1. Dr Claire Faulkner Trainee Project
Cystic Fibrosis Newborn Screening in the Bristol Genetics Laboratory: Development & One Year’s Experience
Dr Claire Faulkner
Trainee Project

2. CF Newborn Screening
National Protocol designed to pick up 95% CF cases
Enables early diagnosis and treatment
Multi-stage protocol involving IRT and DNA analysis on bloodspot
IRT (Immunoreactive trypsinogen)
- serum levels elevated in newborns with CF
- normal passage of trypsinogen from pancreas to duodenum blocked
- not 100% specific
- samples raised IRT >99.5th centile (70ng/L) referred to DNA
DNA analysis
- CF4: p.Phe508del, p.Gly551Asp, p.Gly542X, c.489+1G>T
80% coverage
- CF29: 90% coverage
The CF newborn screening national protocol is designed to pick up at least 95% of CF cases enabling early diagnosis and treatment in these babies.
This is a multi-stage protocol involving IRT and DNA analysis on a blood spot taken from the baby at around 5 days old.
Immunoreactive trypsinogen levels are tested by the Biochemistry department since serum IRT levels are raised in newborns with CF. This is because the normal passage of trypsinogen from the pancreas to the duodenum is blocked, leading to leakage of elevated trypsinogen levels in the blood.
This assay is not 100% specific so if the IRT is raised above the cutoff set at the 99.5th centile, which is 70ng/L, the blood spot is referred for DNA analysis.
DNA analysis is a 2-step process. For all samples, an assay for the 4 most common severe CFTR mutations is undertaken. This covers 80% of mutations in our local population.
If two mutations are detected the baby is referred with a diagnosis of CF. If only one mutation is detected then the sample is tested for the 29 most common CFTR mutations covering 90% of mutations in the population.

3. Aims of Project
To set up and run Molecular CF Newborn Screening Service for the South West
Develop and validate a method to extract DNA from bloodspots
Optimise and validate 4 mutation assay using real-time PCR
- quick (setup-run-analysis in 3hrs)
- cost-effective (£1 per mutation per sample)
- equipment in-house
- validated by Cardiff laboratory
Write protocols, assess H&S, train technicians, liaise with Biochemistry, day-to-day management of service
Go-live Jan 07
Compile audit data to monitor and improve service
The aim of my project was to set up and run molecular aspects of CF newborn screening service for the south west
In order to do this, I had to:
The choice of method used to test for the 4 mutations is up to each individual laboratory. The Bristol laboratory decided to use a real-time PCR method since this is quick, cost-effective and had previously validated by the Cardiff Genetics laboratory

4. DNA Extraction from Bloodspots
Chelex 100 resin
EZ1 robot
DNA quality
Poor
260/280 ~1.6
Excellent
260/280 ~1.8
Performance on real-time PCR
Poor – differs to control DNA
Good – comparable to controls
Performance on CF29
Requires dilution
> 1/10
Good
Reagent cost
(per sample)
£0.14
£4.97
Staff time
(per batch of 6)
3 hrs
0.5hrs
Total cost
£5.09
£5.77
The first method I optimised for extracting DNA from bloodspots involved using Chelex resin.
Chelex = ion exchange resin
Binds polyvalent metal ions that may breakdown DNA and inhibit PCR.
DNA was of poor quality and purity so behaved differently to control DNA from our DNA bank on real-time assay. This meant that bloodspot extracted DNA would have to be used as controls for real-time and we didn’t have bloodspots available for all the controls needed. Also required dilution to work optimally on CF29.
I then tried the automated EZ1 system which uses magnetic bead technology to capture DNA and purify it. This resulted in excellent quality DNA which performed well on both real-time PCR and CF29. The cost is more expensive per sample but savings are made on staff time compared to the Chelex method.
The EZ1 System is now in routine use for bloodspot extractions.

5. Example Real-Time PCR Results
TaqMan allelic discrimination (AD) for p.Phe508del
Mut/Mut
N/Mut
N/N
Negative
controls
This is an example of the allelic discrimination plot obtained for Phe508del using the real-time TaqMan assay. This is an end-point assay plotting the overall change in normal fluorescence against the overall change in mutant fluorescence for each sample.

