Introduction

Cystic fibrosis (CF) is one of the most common life-shortening autosomal recessive diseases in Caucasian population, affecting about 1 person in every 2500–3500.1 Previously, the incidence of CF in Poland equalled 1/2500 as calculated on the basis of the available epidemiological data.2 However, only 1440 CF patients in Poland (of over 38 million inhabitants) were registered in Polish Cystic Fibrosis Registry (data from 30 September 2010).3 According to the Wisconsin Cystic Fibrosis Neonatal Screening Study Group and others,4, 5 newborn screening for CF (NBS CF) provides an immediate and accurate diagnosis, often before the onset of clinical symptoms and is associated with significantly improved physical development and a lower prevalence of lung infections. Neonatal screening for CF has been widely implemented and accepted, however, ethical aspects with regard to possible benefits and risks are still a matter of debate.6

CF screening programmes are based on the immunoreactive trypsin (IRT) assay on dried blood spots.7 The elevated levels of IRT have a relatively poor positive predictive value in the neonatal period, therefore earlier protocols relied on a second sample to confirm the initial screening result (IRT/IRT protocol),8 thereby enhancing sensitivity of affected newborn detection during the NBS CF to 80%.9 The identification of the CFTR gene and pathogenic mutations subsequently have opened a gateway for the genetic testing of the DNA from the original blood spot. A two-tier neonatal screening programme (IRT/DNA) was first performed in South Australia in 1989.10 Furthermore, a reduction of molecular techniques costs has allowed the possibility of direct gene analysis as a second analytic stage in many national screening programmes. Unfortunately, in the CFTR gene (188 kb, 27 exons) nearly two thousand sequencing variants have been identified to date (precise count for 31 October 2011: 1893 according to the CFTR Mutation Database, http://www.genet.sickkids.on.ca/app). The relative frequencies of mutations vary in different ethnic groups and populations11. Only one mutation, the F508del accounts for ∼70% of CFTR mutant alleles in Europe but this also differs between populations.12, 13 In Polish CF patients the frequency of the F508del mutation is estimated as 53–57%.14, 15

According to the a panel of the most frequent mutations in Polish CF patients published in 200916 and our personal experience (unpublished data), mutations such as K710X and 2184insA have a frequency >0.45%. As these sequence variants, are not included in commercial assays (eg, InnoLipa Innogenetics, CF assay Abbott, Tepnel Diagnostics Elucigene), we decided to apply the direct sequencing of selected CFTR gene regions as part of the genetic protocol of NBS CF, in order to avoid the poor effectiveness of the IRT/DNA protocol using commercial kits.

The first pilot NBS CF study was introduced in Poland in 1999 and continued until 2002, covering one-quarter of the Polish population.17 The current programme was implemented in 2006 and was expanded to cover whole country in summer 2009 (Figure 1a).

Figure 1
figure 1

NBS CF in Poland. (a) Spectrum of mutations analysed; (b) strategy and summary of obtained results.

Full size image

The aim of this summary is to evaluate the strategy of the NBS CF programme in Poland with respect to the usefulness of the DNA sequencing, as well as to revise the panel of most frequent CFTR mutations in the Polish population.

Patients and methods

Study population

The study group includes neonates born between September 2006 to December 2010. The screening programme was gradually introduced into Polish territory and since the summer of 2009, all neonates born in Poland have been included into NBS CF programme. A total of 1 212 487 newborns were screened for CF.

IRT analysis protocol

IRT was measured in dried blood spot samples from neonates (aged 3–5 days). To determine an IRT concentration (ng/ml), the IRT Neonatal Screening ELISA colorimetric assay (IBL International, Hamburg, Germany) was used. According to results from the pilot NBS CF programme completed in 2002, the IRT concentration cutoff was established as >99.4 percentile. In addition in all neonates with meconium ileus (MI), the DNA analysis was performed regardless of IRT value.

DNA analysis

Genomic DNA was eluted from two 3-mm diamater discs using the Extract Blood PCR Kit (Sigma Aldrich, St Louis, MO, USA). According to the Polish Society of Cystic Fibrosis recommendation,16 for the purpose of NBS CF, a panel of the CFTR gene mutations showed in Figure 1a was implemented. This panel and the examined regions of the CFTR gene were slightly altered during the screening programme.

