected as soon as possible, the first samples for selected patients were between day 2 and 3. This may be another factor contributing to the variation seen in our analysis.  The strength of our study is that we explored various linear combinations of CRP and NT-proBNP logistic regression models on a medium-sized and well-characterized COPD exacerbation cohort. We determined that the LOOCV linear combinatorial model provided substantial discriminatory power to differentiate exacerbation versus stable COPD patients, and that the model is replicated in an external LEUKO dataset with the best accuracy. The combinatorial method is not a new concept, and has been utilized in other diseases (109). Here in our study, we proposed the use of a systemic inflammatory biomarker along with a cardiac stress biomarker as the starting point to diagnose COPD exacerbations. This combinatorial approach could potentially be expanded to more biomarkers in the future, and then the new additions could be assessed on whether they add any significant incremental value.  In summary, a cross-validated combination of CRP and NT-proBNP could almost completely separate patients who were experiencing AECOPD that required hospitalization from stable patients. A combinatorial panel of biomarkers consisting of CRP and NT-proBNP is a promising blood test to diagnose COPD exacerbations.   67  Chapter 5: Relationship of other biomarkers in COPD exacerbation Systemic inflammation and cardiac dysfunction are often associated with AECOPD (72, 80). In this chapter, we determined the temporal relationship and the discriminatory power of 4 biomarkers: cardiac troponin I, pulmonary and activation-regulated chemokine or C-C motif ligand 18 (PARC/CCL18), D-Dimer, and myeloperoxidase (MPO) to AECOPD.   Cardiac troponin T and I have been used in the diagnosis of cardiac injury, such as myocardial infarction since 1990’s (110) due to their great specificity for myocardial necrosis (111). The impact of cardiac dysfunction on AECOPD or vice versa is not completely understood, however, those with cardiovascular diseases have poorer outcome in COPD exacerbations (112). There are studies that reported that elevation in cardiac troponin T is associated with AECOPD severity (103, 113). Although troponin I shares similar specificity for myocardial necrosis as troponin T, diagnostic information on one cannot be directly inferred to the other. A majority of the studies published in relation to COPD exacerbation have focused on troponin T (102, 103, 114). However, one of the disadvantages of using troponin T is that it appeared to be falsely elevated for patients with renal failure without acute myocardial injury (115, 116). To avoid potential issues with troponin T, we alternatively sought to determine the levels of troponin I.   Second biomarker of interest is PARC/CCL18, which is a chemokine mainly expressed in the lungs by monocytes/macrophages and dendritic cells (117). PARC has been shown to be associated with cardiovascular hospitalization and mortality for COPD patients in large cohort studies (118), as well as in a panel of 36 inflammatory biomarkers, which compared exacerbation versus baseline (68). Although the role of PARC and its receptors have not been elucidated, it is 68  a potential biomarker for COPD exacerbation given its specificity for lungs. We aimed to explore its relationship to AECOPD, with more time points in a medium-sized well documented cohort.  Third biomarker of interest is D-Dimer, which is a degradation product of fibrin clot. D-Dimer has clinical utility in ruling out pulmonary embolism (PE) and deep vein thrombosis in an acute clinical setting (119). Pulmonary embolism has been implicated as a possible precipitant of COPD exacerbations (120). It is estimated that approximately 25% of COPD patients hospitalized with unexplained exacerbation might have pulmonary embolism (120). Moreover, even in the absence of an acute pulmonary embolism, D-dimer may be elevated owing to activation of the inflammatory and clotting cascades during AECOPD. We thus sought to determine whether D-Dimer levels are associated with AECOPD.  Lastly, we evaluated MPO concentrations in the context of AECOPD. MPO is a well-characterized enzyme that is stored mainly in the azurophilic granules of neutrophils and monocytes/macrophages, and has been implicated in bacterial infections leading to adverse cardiovascular events (121, 122). Given its role in inflammation and oxidative stress, MPO has been shown to have utility in stratifying patients with acute heart failure (123) and lung function decline in COPD patients (122). Serum MPO has been shown to be elevated in exacerbating patient versus stable COPD patients and healthy controls (71). Furthermore, sputum MPO time course has been published where it was shown to track well with COPD exacerbation progression (17). Therefore, we sought to determine whether similar patterns could be detected in blood samples. 69   Our overarching hypothesis for this chapter was that these 4 biomarkers are elevated in AECOPD, and can be used to distinguish exacerbation versus stable patients.  5.1 Methods 5.1.1 Study subjects We used data from the COPD Rapid Transition Program. The cohort has been described in detail in chapter 3.   5.1.2 Specimens and measurement technique Specimen collection and storage condition have been described in detail in chapter 3. All patients have given informed consent to participate in the study.   Baseline lung function measurements were performed at the time of convalescence (i.e. at day 30 or day 90) for AECOPD patients, and at COPD clinic visits for stable COPD control patients. Spirometry was used to obtain lung function parameters after bronchodilator administration.   Plasma cardiac troponin I was measured from EDTA whole blood specimens using the RAMP® diagnostic rapid kit (Response Biomedical Corp, Vancouver, BC, Canada). The troponin assay has a measurement range of 0.10 to 32 µg/L.  Blood D-Dimer was also measured from EDTA whole blood specimens on the RAMP® diagnostic rapid kit (Response Biomedical Corp, Vancouver, BC, Canada). The kit has a 70  measurement range of 100 to 5000 µg/L FEU (Fibrinogen equivalent unit). Results below the limit of detection (LOD) of 100 µg/L FEU were reported as one-half of LOD at 50 µg/L FEU.  PARC was measured from the plasma of EDTA specimens via enzyme-linked immunosorbent assay (ELISA) manufactured by R & D Systems Inc. Minneapolis, MN, USA. The assay has a measurement range of 3 to 300 ng/mL. Only a small proportion of samples were over-the-range, and these were extrapolated from the calibration standard curve, which was linear. Internal quality control (QC) specimens were run on all ELISA plates to ensure reliable results. The coefficient of variation (CV) for PARC ELISA kits was 4.1%.  Serum MPO was measured using an ELISA kit from R & D Systems Inc. The analytical range of MPO kits was from 7.8 to 500 ng/mL. As the calibration standard curve was non-linear near the upper range, all samples that were over-the-range were diluted appropriately to fall within the measurement range. QC samples were included in all ELISA plates to monitor plate-to-plate variation. The overall CV for MPO kits was 4.1%.  5.1.3 Statistical analysis The study population consisted of AECOPD group and COPD stable control group as reported in the previous chapter. Continuous variables that were not normally distributed were natural-logarithmically transformed prior to Student’s t-test analysis. Categorical dichotomous variables are displayed as counts and compared by chi-square test.   71  Analysis of variance (ANOVA) was performed within all five time points for AECOPD patients followed by Bonferroni’s post-hoc analysis correction for multiple comparisons, where appropriate. A pair-wise t-test was also used to determine significant changes in the biomarkers at each time point versus the control group. The Receiver-operating characteristic (ROC) curves were generated based on logistic regression model, with exacerbation (i.e. day 1 or day 3) and convalescent (i.e. day 30 or day 90) samples used as continuous binary classifiers ranging from 0 to 1. We compared the area under the curve (AUC) of 4 ROC models: 1) Troponin I, 2) PARC, 3) D-Dimer and 4) MPO. Sensitivity is computed according to the formula: Sensitivity = True Positive / (True Positive + False Negative). On the other hand, specificity is calculated as follows: Specificity = True Negative / (True Negative + False Positive). Binomial exact test is used to compute for the 95% confidence interval for the AUC in ROC analysis in Table 5.1. The Youden’s index was defined as J = max (sensitivity + specificity – 1), where it finds a criterion value over the range of all points that maximizes both sensitivity and specificity on the ROC curve. DeLong’s method was used to compute the p-values in comparing pairs of ROC curves in Table 5.2.  