6. Real-Time PCR Validation
86 stored DNA samples
34 anti-coagulated bloodspot samples
Blind trial on 8 stored DNA samples
Blind trial 16 archived coagulated newborn screening bloodspots
88 +ve controls
All samples (136 in duplicate) correctly genotyped except:
i. p.Gly551Asp/p.Arg553X sample called as a p.Gly551Asp homozygote
ii. p.Phe508del homozygous archived bloodspot called as a p.Phe508del homozygote and a c.489+1G>T heterozygote
This included 88 positive control samples
All 136 samples tested in duplicate were correctly genotype except for 2 problem samples:
G551D/R553X sample called as G551D homozygous
F508del homozygous bloodspot correctly called as F508del homozygote but also called as a 489 heterozygote.

7. p.Gly551Asp/p.Arg553X p.Gly551Asp/p.Arg553X amplification curve
cycle no.
change in fluorescence
Blue = mutant
Pink = normal
N/N
N/p.Gly551Asp
p.Gly551Asp/
p.Gly551Asp
N/p.Arg553X
Allelic Discrimination plot shows a p.Gly551Asp/p.Arg553X sample in duplicate called as a p.Gly551Asp homozygote.
Amplification curve shown
Representative amplification curves are shown for each genotype (pink curve = normal, blue curve = mutant). Although called as homozygote on allelic discrimination, the amplification curve for the p.Gly551Asp/p.Arg553X sample is distinct from a true homozygote which was later obtained via EAQ scheme! . p.Arg553X heterozygous samples are correctly genotyped as normal on the p.Gly551Asp assay.

8. p.Gly551Asp/p.Arg553X
Probe incorporates the p.Arg553 site, presence of a mutation at this site likely to decrease binding of the normal probe
p.Arg553X
p.Arg553X N
Mut
p.Gly551Asp
Any apparent p.Gly551Asp homozygote verified with another assay e.g. CF29
The probe does incorporate the Arg553 site so it is likely that the presence of a mutation at this site interferes with binding of the normal probe. This is not a problem for Arg553X heterozygous samples as there is also one normal allele present but becomes apparent in G551D/R553X compound heterozygous samples.
Following this it was decided that amplification plots should be checked for all samples and any sample called as an apparent G551D homozygote should be verified with an additional assay e.g. CF29.

9. p.Phe508del homozygote also positive for c.489+1G>T
Abnormal sample amplification curve
cycle no.
change in fluorescence
Orange = normal
Purple = mutant
N/N
N/c.489+1G>T
Abnormal sample
Normal bloodspot
Sample was correctly identified as a p.Phe508del homozygote but was also called as a c.489+1G>T heterozygote - circled on allelic discrimination plot.
The amplification curve for this sample showed an abnormal shape compared to the Representative amplification curves shown here (orange = normal, purple = mutant).
Quantification of this sample using the Nanodrop showed a large contamination peak at 240nm. A repeat extraction was performed on this sample and subsequent testing gave the correct genotype.

10. Real-Time PCR Thresholds
Both AD and amplification curves should be checked for each sample
Ensures contamination or unusual amplification patterns are detected
Introduced threshold levels as a non-subjective method to check amplification curves
N/N
N/p.Gly551Asp
p.Gly551Asp/p.Gly551Asp
The validation work highlighted the importance of checking both AD and amplification plots for each sample to ensure genotypes are concordant.
This also ensures that any contamination or unusual amplification patterns are detected.
Threshold level shown for p.Gly551Asp assay = green horizontal line. Pink curve = normal, blue curve = mutant. If the curve crosses the threshold then the sample is considered positive for that probe. Threshold levels are consistent for each batch of assay and have been tested at a range of DNA concentrations.
For p.Gly551Asp the same threshold level can be applied to both normal and mutant probes. However for other assays, normal and mutant probes may require different thresholds as we have found that the normal probe shows increased fluorescence.
Blue = mutant
Pink = normal

11. One year Audit Observed Expected Babies screened 40,421 IRT >99.5th
302
202
Over 40,000 babies were screened from Jan 07 to Jan 08. Of these 302 had raised IRT and were referred for molecular testing. This referral rate is approx. 50% higher than expected from the national protocol.

12. One year Audit Observed Expected Babies screened 40,421 IRT >99.5th
302
202
2 mutations 17 14
2 mutations on CF4 9 12
p.Phe508del/p.Phe508del 8
p.Phe508del/p.Gly551Asp 1
1 mutation on CF4, 2nd CF29 2
p.Phe508del/p.Arg117His 5
p.Phe508del/p.Asp1152His
p.Phe508del/p.Arg553X
p.Phe508del/c G>A
Of the 302 babies tested, 17 had 2 CFTR mutations.
In 9 of these both mutations were picked up on the CF4 real-time assay
1n 8 samples one mutation was picked up on CF4 and the second on CF29. This number of babies with two mutations is probably greater than the expected value due to 6 babies being picked up with mild or variable mutations on CF29, including 5 with the R117H mutation making this the second most common mutation after F508del in our population.

