Objectives In cystic fibrosis, mutations of the CFTR gene lead to diffuse bronchiectasis (DB). DB is also associated with other diseases including rheumatoid arthritis (RA) in which the role of genetic factors in the predisposition to DB remains unclear.
Methods A family-based association study was carried out to determine whether the frequency of CFTR mutations was higher in patients with RA-associated DB and to determine whether a causal relationship could be established between the variant and the disease by evaluating its cosegregation with DB within families. Families of probands with RA-DB were included if one first-degree relative had RA and/or DB. The controls comprised healthy subjects requesting genetic counselling because their partner had cystic fibrosis.
Results The frequency of CFTR mutations was higher in family members with RA-DB or DB only than in unaffected relatives (p<0.005 for each comparison) and in unrelated healthy controls (p<0.001 for each comparison) but not in family members with RA only. CFTR mutations were more frequent in family members with RA-DB than in those with RA only (OR 5.30, 95% CI 2.48 to 11.33; p<5×10−5). They cosegregated with RA-DB in the families (sib-TDT=10.82, p=0.005).
Conclusions RA-DB should be added to the list of phenotypes in which CFTR mutations are pathogenic. CFTR mutation is the first genetic defect linked to an extra-articular feature of RA to be described. CFTR mutations in patients with RA appear to be an important marker of the risk of associated DB, which has been linked to a less favourable prognosis.
Statistics from Altmetric.com
Rheumatoid arthritis (RA) is a complex systemic autoimmune disease primarily affecting the joints. Extra-articular features such as lung and airway involvement may also be observed. In particular, an association between bronchiectasis, which is defined as an abnormal irreversible dilation of the airways, and RA is now widely recognised. On the one hand, the prevalence of diffuse bronchiectasis (DB) ranges from 5.6% to 30% of patients with RA in prospective blind studies based on high-resolution CT (HRCT) scanning.1,–,6 On the other hand, RA is found in 2.7–5.2% of patients with DB referred to respiratory medicine departments for investigation.7 8 Patients with RA associated with DB seem to have a poorer prognosis than patients with RA without DB.9,–,11 It is therefore important to identify the factors predisposing patients with RA to DB.
The damage to the airways leading to DB may result from several mechanisms, including chronic inflammatory and/or infectious processes, which may be caused by various acquired or genetic abnormalities. Mutations in both copies of the cystic fibrosis transmembrane conductance regulator (CFTR) gene, severely impairing CFTR protein production and function, have been identified as a major cause of severe DB in cystic fibrosis.12
Patients with disease linked to CFTR mutations displaying residual protein function but not meeting the diagnostic criteria for cystic fibrosis are considered to have CFTR-related disease.12 13 Thus, some CFTR mutations resulting in the production of a CFTR protein with residual function have been implicated in diseases affecting a single organ, such as congenital bilateral absence of the vas deferens (CBAVD),14 idiopathic chronic pancreatitis15 16 and non-classic lung disease, including DB.17
The higher prevalence of CFTR mutations found in some studies of 16–55 patients with DB of unknown origin18,–,23 remains a matter of debate. It was not confirmed in larger studies.8 24 Some of these studies20 21 did not exclude patients with cystic fibrosis according to current criteria25 26 or did not carry out complete screening of the CFTR gene.8 18 24 Moreover, most of these studies included no control group8 18,–,20 23 24 and none included family-based controls to allow matching for potential confounders.
We previously reported a higher than normal prevalence of CFTR mutations in a more homogeneous group of patients suffering from RA-associated DB.27 We also showed, in patients with both RA and DB carrying at least one abnormal CFTR allele but not fulfilling diagnostic criteria for cystic fibrosis, more frequent chronic sinusitis, a trend towards more severe pulmonary involvement and a lower nasal potential difference than in patients with both RA and DB without CFTR mutation, consistent with possible CFTR protein dysfunction.27 These findings support the hypothesis that CFTR mutations may predispose patients with RA to RA-associated DB.
We carried out a family-based association study to determine whether the frequency of CFTR gene mutations was higher in patients with RA-associated DB and to determine whether a causal relationship could be established between the variant and the disease.
