Article Text


Cystic fibrosis patients with the 3272-26A>G splicing mutation have milder disease than F508del homozygotes: a large European study
  1. Margarida D Amarala,b,
  2. Paula Pachecoa,
  3. Sebastian Becka,
  4. Carlos M Farinhaa,b,
  5. Deborah Penquea,
  6. Paulo Nogueirac,
  7. Celeste Barretod,
  8. Beatriz Lopese,
  9. Teresa Casalsf,
  10. Javier Dapenag,
  11. Silvia Gartnerh,
  12. Carlos Vásquezi,
  13. Javier Pérez-Fríasj,
  14. Casilda Olveiraj,
  15. Rodrigo Cabanask,
  16. Xavier Estivillf,
  17. Maria Tzetisl,
  18. Emmanuel Kanavakisl,
  19. Stavros Doudounakisl,
  20. Thilo Dörkm,
  21. Burkhard Tümmlerm,
  22. Emmanuelle Girodon-Boulandetn,
  23. Cécile Cazeneuven,
  24. Michel Goossensn,
  25. Martine Blayauo,
  26. Claudine Verlinguep,
  27. Isabel Vieirap,
  28. Claude Férécp,
  29. Mireille Claustresq,
  30. Marie des Georgesq,
  31. Christine Clavelr,
  32. Philippe Birembautr,
  33. Dominique Huberts,
  34. Thierry Bienvenut,
  35. Michèle Adounu,
  36. Jean-Claude Chomelu,
  37. Kris De Boeckv,
  38. Harry Cuppensw,
  39. João Lavinhaa
  1. aCentro de Genética Humana, Instituto Nacional de Saúde Dr Ricardo Jorge, Av Padre Cruz, 1649-016 Lisboa, Portugal, bDepartmento de Química e Bioquímica, Faculdade de Ciências da Universidade de Lisboa, Lisboa, Portugal, cObservatório Nacional de Saúde, Instituto Nacional Saúde Dr Ricardo Jorge, Lisboa, Portugal, dUnidade de Fibrose Quística, Serviço de Pediatria, Hospital de Stª Maria, Lisboa, Portugal, eHospital D Estefânia, Lisboa, Portugal, fMolecular Genetics Department-IRO, Hospital Duran I Reynals, Barcelona, Spain, gCystic Fibrosis Unit, Hospital Universitario Virgen del Rocio, Sevilla, Spain, hCystic Fibrosis Unit, Hospital Universitario Materno Infantil Vall d'Hebrón, Barcelona, Spain, iCystic Fibrosis Unit, Hospital Infantil de Cruces, Barakaldo, Vizcaya, País Vasco, Spain, jCystic Fibrosis Unit, Hospital Carlos Haya, Malaga, Spain, kPaediatric Service, Hospital Xeral, Galicia, Spain, lFirst Department of Paediatrics and Choremio Research Laboratory, Unit of Molecular Medicine, St Sophia Children's Hospital, Athens, Greece, mKlinische Forschergruppe, OE 6711, Medizinische Hochschule, Hannover, Germany, nService de Biochimie et de Génétique, Hôpital Henri Mondor, Créteil, France, oService de Génétique Moléculaire et Hormonologie, Centre Hospitalier Regional et Universitaire de Rennes Pontchaillou, Rennes, France, pLaboratoire de Génétique Moléculaire et d'Histocompatibilité, Centre Hospitalier Universitaire de Brest, France, qLaboratoire de Génétique Moléculaire et Chromosomique, Institut de Biologie, Montpellier, France, rLaboratoire Pol Bouin et Unité Inserm U514, Hôpital Maison Blanche, CHU Reims, France, sLaboratoire de Biochimie et Biologie Moléculaire, Groupe Hospitalier Cochin, Paris, France, tService de Pneumologie, Groupe Hospitalier Cochin, Paris, France, uLaboratoire de Génétique Céllulaire et Moléculaire, CHU de Poitiers, France, vDepartment of Paediatrics, UZ Gasthuisberg, Leuven, Belgium, wCentre for Human Genetics, University of Leuven, Campus Gasthuisberg, Leuven, Belgium
  1. Dr Amaral, mdamaral{at}