13. One year Audit Observed Expected Babies screened 40,421 IRT >99.5th
302
202
2 mutations 17 14
1 mutation 13 22
Normal 2nd IRT 12 20
N/p.Phe508del
Raised 2nd IRT 1 2
p.Phe508del/p.Arg352Gln
13 babies were identified with one mutation and all were carriers of F508del. This is much less than the expected value.
A second IRT sample is requested for these babies to help differentiate carriers from babies with a second unidentified mutation.
One Baby had a second raised IRT and went on to receive a clinical diagnosis of CF following an equivocal sweat test and low faecal elastase demonstrating pancreatic insufficiency. The second mutation was identified in Exeter following full gene sequencing.

14. One year Audit Observed Expected Babies screened 40,421 IRT >99.5th
302
202
2 mutations 17 14
1 mutation 13 22
No mutations
272
165
IRT >99.9th 34 12
Normal 2nd IRT 26
No 2nd sample 2 -
Raised 2nd IRT 6*
<1
For the remainder no mutations were detected.
Repeat IRT samples were requested for 34 babies as these had a very high initial IRT above the 99.9th centile. This step is included in the protocol to detect babies with rare mutations not in the screening panel such as babies from different ethnic populations.
Of these, 6 babies had a significantly raised second IRT. Majority were non British but had congenital abnormalities which can be associated with raised IRT levels and so are unlikely to have CF.
* 5/6 non-British: Pakistani, African, Bangladeshi, mixed, other
5/6 congenital abnormalities, 1/6 hypoxia at birth

15. Case Study: p.Phe508del/p.Arg117His
Uncertain outcome: PS-CF / late-onset CF / CBAVD / asymptomatic
PolyT testing as reflex for p.Arg117His?
Report issued with interpretative comments and recommending further molecular testing following genetic counselling
PolyT testing later requested: 9T/7T
Clinical follow-up:
- PS, normal sweat test
- respiratory pathogens isolated, antibiotics
- difficult to counsel parents
- consultant paediatrician will continue to monitor clinically but label him as atypical or non-classical CF rather than CF
We have picked up 6 babies with mild or variable CFTR mutations and I thought I would briefly describe one case of a baby boy with the genotype F508del/R117H
The R117H mutation is associated with an uncertain outcome ranging from pancreatic sufficient CF to asymptomatic
The phenotype at least in part depends on the length of the polyT tract in intron 8 which affects splicing of the RNA. 5T cases likely have PS phenotype whereas 7T cases generally have a milder phenotype.
For diagnostic cases identified with this mutation we would undertake polyT testing as a reflex test. However there are no guidelines as to whether we should be doing this for newborn screening cases and the predictive value of this test is not particulary clear-cut.
There was a lot of discussion about how to report these cases and it was decided that the best approach was to issue a report with interpretative comments and to recommend further molecular testing following genetic counselling.
PolyT testing was later requested for this baby and has been requested for all cases
Clinical follow-up of this baby has shown that he is pancreatic sufficient and has a normal sweat test.
However respiratory pathogens have been isolated from cough swabs and he has been given antibiotics. Studies have shown that pulmonary symptoms can occur at a young age in children with the 7T association may benefit from prophylactic antibiotic therapy.
Difficulty in counselling parents due to uncertain outcome
Consultant paediatrician will closely monitor baby for signs of disease but label him as atypical or non-classical CF rather than CF

16. Summary
Real-time PCR is an effective method for mutation detection and an excellent tool for CF Newborn Screening
Essential that both AD and amplification curves checked for each sample, using threshold levels
In one year of running service, over 40,000 cases screened and 17 babies with two CFTR mutations detected
It is important that babies with mild or variable mutations (e.g. p.Arg117His) that are well are not labelled with a diagnosis of CF but are closely monitored for signs of disease

17. Thank you! Molecular Maggie Williams Hilary Sawyer Anne Gardner
Mark Greenslade
Thais Simmons
Rose Salamanca
Jean Worgan
Jenny Coles
Biochemistry
Helena Kemp
Anny Brown
Mark deHora
Cardiff Laboratory
Sarah Maund
Linda Meredith