A family-based association study was conducted to compare CFTR mutation frequencies in affected family members with RA-DB, with DB only, with RA only and with unaffected relatives. In a case–control study analysis, CFTR mutation frequencies in affected family members were compared with those in an independent set of healthy controls. Finally, an association and linkage study was carried out within families, with the sib-TDT, to determine whether a causal relationship could be established between the CFTR mutations and the RA-associated DB phenotype by evaluating their cosegregation with DB within families.
Family-based association study
Families were recruited from a variety of sources including publications as well as other forms of publicity and information provided by the French Society of Rheumatology and French National Society of Internal Medicine. For inclusion in the family study, each family had to include at least one proband with both RA and DB and one affected first-degree relative with RA and/or DB willing to participate in the study. Once a family was deemed eligible for inclusion, all affected subjects and unaffected individuals were interviewed and samples were taken for DNA extraction according to a standardised protocol.
We defined three groups of patients: patients with RA-associated DB (RA-DB), patients with DB only (without RA) and patients with RA only (without DB).
Interview and clinical assessment
A pedigree analysis was carried out and all family members underwent face-to-face semi-structured interviews and clinical assessment. Given the known association of RA and DB and the possibility of asymptomatic or paucisymptomatic DB, all patients with RA underwent HRCT scanning of the lungs. An HRCT evaluation of the lungs was also conducted for all other participants presenting with at least two of the following symptoms: history of haemoptysis, recurrent or chronic cough, purulent sputum production or dyspnoea.
Confirmation of RA and symptomatic DB
RA was diagnosed according to the American College of Rheumatology criteria.28 DB was confirmed in all cases by two independent observers finding evidence of diffuse (more than one lobe) bronchiectasis on HRCT scanning. The diagnosis was established without knowledge of the genetic status. Discordance between observers was resolved by discussion to reach a final consensus.
Family-based association study analysis
We compared the frequency of CFTR mutations between family members from the various phenotypic groups and unaffected relatives.
To assess possible association and linkage between CFTR mutations and DB in the family members, we investigated the cosegregation of these mutations with DB within families.
Family-based case–control study
Case–control study analysis
We compared CFTR mutation frequencies in affected family members with RA-DB, with DB only and with RA only with that for an independent set of healthy controls.
The control population comprised 120 healthy French subjects of Caucasian origin who requested genetic counselling because their partner had cystic fibrosis (homozygous or compound heterozygous). They were recruited in the region of the study and were well matched for potential confounders such as environmental factors and ethnicity.
A 10 ml blood sample was drawn from each of the participants. Genomic DNA was extracted from nucleated cells by a standard technique (Wizard Genomic DNA purification kit; Promega, Madison, Wisconsin, USA). We tested for the 32 most frequently observed CFTR mutations with the cystic fibrosis assay kit (v3 Genotyping assay; Abbott, Wiesbaden, Germany). This kit detects approximately 83% of the mutations observed in French populations with cystic fibrosis. A complementary analysis for variants and mutations of the whole coding region and exon/intron junctions of CFTR was carried out systematically by denaturing gradient gel electrophoresis (DGGE) (exons 9 and 10) and denaturing high-performance liquid chromatography (DHPLC) (all other exons), as previously described.29 30 This approach detects more than 95% of the CFTR mutations and variants in our cystic fibrosis population. Mutations and variants detected by DGGE and DHPLC were subsequently confirmed by direct DNA sequencing with the ABI PRISM Big Dye Terminator cycle sequencing kit (Applied Biosystems, Foster City, California, USA) and an ABI Prism 3100-Avant sequencer. If a CFTR mutation or variant was detected in a patient, we screened for large CFTR rearrangements by multiplex ligation-dependent probe amplification (MLPA; SALSA091R, MRC-Holland, Amsterdam, The Netherlands).
Nucleotide substitutions were classified into three groups according to their known or predicted functional consequences: Group A, corresponding to severe CFTR mutations, previously identified in patients with classic cystic fibrosis; Group B, corresponding to mild CFTR defects, previously identified in patients with CFTR-related disorders such as CBAVD, DB or chronic pancreatitis; and Group C, variants of unknown significance considered, on the basis of linkage analysis and their frequency in normal individuals, to be rare polymorphisms with incomplete penetrance potentially causing minor changes in the amount and/or function of CFTR protein (such as the 5T allele) (www.genet.sickkids.on.ca/cftr/app).13 The groups of mutations were considered both separately and collectively in the analyses.