All statistical analysis was performed using R version 3.1.3 with R Studio version 0.99.441, as well as MedCalc version 14.8.1 (MedCalc Software, Ostend, Belgium). P-values of less than 0.05 were considered significant.  72   Figure 5.1 Troponin I time course box-plots. The concentration of troponin I for AECOPD patients at five time-points and stable COPD controls. The data are expressed as Tukey boxplots, in which the box represents the 25th, the median, and the 75th percentile. The whiskers extend to 1.5 times of the interquartile range on either side of the box, and the outliers plotted separately. The y-axis is displayed on a natural-log scale. The p-values represent the pair-wise t-test with Bonferroni correction. NS = Not significant. 73   Figure 5.2 PARC time course box-plots. The concentration of PARC values for AECOPD patients at five time-points and stable COPD controls. The data are expressed as Tukey boxplots, in which the box represents the 25th, the median, and the 75th percentile. The whiskers extend to 1.5 times of the interquartile range on either side of the box, and the outliers plotted separately. The y-axis is displayed on a natural-log scale. The p-values represent the pair-wise t-test with Bonferroni correction. PARC = Pulmonary and activation-regulated chemokine, NS = Not significant.   5.2 Results 5.2.1 Time course Boxplots of troponin I were shown in Figure 5.1. The median at all time points compared to the stable control were non-significant 0.00 [0.00-0.00] µg/L (median [IQR]). The boxplots of 25th to 75th percentiles were both 0.00 µg/L, with few outliers of positive troponin I results (> 0.03 µg/L). Within the AECOPD group, there is no significant difference in trend as well.Figure 5.3 D-Dimer time course box-plots. points and stable COPD controls. The data are expressed as Tukey boxplots, in which the box represents the 25median, and the 75th percentile. The whiskers exteand the outliers plotted separately. The ywise t-test with Bonferroni correction. NS = Not significant.  The boxplots for PARC results are shown in Figure points (62.4, 55.8, 60.3, 60.7, and 70.9 ng/mLThe concentration of D-Dimer for AECOPD patients at five timend to 1.5 times of the interquartile range on either side of the box, -axis is displayed on a natural-log scale. The p-values represent the pair 5.2. Similarly, the median at all five time , day 1 to day 90 respectively) compared to the 74    -th, the -stable control (49.5 ng/mL) were nonsignificant trend observed. Figure 5.4 MPO time course box-plots. stable COPD controls. The data are expressed as Tukey boxplots, in which the box represents the 25and the 75th percentile. The whiskers extend to 1.5 times of the interquartile range on either side of the box, and the outliers plotted separately. The y-axis is with Bonferroni correction. MPO = myeloperoxidase, NS = Not significant. Figure 5.3 shows the boxplot results for Dmedians are significantly different than those of stable control (day 3: 560 [312.5discharge: 583 [347.8-872.0] versus stable: 364.0 [220.0-significant. Within the AECOPD group, there was no The concentration of MPO for AECOPD patients at five timedisplayed on a natural-log scale. The p-values represent the pair -Dimer. At day 3 and discharge time points, the -739.5] µg/L FEU respectively). Day 1, 75   -points and th, the median, -wise t-test -1078.8] and 76  day 30, and day 90 are not significant compared to the stable control. Within the 5 time points of AECOPD group, the ANOVA has a significant p value of 0.0067. Only day 3 is significantly different from day 30 (p < 0.05).  Lastly, figure 5.4 showed the time course results of MPO. Median values of day 1, day 30, and day 90 were significantly different compared to the controls (day 1:  212.8 [114.0-362.4], day 30: 71.8 [38.8-112.2], and day 90: 81.4 [45.2-156.8] versus stable: 136.1 [88.5-238.3] ng/mL respectively). On the other hand, day 3 and discharge are both non-significant when compared to the controls. The ANOVA for within the AECOPD group is also significant, with a decreasing trend (p < 0.0001).   5.2.2 ROC curve analysis ROC curves are shown in Figure 5.5 for the 4 biomarkers. The 4 curves depicted in order are: 1) troponin I, 2) PARC, 3) D-Dimer, and 4) MPO. The AUC for the four biomarkers respectively are: 0.540 for troponin I, 0.562 for PARC, 0.612 for D-Dimer, and 0.610 for MPO. The corresponding test sensitivity, specificity, and Youden’s index for optimal cut-off for all 4 models are tabulated in Table 5.1. D-Dimer and MPO showed significant p values when compared to the reference line AUC of 0.5 (p = 0.0021 and 0.0008 respectively). In contrast, troponin I and PARC showed no significant difference compared to the reference line AUC. Table 5.2 showed the AUC comparison across all 4 biomarkers. Only D-dimer showed a significant difference in AUC compared to troponin I (p = 0.0244), whereas the rest of the comparisons are non-significant.  77   Figure 5.5 Plot of ROC curves for the four biomarkers. 4 biomarkers are shown: 1) Troponin I, 2) PARC, 3) D-Dimer, and 4) MPO. The ROC curve is used in discriminating patients with AECOPD. The AUC of each biomarker is displayed respectively. The diagonal line represents an AUC of 0.5. PARC = Pulmonary and activation-regulated chemokine, MPO = myeloperoxidase.      78  Table 5.1 ROC curve characteristics of the four biomarkers. Variable Sample size (cases : controls) AUC AUC  95% CI Sens Spec Youden’s index P-value of comparison to reference line (AUC = 0.5) Troponin I 251:25  0.540  0.483-0.596  0.0797  1.0000  0.30  0.1721  PARC 261:12 0.562  0.365-0.760  0.8391  0.4167  34.6 0.5385  D-Dimer  369:75 0.612 0.541-0.683  0.7859  0.4667  291 0.0021  MPO 413:76 0.610 0.546-0.675  0.5036 0.7237 204.5 0.0008  Variables in the ROC curve analysis with performance characteristics. AUC = Area under the curve, CI = confidence interval computed via the binomial exact method, Sens = sensitivity, Spec = specificity, Youden’s index = Optimal cut-off, PARC = Pulmonary and activation-regulated chemokine, MPO = myeloperoxidase.  Table 5.2 ROC curve AUC comparisons amongst the four biomarkers.  ROC curves AUC1 – AUC2  Absolute AUC difference P-value Troponin I vs PARC 0.5395 – 0.5619  0.0224 0.8393  Troponin I vs D-Dimer  0.5395 – 0.6120  0.0725 0.0244  Troponin I vs MPO 0.5395 – 0.6102 0.0707 0.2427  PARC vs D-Dimer  0.5619 – 0.6120  0.0501  0.1705  PARC vs MPO 0.5619 – 0.6102  0.0483 0.9675  D-Dimer vs MPO 0.6120 – 0.6102  0.0018 0.9012  ROC curve comparisons with AUC differences and pair-wise comparisons. AUC = Area under the curve, PARC = Pulmonary and activation-regulated chemokine, MPO = myeloperoxidase. P-values are computed by pair-wise comparison via the DeLong method.   5.3 Discussion In this study, we made several novel observations: 1) troponin I, and PARC, showed no temporal changes during hospitalization and were no different between those patients with AECOPD and 79  those with stable COPD. 2) D-Dimer showed significant differences only between day 3 and discharge time point versus stable controls. 3) MPO showed the most significant changes between the onset versus convalescent time points. 4) Only D-Dimer and MPO have significant AUC versus the reference line, whereas troponin I and PARC are no different from reference.  First of all, from the observation that the majority of our patients have negative troponin I results (< 0.03 µg/L), we inferred that most of our patients did not have cardiac injury throughout the duration of the study. It is an intriguing finding owing to the fact that our patient enrolment selection criteria perhaps effectively excluded patients with presentation of cardiac injury. On the contrary, previously published studies reported that troponin T elevation can be used in conjunction with BNP/NT-proBNP in AECOPD to predict outcomes of hospitalization (47, 103, 114). Based on the time course profile, troponin I is not useful in tracking AECOPD progression. One explanation for this discrepancy could be that the selection criteria differed among various studies, thus that one cohort might be recruiting COPD patients with a slightly variable phenotype. Therefore, our troponin I results reflected that we recruited COPD patients without cardiac injury into our Rapid Transition program.  