Statistics from

Editor—Cystic fibrosis (CF, MIM 219700) is a common, severe, autosomal recessive disease caused by mutations in the CF transmembrane conductance regulator (CFTR) gene cloned in 1989.1-3The disease, characterised by chronic lung disease which is the main cause of morbidity and mortality, pancreatic dysfunction, raised electrolyte levels in sweat, and male infertility, is caused by altered chloride (Cl) secretion across the apical membrane of epithelial cells.4 There is, however, substantial variability in the clinical manifestations affecting the various organs.4 5

One single mutation, F508del, generally associated with severe disease, accounts for about 70% of CF chromosomes world wide, although with a heterogeneous geographical distribution.5 Patients homozygous for the F508del mutation have the classical severe form of the disease which includes chronic mucous obstruction of the lung and conducting airways, followed by recurrent infections mostly byPseudomonas aeruginosa (Pa) andStaphylococcus aureus (Sa), exocrine pancreatic insufficiency (PI), resulting in failure to gain weight and height, and raised levels of Cl, sodium, and potassium in exocrine sweat.5 However, almost 1000 genetic alterations have been detected in the CFTR gene (CFTR Mutation Database), most presumed to be disease causing mutations. About half of these are amino acid substitutions (missense mutations) and about 20% are splicing mutations. The remainder are nonsense, frameshift (including small deletions and insertions), and a small proportion of promoter mutations.

The relationship between genotype, that is, the mutations in theCFTR gene, and the clinical phenotype of CF patients has been difficult to establish, in particular for lung disease.

It was previously shown that the 3272-26A>G mutation leads to the creation of an alternative acceptor splice site competing with the normal one during RNA processing and resulting in the occurrence of an alternatively spliced mRNA with 25 extra nucleotides from intron 17a and a premature stop codon soon thereafter.6 Previously it was reported that three patients carrying the 3272-26A>G mutation in one of their CFTR alleles had mild CF.7 Another patient with the 3272-26A>G/F508del genotype was reported to have severe CF.8 Here, we report the clinical phenotypes of 60 CF patients from several European centres, with the 3272-26A>G mutation on oneCFTR allele, and mostly another severe mutation (73% F508del) on the other allele. We compare them, by statistical methods, with the clinical phenotypes of F508del homozygotes (n=89) from the same centres, matched for age and sex as exactly as possible.

Subjects and methods


Patients were clinically and genetically characterised in the various CF centres involved in this study (table 1). For the control group, one (two, when possible) F508del homozygous patient for each patient with the 3272-26A>G mutation was selected from the same centres, matching for age and sex with the 3272-26A>G patients as far as possible. However, CF patients with 3272-26A>G were generally among the oldest CF patients in each centre. Therefore, for some cases it was not possible to have two control F508del homozygotes exactly matched. In such cases only one was included and this was chosen as the oldest F508del homozygote from the same centre. The mean current age (SD) of patients in the groups of 3272-26A>G (n=60) and F508del homozygous (n=89) patients were 20.5 (SD 17.5) and 17.0 (SD 11.5) years, respectively. It was found, however, that this difference was not significant (data not shown).

Table 1

Distribution of CF patients in this study according to their genotype and country


Genomic DNA was isolated from peripheral blood lymphocytes according to standard protocols. The 3272-26A>G mutation (as well as non-F508del mutations on the other allele) were detected either by single stranded conformation analysis (SSCA)9 or by denaturing gradient gel electrophoresis (DGGE)10 after PCR amplification of genomic DNA in the region of the correspondingCFTR exon. Amplicons with abnormal patterns were sequenced either by the ABI PRISM™ Dye Terminator Cycle Sequencing System (Perkin-Elmer, Norwalk, CA) or by the dideoxy manual method with a [35S] nucleotide.

