UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Clinical characteristics and predictors of reduced survival for adult-diagnosed cystic fibrosis patients… Desai, Sameer 2017

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

Item Metadata

Download

Media
24-ubc_2017_september_desai_sameer.pdf [ 1.1MB ]
Metadata
JSON: 24-1.0349192.json
JSON-LD: 24-1.0349192-ld.json
RDF/XML (Pretty): 24-1.0349192-rdf.xml
RDF/JSON: 24-1.0349192-rdf.json
Turtle: 24-1.0349192-turtle.txt
N-Triples: 24-1.0349192-rdf-ntriples.txt
Original Record: 24-1.0349192-source.json
Full Text
24-1.0349192-fulltext.txt
Citation
24-1.0349192.ris

Full Text

Clinical Characteristics and Predictors of Reduced Survival for Adult-Diagnosed Cystic Fibrosis Patients – A population based study    by   Sameer Desai   A THESIS SUBMITTED IN PARTIAL FULFILLMENT  OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE  in   The Faculty of Graduate and Postdoctoral Studies  (Population and Public Health)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)    July 2017  									© Sameer Desai, 2017 	 ii	Abstract	 Background: Approximately 5-10% of cystic fibrosis (CF) diagnoses are made during adulthood. These patients are a minority, and there is a paucity of literature describing their characteristics and prognosis. The objectives of this study are to describe the clinical characteristics, estimate survival, and identify clinical predictors of reduced survival at the time of diagnosis in adult-diagnosed CF patients. Methods: There were 362 newly diagnosed adult CF (≥18 years) patients from 1990 to 2014 in the Canadian CF Patient Registry. Clinical characteristics were described, the Kaplan-Meier method was employed for 10- and 15-year lung transplant-free survival estimates and multivariable Cox regression analysis was conducted to identify significant predictors of reduced survival at baseline. Adjusted survival curves were used to illustrate the impact of the significant predictors on lung transplant-free survival. Results: The median follow-up time observed was 7.7 years (range: 0.0-23.6) and included 33 deaths and 15 transplants for a total of 48 events in 3,106 patient years (15.5 events per 1,000 PYs). The median age at diagnosis was 34.3 years (range: 18.0-73.8), with the majority presenting with pulmonary and/or gastrointestinal symptoms (70%) and a nearly equal distribution of males and females. During the study period, 15% were diagnosed with CF-related diabetes (CFRD), 35% with pancreatic insufficiency and 50% were culture positive for P. aeruginosa. The most common genotype identified was ∆F508 heterozygous (38%). Lung transplant-free survival was 88% at 10 years and 86% by 15 years. Age at diagnosis (HR: 1.32 per 5-year increase, 95% CI: 1.13-1.54), CFRD (HR: 7.86, 95% CI: 2.09-29.55) and lower lung function (HR: 0.76 per 5% increase, 95% CI: 0.69-0.83) at baseline were significant predictors of reduced survival. In terms of clinical utility, low lung function (FEV1 % predicted < 60%) and CFRD were predictors that impacted lung transplant-free survival substantially. Conclusions: Adult-diagnosed CF patients have a milder phenotype of disease and a better prognosis than previously reported. Older age at diagnosis, lower lung function, and CFRD were important predictors of reduced survival. Adult CF clinicians and other CF caregivers can use this information to educate patients about their prognosis and to guide treatment. 	 iii	Lay	Abstract	 Cystic fibrosis (CF) is the most common genetic disease affecting children and young adults. More adults are being diagnosed with CF due to increased awareness among doctors and more sensitive diagnostic methods. Being diagnosed with CF as an adult is very rare; currently, they represent only 5-10% of all CF patients in Canada. These patients face unique challenges, issues, and needs compared to other CF patients diagnosed at younger ages. Information about life expectancy and prognosis is important, particularly for mature, older diagnosed patients. However, due to the rarity of this disease, doctors have little information to inform them. The main purpose of this study was to evaluate disease prognosis in this group of CF patients and we found that these patients have higher survival rates than previously reported. Also, being older at diagnosis and having low lung function at the time of diagnosis are important factors for poor survival. Having CF-related diabetes is also another potentially important factor. Doctors can now use this information to educate patients and guide treatment.                  	 iv	Preface	 I was responsible for reviewing the literature, protocol development, ethics registration, statistical data analysis, interpretation of the analysis, and drafting of the manuscript (thesis).   Dr. Joel Singer was involved with the concept and design of the project, statistical advice and interpretation of the analysis. Dr. Bradley Quon was involved with the concept and design of the project, acquisition of data, interpretation of analysis, and overall project coordination. Dr. Hubert Wong was involved with providing statistical advice and interpretation of the analysis.  This project, titled “Clinical Characteristics and Predictors of Reduced Survival for Adult-Diagnosed Cystic Fibrosis Patients” was approved by the Research Ethics Board at St Paul’s Hospital, Vancouver, Canada (REB # H15-01439).                  	 v	Table	of	Contents	Abstract	..................................................................................................................................	ii	Lay	Abstract	...........................................................................................................................	iii	Preface	...................................................................................................................................	iv	Table	of	Contents	....................................................................................................................	v	List	of	Tables	..........................................................................................................................	vi	List	of	Figures	........................................................................................................................	vii	List	of	Abbreviations	and	Acronyms	.....................................................................................	viii	Acknowledgements	................................................................................................................	ix	1									Introduction	..................................................................................................................	1	1.1	Review	of	the	literature	..............................................................................................................	2	1.2	Rationale	and	objectives	of	the	study	.........................................................................................	5	2									Methods	.......................................................................................................................	6	2.1	Data	source	................................................................................................................................	6	2.2	Study	design	and	sample	............................................................................................................	6	2.3	Study	variables	...........................................................................................................................	6	2.4	Statistical	analysis	......................................................................................................................	7	3									Results	.........................................................................................................................	10	3.1	Study	sample	characteristics	.....................................................................................................	10	3.2	Descriptive	statistics	by	event	status	........................................................................................	10	3.3	Survival	analysis	.......................................................................................................................	11	3.4	Sensitivity	analyses	...................................................................................................................	12	4									Discussion	....................................................................................................................	14	4.1	Summary	of	findings	.................................................................................................................	14	4.2	Key	findings	in	the	context	of	existing	literature	.......................................................................	14	4.3	Weaknesses	of	study	................................................................................................................	18	5	Conclusion	..........................................................................................................................	20	6	Future	Directions	................................................................................................................	21	7	Figures	................................................................................................................................	22	8	Tables	.................................................................................................................................	30	References	............................................................................................................................	37	 		 	 vi	   List	of	Tables		Table	1		 Transplantation	Criteria………………………………………………………………………………….30	Table	2		 Descriptive	statistics……………………………………………………………………………………….31	Table	3		 Univariable	and	multivariable	modelling…………………………………………………….....33	Table	4		 Predicted	lung	transplant-free	survival	probabilities……………………………………….34	Table	5		 Sensitivity	Analysis	substituting	P.	aeruginosa	at	baseline	to	up	to		5	years	post-diagnosis	…………………………………………………………………………………...35	Table	6	 Mean	age	at	diagnosis	and	FEV1	%	Predicted	for	individuals	with		and	without	P.	aeruginosa	at	baseline	……………………………………………………………36		 		 vii				List	of	Figures	 Figure	1		 Distribution	of	Age	at	diagnosis………………………………………………………………………22	 Figure	2		 Study	Flow	Chart…………………………………………………………………………………………….23	 Figure	3		 Lung	transplant-free	survival	curve	of	entire	cohort……………………………………….24	Figure	4	 Plot	of	Schoenfeld	residuals	for	age	at	diagnosis	variable……………………………….25	Figure	5	 Adjusted	survival	curves	for	age	at	diagnosis……………………………….	