Secondly, PARC also did not demonstrate strong temporal changes throughout the time course. One reason could be due to the relatively smaller sample size for the stable control group (n = 12), as not all control patients have had PARC levels measured at the time of ELISA. PARC has been shown to be elevated in exacerbation by as much as 10% change (68), and is associated with undesirable clinical outcomes (118). In our cohort we did not observe 10% change comparing exacerbation onset to convalescent time points (day 30 and 90). Although statistically 80  non-significant, the levels presented in the time course are in fact, higher than young, healthy non-COPD controls (33.6 ng/mL; n = 5). With the limited n for PARC, drawing conclusions based on current observation need to be interpreted with caution. Perhaps, measuring the rest of the cohort to obtaining better statistical power would be a plausible future direction.  D-Dimer showed slightly better temporal changes than the previous two biomarkers, having both day 3 and discharge time points significantly different from those of stable controls. There is approximately 1.5 fold decrease from those of exacerbation onset levels to convalescent/stable time points. Being a degradation product of cross-linked fibrin, D-Dimer reflects the ongoing hemostatic changes. In our cohort, approximately half of the patients have a D-Dimer concentration of greater than 500 µg/L FEU, which is the reference range for positive D-Dimer. This could in turn support the notion that these patients might have PE. However, having positive D-Dimer results does not always indicate PE, as it has low positive predictive value in AECOPD due to other inflammatory factors (120). In respect to AUC data shown in Figure 5.5 and Table 5.1, D-Dimer is significantly different from the reference line. D-Dimer has the highest AUC among the four biomarkers measured in the analysis, and is significantly different than those of troponin I, which has the lowest AUC of the biomarkers. This can be interpreted as having some discriminatory power for D-Dimer, although not as strong as biomarkers such as CRP and NT-proBNP. One approach to increase the diagnostic performance of D-Dimer could be the integration of computed tomography (CT) results. Studies have shown certain utility in incorporating CT scans with D-Dimer results in AECOPD diagnosis, due to their negative predictive value (124, 125). This warrants a look into CT scan data, which might resolve and 81  elucidate the prevalence of PE in our cohort, and henceforth, we could sub-phenotype our patients for future analysis.  Lastly, MPO showed the most significant fold changes between the onset versus convalescent time points, with approximately 2.8 fold difference (See Figure 5.4). The biomarker seemed to track well with the progression of AECOPD treatment as the boxplot shared similar trend with those from CRP and NT-proBNP. However, on the discriminatory analysis it did not outperform D-Dimer. One can notice the stable COPD group having an elevated level of MPO, which in turn influenced the AUC. By integrating this group into the binary classifier of the ROC analysis, it reduced the fold changes observed between onset and stable. It is unclear at this point of interpretation why the stable COPD group has MPO that is significantly different than those in day 30/day 90 time points. One explanation could be due to the use of serum versus plasma samples. MPO leakage from leukocytes has been reported during the pre-analytical phase of serum clotting (126). It is speculated that MPO in blood are released from both neutrophils and monocytes/macrophages in acute inflammatory events including exacerbation (71). We cannot rule out the possibility that the neutrophils were activated during the clotting process, and thereafter released MPO into the serum. Nevertheless, the time course relationship is a true representation of the cohort. The ELISA experiments were all measured in same lot number kits, with little variation in intra-plate control (111.1[105.787-115.171] ng/mL). There were no systemic trends observed in the distribution of measurements, and as well as the samples were all randomized on the ELISA plate layout. There were no changes to the figure even when we normalized the MPO results to the respective intra-plate control, nor when we normalized to the neutrophil counts per sample. Various studies have been published measuring serum MPO 82  levels, and thus that the measurement in serum is well documented and not uncommon (17, 71, 122). The variability seen in the stable COPD group will need to be examined in the future, and perhaps a look into co-morbidity of the stable COPD group could facilitate the interpretation of the particular boxplot.  There are few limitations in this portion of the study. Pertaining to the biomarkers, the sample size for troponin I and PARC were relatively smaller in comparison to the overall cohort. The measurement of troponin I was discontinued partly through the patient recruitment process due to minimal measured signals. For those with positive troponin signals, they have not been confirmed with electrocardiogram (ECG) as these were not fully reviewed at the time of the analysis. Perhaps the ECG data would provide further explanation for these patients with elevated troponin I values, and in turn help with understanding this particular sub-group of AECOPD patients. Secondly, the relationship of MPO in serum versus plasma procurement will need to be elucidated as mentioned in the discussion section.   In our study, we have demonstrated that troponin I, and PARC, showed no temporal changes during hospitalization and were no different between those patients with AECOPD and those with stable COPD. D-Dimer showed slightly better signal in temporal changes and in AUC. Lastly, MPO showed the most significant temporal changes between onset and convalescent time points, with comparable AUC to those of D-Dimer. Out of these four biomarkers, only MPO has the potential to be a biomarker for AECOPD.   83  Chapter 6: Conclusion In this thesis, we identified an issue faced by physicians, health care professionals, and COPD patients, namely in the diagnosis and treatment of acute exacerbations requiring hospitalization. Given the variation seen in COPD phenotypes, treating acute exacerbation events have been suboptimal. Current treatment of corticosteroids, often times is not helpful, as some patients respond poorly to such therapy. There is a need for biomarkers in this area to facilitate healthcare providers in making the best decisions for treating AECOPD patients. Based on literature review, we identified a selection of blood proteins that we believe to have potential use as surrogate markers in acute exacerbations. From our study, CRP and NT-proBNP showed the most promise in tracking AECOPD.  First, we explored the temporal relationship of CRP and NT-proBNP to COPD exacerbations, and found that these two proteins tracked well with treatment. Both biomarkers were significantly increased at hospitalization and decreased while the patients are being treated. The concentrations decreased and stabilized once the patients were discharged and recovered, with approximately 3-4 fold change. The time course experiment suggested that CRP and NT-proBNP are modifiable by therapeutic treatment and thus supported the notion that these two biomarkers can be potentially used as surrogate biomarkers.  Secondly, we provided further evidence to support that CRP and NT-proBNP have discriminatory power when used in combination. The LOOCV model gave an AUC of 0.80 in separating COPD patients with exacerbation and without. We computed the optimal cut-off point from the ROC curves, which is useful when we wanted to categorizing our cohort further. From 84  the survival and the linear regression analyses, we found that those with higher initial NT-proBNP values also have higher mortality rates. Although mechanism for NT-proBNP elevation is unknown, there appears to be some involvement of the heart in COPD exacerbation.  