Statistics from Altmetric.com
Editor—Neonatal screening for cystic fibrosis (CF) involves measurement of neonatal blood spot immunoreactive trypsinogen (IRT),1 followed by gene mutation analysis in those with a raised (>99th centile) immunoreactive trypsinogen (IRT).2 Screening with this IRT/DNA protocol has been shown, from a number of centres, to detect a greater number of ΔF508 heterozygotes than expected from the known carrier frequency of ΔF508.1-3 The reason for this is unknown but may be explained if ΔF508, on its own, affects pancreatic function. If this is true, then the frequency of ΔF508 should increase with increasing values of neonatal trypsinogen.
It has also been shown that the intron 8 polythymidine tract sequence 5T (IVS8-5T) is more frequent in neonates with hypertrypsinaemia.4 5 The intron 8 polythymidine sequence regulates the splicing of exon 9 in transcription of the cystic fibrosis transmembrane conductance regulator protein (CFTR).6 IVS8-5T is associated with the least efficient splicing of CFTR, but whether the frequency of IVS8-5T increases with increasing level of IRT, in the absence of another mutation, is unknown.
The relationship between neonatal immunoreactive trypsinogen, ΔF508, and IVS8-5T has not been defined. To answer this question we studied the frequency of ΔF508 and IVS8-5T from neonatal blood spots systematically in IRT strata below the 99th centile cut off normally used for DNA testing.
Newborn screening for cystic fibrosis using day 4 blood spots on filter paper cards is routine in Victoria, Australia. The measurement of immunoreactive trypsinogen (IRT) by fluoroimmunoassay is the primary screen, with the IRT values being normally distributed, and the top 1% being selected for ΔF508 mutation analysis.7 We divided newborn screening cards into six strata, based on IRT value below the 99th centile, and 105 cards were selected at random from each strata. We chose an IRT value of 30 μg/l as the upper limit, since the 99th centile is almost always above this figure. Cards from the most recent complete year (1996) were selected to ensure adequate freshness of the blood spot. The study was conducted anonymously and 3 mm punched blood spots labelled to allow correlation between the ΔF508 mutation analysis and the intron 8 polythymidine sequence. DNA was eluted from the blood spot and ΔF508 mutation analysis performed by polymerase chain reaction (PCR) using specific primers and the products detected by gel electrophoresis.8 9 The polythymidine sequence was determined by nested PCR and the products sized by gel electrophoresis.10 11 The study was approved by the ethics committee of the Royal Children's Hospital, Melbourne.
The strength of the association between IRT level and genotype frequency was assessed by chi-squared test of trend.12 In order to determine whether the total number of samples with either ΔF508 or IVS8-5T was comparable to the known frequency of these mutations in the population, we calculated a stratum weighted prevalence estimate, using weights based on the total of infants in each IRT stratum for the year of the study.
The results of the relationship between IRT, ΔF508, and the intron 8 polythymidine tract are presented in table 1. In some cases, DNA did not amplify and only subjects for whom both ΔF508 and IVS8-5T results were available are presented.
There were no ΔF508 homozygous infants detected below the 99th centile, the cut off that is currently used for newborn screening. There were 22 ΔF508 heterozygotes detected, representing 3.6% of the study sample. The frequency of ΔF508 increased with increasing level of IRT (χ2 for trend=4.3, p=0.04). The stratum weighted prevalence estimate of the frequency of ΔF508 in the population was 2.5% (95% CI 0.9-4%) which is not different from the expected number, had we selected 605 blood spots cards at random, based on the carriage frequency for ΔF508 of 1/33 in the Victorian population.
Only one ΔF508 heterozygous infant also had the IVS8-5T allele (genotype 5T/7T), and this infant was in the 26-30 IRT cohort. Of the remaining ΔF508 infants, 18 had the 7T/9T genotype, two 7T/7T, and one 9T/9T. Heteroduplex band analysis indicated that one of the subjects with a 7T/7T background was a ΔI506/7 heterozygote (fig 1) while the other 7T/7T subject and the 5T/7T subject were both ΔF508 heterozygotes. Because ΔF508 has always been reported to be in cis with 9T, we performed gene sequencing on the two ΔF508 subjects with a non-9T background which confirmed the exon 10 mutation as ΔF508 (fig 2).
There were no homozygous 5T/5T infants detected, but 40 had the 5T/7T genotype and three the 5T/9T genotype. The most frequent intron 8 polythymidine genotype was 7T/7T (n=447), with 7T/9T (n=107) the second most common, while only eight infants had the 9T/9T background. The total number of infants with IVS8-5T detected was 7.1% of the group, and the stratum weighted prevalence estimate of the frequency of IVS8-5T in the study was 6% (95% CI 3.6-8.4%). There was no clear increase in the frequency of IVS8-5T with increasing IRT (χ2 for trend=2.4, p=0.12).