………………..26	Figure	6	 Adjusted	survival	curves	for	FEV1	%	predicted……………………………….	………………27	Figure	7	 Adjusted	survival	curves	for	CFRD	status………………………………………………………..28	Figure	8		 Adjusted	survival	curves	for	standardized	comparison	for	FEV1	%	predicted	and	Age	at	diagnosis……………………………………………………………………………………………..29			 		 viii		List	of	Abbreviations	and	Acronyms	 Abbreviation Meaning BMI Body mass index CCFR Canadian Cystic fibrosis registry CF Cystic fibrosis CFFPR Cystic fibrosis foundation patient registry CFRD Cystic fibrosis related diabetes CFTR Cystic fibrosis transmembrane conductance regulator CI Confidence interval COPD Chronic obstructive pulmonary disease FEV1 Forced expiratory volume in one second GI Gastrointestinal GLI Global lung initiative HR Hazard ratio LTFU Lost to follow-up MRSA Methicillin-resistant staphylococcus aureus NTM Non-tuberculosis mycobacteria PH Proportional hazards PY Patient years US United States           	 ix	 Acknowledgements	 I offer my immense gratitude to Drs. Bradley Quon, Joel Singer, and Hubert Wong for guiding me throughout my master’s education and thesis formulation. I was fortunate to be mentored by such knowledgeable, insightful and supportive individuals.  I also thank Dr. Terry Lee for providing assistance on programming during the analysis and also Dr. Denise Mak for invaluable assistance on my endless questions about CF Canada Registry data.   Special thanks are owed to my parents, who encouraged me to continue my studies and provided boundless support as I took on this journey.  Last, and most importantly, I cannot thank enough my loving wife, Mariyam and my two sons, Isa and Musa, for being there for me at every step of the way. They are my inspiration and motivation to continue my work in this field.   This work was supported in part by the Rare Disease Foundation microgrant (#18-45).              	 1	1									Introduction	 Cystic fibrosis (CF) is the most common, fatal genetic disease affecting Canadian children and young adults. Approximately 4,100 Canadians are currently living with CF and it has an annual incidence of 1 in 3600 live births (1). CF is caused by a defective protein called the CF transmembrane conductance regulator (CFTR), responsible for secretion of chloride in the epithelial cells of our bodies. Defective chloride secretion leads to the development of thick mucus in the airways which results in chronic pulmonary infections. Other common clinical manifestations include difficulty digesting fats and proteins, CF-related diabetes (CFRD) and sinus infections. Ultimately, recurrent pulmonary infections cause permanent lung destruction and continual loss of lung function over time, resulting in the need for a lung transplant or premature death in the majority of CF patients.  The current diagnostic criteria for CF includes identification of at least one clinical feature of CF (e.g. family history, bronchiectasis, hemoptysis, etc.) and evidence of CFTR dysfunction by a biochemical marker (i.e. elevated sweat chloride) or confirmation of CF-causing mutations on both alleles (2). Approximately 60% of CF cases are diagnosed within the first year of life and about 90% are diagnosed by eighteen years of age (1,3,4) (Figure 1). The remaining 5-10% represent a very rare sub-population of CF cases that are diagnosed in adulthood and will be the focus of the current study. Although historically rare, recently researchers have noted a rise in adult-diagnosed cases, possibly attributed to increased awareness of disease heterogeneity among physicians, more wide-spread genotyping availability, and more streamlined diagnostic criteria (5,6). Though, it is likely that implementation of routine newborn and prenatal screening programs will reduce the incidence of adult-diagnosed CF in the future, thus limiting the ability to discriminate this rare sub-population from the overall CF population (7). Despite being in the minority, adult-diagnosed patients face unique challenges, issues, and needs compared to childhood-diagnosed patients. For instance, they are more likely to be married, attending educational institutions, working full-time and raising a family at the time of diagnosis (8). For most adult-diagnosed patients, the diagnosis of CF is accompanied by shock, fear, and depression. For others, the diagnosis could be an explanation for symptoms they had experienced for a long time and bring a sense of relief (9). Regardless, they are concerned about the prognostic implications of the disease and it is not surprising that the most common question at the time of diagnosis is about life expectancy (9). However, due to the relative rarity of adult-diagnosed CF cases and the paucity of available prognostic 	 2	literature for this population, this question remains difficult to answer by most physicians leaving the patient confused and frustrated (9,10). Some patients resort to online resources and material which focuses on childhood-diagnosed CF and cites life spans that are relatively short and sometimes briefer than the patients' ages, adding to their perplexity. Others are unwilling to attend CF care centers and adhere to rigorous CF treatment modalities (many of which are nebulized and thus time-consuming) due to limited data on expected disease progression. As a result, clinicians often also find it challenging to educate these patients on the need for treatments (5).  	1.1	Review	of	the	literature	 Clinical Characteristics  The first published report of an adult-diagnosed CF patient occurred in 1946 (11). Since then, there have been numerous case studies describing these patients in further detail (12–15). Most of these studies conclude that adult-diagnosed patients are atypical with a milder form of the disease than what is typically observed in childhood-diagnosed CF. Case studies are an important first step in research. However, they lack the ability to generalize the findings due to small sample sizes, thus preventing researchers from making meaningful inferences regarding the population of interest.  The first population-based study on this population was conducted by Widerman et al. in 2000 (8). They sought to describe clinical characteristics and socio-demographic differences between adults diagnosed in childhood (<18 years) and adulthood (>18 years) using the United States (US) Cystic Fibrosis Foundation Patient Registry (CFFPR) for the year of 1996. There were 786 adult-diagnosed patients included, representing approximately 10.9% of all CF patients for that year. In terms of social factors, they found that they were more likely to be college graduates, employed full-time, and married, as expected. They also experienced fewer complications, cultured less severe bacterial organisms, had a lower prevalence of common mutations, fewer hospitalizations, less oxygen use, less enzyme use, and lived longer than childhood-diagnosed patients. Based on these findings, Widerman et al. concluded that adult-diagnosed patients are a unique subgroup of adults with CF. Although this study offered great insight into the characterization of adult-diagnosed CF patients using a large sample size, there were several limitations. Firstly, they queried the CFFPR registry which does not adequately cover all CF patients, particularly those diagnosed in adulthood (5). Secondly, Widerman et al. did not employ multivariable modeling to account for confounders when comparing childhood-diagnosed to adult-	 3	diagnosed CF patients, which could affect their conclusions regarding some statistically significant bivariable associations. Lastly, many comparisons were examined in which an excess of 10% of missing data for at least one of the variables was noted, possibly biasing the results. Gilljam et al. (6) conducted the first Canadian study examining adult-diagnosed CF patients using a retrospective comparative cohort study design from 1960 to 2001, with an emphasis on the period from 1990 to 2001. They observed an exponential increase in adult diagnoses from 1960 to 2001, possibly due to increased physician and patient awareness. They further confirmed Widerman et al.'s findings that adult-diagnosed CF patients exhibited a milder form of the disease compared to the childhood-diagnosed patients based on clinical characteristics. Moreover, Gilljam et al. (6) emphasized that these patients are a unique subset of the overall CF population with a wider heterogeneity of disease. Notably, one-third of the patients did not have a conclusive sweat test and needed additional diagnostic testing to confirm their diagnosis. However, this study only examined patients from two clinics in Toronto, Canada and thoroughly investigated just 46 adult-diagnosed patients from 1990 and onwards due to incomplete data for patients diagnosed prior to that year – thus limiting the generalizability of the results. Nick et al. investigated long-term survivors of CF (>40 years of age) and compared those diagnosed in childhood to those diagnosed in adulthood. This study was single center and identified 55 patients between 1992 and 2004 who were 40 years and older, and of those 27, were diagnosed as adults. The findings supported the conclusions of previous reports. Namely, adult-diagnosed CF patients have less severe lung disease, were more likely to have pancreatic sufficiency, lower prevalence of P. aeruginosa, and lower prevalence of severe mutations (16). Interestingly, the authors also noted an unusually high prevalence of non-tuberculous mycobacteria (NTM) in the adult-diagnosed group relative to the childhood-diagnosed group; a novel finding for this cohort but potentially biased as the National Jewish Hospital is a regional referral center for difficult to treat cases of NTM. The greatest strength of this study was comparing a homogenous group of age-matched childhood-diagnosed CF patients to adult-diagnosed CF patients to account for possible age-associated effects. Adult-diagnosed patients are not diagnosed until adulthood, whereas childhood-diagnosed patients are diagnosed in early childhood leading to differential follow-up characteristics and the heightened potential for biased conclusions if age is not taken into account. Though, the authors acknowledged that the included patients consist of many out-of-state referrals for specialized assessment at their institution, therefore the 	 4	generalizability of the results are greatly affected since it most likely does not represent the general adult-diagnosed CF population.  