The frequencies of the various CFTR mutations were determined by gene counting or using a quasi-likelihood estimator accounting for the relatedness of individuals from the same family.31 32 Indeed, when individuals are related, their different alleles are not independent and using a naïve allele counting estimator would result in an inflation of the variance. The Case-Control Quasi-Likelihood Score program was used to obtain an estimator that accounts for the familial relationship and its 95% CI.31
Association between the various phenotypes and CFTR mutations was assessed by logistic regression analysis with a robust variance estimator based on family clusters when appropriate. Additive models were used where genotypes were coded as 0, 1 or 2 depending on whether they carried 0, 1 or 2 CFTR mutations. If the sample size was too small for this type of analysis, an exact logistic procedure was used. We also calculated ORs and 95% CIs. Stata V.10 (Stata Statistical Software 2007, Release 10, College Station, Texas, USA) was used for these analyses.
Family-based association tests were carried out, with the sib-TDT, implemented in the dfam procedure of PLINK v1.06.33 Briefly, dfam within PLINK implements the sib-TDT, where the number of mutated alleles carried by affected siblings is compared with those carried by unaffected siblings.34 Full details are given in the online supplement. Both asymptotic and permutation-based p values were reported.
Characteristics and CFTR mutations of the collection of families
We identified 24 families with a proband displaying RA-DB and a first-degree relative with RA and/or DB. In total, 138 participants were included in the study, including 30 with RA-DB, 8 with DB only and 24 with RA only (table 1). All were of Caucasian origin.
A CFTR mutant allele was found in at least one family member for 14 of the 24 families (table 1). In these 14 families a CFTR mutation was detected in 18 of 19 affected members with RA-DB, 6 of 8 with DB only and 5 of 14 with RA only.
Five patients with either RA-DB (3 patients) or DB only (2 patients) were found to carry Group A or B CFTR mutations in both alleles. These 5 patients were compound heterozygotes and the investigations carried out in this study led to the diagnosis of cystic fibrosis in two sisters, one with RA-DB and the other with DB only (family 8). These sisters presented with productive chronic cough and colonisation with Staphylococcus aureus in one case and Pseudomonas aeruginosa in the other. Repeated chloride test values and nasal potential differences were within the normal range for both. Both sisters carried the p.Asp1152His mutation, which has previously been identified as a Group A or Group B mutation, and the Group A c.262 263delTT mutation.13
Thus, 18 of the 30 patients with RA-DB (60%) carried at least one mutant CFTR allele and 3 of these 30 patients (10%) were compound heterozygous (table 1). Similarly, 6 of the 8 patients with DB only (75%) carried at least one mutant CFTR allele and two of these patients (25%) were compound heterozygous. This corresponds to a mutant allele frequency of 35% (21/60 chromosomes) for patients with RA-DB and 50% (8/16 chromosomes) for patients with DB only, based on a naïve estimate not taking into account the relatedness between individuals of the same family. This frequency is 32.9% for patients with RA-DB and 50% for patients with DB only if relatedness is taken into account using a quasi-likelihood method (table 2).
CFTR mutations in healthy controls
Screening of the whole coding sequence and adjacent intronic regions in the 120 controls from the healthy population revealed that five (4.2%) were heterozygous for a Group A mutation (mutant allele frequency 0.021), eight (6.7%) were heterozygous for a Group B mutation (mutant allele frequency 0.033) and none were found to have mutations affecting both chromosomes. Nine (7.5%) polymorphisms or sequence variations with possible minor effects on CFTR protein function (Group C) were also identified (mutant allele frequency 0.037). This corresponds to a cumulative mutant allele frequency of 9.2% (22/240 chromosomes) (table 2).
Comparison of CFTR mutation frequencies between the study population and healthy controls
The frequency of CFTR mutations in the 30 patients with RA-DB (p=1.9×10−5, OR 6.53) and in the 8 patients with DB only (p=7.5×10−4, OR 14.77) was significantly higher than that in healthy controls (table 2). In the 24 patients with RA only and in the 76 unaffected family members, the frequency of CFTR mutations was similar to that in the control group.