Detection of the F508del mutation was either by dot blotting, by amplification refractory mutation system (ARMS),11heteroduplex analysis on polyacrylamide gel electrophoresis (HA-PAGE),12 or oligonucleotide ligation assay (OLA).13

Microsatellites IVS8(CA)n were investigated as described previously.14 The IVS8 (TG)nTmpolymorphic tract was also analysed as previously described15 and sequenced with the following primer: 5′GAAATTACTGAAGAAGAGGC3′. The GATT tetranucleotide in intron 6a IVS6a-(GATT)n was analysed by PCR amplification and electrophoresis of the products in a 12% (w/v) polyacrylamide gel.16 The diallelic markers XV-2c/TaqI, KM.19/PstI, MP6-D9/MspI, J44/XbaI, M470V/HphI (1540A>G), and T854/AvaII (2694G/T) were analysed by PCR amplification and digestion with appropriate restriction enzymes as described previously.17


The clinical data included in this study were age at diagnosis, sweat test values, pulmonary status assessed by forced expiratory volume in one second (FEV1) % predicted, and forced vital capacity (FVC) % predicted (which are predicted values for the non-CF population by Knudson et al 18), lung colonisation with bacterial pathogens, pancreatic status, history of meconium ileus, weight and height centiles (for patients under 18) or body mass index (BMI, for patients over 18), Chrispin-Norman (CN) chest radiological score (from 0 to 38, with 0 being the best score, as defined by Conway and Littlewood19), Shwachman-Kulczycki (SK) general status score (100 is the best score, also as defined by Conway and Littlewood19), nasal polyposis, and other clinical complications or abnormalities.


Data for all clinical parameters are presented (table 2) as the mean (SD) as well as median and interquartile deviation (QD). For distributions of quantitative measurements, the hypothesis of equal variances between the two groups of CF patients under study was tested using Levene's test and there was no evidence to reject it (p>0.05), except for current age, age at diagnosis, and weight centile (table 3). Thus, for these two distributions, the Mann-Whitney U test (Wilcoxon) for two independent samples20 was applied. For the other distributions with equal variances, statistical significance comparisons between the two groups of CF patients were performed using the parametric Student's t test for two unpaired samples.20 For qualitative distributions, the hypothesis of association was tested with each of the two groups of CF patients, considered as two independent samples in 2 × 2 crosstabs (1 degree of freedom), using both the Pearson chi-square calculation with continuity correction and Fisher's exact test.20Coefficients with a p value less than 0.05 were considered to be statistically significant. The SPSS® for Windows software (SPSS Inc, Chicago, IL) was used for all statistical calculations.

Table 2

Clinical features the two groups of CF patients

Table 3

Significance tests for comparisons of clinical features between cystic fibrosis patients with 3272-26A>G/any mutation and F508del/F508del genotypes


We have compared the clinical phenotypes of 60 CF patients with the CFTR genotype 3272-26A>G/any mutation with those of 89 patients who were F508del homozygotes, that is, presenting classical CF disease. Patients with the 3272-26A>G mutation were from six different countries, namely France, Spain, Greece, Germany, Portugal, and Belgium (table 1). Among these, 44 (73%) patients were F508del compound heterozygotes (one of them died at the age of 59 years and another one received a lung transplant at the age of 34), four patients were heterozygotes for R1162X (two sib pairs) and two (also sibs) for E822X (table 1). Nine patients carried the following mutations in the otherCFTR allele (one of each): W1282X, 2869insG, L206W, N1303K, 1717-1 G>A, G542X, 4218insT, W846X, P99L. Except for the P99L and L206W substitutions, all the mutations are known to represent severe CF alleles. For one of the patients carrying the 3272-26A>G mutation the second mutation was not identified (table 1). However, the clinical phenotype was clearly CF, although mild, so the patient was included in the study.


Most of the patients analysed (80%) had the haplotype D (2,2) at the XV-2c/KM19 loci associated with the 3272-26A>G mutation. For these, an extended haplotype analysis, when done, showed the same variants, being thus consistent with the presence of the same mutant allele in all these patients. The 3272-26A>G mutation was also found in alleles with haplotypes B (1,2) and C (2,1) at XV-2c/KM19 in one French patient and in another previously described Belgian patient,21 respectively. The other markers in the extended haplotype for the Belgian patient were all the same as for D.

Four other patients, representing about 13% of patients analysed (one in Greece, one in Germany, and two in France) had the haplotype A (1,1) at XV-2c/KM19 in linkage disequilibrium with the 3272-26A>G mutation. An extended haplotype with other markers was determined for the German and the Greek patients, differing substantially from those present in the D allele.