Prognosis  Nick et al. conducted another study, focusing again on long-term survivors, aged 40 years and older from nearly the same period (1992 to 2008) but now the goal was to evaluate disease progression in this population using the CFFPR (17). It was found that adult-diagnosed patients exhibited similar age-related lung function decline and frequency of death from respiratory failure from age 40 and onwards compared to childhood-diagnosed CF patients (17). This result was unexpected because previous literature established adult-diagnosed patients to have a milder form of CF. However, this was the first prognostic study in this population at that time, and it appeared that the pulmonary manifestations at the time of diagnosis are not mild but delayed and represent an equally serious form of CF compared to age-matched childhood-diagnosed CF patients. Since this study specifically selected CF patients who lived until the age of 40, the results from this cohort would be subject to survivor bias. In other words, the results would only apply to those patients who live until 40 years of age, with no information regarding those who were diagnosed as adults but died before the age of 40.    The most recent prognostic study by Keating et al. (18) used the CFFPR to study lung function decline and survival in different age groups, including adult-diagnosed CF patients from 1993 to 2003 and followed up until 2010. They found the rate of lung function decline in adult-diagnosed patients to be -1.13% per year, almost double the rate found in Nick et al.'s study (17). Unadjusted 10 and 15-year survival was 76% and 65%, respectively, with P. aeruginosa and Burkholderia complex infections as the only significant factors associated with survival. Additional regression models with lung function decline as the outcome also found that baseline lung function, body mass index (BMI), the presence of a ∆F508 mutation and P. aeruginosa were associated with faster rates of decline. This study, however, had many drawbacks that may affect the conclusions. Firstly, they only included patients with complete follow-up, which resulted in approximately 35% of the original sample to be excluded from the final analysis. Generally, patients with complete follow-up are systematically different than those loss to follow-up. Also, Nick et al cross-referenced adult-diagnosed patients in their institutional database to the CFFPR and found missing deaths in up to 50% of patients beyond the age of 45 years and concluded that long-term survivors of CF, with majority being adult-diagnosed patients, are not frequently seen at CF centers and many die without being accounted for in the CFFPR. The large number of patients lost to 	 5	follow-up, together with the potential of missing deaths may be biasing the results from Keating et al.'s study. In fact, Sykes et al empirically assessed the impact of missing deaths and loss to follow-up in prognostic studies in CF and determined by simulations that missing just 15% of deaths can result in overestimation of survival by 3.5 years and having 25% of patients loss to follow-up can underestimate survival by 3.3 years (19).  Overall, the current body of evidence supports adult-diagnosed CF patients to be a unique subset of the overall CF population that exhibit a milder form of disease in comparison to childhood-diagnosed patients, while recognizing the wide disease heterogeneity among adult-diagnosed patients. However, current evidence concerning life expectancy and prognosis is limited and remains poorly characterized.   1.2	Rationale	and	objectives	of	the	study			 The lack of high-quality evidence specifically about the prognosis of adult-diagnosed CF patients in the current literature and the recent increase in the frequency of this diagnosis provides the rationale for this study. The objectives of this population-based study are to describe the clinical characteristics, estimate lung transplant-free survival, and identify significant clinical predictors of reduced survival at the time of diagnosis in adult-diagnosed CF patients. The findings from this study will help CF clinicians educate these patients on the need for treatments, expected disease course, and life expectancy.   	 6	2									Methods	 2.1	Data	source	 The Canadian Cystic Fibrosis Registry (CCFR) was created in the 1970s with the goal of monitoring important clinical trends in the Canadian CF population. The CCFR captures individuals with a confirmed diagnosis of CF based on current diagnostic guidelines (20). It contains detailed demographic and clinical information on CF patients receiving care in 42 accredited CF centers across Canada (1). These data undergo routine validation checks and discrepancies are resolved by contacting the reporting CF Center to cross-reference with original data. Additionally, CF Canada provides funding to centers dependent on data submission to the registry and CF patients must attend CF centers as many expensive medications are covered by provincial drug plans if prescribed by a CF physician. For these reasons, the CCFR is recognized for its completeness, accuracy and comprehensive coverage of CF patients across Canada (21).   	2.2	Study	design	and	sample	 Population-based cohort study using the CCFR of newly diagnosed adult CF patients ≥ 18 years of age between the years 1990 and 2014 (Figure 2). The year 1990 was chosen as the study start date because it was after the discovery of the CF gene in 1989 (22) which allows for consistent diagnostic pathways during the study period. Additionally, prior to 1990 the CCFR had very limited clinical data which made it more appropriate to start from 1990 given the objectives of the present study.  2.3	Study	variables		Demographic variables of age at diagnosis, sex (male or female) and race (caucasian, black, asian, south asian, other or unknown) were included in the descriptive analysis. Other clinical variables such as symptoms at diagnosis (asymptomatic, gastrointestinal, minor manifestations, pulmonary, pulmonary and GI, other or unknown), genotype (∆F508 heterozygous, ∆F508 homozygous, one mutation only, two non-∆F508 or unknown), CFRD (yes or no) which was defined as hyperglycemia based on oral glucose tolerance test results or random blood sugar measurements requiring insulin therapy (23), pancreatic insufficiency (yes or no)  based on enzyme usage, P. aeruginosa culture positive (yes or no), and cause of death (pulmonary, cardiovascular, gastrointestinal, transplant-related, infection/sepsis or other) were also described. For the characterization of the disease, CFRD, pancreatic insufficiency, and P. 	 7	aeruginosa culture positive were considered ever variables, which means that if an individual was assigned positive status (yes), for these variables at any time point during the study period, they remained that status for the entire study period. The reason for this distinction was due to limitations in the CCFR database, in which the start/end date was not included for patients diagnosed with CFRD or administered enzymes (pancreatic insufficiency) and therefore it can be considered the approximate life-time prevalence of these characteristics. 	 Baseline variables available at the time of diagnosis (defined as ≤ two years from diagnosis date) were also evaluated. These variables included the demographic factors, symptoms at diagnosis, genotype, CFRD, BMI, stable lung function measurements, and P. aeruginosa culture. Lung function was measured by forced expiratory volume in one second (FEV1) and expressed as a percentage of predicted based on the GLI equation (24,25). For variables which had multiple measurements within the baseline period, the one closest to the diagnosis date was selected. Additionally, it was assumed that any patient with no cultures tested was negative for P. aeruginosa culture. The study variables of race, symptoms at diagnosis, and genotype were collapsed into broader categories during modelling because many of their categories contained small cell numbers affecting the convergence of the regression models.  The outcome variable was death or transplant (i.e. lung transplant-free survival), whichever occurred first. The selection of transplant recipients is based on the recommendations by the Pulmonary Council of the International Society for Heart and Lung Transplantation (ISHLT) and shown in Table 1 (26). Generally, patients are considered for transplantation when their condition is worsening and have a 2-year predicted survival of <50%. The survival time was calculated from the diagnosis date to the outcome date and since the objective of the analysis was to identify variables predictive of lung transplant-free survival only, time since diagnosis was the time scale selected (27). If the outcome did not occur or the patient was loss to follow-up (defined as those patients alive with their last verified contact occurring more than two years before the study end date), the patient was censored at their last reported visit.  