The higher frequency of CFTR mutations than in the control group was principally due to a higher frequency of Group C mutations in patients with RA-DB and a higher frequency of Group A mutations in patients with DB only (table 2).
Frequency of CFTR mutations in family members from different phenotypic groups
Importantly, CFTR mutations were more frequent in patients with RA-DB than in patients with RA without DB (OR 5.30, p=1.6×10−5; table 3). Similarly, in patients without RA, CFTR mutations were significantly more frequent in patients with DB than in those without DB (OR 4.91, p=2.3×10−5; table 4). These results remained significant even after the exclusion of family 8 which included two affected individuals diagnosed with cystic fibrosis during the study (OR 5.14, 95% CI 2.33 to 11.32, p=5.1×10−5 for comparison of RA-DB patients with patients with RA only and OR 5.04, 95% CI 2.09 to 12.13, p=10−5 for comparison of DB patients with unaffected relatives).
The patients with RA-DB also had a significantly higher frequency of CFTR mutations than unaffected relatives (OR 3.42, p=0.003; table 4). No difference in the frequency of CFTR mutations was found between patients with RA and unaffected relatives. Again, this higher frequency of CFTR mutations than in unaffected relatives was due principally to a higher frequency of Group C mutations in patients with RA-DB and a higher frequency of Group A mutations in patients with DB only (table 4).
Group C CFTR mutations were more frequent in patients with RA-DB than in those with RA only (table 3). By contrast, Group A CFTR mutations were less frequent in patients with RA-DB than in those with DB only, but the cumulative frequency of CFTR mutations did not differ significantly between the groups (table 3).
Association and linkage of CFTR mutations with DB
The sib-TDT analysis showed preferential transmission of mutant CFTR alleles with the RA-DB phenotype (p=0.005), but not with the DB-only phenotype or the RA-only phenotype (table 5). However, the number of informative allele transmissions was small for the DB-only and RA-only phenotypes. The exclusion of family 8, which included two patients diagnosed with cystic fibrosis during the study, did not affect the conclusions of the sib-TDT analysis (p=0.004 for RA-DB).
In multiplex families recruited through patients with RA-associated DB, we found that the probability of finding a CFTR mutation in family members with RA-DB or DB only was 6–15 times higher than that for healthy population controls and 3–5 times higher than that for unaffected relatives. CFTR mutations were also five times more frequent in RA-DB patients than in patients with RA without DB. Furthermore, the CFTR mutation cosegregated with RA-DB in the families. This family-based association study provides the first demonstration of an association and linkage between non-cystic fibrosis-related DB and CFTR mutations. These data indicate that RA-DB should be added to the list of phenotypes in which mutations in the CFTR gene are involved.
Heterozygosity for a CFTR mutation is the first genetic defect associated with an extra-articular feature of RA to be described, and the evidence of the linkage strongly suggests that it may predispose patients with RA to DB. The frequency of CFTR mutations was not higher in patients with RA without DB and no association was found with RA alone. Interestingly, the higher frequency of CFTR mutations in patients with RA-DB than in healthy controls or unaffected relatives resulted principally from a higher frequency of Group C minor mutations. By contrast, the higher frequency of CFTR mutations in patients with DB only than in healthy individuals or unaffected controls was principally due to a higher frequency of Group A severe mutations. Taken together, these findings suggest that even minor mutations seem to facilitate the development of DB in patients with RA whereas, in the absence of RA, the DB phenotype may result from more severe mutations. Although the frequency of CFTR mutations was roughly equivalent for patients with RA-DB and DB only, the discrepancies in the repartition of the severity of mutations between those patients might suggest that minor CFTR mutations may be relevant only in patients with RA rather than reflecting a general association with DB. Thus, common CFTR genetic variants which have unclear clinical significance independently may have modifying effects on other frequent diseases like RA. Given the poorer prognosis of patients with RA with DB, it is possible to study CFTR mutations in this particular population and genetic evaluation of the risk of associated DB should include complete screening of the CFTR gene.
