CFTR Genotype as a Predictor of Prognosis in Cystic Fibrosis*

Transcription

1 Original Research CYSTIC FIBROSIS CFTR Genotype as a Predictor of Prognosis in Cystic Fibrosis* Edward F. McKone, MD, MS; Christopher H. Goss, MD, MS, FCCP; and Moira L. Aitken, MD, FCCP Study rationale: Certain CFTR genotypes are associated with reduced mortality. The accuracy of using CFTR genotype as a predictor of survival and the mechanisms through which CFTR genotype influences survival are unknown. Participants: All patients with cystic fibrosis (CF) enrolled in the US Cystic Fibrosis Foundation national registry between 1993 and Design: We examined the prognostic value of CFTR genotype, grouped into high-risk and low-risk categories based on the effect of their CFTR genotype on phenotype and protein production. Measurements and results: Clinical and genetic data were available from 15,651 patients with CF. Patients with a high-risk CFTR genotype had a greater than twofold increased risk of death compared to patients with a low-risk CFTR genotype (relative risk, 2.25; 95% confidence interval [CI], 1.77 to 2.84; p < 0.001). This association was partly explained by lung function, nutritional status, pancreatic insufficiency, and Pseudomonas aeruginosa colonization. Of the 1,672 patients who died, median age at death for the high-risk CFTR genotype was 24.2 years (interquartile range, 18.4 to 32.0 years) and for the low-risk CFTR genotype was 37.6 years (interquartile range, 28.8 to 47.9 years; p < 0.001). The positive predictive value of this classification method as a test to identify patients who died before or after their 30th birthday was 69% (95% CI, 67 to 72%) with a negative predictive value of 71% (95% CI, 60 to 80%). Conclusions: Grouping patients into high-risk and low-risk CFTR genotype categories is associated with significant differences in survival and median age at death. These differences are not fully explained by lung function, nutritional measures, pancreatic insufficiency, or P aeruginosa colonization. Modest reassurance about the likelihood of a milder than average course can be provided for CF patients with a low-risk CFTR genotype, although it should be acknowledged that substantial phenotypic variability exists. (CHEST 2006; 130: ) Key words: epidemiology; genetics; survival Abbreviations: BMI body mass index; CF cystic fibrosis; CFTR cystic fibrosis transmembrane conductance regulator; CI confidence interval; NPV negative predictive value; PPV positive predictive value Since the discovery of the cystic fibrosis transmembrane conductance regulator (CFTR) gene that causes cystic fibrosis (CF), 1,000 mutations have been identified. 1 The most common mutation, F508, results in a deletion of phenylalanine at *From the Division of Pulmonary and Critical Care Medicine and Adult Cystic Fibrosis Center, University of Washington, Seattle, WA. Dr. McKone is supported by the Cystic Fibrosis Foundation and National Institutes of Health grant K23 HL/ The authors have no conflicts of interest to disclose. Manuscript received May 10, 2005; revision accepted April 18, position 508 in the transcribed CFTR protein and is found in approximately 70% of CF patients. 2 CF patients homozygous for F508 generally have a more severe clinical phenotype than F508 heterozygotes and patients with no F508 allele. 3 5 In Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: Edward McKone, MD, University of Washington Medical Center, BB1253 Health Sciences Building, Box , Seattle, WA ; emckone@u.washington. edu DOI: /chest CHEST / 130 / 5/ NOVEMBER,

2 addition, a small number of CFTR genotypes have been associated with a milder clinical phenotype Recently, we carried out a large retrospective cohort study using the US Cystic Fibrosis Foundation patient registry. We found that classifying CFTR genotype according to the functional effects of CFTR mutation on CFTR protein production 12,13 is associated with significant differences in mortality rates. 14 Patients with CFTR genotypes associated with severely reduced CFTR production (class I to III) had a very similar severe phenotype with higher mortality rates than patients with a CFTR genotype associated with some residual CFTR function (class IV to V). This suggested that patients with CF could be grouped into two broader genetic risk categories: a high-risk group of CF patients with a severe CFTR genotype, and a low-risk group of CF patients with a milder CFTR genotype. 