2.4	Statistical	analysis		 All analyses were conducted using R statistical programming (28). Patient demographics and clinical characteristics were described using simple descriptive statistics (medians, means and proportions). A Kaplan-Meier Curve was computed to estimate the overall 10 and 15-year lung transplant-free survival probability of the entire cohort. Cox proportional hazards (PH) models were used to determine the 	 8	predictive utility of clinical factors after accounting for the other model variables. The PH assumption was tested for each variable (and was found to be violated for the age at diagnosis variable) with a plan to split the follow-up period at a time point where the assumption appeared to begin to fail and a time interaction term was included to assess the change in hazard ratio. Moreover, the functional form of the continuous covariates was also checked by examining the martingale residuals and adding a quadratic term. Based on the final model, adjusted survival curves were then generated using the corrected group prognosis method (29). Briefly, this method calculates survival curves for each patient in the study assuming each patient was exposed for a given variable; then the predictions are averaged. The same predictions are obtained and averaged assuming each patient is now non-exposed for that same given variable. This method allows for an adjusted assessment of the potential absolute impact of a given predictor by considering differences in the survival curves arising from varying the values of each predictor. The associated 95% CIs for the predicted survival probabilities were calculated by non-parametric bootstrapping with 1000 replications using the percentile interval method.  Multiple imputation using the fully conditional specification algorithm was employed to account for missing data, under the assumption that missing data was missing at random. Eleven imputed datasets were created and then pooled using Rubin’s rules (30) for Cox regression analysis. Eleven imputed datasets were preferred based on White and Bodner’s recommendation of using the same number of imputations as to the percentage of incomplete cases missing in the dataset (31,32). Satisfactory convergence with no definitive trends was reached for the incomplete variables during imputation.   All variables that were found to have p-values below the a priori defined cut-off (p<0.25) in univariable testing or those with strong a priori hypotheses were added to the final multivariable model. BMI is a surrogate for the nutritional status of a patient and it is well-established in the overall CF population that increased BMI and nutritional status leads to better clinical outcomes, including survival (33–35). Thus, it is considered a pivotal metric of health among patients with CF and included in the final multivariable model. Moreover, numerous studies postulate a significant “gender gap” in CF where females have been found to have worse survival for a variety of reasons, some of which include genetic modifications, socioenvironmental factors as well as psychosocial factors (36,37). Therefore, sex was also added to the final multivariable model. Finally, the present study spanned a long time-period, thus diagnosis year was added into the final model to mitigate any cohort effects.  The level of significance was p < 0.05 for all statistical analyses, and all reported p-values reflect 	 9	two-tailed tests. All patients within the registry provided informed consent to have their data collected and be used for research purposes. The study was approved by the Research Ethics Board at St Paul’s Hospital, Vancouver, Canada (PHC-REB# H15-01439).   	 10	3									Results		3.1	Study	sample	characteristics	In total, 362 patients were diagnosed with CF as adults between 1990 and 2014. The median follow-up time observed was 7.7 years (range: 0.0-23.6) and included 33 deaths and 15 transplants for a total of 48 events in 3,106 patient years (15.5 events per 1000 PYs). Forty-eight patients (13.3%) were lost to follow-up (defined as those patients not seen for at least 2 years before study end date) over the study period.  Table 1 describes the clinical characteristics of the overall sample during the study period. The study sample was nearly equally distributed among males and females with a median age at diagnosis of 34.3 (range: 18.0-73.8). The majority of the sample was caucasian (90.3%) and patients primarily presented with pulmonary and/or gastrointestinal (GI) symptoms (70.5%). Sixty-two patients (17.4%) also presented with no symptoms or minor manifestations (e.g. clubbing on its own, hyponatremia, nasal polyps, etc.) and the remaining proportion (12.0%) either had other symptoms that could not be categorized or it was not clear or known what the presenting symptoms were. The most common genotype was ∆F508 heterozygous (38.1%), followed closely by one mutation only (31.2%). Only seventeen patients (4.7%) were ∆F508 homozygous. During the study period, the prevalence of CFRD (ever) was 15%, 35% of the sample was pancreatic insufficient (ever), and 50% were culture positive for P. aeruginosa. At baseline, the mean BMI was 23.6 ± 4.3 kg/m2, the mean FEV1 % predicted was 77.6 ± 23.7%, 3.0% presented with CFRD and 30.4% were culture positive for P. aeruginosa. 3.2	Descriptive	statistics	by	event	status		Table 2 also presents patient demographics and clinical characteristics by event status at the end of the follow-up period. The most common cause of death was due to pulmonary issues (54.5%). Patients who died or transplanted were more likely to be older at diagnosis, have pulmonary/GI symptoms, CFRD, and culture positive for P. aeruginosa at baseline. Additionally, lung function at baseline was substantially lower and the prevalence of CFRD (ever) and pancreatic insufficiency (ever) was also higher in these patients. Overall, patients who died or transplanted during the study period demonstrated a more severe manifestation of CF at baseline.    	 11	3.3	Survival	analysis		 Figure 3 displays the Kaplan-Meier survival curve for all patients in the study. The 10-year and 15-year lung transplant-free survival probability was 87.7% (95% CI: 82.9%, 91.3%) and 86.1% (95% CI: 80.7%, 90.1%).  Table 3 presents the univariable and the final model Cox regression hazard ratios (HRs) for all variables available at baseline. During univariable testing, age at diagnosis, P. aeruginosa culture positive, CFRD and FEV1 % predicted all reached statistical significance (p<0.05) and were included in the final model. Additionally, symptoms at diagnosis was below the a priori defined p-value cut-off (p<0.25) and added to the final model, whereas sex and BMI were considered strong a priori hypothesized variables and also added to the final model. The PH assumption was tested for each variable and found to be unsatisfied for the age at diagnosis variable and therefore, the follow-up period was split at the 10-year time point (Figure 4), with a time interaction included in the final model to satisfy assumptions. In the final multivariable model, adult-diagnosed patients were more likely to die or be transplanted in the first 10 years following diagnosis if they were older at diagnosis (HR: 1.32 per 5 years, 95% CI: 1.13, 1.55). However, the effect of age at diagnosis was not sustained after the 10-year period (HR: 0.98, 95% CI: 0.75, 1.29). In addition, those with lower lung function (HR: 0.76 per 5% increase, 95% CI: 0.69, 0.83) at baseline were also more likely to die or be transplanted during the study period. CFRD was also identified as a significant predictor of reduced survival (HR: 7.86, 95% CI: 2.09, 29.55), however, the confidence interval indicates imprecision with a very wide width. The influence of the significant predictors on reduced survival identified in the final multivariable model were illustrated by considering differences in adjusted survival curves and 15-year lung-transplant free survival probabilities using the corrected group prognosis method.  Scenario 1: Age at diagnosis was increased by 10 year increments (30 vs 40 vs 50 vs 60) Scenario 2: Baseline FEV1 % predicted was increased by 10% (40% vs 50% vs 60% vs 70% vs 80%) Scenario 3: CFRD positive patients compared with CFRD negative patients Scenario 4-5: The values of the first and third quartiles for age at diagnosis (27.1 years and 44.1 years) and baseline FEV1 % predicted (62.4% and 94.3%)  Table 4 presents the predicted survival probabilities for each scenario from above. The adjusted survival curves for scenario 1 demonstrated that increasing age at diagnosis by 10 years reduced survival (Figure 	 12	5), although the survival difference between each 10-year increment appears to be minimal for the first 10 years post-diagnosis while becoming more pronounced over time. The predicted 15-year lung transplant-free survival probabilities for age at diagnosis were high for the younger ages at diagnosis (30 and 40) but reached below 60% for a diagnosis made in the oldest age at diagnosis (60 years). For scenario 2, the adjusted survival curves for each 10% increase in baseline FEV1 % predicted were distinctive immediately from diagnosis, with the survival difference becoming even larger as the baseline FEV1 % predicted decreased further. The predicted 15-year lung transplant-free survival probabilities for FEV1 % predicted decreased from 84.0% (95% CI: 43.2%, 96.5%) for having 80% FEV1 % predicted (high lung function) to 43% (95% CI: 22.2%, 75.3%) survival for having 40% FEV1 % predicted (low lung function) equating to a 41% difference in survival. Patients with CFRD also experienced a marked decrease in lung transplant-free survival, with the adjusted difference being 33% at 15-years post-diagnosis (Figure 7). Standardized comparisons for age at diagnosis and FEV1 % predicted showed that the difference in adjusted curves were larger for FEV1 % predicted compared to age at diagnosis (Figure 8). Additionally, the predicted 15-year lung transplant-free survival probability difference for FEV1 % predicted was 23.1% (from 91.9% to 68.8%) compared to 11.2% (from 83.9% to 73.7%) for age at diagnosis.  3.4	Sensitivity	analyses		 Since the present study assumed patients who had no cultures examined for P. aeruginosa were negative at baseline, a sensitivity analysis was conducted for possible misclassification. P. aeruginosa culture status was extended to five years after diagnosis and included in the final model as a substitute for the original P. aeruginosa at baseline variable.  