Lastly, we presented four additional biomarkers: troponin I, PARC, D-Dimer and MPO, which showed weaker temporal responses and performed poorer than the former two biomarkers. Out of these four biomarkers, only MPO showed the most potential in association with time course, as well as in its discriminatory power. D-Dimer showed modest fold changes and had comparable AUC to MPO, but it was still less superior to CRP and NT-proBNP.  All in all, we demonstrated that surrogate markers can be used to objectively diagnose and treat COPD exacerbations. Out of all potential biomarkers, we provided evidence to show support for the use of CRP in conjunction with NT-proBNP. Being clinically available in the laboratory, we propose the use of these existing, orderable laboratory tests in the context of AECOPD, as an alternative approach to biomarker discovery. Collectively as presented in this thesis, we showed these two biomarkers are easy to measure, respond well to reflect exacerbation treatment progression, and can be replicated in an external cohort. Therefore, this thesis adds to the field by showing potential in repurposing currently available blood biomarkers for patients with known COPD history undergoing exacerbation.   85  References  1. Vestbo J, Hurd SS, Agusti AG, Jones PW, Vogelmeier C, Anzueto A, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med. 2013;187(4):347-65. 2. Pauwels RA, Rabe KF. Burden and clinical features of chronic obstructive pulmonary disease (COPD). Lancet. 2004;364(9434):613-20. 3. Mathers CD, Loncar D. Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med. 2006;3(11):e442. 4. Celli BR, Barnes PJ. Exacerbations of chronic obstructive pulmonary disease. Eur Respir J. 2007;29(6):1224-38. 5. Rodriguez-Roisin R. Toward a consensus definition for COPD exacerbations. Chest. 2000;117(5 Suppl 2):398S-401S. 6. CIHI. Health Indicators. 2008. July 23, 2014:[21 p.]. Available from: https://secure.cihi.ca/free_products/HealthIndicators2008_ENGweb.pdf. 7. Najafzadeh M, Marra CA, Lynd LD, Sadatsafavi M, FitzGerald JM, McManus B, et al. Future impact of various interventions on the burden of COPD in Canada: a dynamic population model. PLoS One. 2012;7(10):e46746. 8. Pauwels R, Calverley P, Buist AS, Rennard S, Fukuchi Y, Stahl E, et al. COPD exacerbations: the importance of a standard definition. Respir Med. 2004;98(2):99-107. 9. Cazzola M, Novelli G. Biomarkers in COPD. Pulm Pharmacol Ther. 2010;23(6):493-500. 10. Snell N, Newbold P. The clinical utility of biomarkers in asthma and COPD. Curr Opin Pharmacol. 2008;8(3):222-35. 11. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339. 12. Sin DD, Hollander Z, DeMarco ML, McManus BM, Ng RT. Biomarker Development for Chronic Obstructive Pulmonary Disease. From Discovery to Clinical Implementation. Am J Respir Crit Care Med. 2015;192(10):1162-70. 13. McShane LM, Altman DG, Sauerbrei W, Taube SE, Gion M, Clark GM, et al. Reporting recommendations for tumor marker prognostic studies. J Clin Oncol. 2005;23(36):9067-72. 14. Bafadhel M, McKenna S, Terry S, Mistry V, Reid C, Haldar P, et al. Acute Exacerbations of Chronic Obstructive Pulmonary Disease Identification of Biologic Clusters and Their Biomarkers. Am J Respir Crit Care Med. 2011;184(6):662-71. 15. Andelid K, Andersson A, Yoshihara S, Ahren C, Jirholt P, Ekberg-Jansson A, et al. Systemic signs of neutrophil mobilization during clinically stable periods and during exacerbations in smokers with obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2015;10:1253-63. 16. Gumus A, Altintas N, Cinarka H, Kirbas A, Haziroglu M, Karatas M, et al. Soluble urokinase-type plasminogen activator receptor is a novel biomarker predicting acute exacerbation in COPD. Int J Chron Obstruct Pulmon Dis. 2015;10:357-65. 17. Chang C, Yao W. Time course of inflammation resolution in patients with frequent exacerbations of chronic obstructive pulmonary disease. Med Sci Monit. 2014;20:311-20. 18. Chang C, Guo Z, Shen N, He B, Yao W, Zhu H, et al. Dynamics of inflammation resolution and symptom recovery during AECOPD treatment. Sci Rep. 2014;4:5516. 19. Fattouh M, Alkady O. Inflammatory biomarkers in chronic obstructive pulmonary disease. Egypt J Chest Dis Tuberc. 2014;63(4):799-804. 20. Johansson SL, Roberts NB, Schlosser A, Andersen CB, Carlsen J, Wulf-Johansson H, et al. Microfibrillar-associated protein 4: a potential biomarker of chronic obstructive pulmonary disease. Respir Med. 2014;108(9):1336-44. 86  21. Labib S, Younes S, Riad E, Abdallah A. Study of blood desmosine level in patients with COPD exacerbation in relation to severity. Egypt J Chest Dis Tuberc. 2014;63(3):569-73. 22. Lee SJ, Kim SH, Kim W, Lim S, Lee SH, Kim YE, et al. Increased plasma osteopontin in frequent exacerbator and acute exacerbation of COPD. Clin Respir J. 2014;8(3):305-11. 23. Liu HC, Lu MC, Lin YC, Wu TC, Hsu JY, Jan MS, et al. Differences in IL-8 in serum and exhaled breath condensate from patients with exacerbated COPD or asthma attacks. J Formos Med Assoc. 2014;113(12):908-14. 24. Liu Y, Liu X, Lin G, Sun L, Li H, Xie C. Decreased CD34+ cell number is correlated with cardiac dysfunction in patients with acute exacerbation of COPD. Heart Lung Circ. 2014;23(9):875-82. 25. Meng DQ, Li XJ, Song XY, Xin JB, Yang WB. Diagnostic and prognostic value of plasma adrenomedullin in COPD exacerbation. Respir Care. 2014;59(10):1542-9. 26. Nikolakopoulou S, Hillas G, Perrea D, Tentolouris N, Loukides S, Kostikas K, et al. Serum angiopoietin-2 and CRP levels during COPD exacerbations. COPD. 2014;11(1):46-51. 27. Nishimura K, Nishimura T, Onishi K, Oga T, Hasegawa Y, Jones PW. Changes in plasma levels of B-type natriuretic peptide with acute exacerbations of chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2014;9:155-62. 28. Omar MM, Issa HA, Mohammad AA, Abd-Elhamid MA-E. Serum adiponectin level in obese and non-obese COPD patients during acute exacerbation and stable conditions. Egypt J Chest Dis Tuberc. 2014;63(2):313-9. 29. Oraby SS, Ahmed ES, Farag TS, Zayed AE, Ali NK. Adiponectin as inflammatory biomarker of chronic obstructive pulmonary disease. Egypt J Chest Dis Tuberc. 2014;63(3):583-7. 30. Urban MH, Valipour A, Kiss D, Eickhoff P, Funk GC, Burghuber OC. Soluble receptor of advanced glycation end-products and endothelial dysfunction in COPD. Respir Med. 2014;108(6):891-7. 31. Zhang Y, Li S, Wang G, Han D, Xie X, Wu Y, et al. Changes of HMGB1 and sRAGE during the recovery of COPD exacerbation. J Thorac Dis. 2014;6(6):734-41. 32. Zhao YF, Jiang YP, Zhou LF, Wu XL. The value of assessment tests in patients with acute exacerbation of chronic obstructive pulmonary disease. Am J Med Sci. 2014;347(5):393-9. 33. Adnan AM, Ammar AZ, Khalil K. Role of eotaxin in chronic obstructive pulmonary disease (COPD). Int J Pharm Sci Rev Res. 2013;21(1):10-4. 34. Carter RI, Ungurs MJ, Mumford RA, Stockley RA. Aalpha-Val360: a marker of neutrophil elastase and COPD disease activity. Eur Respir J. 2013;41(1):31-8. 35. Gao P, Zhang J, He X, Hao Y, Wang K, Gibson PG. Sputum inflammatory cell-based classification of patients with acute exacerbation of chronic obstructive pulmonary disease. PLoS One. 2013;8(5):e57678. 36. Jin Q, Chen Y, Lou Y, He X. Low Serum retinol-binding protein-4 levels in acute exacerbations of chronic obstructive pulmonary disease at intensive care unit admission is a predictor of mortality in elderly patients. J Inflamm (Lond). 2013;10(1):31. 37. Mohamed NA, Fawzy MA, Elgamry R, Gad DM, Ibraheem HA. Role of adiponectin and other inflammatory biomarkers in COPD patients. Egypt J Chest Dis Tuberc. 2013;62(1):45-50. 38. Scherr A, Graf R, Bain M, Christ-Crain M, Muller B, Tamm M, et al. Pancreatic stone protein predicts positive sputum bacteriology in exacerbations of COPD. Chest. 2013;143(2):379-87. 39. Shoukry A, Said NS, Ayman Abd-Elrahman MN, Saber T, Amany Fawzy M, Shalaby S. Thyroid dysfunction and inflammatory biomarkers in chronic obstructive pulmonary disease: Relation to severity and exacerbation. Egypt J Chest Dis Tuberc. 2013;62(4):567-74. 40. Stanojkovic I, Kotur-Stevuljevic J, Spasic S, Milenkovic B, Vujic T, Stefanovic A, et al. Relationship between bone resorption, oxidative stress and inflammation in severe COPD exacerbation. Clin Biochem. 2013;46(16-17):1678-82. 87  41. Chen H, Wang Y, Bai C, Wang X. Alterations of plasma inflammatory biomarkers in the healthy and chronic obstructive pulmonary disease patients with or without acute exacerbation. J Proteomics. 