15 This method of genetic risk stratification could be of clinical value as a prognostic tool for patients with newly diagnosed CF, as CFTR genotype is one of the few data available at time of diagnosis. The mechanisms through which CFTR genotype influences survival and the accuracy of using CFTR genotype to predict survival are unknown. The purpose of this study was to quantify the differences in survival associated with CFTR genotype grouped into high-risk and low-risk categories. We then examined which phenotypic characteristics explained the association between CFTR genotype and survival and whether CFTR genotype is an independent predictor of survival. Finally, we tested whether this method of risk stratification of CFTR genotype was an accurate predictor of age at death and whether CFTR genotype might be a useful tool for predicting long-term prognosis. Materials and Methods The study design is a retrospective cohort study using the Cystic Fibrosis Foundation patient registry, 16,17 measuring risk of death during an observation period from 1993 to The primary outcome of interest was all-cause mortality including patients who died after transplantation. Secondary outcome was all-cause mortality (excluding transplant-related deaths). All procedures were approved by the University of Washington Institutional Review Board. Genotypes were classified into high-risk or low-risk groups based on the effects of the functional class of their CFTR mutations 12,13 on phenotype 6 11,14,18,19 and mortality. 14 Analysis was limited to CFTR genotypes that were recorded in the database, had a known functional class, and whose allele frequency was 0.1%. Patients with both mutant alleles in either class I, class II, or class III were grouped together as a high-risk genotype, 3,7,14 while patients with at least one mutant allele in class IV and V were considered low-risk genotypes. 6 11,14,19 Risk stratification of mutations is shown in Table 1 and includes 82% of the CF population CFTR alleles. To determine the mortality risk associated with CFTR genotype, patients were considered at risk at age of entry at any time during the follow-up period between from 1993 to Patients continued to be at risk until the end of the follow-up period or if they died or were unavailable for follow-up. To determine possible mechanisms through which CFTR genotype might influence survival, we repeated the analysis while adjusting for various phenotypic characteristics. We assumed that adjusting for phenotypic characteristics that were within the causal pathway between CFTR genotype and survival would lead to a significant reduction in the strength of association between CFTR genotype and survival. Phenotypes of interest were lung disease (FEV 1 percentage of predicted), nutritional status (body mass index [BMI]), and presence or absence of pancreatic insufficiency and Pseudomonas aeruginosa colonization. All clinical measures were collected during the year of entry into the cohort. Pancreatic insufficiency was considered present if the patient was receiving pancreatic enzyme supplementation during the year of entry into the cohort. P aeruginosa colonization was considered present if a sputum sample was positive within the previous year. Clinical variables such as Burkholderia cepacia infection, CF-related diabetes, and exacerbation rate were not included due to concerns about case definition, misclassification, or due to insufficient data being present in the genotyped cohort. To avoid biases related to birth cohort and size of the CF center, we also adjusted for year of entry into the cohort and CF center population. In addition, to determine whether CFTR genotype is a reliable tool for predicting age at death, we calculated the sensitivity, specificity, positive predictive values (PPV), and negative predictive value (NPV) of our risk-stratification method using a variety of age at death cutoff points. Table 1 Risk Stratification of CFTR Alleles* Risk Stratification Mutation Functional Class Subjects, No. Alleles High-risk CFTR genotype Class I 2,131 G542X, R553X, W1282X, R1162X, 621 1G3T, G3A, 1078 T, 3659 C Class II 11,231 F508, I507, N1303K, S549N, G85E Class III 783 G551D, R560T Low-risk CFTR genotype Class IV 391 R117H, R334W, R347P Class V KbC3T, G3A, A455E *Patients with both CFTR alleles in either class I, class II, or class III were grouped together as a high-risk genotype, while patients with at least one mutant allele in class IV and V were considered to have low-risk genotypes; 380 patients had both mutations in either class I, II, or III, while 314 patients had both mutations in either class IV or V (total, n 15,651) Original Research