It resulted in no changes in the overall conclusions of the model (Table 5). Additionally, CFRD status at baseline was identified as a significant predictor in the present study; however, the number of patients with CFRD was small (n=10, 3%) which resulted in wide confidence intervals and a large standard error (Table 3). Therefore, an investigation of the final model (data not shown) which substituted CFRD at baseline with the CFRD ever variable was conducted to see if this association persisted. The HR for the CFRD ever variable was attenuated and variable (HR: 1.51, 95% CI: 0.81, 2.81) compared with CFRD at baseline (HR: 7.86, 95% CI: 2.09, 29.55), but was still demonstrating harmful effects. This finding corroborates the conclusions of CFRD status at baseline being a significant predictor of reduced survival in the adult-diagnosed population, but it likely is less pronounced than what the multivariable model indicates. Lastly, the patients lost to 	 13	follow-up (LTFU) had significantly less pulmonary/GI symptoms, less prevalence of P. aeruginosa positivity, one mutation only, higher baseline lung function and more likely to be male. These findings suggest that the LTFU patients appear to be healthier than those not lost to follow-up and most likely not attending their visits because they are feeling well. However, these patients will have little to no impact on the current study results as studies show LTFU rates of 10% or lower do not change results (19). At most, the HRs in the present study may be slightly over-estimated for some predictors as the LTFU patients are healthier but not enough to change the overall conclusions of the model.  	 14	4									Discussion			4.1	Summary	of	findings		The current study used a comprehensive, national population-based CF patient registry known for its completeness and accuracy to describe the characteristics of adult-diagnosed CF patients, estimate their 10 and 15-year lung transplant-free survival, and identify significant predictors of reduced survival. The clinical characteristics of this population were generally consistent with the literature, however they had less severe disease based on lower rates of pancreatic insufficiency, CFRD, P. aeruginosa culture positive and a reduced prevalence of severe mutations compared to previous studies examining adult-diagnosed CF patients (6,8,16,18). They also had 10-and 15-year transplant-free survival of 88% and 86%, respectively.  The study also identified older age at diagnosis, lower lung function, and CFRD to be significant predictors of reduced survival	in this population.  However, adjusted survival curves demonstrated lung function and CFRD to have the most clinical utility for prognosis. 	 4.2	Key	findings	in	the	context	of	existing	literature		Although, the clinical characteristics of the study sample were generally consistent with prior studies in the literature, adult-diagnosed CF patients followed in the CCFR have a milder disease at the time of diagnosis (6,38). Particularly, rates of pancreatic insufficiency, CFRD, severe mutations and P. aeruginosa culture positive were markedly lower in the present study in comparison to previous evidence for this population. This might be explained by most studies using the CFFPR from the US as their data source which has been reported to underrepresent adult-diagnosed patients (17) and potentially biasing the clinical characteristics toward more severe adult-diagnosed patients who are more likely to be followed due to more active disease. The present study utilized a registry known for comprehensive coverage of CF patients in Canada, and the current results may indicate that the CCFR follows patients with a wider spectrum of the disease compared to the CFFPR that tends to capture the severest.   Life expectancy is a major concern for adult-diagnosed patients at the time of diagnosis (9) but there are limited studies investigating this. Recently, Keating et al reported that 10- and 15-year survival rates using the CFFPR were estimated at 76% and 65%, respectively, compared to the present study which estimated the 10- and 15-year survival rates to be much higher at 87.7% (95% CI: 82.9%, 91.3%) and 86.1% (95% CI: 80.7%, 90.1%). Even the lower bound of the CIs for the present study’s survival 	 15	rates exceed Keating study’s findings. The higher survival rate in the present study further supports the above hypothesis that CCFR follows a more diverse spectrum of adult-diagnosed CF patients who have a milder phenotype compared to the CFFPR which includes more severe adult-diagnosed patients. Nick et al also investigated survival in this population and concluded that the median survival in their adult-diagnosed cohort was 76.9 years, higher than childhood manifested patients (17). However, Nick et al only studied long-term survivors of CF who survived up to 40 years of age, resulting in biased estimates and therefore, their findings are not comparable to Keating et al (18) and the present study. Nonetheless, it should be reassuring for newly diagnosed adult CF patients that their survival is lengthier than previously reported in the literature.  Increased survival in this population could be attributable to an attenuated disease course because of less CFTR dysfunction compared to the more classic childhood-diagnosed CF patients. Typically, severe mutations in the CFTR protein disrupt either its quantity or function leading to minimal CFTR activity, however, milder mutations often result in residual CFTR function and milder phenotype including preserved pancreatic function (39). In the present study, 84.2% of the patients had genotypes with at least one non-∆F508 mutation and another 11% had unknown mutations, whereas only 5% were homozygous for the ∆F508 mutation, a severe genotype. Homozygous ∆F508 prevalence in the present study was less than what is reported in the literature for the adult-diagnosed population (8,17) and substantially less than the overall CF population which is primarily made up of patients diagnosed in childhood.  For instance, the CFFPR reports 50.6% of their overall CF population that are ∆F508 homozygote and the CCFR also reports 48.1% of their patients that are ∆F508 homozygote (3,21); a stark increase relative to the adult-diagnosed patients in the present study.   Many of the factors associated with survival in adult-diagnosed CF population are derived from childhood-diagnosed CF studies and since adult-diagnosed patients are less severe, these previous studies may not be relevant. Only one study to date has analyzed predictors of survival in this specific adult-diagnosed population and identified P. aeruginosa positive, Burkholderia complex positive and Methicillin-resistant staphylococcus aureus (MRSA) positive as the only significant predictors of reduced survival (18). The present study identified FEV1 % predicted, CFRD and age at diagnosis to also be significant predictors of reduced survival. While this study evaluated P. aeruginosa as a potential prognostic factor, it did not find sufficient evidence to conclude that an association exists between P. aeruginosa and survival (HR: 0.74, 95% CI: 0.39, 1.42). The reason for the conflicting results may be related to the selection of variables in the Cox regression model. Keating et al’s study 	 16	which identified P. aeruginosa as an important prognostic factor, did not include age at diagnosis and lung function as covariates in their model. Although the present study showed a strong association between P. aeruginosa and survival during univariable testing (Table 2), this effect was no longer present after including age at diagnosis and lung function in the multivariable model. It appears that P. aeruginosa at baseline in this population may be redundant once extent of lung function impairment and age at diagnosis have been accounted for. Furthermore, a post-hoc analysis in the present study (Table 6) confirmed that patients with P. aeruginosa culture positive were older and had lower lung function, suggesting overlap in the predictive utility of these two variables. It is also plausible that milder adult-diagnosed patients may be less affected by the infection, particularly if it is a less virulent, non-mucoid strain potentially resulting in less pathogenic consequences or easier eradication than what is observed in the general CF population. However, the severity of infection could not be evaluated in the current study because P. aeruginosa was considered a binary variable and perhaps not sufficient to capture an association.   The present study identified older age at diagnosis during the first 10 years as a significant predictor of reduced survival in this population, but this effect subsides after 10 years.  This finding means that older aged patients are at the highest risk of dying or being transplanted during the first 10 years post-diagnosis. After 10 years, many of the older patients will have died or transplanted, while the remaining patients will have been diagnosed in early adulthood and within this group of younger patients, the effect of age at diagnosis is not as strong. The comparison of characteristics by event status in Table 1 supports this explanation. Patients who died or transplanted were diagnosed approximately 8 years later with a substantially lower lung function at baseline and more severe symptoms. Furthermore, the median age at death or transplant in this cohort was 54.8 years; a relatively old age. These findings could mean that a combination of natural age and CF-related lung function decline may be a possible reason older aged adult-diagnosed patients are more likely to die or be transplanted.  CFRD at baseline was also another significant predictor of reduced survival, but the HR and confidence intervals suggest a great deal of uncertainty for this finding. The large HR and wide confidence intervals might be due the small number of patients with CFRD at baseline (n=10) and shortened follow-up compared to patients without CFRD as well as collinearity with other variables in the model. The results from the present study need to be replicated in future studies with larger sample sizes. However, the present study conducted a sensitivity analysis by substituting the CFRD at baseline variable with the CFRD ever variable in the final model increasing the number of patients with CFRD 	 17	and there was a trend towards significance in the relationship. Nonetheless, the current results warrant physicians to be cognizant of adult-diagnosed patients presenting with CFRD at baseline and its probable harmful effects on survival. Many previous studies have associated CFRD with increased mortality (40), however, there are no studies to date that have studied CFRD in the adult-diagnosed population. The mechanism by which developing CFRD impacts survival is unclear but it is hypothesized that CFRD adds an additional inflammatory burden that together with CF-related pulmonary inflammation creates a greater risk of death (40).   