2012;75(10):2835-43. 42. Falsey AR, Becker KL, Swinburne AJ, Nylen ES, Snider RH, Formica MA, et al. Utility of serum procalcitonin values in patients with acute exacerbations of chronic obstructive pulmonary disease: a cautionary note. Int J Chron Obstruct Pulmon Dis. 2012;7:127-35. 43. Huang JT, Chaudhuri R, Albarbarawi O, Barton A, Grierson C, Rauchhaus P, et al. Clinical validity of plasma and urinary desmosine as biomarkers for chronic obstructive pulmonary disease. Thorax. 2012;67(6):502-8. 44. Ju CR, Liu W, Chen RC. Serum surfactant protein D: biomarker of chronic obstructive pulmonary disease. Dis Markers. 2012;32(5):281-7. 45. Koczulla AR, Noeske S, Herr C, Koepke J, Jorres RA, Nell C, et al. Alpha-1 antitrypsin is elevated in exhaled breath condensate and serum in exacerbated COPD patients. Respir Med. 2012;106(1):120-6. 46. Kwiatkowska S, Noweta K, Zieba M, Nowak D, Bialasiewicz P. Enhanced exhalation of matrix metalloproteinase-9 and tissue inhibitor of metalloproteinase-1 in patients with COPD exacerbation: A prospective study. Respiration. 2012;84(3):231-41. 47. Marcun R, Sustic A, Brguljan PM, Kadivec S, Farkas J, Kosnik M, et al. Cardiac biomarkers predict outcome after hospitalisation for an acute exacerbation of chronic obstructive pulmonary disease. Int J Cardiol. 2012;161(3):156-9. 48. Mohamed KH, Abderabo MM, Ramadan ES, Hashim MM, Sharaf SM. Procalcitonin as a diagnostic marker in acute exacerbation of COPD. Egypt J Chest Dis Tuberc. 2012;61(4):301-5. 49. Pazarli AC, Koseoglu HI, Doruk S, Sahin S, Etikan I, Celikel S, et al. Procalcitonin: Is it a predictor of noninvasive positive pressure ventilation necessity in acute chronic obstructive pulmonary disease exacerbation? J Res Med Sci. 2012;17(11):1047-51. 50. Rohde G, Radsak MP, Borg I, Buhl R, Schultze-Werninghaus G, Taube C. Levels of soluble triggering receptor expressed on myeloid cells 1 in infectious exacerbations of chronic obstructive pulmonary disease. Respiration. 2012;83(2):133-9. 51. Shaker A, El-Shora A, El-Gammal M, Labib HA. Endocrinal disturbances and systemic inflammatiom in chronic obstructive pulmonary disease (COPD). Egypt J Chest Dis Tuberc. 2012;61(3):81-8. 52. Yerkovich ST, Hales BJ, Carroll ML, Burel JG, Towers MA, Smith DJ, et al. Reduced rhinovirus-specific antibodies are associated with acute exacerbations of chronic obstructive pulmonary disease requiring hospitalisation. BMC Pulm Med. 2012;12:37. 53. Chen H, Song Z, Qian M, Bai C, Wang X. Selection of Disease-Specific Biomarkers by Integrating Inflammatory Mediators with Clinical Informatics in Patients with Aecopd: A Preliminary Study. Respirology. 2011;16:126-. 54. Lacoma A, Prat C, Andreo F, Lores L, Latorre I, Perez M, et al. Usefulness of mid regional pro-atrial natriuretic peptide in the exacerbations of chronic obstructive pulmonary disease. Clin Chim Acta. 2011;412(5-6):470-5. 55. Lacoma A, Prat C, Andreo F, Lores L, Ruiz-Manzano J, Ausina V, et al. Value of procalcitonin, C-reactive protein, and neopterin in exacerbations of chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2011;6:157-69. 56. Lim SC, Ju JY, Chi SY, Ban HJ, Kwon YS, Oh IJ, et al. Apoptosis of T lymphocytes isolated from peripheral blood of patients with acute exacerbation of chronic obstructive pulmonary disease. Yonsei Med J. 2011;52(4):581-7. 57. Markoulaki D, Kostikas K, Papatheodorou G, Koutsokera A, Alchanatis M, Bakakos P, et al. Hemoglobin, erythropoietin and systemic inflammation in exacerbations of chronic obstructive pulmonary disease. Eur J Intern Med. 2011;22(1):103-7. 88  58. Krommidas G, Kostikas K, Papatheodorou G, Koutsokera A, Gourgoulianis KI, Roussos C, et al. Plasma leptin and adiponectin in COPD exacerbations: associations with inflammatory biomarkers. Respir Med. 2010;104(1):40-6. 59. Quint JK, Donaldson GC, Goldring JJ, Baghai-Ravary R, Hurst JR, Wedzicha JA. Serum IP-10 as a biomarker of human rhinovirus infection at exacerbation of COPD. Chest. 2010;137(4):812-22. 60. Koutsokera A, Kiropoulos TS, Nikoulis DJ, Daniil ZD, Tsolaki V, Tanou K, et al. Clinical, functional and biochemical changes during recovery from COPD exacerbations. Respir Med. 2009;103(6):919-26. 61. Kythreotis P, Kokkini A, Avgeropoulou S, Hadjioannou A, Anastasakou E, Rasidakis A, et al. Plasma leptin and insulin-like growth factor I levels during acute exacerbations of chronic obstructive pulmonary disease. BMC Pulm Med. 2009;9:11. 62. Shakoori TA, Sin DD, Ghafoor F, Bashir S, Bokhari SN. Serum surfactant protein D during acute exacerbations of chronic obstructive pulmonary disease. Dis Markers. 2009;27(6):287-94. 63. Karadag F, Karul AB, Cildag O, Yilmaz M, Ozcan H. Biomarkers of systemic inflammation in stable and exacerbation phases of COPD. Lung. 2008;186(6):403-9. 64. Stolz D, Breidthardt T, Christ-Crain M, Bingisser R, Miedinger D, Leuppi J, et al. Use of B-type natriuretic peptide in the risk stratification of acute exacerbations of COPD. Chest. 2008;133(5):1088-94. 65. Groenewegen KH, Dentener MA, Wouters EFM. Longitudinal follow-up of systemic inflammation after acute exacerbations of COPD. Respir Med. 2007;101(11):2409-15. 66. Perera WR, Hurst JR, Wilkinson TM, Sapsford RJ, Mullerova H, Donaldson GC, et al. Inflammatory changes, recovery and recurrence at COPD exacerbation. Eur Respir J. 2007;29(3):527-34. 67. Pinto-Plata VM, Livnat G, Girish M, Cabral H, Masdin P, Linacre P, et al. Systemic cytokines, clinical and physiological changes in patients hospitalized for exacerbation of COPD. Chest. 2007;131(1):37-43. 68. Hurst JR, Donaldson GC, Perera WR, Wilkinson TMA, Bilello JA, Hagan GW, et al. Use of plasma biomarkers at exacerbation of chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2006;174(8):867-74. 69. Phua J, Koay ES, Zhang D, Tai LK, Boo XL, Lim KC, et al. Soluble triggering receptor expressed on myeloid cells-1 in acute respiratory infections. Eur Respir J. 2006;28(4):695-702. 70. Roland M, Bhowmik A, Sapsford RJ, Seemungal TA, Jeffries DJ, Warner TD, et al. Sputum and plasma endothelin-1 levels in exacerbations of chronic obstructive pulmonary disease. Thorax. 2001;56(1):30-5. 71. Fiorini G, Crespi S, Rinaldi M, Oberti E, Vigorelli R, Palmieri G. Serum ECP and MPO are increased during exacerbations of chronic bronchitis with airway obstruction. Biomed Pharmacother. 2000;54(5):274-8. 72. Patel AR, Kowlessar BS, Donaldson GC, Mackay AJ, Singh R, George SN, et al. Cardiovascular risk, myocardial injury, and exacerbations of chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2013;188(9):1091-9. 73. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143(1):29-36. 74. Akobeng AK. Understanding diagnostic tests 3: Receiver operating characteristic curves. Acta Paediatr. 2007;96(5):644-7. 75. Hajian-Tilaki K. Receiver Operating Characteristic (ROC) Curve Analysis for Medical Diagnostic Test Evaluation. Caspian J Intern Med. 2013;4(2):627-35. 76. Fischer JE, Bachmann LM, Jaeschke R. A readers' guide to the interpretation of diagnostic test properties: clinical example of sepsis. Intensive Care Med. 2003;29(7):1043-51. 77. Metz CE. Basic principles of ROC analysis. Semin Nucl Med. 1978;8(4):283-98. 89  78. Airola A, Pahikkala T, Waegeman W, De Baets B, Salakoski T. An experimental comparison of cross-validation techniques for estimating the area under the ROC curve. Comput Stat Data Anal. 2011;55(4):1828-44. 79. Taylor JM, Ankerst DP, Andridge RR. Validation of biomarker-based risk prediction models. Clin Cancer Res. 2008;14(19):5977-83. 80. Thomsen M, Ingebrigtsen TS, Marott JL, Dahl M, Lange P, Vestbo J, et al. Inflammatory biomarkers and exacerbations in chronic obstructive pulmonary disease. JAMA - Journal of the American Medical Association. 2013;309(22):2353-61. 81. Gan W, Man S, Senthilselvan A, Sin D. Association between chronic obstructive pulmonary disease and systemic inflammation: a systematic review and a meta-analysis. Thorax. 2004;59(7):574-80. 82. Pinto-Plata VM, Mullerova H, Toso JF, Feudjo-Tepie M, Soriano JB, Vessey RS, et al. C-reactive protein in patients with COPD, control smokers and non-smokers. Thorax. 2006;61(1):23-8. 83. Man SF, J.E. C, Anthonisen NR, Wise RA, Tashkin DP, Sin DD. C-reactive protein and mortality in mild to moderate chronic obstructive pulmonary disease. Thorax. 2006;10:849-53. 