3 Statistical Analysis The Cox proportional hazards model was used to determine age-adjusted risk of death by CFTR genotype. Univariable analysis was initially performed with the CFTR genotype as the primary predictor variable. Bivariable analysis was then performed with CFTR genotype and each clinical measure to determine the effect of each variable on the association between the CFTR genotype and survival. All clinical variables that were associated with worse survival on bivariable analysis (p 0.2) were then included in the full model, and backward stepwise regression was used to generate a final model. The stphtest command in Stata 8.0 (StataCorp; College Station, TX) was used to test for the Cox proportional hazards assumption. Analyses were subsequently repeated using 50% of the cohort who were randomly selected to internally validate the initial findings. Summary data for CFTR genotype as a predictive test for age at death were calculated for the patients who died using the diagt option in Stata 8.0. All statistical analysis was performed using Stata 8.0. Results The patient population is outlined in Figure 1. There were a total of 30,396 patients included in the database, with 193,856 person-years at risk; 21,353 patients (70%) were genotyped. Of the genotyped patients, 15,651 patients had a CFTR genotype containing mutations with a known functional class. Of these, 14,525 patients (93%) had a high-risk CFTR genotype and 1,126 patients (7%) had a low-risk CFTR genotype. Median follow-up was 8.6 years for patients with a high-risk CFTR genotype vs 5.1 years for patients with a low-risk CFTR genotype. There were a total of 1,672 deaths during the 10-year follow-up period. Clinical characteristics of the cohort are shown in Table 2. Figure 1. Study population flow chart. CHEST / 130 / 5/ NOVEMBER,

4 Table 2 Clinical Characteristics of Cohort* Characteristics Severe CFTR Genotype (n 14,525) Mild CFTR Genotype (n 1,126) Age at entry, yr 10.2 (10.2) 19.0 (15.5) Year of entry 1995 (3) 1996 (3) Male gender, % FEV 1 % predicted 72.9 (27) 73.7 (25.9) Weight, kg 27 (20) 45 (26) Height, cm 117 (40) 140 (38) Pancreatic insufficiency, % Pseudomonas colonization, % *Data are presented as mean (SD) unless otherwise indicated. n 7,229. n 779. Figure 3. Box plots of median age at death by CFTR genotype. Patients with a high-risk CFTR genotype were 2.25 (95% confidence interval [CI], 1.77 to 2.84) times more likely to die during the follow-up period than those with a low-risk CFTR genotype. Survival curves are shown in Figure 2. Median survival for patients with a high-risk genotype was 36.3 years (95% CI, 35.5 to 37.6 years), while median survival for patients with a low-risk genotype was 50 years (95% CI, 47.1 to 55.9 years). Of those who died (n 1,672), median age at death for the patients with a high-risk CFTR genotype was significantly lower than that of patients with a low-risk CFTR genotype (24.2 years [interquartile range, 18.4 to 32.0 years] vs 37.6 years [interquartile range, 28.8 to 47.9 years]; p 0.001; Fig 3). These findings were also observed for all-cause mortality (excluding transplant-related deaths) and in the randomly selected internal validation cohort. To determine possible mechanisms through which the CFTR genotype influences survival, we repeated the survival analysis while adjusting for phenotypic Figure 2. Survival curves by CFTR genotype grouped into high-risk and low-risk categories. characteristics. On bivariable analysis, lung function, BMI, and pancreatic insufficiency had the greatest effect on the association between CFTR genotype and survival, although the effect of each of the strength of association was small (Table 3). On multivariable analysis, lung function and BMI were both associated with worse survival, although CFTR genotype remained an independent predictor of mortality (Table 3). Pancreatic insufficiency was not a predictor of mortality after adjusting for CFTR genotype, lung function, and BMI. The results were Table 3 Influence of Various Phenotypic Characteristics on the Association Between CFTR Genotype and Survival* Variables Relative Hazard (95% CI) p Value Bivariable pair CFTR genotype (n 15,651) 2.25 ( ) Unadjusted CFTR genotype (n 8,008) 1.90 ( ) FEV 1 % predicted 0.96 ( ) CFTR genotype (n 14,938) 1.78 ( ) BMI 0.78 ( ) CFTR genotype (n 13,536) 2.04 ( ) Pseudomonas colonization ( / ) 1.20 ( ) CFTR genotype (n 15,196) 1.95 ( ) Pancreatic insufficiency ( / ) 1.45 ( ) Multivariable analysis (n 7,305) CFTR genotype 1.60 ( ) FEV 1 % predicted 0.96 ( ) BMI 0.88 ( ) Pseudomonas colonization ( / ) 1.06 ( ) Pancreatic insufficiency ( / ) 1.01 ( ) *Hazard ratio for FEV 1 % predicted and BMI is that observed for a change of 1 U. The effect of each clinical characteristic on the association between CFTR genotype and survival can be determined by calculating the difference between adjusted and unadjusted CFTR genotype hazard ratios. Effect of single phenotypic characteristics on CFTR genotype hazard ratio Original Research