Lung function is considered one of the most important measures of disease severity, as well as progression, and therapeutic efficacy in the CF population (41). The present study is consistent with the existing evidence and supports the importance of lung function, as measured by FEV1% predicted to be a prominent prognostic factor in the adult-diagnosed CF population (17,18). To put this into perspective, the mean baseline lung function in the present study of adult-diagnosed CF patients was 77.6 ± 23.7 % predicted whereas the mean lung function from two population-based studies of healthy, non-smoking caucasian adult subjects in Canada was 98.7 ± 13.9 % predicted for women and 99.4 ± 13.0 % predicted for men (42). Furthermore, the lung function in the present study was also much worse than subjects suffering from another respiratory disease, chronic obstructive pulmonary disease (COPD) from the Canadian Obstructive Lung Disease study, in which approximately 6% of never smoking and 15% of ever smoking COPD subjects had FEV1 % predicted below 80%, compared to 47.9% (data not shown) in the current study (43). Thus, adult-diagnosed CF patients may be considered milder and less severe compared with the overall CF population but their lung function is much worse compared to the general population and other diseased populations.  Based on the comparisons of adjusted survival curves and predicted 15-year lung transplant-free survival probabilities, FEV1 % predicted and CFRD were strong predictors of reduced survival and may have clinical utility and importance (Table 4). CFRD positive patients exhibited a 33% decrease in predicted 15-year lung transplant-free survival compared with CFRD negative patients, whereas patients with low lung function (40% predicted) had a 23.3% and 41% decrease in 15-year predicted lung transplant-free survival probability compared with moderate lung function (60%) and high lung function at baseline (80%), respectively. On the contrary, it appears that age at diagnosis adds only moderate value to the final multivariable model, particularly for younger ages at diagnosis for which lung transplant-free survival was high. The adjusted curves and predicted survival probabilities only showed modest decreases in survival as the age at diagnosis increased but the survival differences become more 	 18	noticeable at the older ages (age at diagnosis > 50 years). This is not surprising as older aged individuals would have a high baseline risk because of their age and are more likely to die regardless.  It is important to note that the survival probabilities are derived from a model that does not include potentially important predictors (e.g. treatment) and may change substantially if that information were considered. The uncertainty of the survival probabilities is also reflected by the wide confidence intervals and therefore, the results from these models are best interpreted as a rough guide on the potential usefulness of each predictor in the model.  4.3	Weaknesses	of	study	 There were some limitations to the present study. Firstly, since data was extracted from a registry, there were variables unavailable or only recently began to be captured in the CCFR, such as frequency of hospitalizations, sweat chlorides, and detailed specific microbiology (MRSA, S. maltophilia). Therefore, not all potential predictors were assessed in the present study and future studies should incorporate these variables in their analysis. Additionally, CCFR only recently began to capture treatment data and as a result, they were also not included in the model. Therefore, it is difficult to know if the survival analysis reflects the natural history of the disease or is confounded by the effects of treatment. Secondly, since our patient population only includes Canadian adult-diagnosed CF patients, the generalizability of the present study is limited to Canada. There may be differences in outcome due to healthcare delivery and policy differences for care of CF patients between countries (44). Thirdly, adult-diagnosed CF patients are rare, and therefore the present study was restricted to a relatively small sample size and number of events. Ultimately, this affects the power of the final multivariable model and careful attention must be given to the bounds of the confidence intervals for non-significant predictors; in some cases, the present study was unable to rule out important effects. Likewise, one must be cautious in extrapolating these results to individual patients because the current multivariable model was not validated in a training dataset or an independent cohort. Additionally, the results of the current study span a 24-year period starting from 1990, raising the possibility of cohort effects, but the present study did include diagnosis year in the final multivariable model in an attempt to adjust for these effects. Despite this, the relationships identified in the present study could have evolved and may not be entirely reflective of recently adult-diagnosed CF patients. Lastly, baseline measurement was accepted to occur within two years of diagnosis, allowing the use of data from as many patients as possible. Many adult-diagnosed patients, particularly the milder and asymptomatic, may not undergo routine clinical care during their 	 19	first visit to a CF clinician. As such, it does not truly reflect the raw baseline values, although it is unlikely that these measures would have changed drastically within the designated period. Nonetheless, epidemiological studies provide tremendous insight into understanding the natural history of a disease and more precisely, to identify risk-factors for outcomes at the population level. 	 		 20	5	Conclusion	 The present study took advantage of a reliable and comprehensive data source to describe the clinical characteristics and evaluate the prognosis of adult-diagnosed CF patients. It found that these patients have a milder phenotype of the disease and better prognosis than previously reported. Additionally, older age at diagnosis, lower lung function at baseline, and CFRD were identified as significant predictors of reduced survival in this population. In terms of clinical utility, older age at diagnosis (> 50 years), baseline lung function (FEV1 % predicted < 60%), and CFRD were the important predictors of lung transplant-free survival. Adult CF clinicians and other CF caregivers can use this information to educate their patients about their prognosis and to guide treatment, specifically for those patients at high-risk for a worse prognosis in this population.   	 		 21	6	Future	Directions	Future studies in this population should closely examine longitudinal lung function decline, an important predictor in the present study, using more sophisticated techniques such as mixed modelling approaches that enable estimation of population average rates of decline as well as individual patient variation. A longitudinal analysis of lung function may provide further insight into the prognosis of adult-diagnosed CF patients. Additionally, studies should incorporate more potentially important predictors of survival with larger sample sizes that will yield more accurate and reliable predictions of survival probabilities, possibly providing greater clinical utility than the present study. Lastly, studies should begin to explore the eligibility and efficacy of promising new therapies such as CFTR modulators to determine if this population may stand to benefit, particularly due to the advent of newborn screening in many countries that will enable early identification of CF disease in patients with milder genotypes and likely asymptomatic. Initiating therapies early or later in this population will be an important clinical question for future studies.   	 22	7	Figures	 Figure 1. Distribution of CF patients by age at diagnosis (from CF Canada Registry Annual Report, 2015)            	 23	Figure 2. Study Flow Chart             	 24	Figure 3. Lung transplant-free survival curve of entire cohort, 1990-2014   	 25	Figure 4. Plot of schoenfeld residuals for age at diagnosis variable (p<0.05)   	 26	Figure 5. Adjusted survival curves from final multivariable model:  Age at diagnosis varied by 10 years from age 30 to 60    	 27	Figure 6. Adjusted survival curves from final multivariable model:  FEV1 % predicted at baseline varied by 10%, starting from 40%    	 28	Figure 7. Adjusted survival curves from final multivariable model:  CFRD positive versus CFRD negative at baseline    	 29	  Figure 8. Adjusted survival curves from final multivariable model:  Standardized to values of first and third quartile                         	 30	8	Tables	   Table 1: Indications for transplantation in CF patients (from ISHLT Consensus Document, 2015)  Timing of Referral Timing of Listing  FEV1 that has fallen to 30% or a patient with advanced disease with a rapidly falling FEV1 despite optimal therapy (particularly in a female patient), infected with non-tuberculous mycobacterial (NTM) disease or B cepacia complex (see previous comment on B cenoce- pacia and subsequently) and/or with diabetes.   Chronic respiratory failure. 
 ▪ With hypoxia alone (partial pressure of oxygen [PaO2] 
o8 kPa or o60 mm Hg). 
 ▪ With hypercapnia (partial pressure of carbon dioxide 
[PaCO2] 46.6 kPa or 450 mm Hg).

  A 6-minute walk distance <400 m
  Long-term non-invasive ventilation therapy  Development of pulmonary hypertension in the absence 
of a hypoxic exacerbation (as defined by a systolic pulmonary arterial pressure (PAP) 435 mm Hg on echocardiography or mean PAP 425 mm Hg measured by right heart catheterization)  Pulmonary hypertension
   Clinical decline characterized by increasing frequency of exacerbations associated with any of the following:  ▪ An episode of acute respiratory failure requiring non- invasive ventilation. 
 ▪ Increasing antibiotic resistance and poor clinical recovery from exacerbations. 
 ▪ Worsening nutritional status despite supplementation. 
 ▪  Pneumothorax. 
 ▪  Life-threatening hemoptysis despite bronchial embolization. 
