Genomics England recently announced its plans for the Generation Study, an NHS-embedded programme seeking to sequence the genomes of 100,000 newborns. One of the main aims of the study is to investigate the use and possibility of screening newborns for several childhood onset rare genetic conditions using genomic technologies (see BioNews 1210). In October they published an initial list of 223 rare genetic conditions, which are caused by genetic changes in around 500 different genes, that they intend to investigate (see BioNews 1213).
How can we improve newborn screening? Is whole genome sequencing (WGS) the only, or even the best way, to do this?
WGS has proven clinical utility in the diagnosis of rare disease. However, assays that perform very well in diagnostic medicine, where the chance of underlying disease is high, often perform poorly in a screening context. This is especially true when the chance of disease is extremely low, as it fortunately is for rare diseases in apparently healthy newborns.
Most screening programmes start by evaluating each condition in detail to decide whether it is suitable for screening. A crucial first question is whether there is evidence that intervention, after identifying the condition presymptomatically, improves outcomes. Next is working out the best assay to use for each condition. However, Genomics England state that their intention is 'Delivering the NHS-embedded Generation Study, to explore the benefits, challenges, and practicalities of sequencing and analysing a newborns' genome'. Thus, the choice to use genomes for this purpose has already been made. So, the study has been designed to explore the value of sequencing genomes in newborns rather than to optimise newborn screening.
There are often alternative tests of better sensitivity and specificity than WGS; screening for the consequence of a genomic variant is usually more informative than making predictions based on sequence.
Phenylketonuria (PKU) is one of the nine rare conditions screened for by the newborn heelprick test on day five of life. It is one of the most treatable of all inborn errors of metabolism, as it can be managed with the right diet. It is more accurately diagnosed using biochemical methods than genomic sequencing. Indeed, sequencing under performs in the detection of PKU to the extent that biochemical testing is needed for detection because the consequences of missing it in the newborn period are so devastating. Biochemical testing will also be required as it is the phenylalanine level, and not the genetic mutation in PAH, which determines the need for treatment.
Similarly, spinal muscular atrophy is one of the more frequent serious genetic disorders diagnosed in the first year of life. It is a degenerative disease of motor neurons, the nerve cells that control voluntary muscles used for movement. Due to faulty SMN1 genes, motor neurons lack a protein called SMN and progressively die, leading to muscle wasting, difficulties with movement, breathing and swallowing. Recently new treatments have become available (see BioNews 1087). WGS is poor at detecting this fault and again, in order not to miss cases, it is likely that a quantitative PCR test will have to be run alongside the WGS test.
Likewise, newborns will continue to have a hearing test rather than trying to infer whether they have deafness based on genomic analysis of more than 100 genes, all of which will have been fully sequenced by the WGS assay. A hearing test is a much more reliable way of diagnosing deafness than trying to make predictions based on the interpretation of genomic variants with many inherent background differences. WGS won't be able to go it alone, targeted testing will remain necessary for accurate diagnosis.
How could we advance newborn screening?
If we want to improve newborn screening in the UK, we need to start by selecting conditions that have both a clear pre-symptomatic phase and an intervention that can be offered to improve the outcome of the disease. The assay that is best suited to diagnosing each of those conditions will then need to be chosen.
If we were planning a scientific study to address this, we would compare WGS with existing newborn screening and with extended newborn screening. Extended newborn screening would include additional diseases assayed by the best, most relevant approaches, which might include targeted testing for well-understood variants and biochemical assays (eg, tandem mass spectroscopy) for additional metabolic diseases. Tandem mass spectroscopy is already widely used in many countries for newborn screening, so there is plenty of evidence to draw on.
Might parental concerns about data privacy deter parents from newborn screening?
Very little of a baby's genome (less than 0.05 percent) will be used for the Generation Study's newborn screening analysis. The remaining 99.95 percent will be used for research and commercial purposes that are not of direct immediate benefit to the baby.
The recent theft of data with details of around one million people, overwhelmingly including people with Ashkenazi Jewish ancestry, as well as many of Chinese descent, was stolen from genetic testing company 23andMe (see BioNews 1211). Similar data breaches may occur in the future and could damage the public's trust in centralised databases holding genomic data.
The heelprick test on day five of life currently has a very high uptake (>99 percent) and enables the rapid diagnosis of several rare but serious and treatable conditions where early diagnosis is key to a good outcome. It would be a very detrimental public health outcome if, in the future, concerns about data privacy deter parents from accepting the offer of newborn sequencing.
Changing the study design would provide more opportunity to advance newborn screening.
The main problem with the Generation Study is not the high cost of the study, £105 million pounds, but that there are no comparator groups. Typical studies addressing such an important question would have a comparator group and ideally be randomised to best ensure valid comparison. The Generation Study may answer the question: 'Can we detect these 223 rare genetic diseases in neonates using WGS?' but it will not tell us whether WGS is the best way of screening for any of them.
It is difficult to see how we will be able to determine the sensitivity and specificity of newborn screening by WGS (parameters which are essential for the evaluation of any screening programme), if we do not include an alternative, validated screening method where it is available.
With the current study design, there is a risk that at the end of the Generation Study, we will not have the evidence required to introduce newborn screening for any of these conditions and we will have missed a big opportunity to advance the health of babies in the UK.
Leave a Reply
You must be logged in to post a comment.