5 unchanged after adjusting for center size and year of entry into the cohort. Due to deviations from the proportional hazards assumption in certain models, we repeated the analysis either stratifying by clinical variable or testing these variables as time-varying coefficients. This did not lead to a significant change in the results. The accuracy of the CFTR genotype as a predictor of age at death is shown in Table 4. Of the patients who died and had a high-risk CFTR genotype, 69% (95% CI, 67 to 72%) died before reaching their 30th birthday; of those with a low-risk genotype, 71% (95% CI, 59 to 80%) died after reaching their 30th birthday. An age-at-death cutoff at 30 years had the highest combination of PPV and NPV for CFTR genotype as a predictive test, although the accuracy of CFTR genotype as a predictor of age at death for other age-at-death cutoffs is also shown. Discussion The main findings of this study are that patients with CF can be classified into high-risk and low-risk genetic groups that have significant differences in survival and median age at death. These differences in survival are not fully explained by clinical measures of lung function, nutrition, and pancreatic insufficiency, suggesting that the CFTR genotype is an independent predictor of survival. In addition, using the CFTR genotype for risk stratification in CF may have some prognostic value, although substantial phenotypic variability is seen. Since the discovery of the gene that causes CF, there has been substantial interest in the influence of the CFTR genotype on phenotype. 18 Patients homozygous for F508 have more severe clinical manifestations compared to heterozygotes and genotypes without F508, 3 5 although these differences are highly variable. 14,20 In addition, a small number of individual CFTR genotypes have been described that are associated with milder phenotype. 6,8 11,14,18,19 The main differences in phenotype appear to be related to the influence of CFTR genotype on pancreatic function, with F508 homozygotes usually Table 4 Accuracy of CFTR Genotype as a Predictor of Age at Death at Various Age-at-Death Cutoffs* Age-at-Death Cutoff Sensitivity Specificity PPV NPV 25 yr 98 (97 99) 8 (6 10) 53 (51 56) 81 (70 88) 30 yr 98 (97 99) 11 (8 14) 69 (67 72) 71 (60 80) 35 yr 97 (96 98) 14 (10 18) 82 (80 84) 57 (46 68) 40 yr 97 (96 98) 20 (14 26) 91 (89 92) 44 (33 55) *Data are presented as % (95% CI). pancreatic insufficient at an earlier age and having worse nutritional measures and lung function. 3 5 The reason for these phenotypic differences is likely to be due to the effect of CFTR mutation on protein production, with very low levels of CFTR associated with a severe phenotype and intermediate levels associated with milder CF. 8,11,21,22 We found that grouping CFTR alleles according to the functional classification system proposed by Tsui 12 and Welsh and Smith 13 is associated with differences in mortality rates. 14 Patients with two mutations in class I to III had very similar mortality rates that are significantly higher than that observed for patients who have at least one mutation in class IV and V. 14,15 This suggests that the relationship between mortality and CFTR genotype exhibits a threshold effect, and patients with CF could be grouped into two broader risk categories, with class I to III mutations a high-risk group and class IV to V a low-risk group. 15 The results of this study show that patients with CFTR genotypes grouped according to this riskstratification system have significant differences in survival and median age at death. We believe that these data suggest that the CFTR genotype may provide some prognostic information. In particular, modest reassurance about the likelihood of a milderthan-average course can be provided for the patients with a low-risk genotype, although counseling should acknowledge the considerable variation in the CF phenotype than occurs between patients with the same CFTR genotype. 14,20 As there are many factors that influence CF phenotypic variation, it is also important to emphasize that a genetic predisposition to a milder disease course could be negated by nongenetic factors associated with poorer outcomes such as medical noncompliance. Hopefully, as other genetic factors that modify the clinical course in CF are identified, 23 the ability to accurately predict long-term outcomes using genetic data should improve further. The reasons for the strong association between CFTR genotype and survival are unclear. Earlier studies 3,24 26 have suggested that preservation of pancreatic function is the main mechanism through which milder CFTR genotypes influences phenotype. While it is known that pancreatic sufficiency is more common in patients with milder CFTR mutations and is associated with improved survival, 26 this does not fully explain the survival differences we observed. Although our method of classifying pancreatic insufficiency may have led to misclassification, it is unlikely to be severe enough to fully explain the association between our classification of the CFTR genotype and survival. In addition, including lung function and nutritional measures, both of which have been independently associated with CHEST / 130 / 5/ NOVEMBER,