  Frequent hospitalization

   Rapid lung function decline

   World Health Organization Functional Class IV 
                  	 31	     Table 2:  Clinical characteristics for adult-diagnosed CF patients* in the CF Canada Registry from 1990 –  2014  Variable Total Sample  (n=362)         Event  (n= 48) Censored  (n= 314) Median Age at Diagnosis (years) 34.3 (18.0-73.8) 43.8 (18-73.8) 33.9 (18-71.7) Sex      Male 189 (47.8) 24 (50.0) 165 (47.5)   Female 173 (52.2) 24 (50.0) 149 (52.6) Race       Caucasian 327 (90.3) 45 (93.8) 282 (89.8)   Non-Caucasian 11 (3.0) 2 (0.04) 9 (2.9)   Unknown 24 (6.7) 1 (0.02) 23 (7.3) Symptoms at Diagnosis      Asymptomatic 37 (10.2) 0 (0.0) 37 (12.3)   Gastrointestinal 21 (5.8) 0 (0.0) 21 (7.0)   Pulmonary  166 (45.9) 28 (63.6) 138 (46.0)   Pulmonary and GI 68 (18.8) 13 (29.6) 55 (18.3)   Minor Manifestations 26 (7.2) 0 (0.0) 26 (8.7)   Other 18 (5.0) 1 (2.3) 17 (5.7)   Unknown 26 (7.2) 2 (4.6) 6 (2.0) Mutation      Homo ∆F508 17 (4.7) 1 (2.1) 16 (5.1)   Hetero ∆F508 138 (38.1) 17 (35.4) 121 (38.5)   One mutation only 113 (31.2) 15 (31.3) 98 (31.2)   Two Non-∆F508 54 (14.9) 10 (20.8) 44 (14.0)   Unknown 40 (11.1) 5 (10.4) 35 (11.2) CFRD (ever)       Yes 56 (15.5) 18 (37.5) 38 (12.1)   No  306 (84.5) 30 (62.5) 276 (87.9) Pancreatic Insufficiency (ever)       Yes 127 (35.1) 22 (45.8) 105 (33.4)   No 235 (64.9) 26 (54.2) 209 (66.6) P. aeruginosa Positive      Yes 181 (50.0) 36 (75.0) 145 (46.2)   No 181 (50.0) 12 (25.0) 169 (53.8) Cause of Death      Pulmonary  18 (54.5)    Cardiovascular  3 (9.0)    Gastrointestinal  1 (3.0)    Infection/Sepsis  1 (3.0)    Other  7 (21.6)    Unknown  3 (9.0)  Median Age at Death or Transplant (years)  52.9 (26.8-78.8)  Median Follow-up time (years) 7.7 [0.0-23.6]   5.8 [0.4-21.7]  7.9 [0.0-23.6] Lost to follow-upa 48 (13.3)   Baseline variablesb    	 32	Mean BMI (kg/m2) 23.6 ± 4.3 22.4 ± 4.6 23.8 ± 4.3 Mean FEV1 (% Predicted) 77.6 ± 23.7   50.1 ± 18.1 81.7 ± 21.6  CFRD       Yes 10 (3.0) 4 (9.3) 6 (2.1)   No 319 (97.0) 39 (90.6) 280 (97.9) P. aeruginosa Positive      Yes 110 (30.4) 24 (50.0) 86 (27.3)   No 252 (69.6) 24 (50.0) 228 (72.7) *	Numeric	summaries	are	median	(range)	or	mean	±	SD	and	categorical	summaries	are	n	(%)	^	Denominator	for	Cause	of	Death	is	No.	of	deaths	in	study	(n=33)	aLost	to	follow-up	defined	as	those	patients	alive	with	their	last	verified	year	of	contact	occurring	more	than	2	years	before	the	end	of	the	study	period		(Dec	31,	2014)	bBaseline	variables	are	summarised	from	those	patients	with	available	baseline	data,	defined	as	within	2	years	of	CF	diagnosis	                                   	 33	Table 3: Univariable and Final model Cox PH Regression results (Hazard Ratios) for event using baseline# characteristics of adult diagnosed CF Patients (n=362 patients and n=48 events in the final model)   Univariable HRs (95% CIs) Final Model HRs (95% CIs) Age at Dx (per 5 years) 1.39 (1.24, 1.56)*    Time 0 to 10  1.32 (1.13, 1.55)**   Time 10 to 24  0.98 (0.75, 1.29) Symptoms      Non-Pulmonary/GI Reference Reference   Pulmonary/GI 1.69 (0.71, 3.55)* 0.94 (0.38, 2.30) Sex     Female Reference Reference   Male 0.93 (0.53,1.63) 0.84 (0.45, 1.57) Race     Non-Caucasian Reference    Caucasian  0.91 (0.28, 2.96)  Mutation Type     ∆F508 Absent Reference    ∆F508 Present 0.94 (0.48, 1.84)    Unknown 1.90 (0.67, 5.45)  BMI 0.95 (0.87, 1.02) 0.97 (0.89, 1.06) P. aeruginosa      No Reference Reference   Yes 2.01 (1.14, 3.54)* 0.74 (0.39, 1.42) CFRD      No Reference Reference   Yes 5.14 (1.79, 14.74)* 7.86 (2.09, 29.55)** FEV1 % Pred (per 5%) 0.75 (0.69, 0.80)* 0.76 (0.69, 0.83)** Diagnosis Year  1.07 (1.00, 1.16) # All variables were available at baseline, defined as any data within 2 years of diagnosis for CF * Statistically significant at alpha level = 0.25 for univariable testing ** Statistically significant at alpha level = 0.05 for multivariable testing           	 34	Table 4: Predicted survival probabilities for various scenarios using the  corrected prognosis method for covariate adjustment  15-year Survival  Probability (%) 95% CI Age at Diagnosis     Age 30 82.5 58.1, 98.3   Age 40 77.3 37.2, 98.3   Age 50 67.5 14.5, 98.4   Age 60 57.3 3.2, 98.8   1st quartile value (27.1) 83.9 63.0, 98.3   3rd quartile value (44.1) 73.7 27.6, 98.2 Lung Function     40% predicted 43.0 22.2, 75.3   50% predicted 56.8 26.4, 82.9   60% predicted 66.3 30.8, 89.3   70% predicted 77.2 36.1, 93.7   80% predicted 84.0 43.2, 96.5   1st quartile value (62.4) 68.8 32.1, 90.6   3rd quartile value (94.3) 91.9 53.4, 98.7 CFRD      Yes 46.8 17.7, 86.8    No 79.9 46.4, 92.3   	 35	Table 5: Sensitivity Analysis substituting P. aeruginosa at baseline to up to  5 years post-diagnosis   Final Model HRs (95% CIs) Sensitivity Analysis Model‡ (95% CIs) Age at Dx (5 years)     Time 0 to 10 1.32 (1.13, 1.55)** 1.30 (1.11, 1.51)**   Time 10 to 24 0.98 (0.75, 1.29) 0.98 (0.75, 1.28) Symptoms      Non-Pulmonary/GI Reference Reference   Pulmonary/GI 0.94 (0.38, 2.30) 0.71 (0.30, 1.69) Sex     Female Reference Reference   Male 0.84 (0.45, 1.57) 1.15 (0.62, 2.13) BMI 0.97 (0.89, 1.06) 0.99 (0.91, 1.08) P. aeruginosa      No Reference Reference   Yes 0.74 (0.39, 1.42) 0.58 (0.30, 1.12) CFRD      No Reference Reference   Yes 7.86 (2.09, 29.55)** 8.67 (2.68, 28.12)** FEV1 % Predicted (5 %) 0.76 (0.69, 0.83)** 0.73 (0.67, 0.80)** Diagnosis Year 1.07 (1.00, 1.16) 1.06 (0.99, 1.14) ‡Sensitivity analysis conducted on one imputed dataset (not pooled) ** Statistically significant at alpha level = 0.05      	 36	Table 6: Mean age at diagnosis and FEV1 % Predicted  for individuals with and without P. aeruginosa at baseline   P. aeruginosa – Yes (n=110) P. aeruginosa – No (n=252) Mean Age at Diagnosis (years) 38.7 ± 15.5 35.6 ± 11.6 Mean FEV1 % Predicted 64.1 ± 22.0 84.5 ± 21.4                 	 37	References	 1.  Foundation CCF. Report of the Canadian Patient Data Registry. Toronto; 2013.  2.  Farrell PMP, Rosenstein BJB, White TTB, Accurso FJ, Castellani C, Cutting GR, et al. Guidelines for Diagnosis of Cystic Fibrosis in Newborns through Older Adults: Cystic Fibrosis Foundation Consensus Report. J …. 2008;153(2):S4–14.  3.  Bathesda. Cystic Fibrosis Foundation Patient Registry: 2007 annual data report to the center directors. 2008.  4.  The Canadian Cystic Fibrosis Registry. 2015 Annual Report. 2015.  5.  Nick J a, Rodman DM. Manifestations of cystic fibrosis diagnosed in adulthood. Curr Opin Pulm Med. 2005;11(6):513–8.  6.  Gilljam M. Clinical Manifestations of Cystic Fibrosis Among Patients With Diagnosis in Adulthood. Chest. 2004;126(4):1215–24.  7.  Lai HJ, Cheng Y, Cho H, Kosorok MR, Farrell PM. Association between Initial Disease Presentation, Lung Disease Outcomes, and Survival in Patients with Cystic Fibrosis. Am J Epidemiol. 