Our study has several strengths that reinforce the evidence of a linkage between CFTR mutations and the DB phenotype in patients with RA. To our knowledge, this is the first study of CFTR-related disorders to make use of a family-based approach. The family-based design has unique advantages over the population-based design in that it is robust against population stratification and thus avoids the possible confounding of genotype-phenotype associations due to an inappropriate choice of controls. It can also be used to test for both linkage and association. Moreover, interviews with relatives provide high-quality data concerning family history. All patients with RA and subjects with bronchial symptoms underwent HRCT scans. DB was confirmed by HRCT in all cases. We screened the entire CFTR coding sequence and adjacent intronic regions. The association identified in our previous population-based case–control study27 was thus replicated in a completely independent sample of patients and strengthened in this family-based association study. All these data are consistent with the existence of a true association, not only because CFTR is clearly involved in pathways relevant to the development of the DB phenotype, but also because most of the CFTR gene mutations detected are known to alter the amount and/or function of CFTR protein.12
The frequency of CFTR mutations in our healthy controls was similar to both estimates and previous reports21 35 and in our study population was similar to that reported in other CFTR-related diseases. For patients with CBAVD, 18.6% (19/102) were found to be compound heterozygotes and the frequency of CFTR mutations was found to be 65.7% (134/204).14 In patients with idiopathic chronic pancreatitis, the frequency of compound heterozygotes has been reported to be between 0% (0/134) and 3.7% (1/27) and the frequency of CFTR mutations between 11.9% (32/268) and 24.1% (13/54).15 16 The frequency of compound heterozygotes (10–25%) in our patients with RA-DB and those with DB only was similar to that for patients with CBAVD14 and higher than that reported for patients with pancreatitis,15 16 whereas the frequency of CFTR mutations (32.9–50%) was intermediate between the estimates for these two groups of patients.
This study confirms that CFTR mutation is not the only factor predisposing patients with RA to DB.27 Despite extensive CFTR analysis, no mutations were detected in 10 of the 24 families. Other genetic and environmental factors are therefore likely to be involved and further research is required to identify them.
Our study has several limitations. Findings for multiple-case families may not apply to the same extent outside that context, since these families could be enriched for high-risk mutations or for other unmeasured risk factors. However, this does not invalidate our results concerning the potential role of CFTR mutations in RA-DB, as significant differences were found within families between patients with RA-DB and those with RA only or their unaffected relatives. The power of the study is limited by the small number of families analysed. However, to our knowledge, this is the largest study to date of patients with RA with DB confirmed by HRCT and, despite its limited power, we detected significant association and linkage. Only the patients with RA and individuals with bronchial symptoms underwent HRCT scanning. We therefore cannot exclude the possibility that some of the family members without RA had undiagnosed asymptomatic DB and may have been falsely classified as unaffected individuals, even if a CFTR mutation was found. Underestimation of the frequency of CFTR mutations in asymptomatic patients with DB would not change our conclusion about the role of CFTR mutations in patients with RA-DB but might explain, together with the small number of patients with DB only in the collection of families, the absence of linkage between CFTR mutations and a DB-only phenotype.
In conclusion, our data for a group of patients with a family history of RA-associated DB show a strong association and linkage between mutations in the CFTR gene and the RA-DB phenotype. RA-associated DB should therefore be added to the list of phenotypes for which mutations in the CFTR gene are of pathogenic importance. Heterozygosity for a CFTR mutation is the first genetic defect linked to an extra-articular feature of RA to be described, and may predispose patients with RA to DB. This study has potential direct clinical applications by providing clinicians with genetic markers linked to a higher risk of associated DB, and thus of a less favourable prognosis, in patients with RA.
We are grateful to the patients and their families for their participation. We thank Nadine Dufeu and Dominique Hubert for identifying families; Paul Legmann for HRCT scan analysis; Maxime Breban for critical reading of the manuscript; Eglantine Rouanet for assistance with data management; the French Society of Rheumatology, the French National Society of Internal Medicine, ‘Association Française des Polyarthritiques’ and ‘Association Rhumatisme et Travail’ for their contribution to the family collection.
Funding This work was supported by grants from the Assistance Publique-Hôpitaux de Paris (Contrat de Recherche et d'Innovation Clinique CRC98046) (to XP) and the Société Française de Rhumatologie (to XP).
Competing interests None.
Ethics approval The study protocol was reviewed and approved by the institutional review board for clinical research of Assistance Publique-Hôpitaux de Paris and the ethics committee of the Medical Faculty of René Descartes University, Paris.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.