Table 2 shows the average values (mean and median) for the incidence of several clinical parameters of the CF patients in this study. Data are shown for a total of 60 patients with theCFTR genotype 3272-26A>G/any mutation and for 89 F508del/F508del patients used as the control group. Differences between these clinical parameters in the two groups of patients were tested for significance (see Methods). Results from the statistical analysis, and reference to the test applied in each case, are shown in table 3. Significant differences were found for the following parameters: age at diagnosis (higher for the 3272-26A>G group), FEV1 and FVC (also higher), incidence of lung colonisation with Pa (lower), occurrence of pancreatic insufficiency (lower), weight centile or BMI (higher), and nasal polyposis (higher). Altogether these results indicate milder CF disease in patients carrying the 3272-26A>G mutation on oneCFTR allele. For none of the other parameters (electrolyte concentration in sweat, colonisation with pathogens other than Pa, occurrence of meconium ileus, height centile, and SK and CN scores) were differences found to be significant.


Patients who have milder symptoms or atypical CF often have one severe and one so called class V mutation.22 This class includes mutations that leave residual levels of normalCFTR transcripts and protein.23Mild CF disease was generally reported for the following class V (splicing) mutations: 3849+10 kb C>T,24IVS8-5T,9 25 and 2789+5G>A.26 However, the severity of the disease resulting from these mutations was never assessed for significance in comparison to typical CF, that is, presented by most F508del homozygotes. Yet this information is important for genetic counselling, in particular for prenatal diagnosis. Here, we report that patients with the 3272-26A>G mutation (and another CFTRmutation, mostly a severe one) exhibit significant differences for various major clinical parameters in comparison to typical CF disease.


Among the 29 patients included in this study whose haplotypes at the XV-2c/KM19 loci were determined and from data on one patient published elsewhere,21 80% have the D haplotype (2,2) in association with the 3272-26A>G mutation.

However, haplotype data also suggest that the 3272-26A>G change must have occurred through a second mutational event. Indeed, four other patients (two French, one German, and one Greek) carry the 3272-26A>G mutation in association with the A haplotype at the XV-2c/KM19 loci. According to data available from the German patient (and also partially from the Greek patient) for other markers, this is a totally different allele, thus strongly suggesting a second origin for the 3272-26A>G mutation. Possibly this happened in populations living in more eastern regions of Europe, considering the origin of two of these patients (one Greek and another from former East Germany). Owing to the lower frequency of allele A (13%) in comparison to D (80%) in association to the 3272-26A>G mutation, the former may have occurred later. Haplotype data suggesting the further change/recombination of the D allele after occurrence of the splicing mutation also support this hypothesis.

The A haplotype is associated with the TG11 repeat (v TG10 for the D haplotype). Since the length of the TG repeat in intron 8 seems to correlate with the extent of exon 9 alternative splicing, that is, the higher the number of TG repeats, the lower the amount of exon 9+transcripts produced,27 it might be expected that the patients with the A haplotype would have a lower amount of exon 9+ transcripts and hence a more severe phenotype. Although only four of the 3272-26A>G patients analysed here had the A haplotype, no significant difference in clinical phenotype was found for these patients in relation to those with haplotype D.


Patients carrying the 3272-26A>G mutation on one allele (and mostly a severe CFTRmutation on the other) were shown here to be diagnosed later, to have better lung function, lower incidence of lung colonisation with Pa, and more often to have normal pancreatic function and higher weight centiles or BMI than patients homozygous for F508del. Altogether, such results indicate milder CF disease in these patients than typical CF. Unexpectedly, nasal polyposis occurs more frequently in these patients (about 37% as opposed to around 10% in F508del homozygotes). It is plausible that nasal polyposis may be underdiagnosed in this study (particularly in the group of F508del homozygotes), since it has been reported to occur generally in about 37% of CF patients28and, in particular, in about 40% of F508del homozygotes.29 The 5T allele, generally associated with mild CF, was also detected with increased frequency in subjects with sinopulmonary disease of ill defined aetiology.30 Thus, the incidence of nasal polyposis, which is generally considered as a minor complication, does not disprove the fact that patients with 3272-26A>G on one allele have milder CF than F508del homozygotes.