The selected CFTR gene regions (Figure 1a) were analysed by direct sequencing. PCR and sequencing primers were either made as described previously18 or designed with the Primer 3 software (Geeknet Inc., Fairfax, VA, USA). The presence of the CFTRdele2,3 mutation was determined according to Dörk et al19 PCR reaction mix and conditions are available on request. PCR products were purified using EXOSAP-IT (USB Corporation, Cleveland, OH, USA) according to the manufacturer’s protocol. Standard sequencing reaction and capillary electrophoresis using ABI PRISM 3730 sequencer (Applied Biosystems, Life Technologies Corporation, Carlsbad, CA, USA) were performed. Fluorochromatograms were analysed using the Mutation Surveyor software (SoftGenetics, State College, PA, USA). Sequence NM_000492.3 was used as a reference. Novel mutations were analysed with prediction module of Alamut version 1.53 software (Interactive Biosoftware, Rouen, France). The CFTR mutations nomenclature is consistent with the CFTR Mutation Database legacy names (http://www.genet.sickkids.on.ca). Other databases used in the study were as follows: NCBI (http://www.ncbi.nlm.nih.gov), Ensembl (http://www.ensembl.org/index.html), 1000 Genomes (http://www.1000genomes.org) and HGMD Professional (https://portal.biobase-international.com/hgmd, http://www.hgmd.cf.ac.uk/ac/index.php).

Clinical confirmation of CF

Neonates with one or two identified mutations were referred directly for clinical assessment and a sweat test at the age of 3–4 weeks to CF Centres in Poland (eight centres). The sweat tests were performed by the classic Gibson and Cook method and Nanoduct method parallely.20 The diagnosis of CF was established according to recommendations.16 Parents of neonates that were identified as affected or carriers were suggested genetic counselling and were offered genetic testing.

Results

During a screening project lasting over 4 years, more than one million newborns were screened (Figure 1b). For each positive IRT result (>99.4 percentile) direct CFTR gene analysis was performed by sequencing of selected regions. The complete analysis using this diagnostic panel allowed us to identify (1) nearly 80% of the mutated alleles in Poland and (2) mutations in one or both alleles in nearly 95% of screened CF patients (according to frequency data published by Bobadilla et al15). Blood trypsin values above the threshold were found in 7532 neonates. After biochemical, genetic and clinical evaluation, 216 affected subjects (194 genotyped and 22 with 1 mutation identified so far) were identified in the NBS CF programme. This group includes 20 cases of (MI; 9.3%, 20/216). In addition, one newborn with MI did not have any genetic analyses performed, because information about MI was not printed on the screening card and IRT was below the cutoff for IRT. This newborn had CFTR molecular analysis performed at the age of 2.5 years (genotype: [F508del];[F508del]) due to clinical symptoms. So far (until March 2012) we have had information about four cases of false-negative results in NBS CF (Table 1). Thus, the total number of CF cases identified in the period of NBS CF in Poland is 221 (216 from NBS CF, 4 omitted by NBS CF and 1 with unreported MI).

Table 1 Characteristic of the cases omitted in the screening for CF programme owing to IRT values <99.4 percentile
Full size table

In five other children, despite two CFTR mutations detected, a clinical evaluation did not confirm the presence of CF. These newborns had the CFTR genotype as follows: [F508del];[D537N], [F508del];[P731L], [F508del];[T1053I] (two unrelated newborns) and [F508del];[L467F]. Mutations D537N and P731L have not been reported to the CFTR Mutation Database so far. Beside these two, several other new defects not registered in the CFTR Mutation Database were detected as well as rare mutations that were not included into commercial assays (Tables 2 and 3).

Table 2 Genotypes of CF newborns with mutations not included into common commercial kits applied in Poland and European countries*
Full size table
Table 3 Frequency of CFTR mutations in Polish CF patients from newborns screening programme
Full size table

The strategy based on sequencing of CFTR regions implemented into NBS CF allowed us to identify a full genotype for 29 newborns (13% of all identified CF cases), which could not be detected by other routinely used molecular methods (Table 2). In addition, in 22 subjects, extended CFTR genetic analysis was suggested owing to abnormal or borderline sweat tests results and clinical suspicion of CF.

