We report findings from a large-scale prospective study of newborn screening for MCADD conducted within a multi-ethnic population where over 20% of the population is non-white, principally Asian or black. This multicentre collaboration across six centres used common protocols for screening and diagnosis and developed a process by which screen positive cases were independently reviewed to assign MCADD diagnoses and infer their underlying clinical severity.
Approximately 1 in 10,000 screened babies were diagnosed with MCADD, approximately 60 per year in England. In the majority, the biochemical and genetic features were consistent with a high risk of clinical severity. The numbers of children with false-positive or carrier diagnoses were minimal reflecting prospective design features, namely thresholds selected for referral as well as the decision to restrict genetic testing to those referred. High specificity and overall PPV were achieved – a key requirement of screening programmes, demonstrating that the majority of babies who were detected were likely to benefit from screening.
Babies who were assigned with a phenotype of definite clinical importance were predominantly of white ethnic origin, whereas nearly a third of those of uncertain phenotype were Asian. Whilst recognizing that the number of Asian babies was small, these marked differences were reflected in the estimated PPV for phenotypes of definite and uncertain clinical importance amongst those of white and Asian ethnicity. Of particular note were seven babies homozygous for c.946-6T>G (IVS10-6T>G). In our study, this mutation was only found amongst Asian babies and has not previously been reported.
Particular strengths of this study include its large scale, adherence to common protocols and high completeness of data. In addition, we developed a novel classification system allowing clinical severity to be inferred, and appropriate for use within a screening programme in which affected children are, in most cases, asymptomatic and/or without family history. By categorizing babies this way, a marker for measuring clinical validity of the screening programme was established.
A limitation of our study is that clinical significance, outcome and benefit for babies assigned as MCADD of uncertain phenotype is not known, since almost all babies identified through newborn screening are currently treated, irrespective of the likely severity of their disease. Thus, the natural history of novel mutations detected only through screening remains difficult to assess. No false-negative diagnoses were reported from this cohort over a two-year follow-up, but longer-term follow-up is required.
We compare our data, as the largest such study published worldwide, with recent, international studies which have reported on the incidence, rate of MCADD diagnoses and number of c.985A>G homozygotes detected through newborn screening, summarized in Table , updating recently published reports.2,12,48
Estimated incidence of MCADD through newborn screening
Early neonatal presentation and death due to MCADD, typically occurring around two to three days of age, has been reported, but is uncommon and unlikely to be prevented by newborn screening, given the timing of bloodspot sample, analysis and reporting.7,9,18,49
In the UK, bloodspot samples for newborn screening are taken between five to eight days of age, later than in many countries.16,24,25,41
However, even within screening progammes which sample blood earlier, typically between 48–72 hours, preventative action would be unlikely as the screening result is reported later. We therefore found no evidence in our study to suggest that a change in UK screening policy to earlier bloodspot sampling would prevent decompensation or death in this early neonatal period.
The C8 cut-off set within England is lower than in many countries which screen at an earlier age and we acknowledge that the chosen cut-off must be appropriate for the population in question and the age at which screening takes place. We have recently examined C8 concentrations in the first two weeks of life46
in an analysis of newborn screening data for 227,098 infants from England and New South Wales, Australia (average postnatal age at testing of 5 and 3 days respectively) in which we demonstrated that median C8 concentrations do not vary significantly by age of sampling in unaffected babies between day three and day 14. Thus the cut-off in our study would be relevant to samples collected over this age range.
MCADD dietary guidelines and other information for parents and professionals, developed specifically through this study, have worked well and are freely available for parents and health-care professionals, along with current UK Standards and Guidelines for Newborn Bloodspot Screening from the UK Newborn Screening Programme Centre, at www.newbornbloodspot.screening.nhs.uk
For most babies, assignment of the diagnostic category through the independent review process was straightforward. However, in a small number this proved difficult. In two babies, both ultimately assigned as carriers, a second mutation (in conjunction with c.985A>G) of unknown clinical significance (c.387 + 40G>A [or IVS5 + 40G>A] and c.388-9C>T [or IVS5-9C>T]) was identified but no biochemical abnormality was detected at diagnosis, including one with normal FAO studies. Furthermore, five babies, also assigned as carriers (simple c.985A > G heterozygotes), showed elevated diagnostic C8 (range 0.54–1.17 µmol/L) despite having no second mutation identified at EMS and no other biochemical abnormality detected including one with normal FAO studies. These issues illustrate the difficulties faced by paediatricians when deciding whether to treat a child with borderline results or only some of the biomarkers associated with MCADD. Whilst completely independent, we are not aware of any discrepancies between assigned classification and the final clinical management of the child.
Nine babies were inappropriately referred with an average C8 result between 0.4 and 0.5 µmol/L. Diagnostic testing revealed three carriers, four ‘Not MCADD’ and two assigned as MCADD of uncertain phenotype. Of the latter two, one had no biochemical abnormalities detected at diagnostic testing and was found to be heteroallelic for c.985A>G and c.127G>A. Previous studies note that this genotype does not express the biomarkers normally associated with MCADD.5,40,41,43
The second case was found to be heteroallelic for c.985A>G and c.199T>C, and, although both C8 results fell below the 0.5 µmol/L cut-off, some organic acid abnormalities were detected. We acknowledge that by strict adherence to an agreed C8 cut off, there may occasionally be cases of (mild) MCADD which are not detected through screening. However, screening is not diagnosis and, by reviewing and reporting these findings, we are able to conclude that at least seven of these nine referrals were made unnecessarily.
Very few carriers were detected in relation to expected total carrier pool. Carrier frequency for c.985A>G in the white UK population has been estimated at 1/65 (95% CI 1/71–1/61)31
assuming Hardy Weinberg equilibrium. Within our screened study population (n
= 1,568,445), an estimated 79.5% or 1,246,914 were white, of which 19,183 (or 1.5%) would be anticipated c.985A>G carriers.31
In total, 23 white, c.985A>G heterozygote carriers were detected over the four-year study, 0.1% of the expected frequency. Moreover, we expect that the number of carriers identified through newborn screening for MCADD to be further reduced, as of April 2010, by the introduction within the screening protocol of an additional C8/C10 ratio measurement with a referral cut off ≥1.0. Use of the C8:C10 ratio has been shown to discriminate between MCADD phenotypes well.41,50–52
Screen positive MCADD referrals in England will now only be made if C8 ≥ 0.5 µmol/L and C8/C10 ≥ 1.0. We have estimated through a specific research audit within this study (results not presented here), that the detection of carriers and those assigned as ‘Not MCADD’ will be reduced by over 90%, minimizing unnecessary referral and anxiety for families and increasing the predicted PPV at screening for all MCADD phenotypes to over 95%.
Our findings demonstrate that prospectively defined, quality assured screening and diagnostic protocols allow identification of children with clinically important disease whilst minimizing the harms of screening related to the detection of children with biochemical or genetic variations of uncertain prognostic significance.
In 2006, interim data from this study were reviewed by the UK National Screening Committee and, as a result, MCADD screening was successfully implemented across England by April 2009.