84. Maisel AS, Krishnaswamy P, Nowak RM, McCord J, Hollander JE, Duc P, et al. Rapid Measurement of B-Type Natriuretic Peptide in the Emergency Diagnosis of Heart Failure. N Engl J Med. 2002;347:161-7. 85. Mueller C, Breidthardt T, Laule-Kilian K, Christ M, Perruchoud AP. The integration of BNP and NT-proBNP into clinical medicine. Swiss Medical Weekly. 2007;137(1-2):4-12. 86. Bozkanat E, Tozkoparan E, Baysan O, Deniz O, Ciftci F, Yokusoglu M. The significance of elevated brain natriuretic peptide levels in chronic obstructive pulmonary disease. J Int Med Res. 2005;33(5):537-44. 87. Kim HN, Januzzi JL, Jr. Natriuretic peptide testing in heart failure. Circulation. 2011;123(18):2015-9. 88. Abroug F, Ouanes-Besbes L, Nciri N, Sellami N, Addad F, Hamda KB, et al. Association of left-heart dysfunction with severe exacerbation of chronic obstructive pulmonary disease: diagnostic performance of cardiac biomarkers. Am J Respir Crit Care Med. 2006;174(9):990-6. 89. El Mallawany H, Mahmoud MI, Morsi TS, El-Shiekh RM. Role of N-terminal pro B-type natriuretic peptide in acute exacerbation of chronic obstructive pulmonary disease. Egyptian Journal of Chest Diseases and Tuberculosis. 2014;63(1):57-65. 90. Chen YR, Tam S, Li Y, Zarzycki M, Ferguson N, Hollander Z, et al. C-Reactive Protein and N-Terminal Prohormone of Brain Natriuretic Peptide as Blood Biomarkers for Acute Exacerbations of Chronic Obstructive Pulmonary Disease.  B38 TALKING ABOUT COPD BIOMARKERS. American Thoracic Society International Conference Abstracts: American Thoracic Society; 2015. p. A2928-A. 91. Crisafulli E, Torres A, Huerta A, Guerrero M, Gabarrus A, Gimeno A, et al. Predicting In-Hospital Treatment Failure (</=7 days) in Patients with COPD Exacerbation Using Antibiotics and Systemic Steroids. COPD. 2015:1-11. 92. Sin DD, Lacy P, York E, Man SF. Effects of fluticasone on systemic markers of inflammation in chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2004;170(7):760-5. 93. McCullough PA, Duc P, Omland T, McCord J, Nowak RM, Hollander JE, et al. B-type natriuretic peptide and renal function in the diagnosis of heart failure: an analysis from the Breathing Not Properly Multinational Study. Am J Kidney Dis. 2003;41(3):571-9. 94. Anwaruddin S, Lloyd-Jones DM, Baggish A, Chen A, Krauser D, Tung R, et al. Renal function, congestive heart failure, and amino-terminal pro-brain natriuretic peptide measurement: results from the ProBNP Investigation of Dyspnea in the Emergency Department (PRIDE) Study. J Am Coll Cardiol. 2006;47(1):91-7. 95. McCullough PA, Hollander Je Fau - Nowak RM, Nowak Rm Fau - Storrow AB, Storrow Ab Fau - Duc P, Duc P Fau - Omland T, Omland T Fau - McCord J, et al. Uncovering heart failure in patients 90  with a history of pulmonary disease: rationale for the early use of B-type natriuretic peptide in the emergency department. (1069-6563 (Print)). 96. Braunstein JB, Anderson GF, Gerstenblith G, Weller W, Niefeld M, Herbert R, et al. Noncardiac comorbidity increases preventable hospitalizations and mortality among medicare beneficiaries with chronic heart failure. Journal of the American College of Cardiology. 2003;42(7):1226-33. 97. Parker CM, Voduc N, Aaron SD, Webb KA, O'Donnell DE. Physiological changes during symptom recovery from moderate exacerbations of COPD. Eur Respir J. 2005;26(3):420-8. 98. Takahashi T, Kubo H. The role of microparticles in chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2014;9:303-14. 99. Agusti A, Sin DD. Biomarkers in COPD. Clinics in Chest Medicine. 2014;35(1):131-41. 100. Woodruff PG, Albert RK, Bailey WC, Casaburi R, Connett JE, Cooper JAD, et al. Randomized Trial of Zileuton for Treatment of COPD Exacerbations Requiring Hospitalization. COPD. 2011;8(1):21-9. 101. Helmy TA, Baess AI, Monsif DAA, Elnasharty AAA. Role of C-reactive protein and interleukin-6 in predicting the prognosis of ICU-admitted patients with acute exacerbation of COPD. Egyptian Journal of Chest Diseases and Tuberculosis. 2014;63(4):829-35. 102. Hoiseth A, Omland T, Hagve TA, Brekke PH, Soyseth V. Determinants of high-sensitivity cardiac troponin T during acute exacerbation of chronic obstructive pulmonary disease: a prospective cohort study. BMC Pulm Med. 2012;12:22. 103. Chang CL, Robinson SC, Mills GD, Sullivan GD, Karalus NC, McLachlan JD, et al. Biochemical markers of cardiac dysfunction predict mortality in acute exacerbations of COPD. Thorax. 2011;66(9):764-8. 104. Zakynthinos E, Kiropoulos T, Gourgoulianis K, Filippatos G. Diagnostic and prognostic impact of brain natriuretic peptide in cardiac and noncardiac diseases. Heart & Lung: The Journal of Acute and Critical Care. 2008;37(4):275-85. 105. Januzzi JL, van Kimmenade R, Lainchbury J, Bayes-Genis A, Ordonez-Llanos J, Santalo-Bel M, et al. NT-proBNP testing for diagnosis and short-term prognosis in acute destabilized heart failure: an international pooled analysis of 1256 patients: the International Collaborative of NT-proBNP Study. Eur Heart J. 2006;27(3):330-7. 106. McKie PM, Burnett JC, Jr. B-type natriuretic peptide as a biomarker beyond heart failure: speculations and opportunities. Mayo Clin Proc. 2005;80(8):1029-36. 107. Wang TJ, Larson MG, Levy D, Benjamin EJ, Leip EP, Omland T, et al. Plasma natriuretic peptide levels and the risk of cardiovascular events and death. N Engl J Med. 2004;350(7):655-63. 108. Ridker PM. Clinical Application of C-Reactive Protein for Cardiovascular Disease Detection and Prevention. Circulation. 2003;107(3):363-9. 109. Reitz D, Gerger A, Seidel J, Kornprat P, Samonigg H, Stotz M, et al. Combination of tumour markers CEA and CA19-9 improves the prognostic prediction in patients with pancreatic cancer. J Clin Pathol. 2015;68(6):427-33. 110. Adams JE, 3rd, Bodor GS, Davila-Roman VG, Delmez JA, Apple FS, Ladenson JH, et al. Cardiac troponin I. A marker with high specificity for cardiac injury. Circulation. 1993;88(1):101-6. 111. Antman E, Bassand J-P, Klein W, Ohman M, Lopez Sendon JL, Rydén L, et al. Myocardial infarction redefined—a consensus document of The Joint European Society of Cardiology/American College of Cardiology committee for the redefinition of myocardial infarction: The Joint European Society of Cardiology/ American College of Cardiology Committee. Journal of the American College of Cardiology. 2000;36(3):959-69. 112. Sin DD, Anthonisen NR, Soriano JB, Agusti AG. Mortality in COPD: Role of comorbidities. Eur Respir J. 2006;28(6):1245-57. 113. Harvey MG, Hancox RJ. Elevation of cardiac troponins in exacerbation of chronic obstructive pulmonary disease. Emerg Med Australas. 2004;16(3):212-5. 91  114. Brekke PH, Omland T, Holmedal SH, Smith P, V Sy. Determinants of cardiac troponin T elevation in COPD exacerbation - a cross-sectional study. BMC Pulm Med. 2009;9:35. 115. Li D, Jialal I, Keffer J. Greater frequency of increased cardiac troponin T than increased cardiac troponin I in patients with chronic renal failure. Clinical Chemistry. 1996;42(1):114-5. 116. Freda BJ, Tang WH, Van Lente F, Peacock WF, Francis GS. Cardiac troponins in renal insufficiency: review and clinical implications. J Am Coll Cardiol. 2002;40(12):2065-71. 117. Gunther C, Bello-Fernandez C, Kopp T, Kund J, Carballido-Perrig N, Hinteregger S, et al. CCL18 Is Expressed in Atopic Dermatitis and Mediates Skin Homing of Human Memory T Cells. The Journal of Immunology. 2005;174(3):1723-8. 118. Sin DD, Miller BE, Duvoix A, Man SF, Zhang X, Silverman EK, et al. Serum PARC/CCL-18 concentrations and health outcomes in chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2011;183(9):1187-92. 119. Righini M, Perrier A, De Moerloose P, Bounameaux H. D-Dimer for venous thromboembolism diagnosis: 20 years later. J Thromb Haemost. 2008;6(7):1059-71. 120. Rizkallah J, Man SF, Sin DD. Prevalence of pulmonary embolism in acute exacerbations of COPD: a systematic review and metaanalysis. Chest. 2009;135(3):786-93. 121. Zhang ZL, Brennan ML, Fu XM, Aviles RJ, Pearce GL, Penn MS, et al. Association between myeloperoxidase levels and risk of coronary artery disease. Jama-Journal of the American Medical Association. 2001;286(17):2136-42. 122. Park HY, Man SFP, Tashkin D, Wise RA, Connett JE, Anthonisen NA, et al. The Relation of Serum Myeloperoxidase to Disease Progression and Mortality in Patients with Chronic Obstructive Pulmonary Disease (COPD). PLoS One. 2013;8(4). 123. Reichlin T, Socrates T, Egli P, Potocki M, Breidthardt T, Arenja N, et al. Use of myeloperoxidase for risk stratification in acute heart failure. Clin Chem. 2010;56(6):944-51. 