The pathogenicy status of novel and rare missense variants was assessed by bioinformatic analysis using the Alamut software. This programme evaluates effect of particular sequence variations using PolyPhen, Align-GVGD and SIFT alghoritms. The D537N variant in exon 11 was designated as possibly pathogenic and P731L as possibly tolerated. Of note, in the case of some other changes, the results were unclear – for example, in the case of already known mutations published in the CFTR Mutation Database: T1053I (according to Align-GVGD: ‘less likely to be pathogenic’, score C0, PolyPhen: ‘possible damaging’, score 0.816 and SIFT: ‘deleterious’, score 0.01), similarly for L467F (Align-GVGD: ‘less likely’, class C0, PolyPhen: ‘probably damaging’, score 0.994 and SIFT: ‘deleterious’, score 0.03).

The evaluation of the most frequent CFTR mutations in the Polish population in examined gene regions showed that three additional mutations reached our adopted cutoff (0.45% of frequency): 3600+2insT, 1898+1G>C and V1240G (Table 3). We assumed that during the period of the NBS CF programme, we detected all F508del homozygotes (90/221 cases, 41%). Based on this assumption, the previously published data of the frequency of this mutation in the Polish population (57%,15), the data from the Polish Cystic Fibrosis Patients Registry3 (56–62%) and the results of the clinical follow-up ending in December 2011, we calculated the CF incidence in our country as 1/4394 and carrier frequency as 1/33. The incidence and carrier frequency are lower (1/5249:5486 and 1/36:37, respectively) when results from NBS CF are taken into account.

Discussion

The screening programme for CF has been implemented for many years in the Europe, USA and Australia despite the discussion over its benefits and risks.21 Here, we present the first summary of the NBS CF in Poland based on the IRT and sequencing of selected CFTR fragments.

Our NBS CF strategy lead to the diagnosis of CF in 216 neonates. The DNA analysis protocol implemented here focused on regions containing the most frequent mutations in the Polish population covering almost 80% of mutated alleles. The 90 F508del homozygotes (41%) were identified during the period of the NBS CF programme and this percentage is in a large convergence with data from the Polish Cystic Fibrosis Registry (38%, according to Stężowska-Kubiak3).

In recent years, the population analysis data are often used for the recalculation of the CF incidence and CFTR mutation carriership. However, the significance of these values is questionable due to the extent of false-negative results, and the incompleteness of data. This particulary important within the first years of a screening programme, when clinical follow-ups are continued. The incidence of the CF in Poland calculated on the number of F508del homozygotes and the previously reported F508del allele frequency15 would be 1/4394. The incidence of disease estimated directly on basis of the NBS CF programme results is 1 per 5248–5486. In past, in the 1970s this parameter has been generally defined in Poland as 1/2500, however, in study of Bozkowa et al2 selection of examined groups was not accurately matched to disease prevalence calculation. CF prevalence is more difficult to ascertain for a number of reasons, such as different medical and scientific data from published literature and from patient registries. The results of other screening programmes also show a lower incidence of the disease, than traditionally adopted 1/2500.22 Data from European countries, published in 2008 point an interregional variation as well (Table 4).23 Similarly, The American College of Medical Genetics with American College of Obstetricians and Gynaecologists published summary from 8 years of CF testing and estimated carrier frequency as 1/37.6 in the tested panethnic population, which allows to calculate the CF frequency as 1/5655.24

Table 4 Incidence of CF in selected European country from the literature, national CF registries and personal information from CF leaders in Europe (according to Farrell,23 modified)
Full size table

It is probable that the incidence of the disease in Poland is lower than the value declared in the 70s, however, its precise calculation can only be possible when the results of the NBS CF programme are summarized with the addition of clinical follow-up. We also cannot rule out that more patients have been omitted by NBS CF, like the four cases reported in this study. According to Fritz formula,9 the predicted false-negative rate for the Polish population is 6–7 cases per year, which gives 1–2 omitted CF cases per 100 000 live births. After a longer follow-up, some children that are currently defined as carriers would turn out to be affected with rather mild CF or CFTRrelated diseases. In addition, according to the tendency of a smaller family model (2+1) in Europe25 and prenatal screening,26 a lower prevalence of the disease will probably be observed in the future. Such tendencies have already been observed.27, 28

Beside analytical and statistical limitations, another possible reason is the under representation of children with MI. In our study, the newborns with MI only represent 9.3%, while the literature data shows about 10–20%. In neonates with MI, normal values of the IRT are often observed.21 Such newborns could have obtained a false-negative result of IRT and had not been redirected for genetic testing.