6 poorer survival, did not fully explain the association between the CFTR genotype and outcome. This may be because of misclassification of data within the registry or that the clinical measures commonly used to assess lung disease and nutritional status are imprecise measures of true disease severity. Alternatively, there may be other mechanisms through which the CFTR genotype influences survival in CF patients. Our classification of CFTR genotypes included the 21 most-common mutations that were available from the Cystic Fibrosis Foundation patient registry and have an allele frequency 0.1%. These mutations account for 82% of all CFTR mutations in the US CF population and include the vast majority of mutations that are currently screened for in at-risk populations. 30 A number of additional mutations that have been associated with milder clinical outcomes 31,32 were not included in our analysis, as the genotype data were not available or the functional class of the mutation is unknown. The affects of these rare mutations on our risk-stratification method will need to be examined in a future study as more information about how these mutations affect mortality becomes available. Using a population-based database to predict prognosis can be problematic, and our study has a number of limitations. 14 As this study was primarily designed to determine the differences in survival across genetic risk categories, it is important to note that the data presented are generated over the entire 10 years of follow-up. In a cohort of patients with newly diagnosed CF, the median age at death is likely to be higher owing to the increase in average life expectancy that occurred during the observation period of our cohort. 33 In addition, the optimum cutoff point at 30 years for age at death, which had the highest combination of PPV and NPV for our test, is likely to be higher in a cohort of patients with newly diagnosed CF followed up prospectively. For these reasons, caution is advised when counseling patients with a more severe CFTR genotype, as these data may underestimate long-term prognosis. Unfortunately, it is impossible to accurately predict the effects of mortality bias on our results until longitudinal data are available on a cohort of patients with CF diagnosed at birth. However, we believe that it is still reasonable to conclude that those with a milder CFTR genotype are more likely to have a less severe clinical course. Another concern is that grouping CFTR genotypes into broad risk categories may excessively weight the effects of genotypes with a large number of patients, such as F508 homozygotes. To account for this potential bias, we repeated the analysis clustering by individuals CFTR genotype after excluding F508 homozygotes (n 5,052 after exclusion of F508 homozygotes). This did not significantly change the association between our classification of CFTR genotype and survival. In addition, the majority of individual CFTR genotypes included in our study have been associated with a mild or severe phenotype in a number of earlier studies 6,8 11,14,18,19 consistent with our classification method. In conclusion, when using the Cystic Fibrosis Foundation patient registry we have shown that this method of risk stratification by CFTR genotype is associated with significant differences in mortality and median age at death. Using this risk-stratification methodology may be of some value in predicting outcomes for patients with newly diagnosed CF and their families. ACKNOWLEDGMENT: We thank Bruce Marshall, Monica Brooks, Preston W. Campbell III, MD, and the Clinical Research Committee of the Cystic Fibrosis Foundation for access to the database. We also thank Wylie Burke MD, PhD, for her assistance in the preparation of this article. References 1 Cystic fibrosis mutation database. Available at: sickkids.on.ca/cftr/. Accessed July 7, Welsh M, Ramsey B, Accurso F, et al. Cystic fibrosis. In: Scriver CR, Beaudet AL, Sly WS, et al, eds. The metabolic and molecular basis of inherited disease. New