124. Rutschmann OT, Cornuz J, Poletti PA, Bridevaux PO, Hugli OW, Qanadli SD, et al. Should pulmonary embolism be suspected in exacerbation of chronic obstructive pulmonary disease? Thorax. 2007;62(2):121-5. 125. Sohne M, Kruip MJHA, Nijkeuter M, Tick L, Kwakkel H, Halkes SJM, et al. Accuracy of clinical decision rule, D-dimer and spiral computed tomography in patients with malignancy, previous venous thromboembolism, COPD or heart failure and in older patients with suspected pulmonary embolism. Journal of Thrombosis and Haemostasis. 2006;4(5):1042-6. 126. Shih J, Datwyler SA, Hsu SC, Matias MS, Pacenti DP, Lueders C, et al. Effect of collection tube type and preanalytical handling on myeloperoxidase concentrations. Clin Chem. 2008;54(6):1076-9.  92  Appendices Appendix A  List of MeSH terms used for the systematic review in Chapter 2. ("pulmonary disease, chronic obstructive"[MeSH Terms] OR ("pulmonary"[All Fields] AND "disease"[All Fields] AND "chronic"[All Fields] AND "obstructive"[All Fields]) OR "chronic obstructive pulmonary disease"[All Fields] OR "copd"[All Fields]) AND (exacerbations[All Fields] OR “acute exacerbations”[All Fields] OR “aecopd”[All Fields]) AND ("biomarkers"[MeSH Terms] OR "biomarkers"[All Fields] OR “biological markers”[All Fields]) AND ("diagnosis"[MeSH Terms] OR "diagnosis"[All Fields] OR "diagnostic"[All Fields]) AND ("blood"[Subheading] OR "blood"[All Fields] OR "blood"[MeSH Terms] OR “serum”[All Fields] OR “plasma”[All Fields])  93  Appendix B  Guidelines for REMARK scores Introduction 1. State the marker examined, the study objectives, and any pre-specified hypotheses. Materials and Methods Patients 2. Describe the characteristics (for example, disease stage or co-morbidities) of the study patients, including their source and inclusion and exclusion criteria. 3. Describe treatments received and how chosen (for example, randomized or rule-based). Specimen characteristics 4. Describe type of biological material used (including control samples) and methods of preservation and storage. Assay methods 5. Specify the assay method used and provide (or reference) a detailed protocol, including specific reagents or kits used, quality control procedures, reproducibility assessments, quantitation methods, and scoring and reporting protocols. Specify whether and how assays were performed blinded to the study endpoint. Study design 6. State the method of case selection, including whether prospective or retrospective and whether stratification or matching (for example, by stage of disease or age) was used. Specify the time period from which cases were taken, the end of the follow-up period, and the median follow-up time. 94  7. Precisely define all clinical endpoints examined. 8. List all candidate variables initially examined or considered for inclusion in models. 9. Give rationale for sample size; if the study was designed to detect a specified effect size, give the target power and effect size. Statistical analysis methods 10. Specify all statistical methods, including details of any variable selection procedures and other model-building issues, how model assumptions were verified, and how missing data were handled. 11. Clarify how marker values were handled in the analyses; if relevant, describe methods used for cutpoint determination. Results Data 12. Describe the flow of patients through the study, including the number of patients included in each stage of the analysis (a diagram may be helpful) and reasons for dropout. Specifically, both overall and for each subgroup extensively examined report the number of patients and the number of events. 13. Report distributions of basic demographic characteristics (at least age and sex), standard (disease-specific) prognostic variables, and tumor marker, including numbers of missing values. Analysis and presentation 14. Show the relation of the marker to standard prognostic variables. 15. Present univariable analyses showing the relation between the marker and outcome, with the estimated effect (eg, hazard 95  ratio and survival probability). Preferably provide similar analyses for all other variables being analyzed. For the effect of a tumor marker on a time-to-event outcome, a Kaplan-Meier plot is recommended. 16. For key multivariable analyses, report estimated effects (eg, hazard ratio) with confidence intervals for the marker and, at least for the final model, all other variables in the model. 17. Among reported results, provide estimated effects with confidence intervals from an analysis in which the marker and standard prognostic variables are included, regardless of their statistical significance. 18. If done, report results of further investigations, such as checking assumptions, sensitivity analyses, and internal validation. Discussion 19. Interpret the results in the context of the pre-specified hypotheses and other relevant studies; include a discussion of limitations of the study. 20. Discuss implications for future research and clinical value. The original table is reproduced with stylistic changes (13). The table summarized the REMARK recommendations used for tumor marker studies.   96  Appendix C  COPD definitions of 59 publications included in the review arranged by the latest published year  Reference  Year COPD Diagnosis Pre/Post Bronchodilator AECOPD Definition Stable COPD Definition Andelid, K., et al. (15) 2015 GOLD criteria Pre Infections in airways or lungs were deemed to be exacerbations 4 weeks free from exacerbation Gumus, A., et al. (16) 2015 GOLD criteria Post W N/A Chang, C. , Yao, W. (17) 2014 GOLD criteria Post W 8 weeks free from exacerbation Chang, C., et al. (18) 2014 GOLD criteria Post W 8 weeks free from exacerbation Fattouh, M. Alkady, O. (19) 2014 GOLD criteria Post N/A 4 weeks free from exacerbation Johansson, S.L., et al. (20) 2014 GOLD criteria Post N/A 4 weeks free from exacerbation Labib, S., et al. (21) 2014 GOLD criteria Post BAP-65 Score One month after exacerbation Lee, S.J., et al. (22) 2014 GOLD criteria Post W 3 months free from exacerbation Liu, H.C., et al. (23) 2014 GOLD criteria Post W N/A Liu, Y., et al. (24) 2014 GOLD criteria Post W No requirement for increased treatment for 12 weeks; GOLD II to IV Meng, D.Q., et al. (25) 2014 GOLD criteria Post W No requirement for increased treatment for 30 days Nikolakopoulou, S., et al. (26) 2014 GOLD criteria Post W 8 weeks free from exacerbation Nishimura, K., et al. (27) 2014 GOLD criteria Post W 4 weeks free from exacerbation Omar, M.M., et al. (28) 2014 GOLD criteria Post W Disappearance of symptoms Oraby, S.S., et al. (29) 2014 GOLD criteria Post W No requirement for increased treatment for 12 weeks Urban, M.H., et al. (30) 2014 GOLD criteria Post W 8 weeks free from exacerbation Zhang, Y., et al. (31) 2014 N/A Post W Disappearance of symptoms Zhao, Y.F., et al. (32) 2014 GOLD criteria Post W N/A   97  Reference  Year COPD Diagnosis Pre/Post Bronchodilator AECOPD Definition Stable COPD Definition Adnan, A.M., et al. (33) 2013 N/A N/A N/A 8 weeks free from exacerbation Carter, R.I., et al. (34) 2013 History of chronic bronchitis Post Anthonisen criteria 8 weeks after exacerbation Gao, P., et al. (35) 2013 GOLD criteria Post N/A N/A Jin, Q., et al. (36) 2013 GOLD criteria Post W & ICU criteria 4 weeks free from exacerbation Mohamed, N.A., et al. (37) 2013 GOLD criteria Post W 3 months free from exacerbation Patel, A.R.C., et al. (72) 2013 GOLD criteria Post W 4 weeks free from exacerbation Scherr, A., et al. (38) 2013 GOLD criteria Post  Anthonisen criteria N/A Shoukry, A., et al. (39) 2013 GOLD criteria Post W 3 months free from exacerbation Stanojkovic, I., et al. (40) 2013 GOLD criteria Post W 30 days after exacerbation Chen, H., et al. (41) 2012 GOLD criteria Post W No requirement for increased treatment for 30 days; 8 weeks free from exacerbation Falsey, A.R., et al. (42) 2012 GOLD criteria Post  Anthonisen criteria N/A Huang, J., et al. (43) 2012 GOLD criteria Post Symptoms of cough and breathlessness 4 weeks free from exacerbation & 2 weeks free from chest infections Ju, C.R., et al. (44) 2012 GOLD criteria Post W 3 months free from exacerbation Koczulla, A.R., et al. (45) 2012 GOLD criteria Post W Free of serum signs of inflammation and radiological signs of pneumonia Kwiatkowska, S., et al. (46) 2012 GOLD criteria Post W N/A Marcun, R., et al. (47) 2012 GOLD criteria Post W N/A Mohamed, K.H., et al. (48) 2012 GOLD criteria