By preparing a newborn screening programme in Poland, we collected information about the CFTR mutations detected in Polish CF patients that are recognized on the basis of the clinical outcome (data from The Polish Cystic Fibrosis Registry and Bobadilla et al15). According to these data, the F508del frequency was estimated as ;56–62%. The frequency of the F508del allele in this group is similar to the value calculated from the NBS CF programme (62%) and previously published data15 (57%). Data from the Polish registry were used to designate the CFTR gene mutations and to prepare a ‘Polish assay’ with mutations such as K710X and 2184insA. The exonic 2184insA mutation in addition to others (eg, CFTRdele2,3, 3849+10kbC>T ) occurs more frequently in Central and Eastern Europe.29 Thus, the main reason for the application of our strategy was the diversity of mutations in the Polish population, for which commercial tests are not suited (Tables 2 and 3). An extended molecular analysis revealed that 3600+2insT, mutations in loci 1898+1 (G>C or G>A) were detected in two and three cases of CF children with extended analysis (thus>cut off 0.45% of mutated alleles detected in NBS CF). These variants, together with V1240G (two alleles, frequency>0.45%) may therefore be candidates for a first line NBS CF molecular test panel in future. However, these results should be confirmed on a larger sample in subsequent years of NBS CF.

A selective CFTR gene scanning strategy was also used in other populations.30 An additional advantage of such diagnostic procedure is its ability to discover new or ultra-rare CFTR alleles. Indeed, we discovered about 30 sequence variants unreported previously.

We realize that the discovery of new defects is not always beneficial owing to interpretational controversies. In example, the detection of a sporadic missense variants (eg, D537N, P731L), still do not provide a clear answer to the biological and clinical significance of the defect. The D537N variant was designated as possibly pathogenic by bioinformatic analysis, whereas the P731L was designated as ‘tolerated’. Both variants were detected in the newborns (current ages: 3 years 2 months and 1 year 5 months, respectively), that had the F508del mutation identified in the second allele, and sweat chloride concentration 14.9 and 31.2 mmol/l, respectively. Until now, the disease based on clinical data has not been confirmed in these children. However, at this time, we cannot exclude the possible development of an atypical CF or CFTR-related disorders (eg, CBAVD – both children are males).

A similar problem was observed while detecting rare known mutations published in the CFTR Mutation Database. For example, the diagnosis of CF has not been confirmed following the identification of the T1053I and L467F mutations. As this is common problem in the NBS CF programmes, the term ‘CFTR-related metabolic syndrome’ was proposed.31

Dorfman et al32 in his work concluded that none of the current generation of in silico software accurately predicted the clinical severity of the CFTR mutations in a reliable way. Therefore, each bioinformatic analysis performed by us only served as an indicator rather than a conclusive diagnostic marker. However, new genotype–phenotype relations are still being discovered that may lead to the requalification of a mutational status. For example, the I148T variant no longer has the qualification of a pathogenic mutation but is still included in the commercial kits (eg, InnoLipa, Innogenetics).33 Prudence is therefore always essential in molecular genetics.

Our experience shows that DNA protocol based on sequencing method is an efficient screening strategy in Poland. In addition to issues described above, this approach seems to be highly cost-effective considering the fact that, there is no need to repeat the analysis of exons examined during NBS CF programme, if extension of the CFTR gene analysis is necessary. However, the limitations of the IRT/DNA test should not be forgotten and each patient with signs or symptoms of CF should be referred to the CF centre for standard diagnostic procedure.

In summary, to our knowledge, this is the world’s first full description of a CF newborn screening strategy based on IRT/DNA-extended genetic analysis method, recently recommended by the Clinical and Laboratory Standards Institute.34