York, NY: McGraw-Hill, Health Professions Division, 2001; Kerem E, Corey M, Kerem BS, et al. The relation between genotype and phenotype in cystic fibrosis: analysis of the most common mutation ( F508). N Engl J Med 1990; 323: Santis G, Osborne L, Knight RA, et al. Independent genetic determinants of pancreatic and pulmonary status in cystic fibrosis. Lancet 1990; 336: Johansen HK, Nir M, Hoiby N, et al. Severity of cystic fibrosis in patients homozygous and heterozygous for F508 mutation. Lancet 1991; 337: Kristidis P, Bozon D, Corey M, et al. Genetic determination of exocrine pancreatic function in cystic fibrosis. Am J Hum Genet 1992; 50: Correlation between genotype and phenotype in patients with cystic fibrosis: the Cystic Fibrosis Genotype-Phenotype Consortium. N Engl J Med 1993; 329: Highsmith WE, Burch LH, Zhou Z, et al. A novel mutation in the cystic fibrosis gene in patients with pulmonary disease but normal sweat chloride concentrations. N Engl J Med 1994; 331: Stern RC, Doershuk CF, Drumm ML kb C T mutation and disease severity in cystic fibrosis. Lancet 1995; 346: Gan KH, Veeze HJ, van den Ouweland AM, et al. A cystic fibrosis mutation associated with mild lung disease. N Engl J Med 1995; 333: Highsmith WE Jr, Burch LH, Zhou Z, et al. Identification of a splice site mutation ( G A) associated with small amounts of normal CFTR mrna and mild cystic fibrosis. Hum Mutat 1997; 9: Tsui LC. The spectrum of cystic fibrosis mutations. Trends Genet 1992; 8: Welsh MJ, Smith AE. Molecular mechanisms of CFTR 1446 Original Research

7 chloride channel dysfunction in cystic fibrosis. Cell 1993; 73: McKone EF, Emerson SS, Edwards KL, et al. Effect of genotype on phenotype and mortality in cystic fibrosis: a retrospective cohort study. Lancet 2003; 361: Lai HJ, Cheng Y, Cho H, et al. Association between initial disease presentation, lung disease outcomes, and survival in patients with cystic fibrosis. Am J Epidemiol 2004; 159: FitzSimmons SC. The changing epidemiology of cystic fibrosis. J Pediatr 1993; 122: Rosenfeld M, Davis R, FitzSimmons S, et al. Gender gap in cystic fibrosis mortality. Am J Epidemiol 1997; 145: Mickle JE, Cutting GR. Genotype-phenotype relationships in cystic fibrosis. Med Clin North Am 2000; 84: Koch C, Cuppens H, Rainisio M, et al. European Epidemiologic Registry of Cystic Fibrosis (ERCF): comparison of major disease manifestations between patients with different classes of mutations. Pediatr Pulmonol 2001; 31: Burke W, Aitken ML, Chen SH, et al. Variable severity of pulmonary disease in adults with identical cystic fibrosis mutations. Chest 1992; 102: Tzetis M, Efthymiadou A, Doudounakis S, et al. Qualitative and quantitative analysis of mrna associated with four putative splicing mutations (621 3A G, T A, 296 1G C, T C-D565G) and one nonsense mutation (E822X) in the CFTR gene. Hum Genet 2001; 109: Ramalho AS, Beck S, Meyer M, et al. Five percent of normal cystic fibrosis transmembrane conductance regulator mrna ameliorates the severity of pulmonary disease in cystic fibrosis. Am J Respir Cell Mol Biol 2002; 27: Accurso FJ, Sontag MK. Seeking modifier genes in cystic fibrosis. Am J Respir Crit Care Med 2003; 167: Gaskin K, Gurwitz D, Durie P, et al. Improved respiratory prognosis in patients with cystic fibrosis with normal fat absorption. J Pediatr 1982; 100: Ahmed N, Corey M, Forstner G, et al. Molecular consequences of cystic fibrosis transmembrane regulator (CFTR) gene mutations in the exocrine pancreas. Gut 2003; 52: Davis PB, Schluchter MD, Konstan MW. Relation of sweat chloride concentration to severity of lung disease in cystic fibrosis. Pediatr Pulmonol 2004; 38: Kerem E, Reisman J, Corey M, et al. Prediction of mortality in patients with cystic fibrosis. N Engl J Med 1992; 326: Liou TG, Adler FR, Fitzsimmons SC, et al. Predictive 5-year survivorship model of cystic fibrosis. Am J Epidemiol 2001; 153: Mayer-Hamblett N, Rosenfeld M, Emerson J, et al. Developing cystic fibrosis lung transplant referral criteria using predictors of 2-year mortality. Am J Respir Crit Care Med 2002; 166: NIH Consensus Development Program. NIH consensus statement: genetic testing for cystic fibrosis. Bethesda, MD: National Institutes of Health, 1997; 15: Kiesewetter S, Macek M Jr, Davis C, et al. A mutation in CFTR produces different phenotypes depending on chromosomal background. Nat Genet 1993; 5: Kulczycki LL, Kostuch M, Bellanti JA. A clinical perspective of cystic fibrosis and new genetic findings: relationship of CFTR mutations to genotype-phenotype manifestations. Am J Med Genet 2003; 116:A262 A Patient registry 2002 annual data report. Bethesda, MD: Cystic Fibrosis Foundation, CHEST / 130 / 5/ NOVEMBER,