Post W N/A Pazarli, A.C., et al. (49) 2012 GOLD criteria Post Anthonisen criteria 4 weeks free from exacerbation 98  Reference  Year COPD Diagnosis Pre/Post Bronchodilator AECOPD Definition Stable COPD Definition Rohde, G., et al. (50) 2012 GOLD criteria Post Anthonisen criteria 4 weeks free from exacerbation; no requirement for increased treatment for 14 days Shaker, A., et al. (51) 2012 GOLD criteria Post W One month after discharge Yerkovich, S.T., et al. (52) 2012 GOLD criteria Post W 6 weeks free from exacerbation Bafadhel, M., et al. (14) 2011 GOLD criteria Post Anthonisen criteria 8 weeks free from exacerbation Chen, H., et al. (53) 2011 N/A N/A W No requirement for increased treatment for 30 days Lacoma, A., et al. (54) 2011 SEPAR guidelines Post Anthonisen criteria 4 weeks free from exacerbation Lacoma, A., et al. (55) 2011 SEPAR guidelines Post Anthonisen criteria 4 weeks free from exacerbation Lim, S.C., et al. (56) 2011 GOLD criteria Post Anthonisen criteria 8 weeks free from exacerbation Markoulaki, D., et al. (57) 2011 GOLD criteria Post Type 1 Anthonisen criteria only 8 weeks free from exacerbation & respiratory tract infection Krommidas, G., et al. (58) 2010 GOLD criteria Post Type 1 Anthonisen criteria only 8 weeks free from exacerbation & respiratory tract infection Quint, J.K., et al. (59) 2010 GOLD criteria Post W > 42 days post-exacerbation and > 14 days pre-exacerbation onset Koutsokera, A., et al. (60) 2009 GOLD criteria Post Type 1 Anthonisen criteria only 10 or 40 days after exacerbation Kythreotis, P., et al. (61) 2009 GOLD criteria Post W 3 months free from exacerbation Shakoori, T.A., et al. (62) 2009 GOLD criteria Post W N/A Karadag, F., et al. (63) 2008 GOLD criteria Post Anthonisen criteria 3 months free from exacerbation Stolz, D., et al. (64) 2008 GOLD criteria Post Anthonisen criteria 14 to 18 days after exacerbation Groenewegen, K.H., et al. (65) 2007 GOLD criteria Post Type 1 Anthonisen criteria only 6 months after discharge Perera, W. R., et al. (66) 2007 GOLD criteria Post W 6 weeks free from exacerbation Pinto-Plata, V.M., et al. (67) 2007 GOLD criteria Post  Anthonisen criteria 8 weeks after discharge 99  Reference  Year COPD Diagnosis Pre/Post Bronchodilator AECOPD Definition Stable COPD Definition Hurst, J.R., et al. (68) 2006 GOLD criteria Post W 4 weeks free from exacerbation Phua, J., et al. (69) 2006 GOLD criteria Post Anthonisen criteria N/A Roland, M., et al. (70) 2001 GOLD criteria Post Anthonisen criteria 3 weeks free from exacerbation Fiorini, G., et al. (71) 2000 GOLD criteria Post Exacerbation requiring hospital admission N/A Anthonisen criteria were based on patients evaluated for dyspnea, sputum production or increased purulence, without radiological consolidation. Anthonisen type 1 criteria defined patients that were presented with all three symptoms of increased dyspnea, sputum volume and sputum purulence. Pre/post bronchodilator indicated whether bronchodilators were used prior to lung function measurements. BAP-65 score definition included BUN level >25 mg/dl, altered sensorium, and pulse rate >109beats/min, and age > 65. Abbreviations: AECOPD = acute exacerbation of chronic obstructive pulmonary disease, GOLD = Global Initiative for chronic obstructive lung disease, SEPAR = Spanish Society of Pneumology and Thoracic Surgery, W = a sustained worsening of the patient’s condition of dyspnea sensation, coughing, or sputum production that can become purulent, from the stable state and beyond normal day-to-day variations, necessitating a change in regular medication in a patient with underlying COPD.100  Appendix D  Biomarkers investigated in one studies Biomarkers Reference Year Direction during AECOPD Statistical Significance Adrenomedullin Meng, D.Q., et al. (25) 2014 ↑ + Alpha-1 antitrypsin Koczulla, A.R., et al. (45) 2012 ↑ + Amphiregulin Hurst, J.R., et al. (68) 2006 ↔ − Angiopoietin-2 Nikolakopoulou, S., et al. (26) 2014 ↑ + Anti-VP1 IgG1 Yerkovich, S.T., et al. (52) 2012 ↓ + Aα-Val360 Carter, R.I., et al. (34) 2013 ↑ + Beta-crosslaps Stanojkovic, I., et al. (40) 2013 ↑ + BPI Groenewegen, K.H., et al. (65) 2007 ↑ + CCL3 Bafadhel, M., et al. (14) 2011 ↓ + CCL13 Bafadhel, M., et al. (14) 2011 ↓ + CCL25 Chen, H., et al. (53) 2011 ↓ − CCL27 Chen, H., et al. (53) 2011 ↓ − CD34+ cells Liu, Y., et al. (24) 2014 ↓ − Cerberus 1 Chen, H., et al. (41) 2012 ↓ − Copeptin Zhao, Y.F., et al. (32) 2014 ↑ + Endothelin-1 Roland, M., et al. (70) 2001 ↑ ± Eotaxin/CCL11 Adnan, A.M., et al. (33) 2013 ↑ + Eotaxin-2/CCL24 Hurst, J.R., et al. (68) 2006 ↓ − EPO Markoulaki, D., et al. (57) 2011 ↑ + Fibronectin Hurst, J.R., et al. (68) 2006 ↑ − FSH Shaker, A., et al. (51) 2012 ↑ + Growth hormone R Chen, H., et al. (41) 2012 ↓ − 101  Hemoglobin Markoulaki, D., et al. (57) 2011 ↓ + HMGB1 Zhang, Y., et al. (31) 2014 ↑ + IgE Fiorini, G., et al. (71) 2000 ↑ − IL-9 Chen, H., et al. (53) 2011 ↓   − IL-12 p40 Hurst, J.R., et al. (68) 2006 ↑ − IL-13 Bafadhel, M., et al. (14) 2011 ↓ + IL-19 Chen, H., et al. (41) 2012 ↓ + LH Shaker, A., et al. (51) 2012 ↑ + LTB4 Pinto-Plata, V.M., et al. (67) 2007 ↑ + Lymphotoxin beta Chen, H., et al. (41) 2012 ↓ − MFAP4 Johansson, S.L., et al. (20) 2014 ↓ + MMP-10 Chen, H., et al. (41) 2012 ↓ − MR-proANP Lacoma, A., et al. (54) 2011 ↑ + Neutrophil elastase Andelid, K., et al. (15) 2015 ↑ + NO Karadag, F., et al. (63) 2008 ↑ + Pancreatic stone protein/regenerating protein Scherr, A., et al. (38) 2013 ↑ + RBP4 Jin, Q., et al. (36) 2013 ↓ + sICAM-1 Hurst, J.R., et al. (68) 2006 ↑ + sTNFR55 Groenewegen, K.H., et al. (65) 2007 ↑ − sTNFR75 Groenewegen, K.H., et al. (65) 2007 ↑ + suPAR Gumus, A., et al. (16) 2015 ↑ + TEAC Groenewegen, K.H., et al. (65) 2007 ↓ + Testosterone Shaker, A., et al. (51) 2012 ↓ + Thrombopoietin Chen, H., et al. (41) 2012 ↓ − 102  T3 Shoukry, A., et al. (39) 2013 ↓ + T4 Shoukry, A., et al. (39) 2013 ↑ − T-Lymphocyte apoptosis Lim, S.C., et al. (56) 2011 ↑ + Toll-like receptor 4 Chen, H., et al. (41) 2012 ↓ − TSH Shoukry, A., et al. (39) 2013 ↓ + Abbreviations: ↑ = biomarker increased during AECOPD, ↓ = biomarker decreased during AECOPD, ↔ = Biomarker showed no change during AECOPD, + = Statistically significant (P-value < 0.05), − = Not statistically significant (P-value > 0.20), ± = Borderline statistically significant (P-value = 0.05-0.20), AECOPD = acute exacerbation of chronic obstructive pulmonary disease, BPI = bactericidal permeability increasing protein, CCL = chemokine C-C motif ligand, CD = cluster of differentiation, EPO = erythropoietin, FSH = follicle stimulating hormone, HMGB = high mobility group box, Ig = immunoglobulin, IL = interleukin, LH = luteinizing hormone, LT = leukotriene, MFAP = microfibrillar associated protein, MMP = matrix metallopeptidase, MR-proANP = mid-regional prohormone of atrial natriuretic peptide, NO = neutrophil elastase, nitric oxide, RBP = retinol-binding proteins, sICAM = soluble intercellular adhesion molecule, sTNFR = soluble tumor necrosis factor  receptor, suPAR = soluble urokinase-type plasminogen activator receptor, TEAC = Trolox equivalent antioxidant capacity, T3 = triiodothyronine, T4 = thyroxine, TSH = thyroid stimulating hormone.  103   Appendix E  Modified REMARK (mREMARK) scores breakdown for the 14 studies listed in Table 2.4 Reference  Total Score Intro Materials and Methods Results Discussion 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Bafadhel, M., et al. (14) 18 X X X X X X   X   X X X X X X X X X X X Lacoma, A., et al. (55)  16 X X X X X X   X   X   X X X X X X   X X Stolz, D., et al. (64) 15 X X     X X X X   X   X X X X X X   X X Jin, Q., et al. (36) 15 X X X X X   X X   X   X X X X   X   X X Hurst, J.R., et al. (68) 14 X X   X X   X X   X X   X X   X X   X X Gumus, A., et al. (16) 13 X X  X X X  X  X   X X X  X  X X Shakoori, T.A., et al. (62) 13 X X   X X X   X   X X   X   X   X   X X Falsey, A.R., et al. (42) 12 X X   X X X         X X X   X   X   X X Quint, J.K., et al. (59) 11 X X   X   X       X X X X       X   X X Pazarli, A.C., et al. (49) 10 X X X   X         X X   X       X   X X Phua, J., et al. (69) 10 X X     X X       X X   X       X   X X Adnan, A.M., et al. (33) 6 X     X X           X               X X This table summarized the REMARK scores in studies from Table 2.4. A score is given to a particular section when the study met the recommendations 

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.24.1-0229571/manifest

Comment

Related Items