2004;159(6):537–46.  8.  Widerman E, Millner L, Sexauer W, Fiel S. Health Status and Sociodemographic Characteristics of Adults Receiving a Cystic Fibrosis Diagnosis After Age 18 Years *. Chest. The American College of Chest Physicians; 2000;118(2):427–33.  9.  Widerman E, Widerman E. Communicating a Diagnosis of Cystic Fibrosis to an Adult : What Physicians Need to Know Communicating a Diagnosis of Cystic Fibrosis to an Adult : What Physicians Need to Know. 2010;(April 2015):37–41.  10.  Widerman E. Knowledge, interests and educational needs of adults diagnosed with Cystic Fibrosis after age 18. J Cyst Fibros. 2003;2(2):97–104.  11.  H H. Cystic fibrosis of the pancreas in an adult. Ohio Med J. 1946;42:616–7.  12.  Gardiner K, Cranley B. Acute presentation of cysticfibrosis in an adult. Postgr Med J. 1989;65(471–471).  13.  Hunt B, Geddes D. Newly diagnosed cystic fibrosis in middle and later life. Thorax. 1984;40:23–6.  14.  Fiel S. Cystic fibrosis in an adult. Emerg Med. 1988;20:109–21.  15.  J G. Cystic fibrosis in adults. J Miss State Med Assoc. 1985;26:161–3.  16.  Rodman DM, Polis JM, Heltshe SL, Sontag MK, Chacon C, Rodman R V., et al. Late diagnosis defines a unique population of long-term survivors of cystic fibrosis. Am J Respir Crit Care Med. 2005;171(6):621–6.  17.  Nick J a., Chacon CS, Brayshaw SJ, Jones MC, Barboa CM, St. Clair CG, et al. Effects of gender and age at diagnosis on disease progression in long-term survivors of cystic fibrosis. Am J Respir Crit Care Med. 2010;182(5):614–26.  18.  Keating C, Poor AD, Liu X, Chiuzan C, Backenroth D, Zhang Y, et al. Reduced survival in adult cystic fi brosis despite attenuated lung function decline ☆. J Cyst Fibros. European Cystic Fibrosis Society.; 2016;  19.  Sykes J, Stanojevic S, Goss CH, Quon BS, Marshall BC, Petren K, et al. A standardized approach to estimating survival statistics for population-based cystic fibrosis registry cohorts. J Clin Epidemiol. Elsevier; 2016;70:206–13.  20.  Farrell PM, Rosenstein B, White T. Guidelines for diagnosis of cystic fibrosis in newborns through older adults: Cystic Fibrosis Foundation consensus report. J Pediatr. 2008;153(2):S4–14.  21.  Stephenson  a. L, Tom M, Berthiaume Y, Singer LG, Aaron SD, Whitmore G a., et al. A 	 38	contemporary survival analysis of individuals with cystic fibrosis: a cohort study. Eur Respir J. 2014;45(3):670–9.  22.  Kerem B, Rommens JM, Iannuzzi MC, Mitchell L, Melmer G, Dean M, et al. Linked references are available on JSTOR for this article : Identification of the Cystic Fibrosis Gene : Chromosome Walking and Jumping. 1989;245(4922):1073–80.  23.  Moran A, Hardin D, Rodman D, Allen HF, Beall RJ, Borowitz D, et al. Diagnosis , screening and management of cystic fibrosis related diabetes mellitus A consensus conference report. 1999;45:61–73.  24.  Wang X, Dockery D, Wypid D. Pulmonary function between 6 and 18 years of age. Pediatr Pulmonol. 1993;15:75–88.  25.  Hankinson J, Odencrantz J, Fedan K. Spirometric reference values from a sample of the general U.S. population. Am J Respir Crit Care Med. 1999;159:179–87.  26.  Weill D, Chairs C, Benden C, Corris PA, Members C, Dark JH, et al. ISHLT CONSENSUS A consensus document for the selection of lung transplant candidates : 2014 — An update from the Pulmonary Transplantation Council of the International Society for Heart and Lung Transplantation. J Hear Lung Transplant. Elsevier; 2015;34(1):1–15.  27.  Bun Y, Gao F, Siong K. Age at diagnosis and the choice of survival analysis methods in cancer epidemiology. 2003;56:38–43.  28.  R core team. A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria;  29.  Therneau TM, Crowson CS, Atkinson EJ. Adjusted Survival Curves. 2015;Section 5.2, 1-24.  30.  Rubin DB. Multiple imputation for nonresponse in surveys. New York, NY: Wiley; 1987. 76-77 p.  31.  White IR, Wood AM. Tutorial in Biostatistics Multiple imputation using chained equations : Issues and guidance for practice. 2011;(July 2010).  32.  Bodner TE, Bodner TE. What Improves with Increased Missing Data Imputations ? What Improves with Increased Missing Data Imputations ? 2017;5511(May).  33.  Levy LD, Durie PR, Pencharz PB, Corey ML. Effects of long-term nutritional rehabilitation on body composition and clinical status in malnourished children and adolescents with cystic fibrosis. J Pediatr. 1985;107(2):225–30.  34.  Pedreira CC, Robert RGD, Dalton V, Oliver MR, Carlin JB, Robinson P, et al. Association of body composition and lung function in children with cystic fibrosis. Pediatr Pulmonol. 2005;39(3):276–80.  35.  Corey M, McLaughlin FJ, Williams M, Levison H. A comparison of survival, growth, and pulmonary function in patients with cystic fibrosis in Boston and Toronto. J Clin Epidemiol. 1988;41(6):583–91.  36.  Rosenfeld M, Davis R, FitzSimmons S, Pepe M, Ramsey B. Gender gap in cystic fibrosis mortality. Am J Epidemiol. 1997;145(9):794–803.  37.  Verma N, Bush  a, Buchdahl R. Is there still a gender gap in cystic fibrosis? Chest. The American College of Chest Physicians; 2005;128(4):2824–34.  38.  Bonadia LC, De Lima Marson FA, Ribeiro JD, Paschoal IA, Pereira MC, Ribeiro AF, et al. CFTR genotype and clinical outcomes of adult patients carried as cystic fibrosis disease. Gene. Elsevier B.V.; 2014;540(2):183–90.  39.  Noone PG, Knowles MR. “CFTR-opathies”: disease phenotypes associated with cystic fibrosis transmembrane regulator gene mutations. Respir Res. 2001;2(6):328–32.  40.  Lewis C, Blackman SM, Nelson A, Oberdorfer E, Wells D, Dunitz J, et al. Diabetes-related 	 39	Mortality in Adults with Cystic Fibrosis Role of Genotype and Sex. 2015;191(2):194–200.  41.  Szczesniak R, Heltshe SL, Stanojevic S, Mayer-hamblett N. Use of FEV 1 in cystic fi brosis epidemiologic studies and clinical trials : A statistical perspective for the clinical researcher. J Cyst Fibros. European Cystic Fibrosis Society.; 2017;  42.  Bourbeau J, Hernandez P, Chapman K, Cowie R. ORiginaL aRtiCLE Canadian prediction equations of spirometric lung function for Caucasian adults 20 to 90 years of age : Results from the Canadian Obstructive Lung Disease ( COLD ) study and the Lung Health Canadian Environment ( LHCE ) study. 2011;18(6):321–7.  43.  Tan WC, Sin DD, Bourbeau J, Hernandez P, Chapman KR, Cowie R, et al. Characteristics of COPD in never-smokers and ever-smokers in the general population : results from the CanCOLD study. 2015;822–9.  44.  Stephenson AL, Sykes J, Stanojevic S, Quon BS, Marshall BC, Petren K, et al. Survival Comparison of Patients With Cystic Fibrosis in Canada and the United States. Ann Intern Med. 2017;   

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.24.1-0349192/manifest

Comment

Related Items