We have shown that the frequency of ΔF508 increases with increasing levels of neonatal IRT and that this is independent of the IVS8-5T allele. We detected the expected number of ΔF508 alleles, but most were found in the higher IRT strata, suggesting that the distribution of ΔF508 heterozygotes is skewed to the higher levels of IRT. It is clear that a single, severe CFTR mutation such as ΔF508 can affect neonatal pancreatic function, and explains why the detection of ΔF508 heterozygotes is increased in an IRT/DNA screening protocol.
The effect of ΔF508 on neonatal IRT is interesting in the light of recent reports of an increased frequency ofCFTR mutations in patients with chronic idiopathic pancreatitis.13 14 This suggests that singleCFTR mutations may be associated with clinical disease, although it is likely that there may be some role for additional environmental exposure such as tobacco smoke or alcohol. We did not detect ΔF508 homozygotes from our group of blood spots taken from IRT values below the cut off normally used for newborn screening. This is reassuring, although we did not sample enough blood spots to be absolutely sure no homozygotes had been missed. We detected the expected number of ΔF508 heterozygotes and it is unlikely that any have a second severe mutation (for example, G551D, G542X, or R553X) as this group of compound heterozygotes have a severe CF phenotype and almost invariably have an IRT above the 99th centile. We studied the group below the 99th centile threshold and it is likely that if a second mutation were present it would be a milder mutation. The milder mutations are rare in the Victorian population and given the large number of possibilities further mutation analysis is impractical. In the highest two IRT cohorts the frequency of ΔF508 was 6% which is twice the expected frequency from the Victorian population overall, but not high enough to warrant lowering the IRT threshold to test for other mutations or arrange a sweat test. Thus, the 99th centile threshold used in the current screening protocol seems justified.
The results of our study suggest that there is not a clear association between the level of neonatal IRT and the IVS8-5T allele below the 99th centile IRT cut off. The number of IVS8-5T alleles detected was consistent with the 5% reported from other centres,11 15 16 and there was only a weak suggestion of higher numbers at the higher IRT levels. This is in contrast to other studies which have suggested the frequency of IVS8-5T is increased over the 99th centile IRT threshold and, by inference, that IVS8-5T can influence neonatal IRT on its own. If this were true, we would expect to have detected a trend, with an increasing frequency of IVS8-5T with increasing IRT, as we have shown with ΔF508.
In our study, we examined the relationship between IRT and IVS8-5T systematically, while in the other studies, a random and small number of subjects below the 99th centile IRT threshold was chosen and compared to subjects with IRT above the 99th centile IRT threshold. The IVS8-5T allele with either a 7T or 9T on the other allele may reduce the production of exon 9 containing CFTR to 30-40%,6 a level of CFTR activity which has not been thought to cause disease.11 17 There are reports of an increased frequency of IVS8-5T in patients with chronic, idiopathic pancreatitis, but we did not find an effect on neonatal IRT.18 With regard to other cystic fibrosis related diseases, IVS8-5T has only been implicated when in association with exonic mutations.10 19
The finding of two ΔF508 heterozygotes from 22 (12%) on a non-9T background challenges current dogma that ΔF508 is always in cis with 9T. The original studies of IVS8 showed that in homozygous ΔF508 subjects only 9T was present on each allele and that in ΔF508 heterozygous subjects at least one IVS8 allele was 9T. Whether the 9T was in cis or trans with ΔF508 in the heterozygous ΔF508 subjects was not determined but was assumed to be in cis, extrapolating from the ΔF508 homozygote data. Our previous experience has suggested that the ΔF508 mutation on a non-9T background may in fact be ΔI507 which has a similar electrophoretic appearance in the polyacrylamide gels used for newborn screening and is known to be associated with a 7T/5T or 7T/7T background. Gene sequencing clearly identified these subjects as having ΔF508, which has not previously been reported on a non-9T background. This makes it unreliable to phase chromosomes using the IVS8 alleles as has been reported.20 It is possible that the infant with the ΔF508 (or ΔI507) 5T/7T genotype, if male, could have congenital absence of the vas deferens (CAVD), although the penetrance of the IVS8-5T allele is variable,18 and no accurate predictions could be made.
We have shown that ΔF508 alone can affect neonatal immunoreactive trypsinogen, and that this is the explanation for the preponderance of ΔF508 heterozygotes detected by the IRT/DNA newborn screening protocol. The IVS8-5T allele does not appear to influence neonatal pancreatic function but further investigation of its role in CF related disorders is required.
The authors wish to thank Professor Bob Williamson for advice on the design of the study and making available the resources of the Murdoch Institute. Dr Massie was supported by a National Health and Medical Research Council postgraduate scholarship.