This mutation has been previously shown6 to create an alternative acceptor splicing site in intron 17a that competes with the normal one but still allows some normalCFTR mRNA to be produced. We postulate that the remaining normalCFTR mRNA still existing in these patients lessens the severity of CF disease. The molecular basis of this significantly milder CF phenotype must thus lie in the existence of remainingCFTR mRNA that is still normally processed, thus giving rise to functional protein.6 This, however, is not enough to completely avoid lung disease. There are reports of other mutations, both in theCFTR gene (reviewed by Kerem and Kerem31) and in other genes such as β globin,32 indicating that reduction in the normal protein levels generally causes milder disease than mutations leading to total absence of or non-functional protein.

The variability and organ involvement in patients carrying this mutation may critically depend on the levels ofCFTR mRNA (and protein) still present. Differences between the normal and the alternative splicing processes can result from differential expression of splicing factors which will thus act as modifying factors ofCFTR expression.33 Indeed, some variability was observed among patients included in this study. Some patients (with F508del on the other allele) were diagnosed at an early age, have Pa colonisation, reduced lung function, and are pancreatic insufficient. Others have very different clinical records. One patient (with F508del on the other allele) was only diagnosed at the age of 32 because of persistent cough during her first pregnancy. Another (also with F508del on the other allele) at the age of 37 had still not developed lung disease, nor pancreatic insufficiency, and was only genotyped because of congenital bilateral absence of the vas deferens (CBAVD).

Owing to the relatively high incidence of this mutation in Europe, it might have been expected that one or more homozygotes for this mutation would be found, as described for other mutations of comparable incidence.24 34 One possible explanation for our failure to identify any homozygotes is that they may not be CF patients. In light of the fact that just one 3272-26A>G allele, in compound heterozygosity with a severe mutation (like F508del), causes a significant reduction in phenotypic severity, it is plausible that two 3272-26A>G alleles would avoid CF totally.

Further characterisation ofCFTR splicing in these patients, namely by quantifying normal messenger still present, is our current research, and it will be useful to estimate the minimum levels of CFTR necessary to avoid CF lung disease. It is hoped that these insights may also provide clues for novel therapies, through the modulation of factors that enhance the normal versus the alternative splicing.

  • We report here clinical phenotypes of 60 cystic fibrosis (CF) patients from six European countries with the 3272-26A>G mutation on one allele of the cystic fibrosis transmembrane conductance regulator (CFTR) gene, and another mutation (mostly F508del) on the other allele. These were compared with the clinical phenotypes of F508del homozygous patients (n=89) from the same centres and matched for age and sex as exactly as possible.

  • Clinical phenotypes of CF patients with 3272-26A>G were found to be significantly milder (p<0.05) than those of F508del homozygotes, namely older age at diagnosis, better pulmonary function, lower incidence of colonisation with Pseudomonas aeruginosa, lower occurrence of pancreatic insufficiency, and higher weight centile or body mass index. Altogether, this information is important for genetic counselling, in particular for prenatal diagnosis.

  • Unexpectedly, incidence of nasal polyposis was found to be significantly higher in 3272-26A>G patients.

  • The heterogeneity of CFTR haplotypes indicates that this mutation, which is spread all over Europe, must have evolved from more than one mutational event. It is shown for the first time that CF patients with one mutation causing alternative splicing (and mostly another severe mutation) have significantly milder disease than patients with two severe mutations. We postulate that the remaining normal CFTR mRNA still existing in these patients alleviates the severity of CF disease.


Electronic database information. Online Mendelian Inheritance in Man (OMIM), http: // (for CF MIM 219700).CFTR Mutation Database, http: // (for genetic alterations detected in theCFTR gene).

This work was supported by PRAXIS XXI P/SAU/55/96 (Portugal), FIS 99/0654 and Instituto Catala de la Salut (Spain), Deutsche Forschungsgemeinschaft (Germany), and Association Française de Lutte contre la Mucoviscidose AFLM (France) research grants. SB and CMF were recipients of BPD/17059/98 and BD/11094/97 fellowships, respectively.


View Abstract

Request permissions

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.