Population studies are key to safe newborn genome screening

The world of newborn screening is quietly undergoing a revolution.

Newborn screening (NBS) first began in 1960, when a dried bloodspot test was developed to detect phenylketonuria, a rare autosomal recessive condition present in around one in 10,000 babies. Early detection enables life-altering treatment that can prevent irreversible brain damage. Over the following decades, numerous other conditions have been added to NBS programmes around the world using the same dried bloodspot.

In the UK, ten different conditions are currently included in the population NBS programme, each of which has undergone extensive individual evaluation by the National Screening Committee to ensure that the benefits of screening outweigh the harms. The conditions included vary between countries, depending partly on their healthcare systems, with an average of 19 conditions included across Europe. Most conditions are rare, monogenic, and have pre-symptomatic treatments that radically improve outcomes.

Although new technologies have improved the balance between sensitivity and specificity, the basic process of selecting and evaluating conditions individually has not changed. Until now.

The advent of affordable whole genome sequencing (WGS) – which determines the sequence of all three billion bases of DNA in an individual, including around five million variants compared with the reference human genome – potentially allows every known genetic condition to be screened for simultaneously. Given there are now more than 4700 single-gene disorders, deciding which conditions to include in NBS is an enormous and daunting task.

Pilot studies, such as the UK's Generation Study, have developed guiding principles to help determine which conditions to include (see BioNews 1210). Almost every study includes penetrance – the proportion of individuals with disease-associated genetic variants who develop the condition – and generally requires high penetrance in childhood for a condition to be considered for screening. However, our research, alongside that of others, shows that incomplete penetrance is common across a range of conditions.

A recent special issue of the European Journal of Human Genetics on 'DNA in public health screening programmes' includes six papers from our group and more than 15 papers overall addressing questions about genomic NBS. Three of our papers investigate different autosomal dominant conditions (tuberous sclerosis complex, retinoblastoma and osteogenesis imperfecta), which are primarily caused by pathogenic loss-of-function variants across five different genes.

Unlike most previous studies of these conditions, which focused on patients and affected families, we used data from two large population cohorts (UK Biobank and All of Us) encompassing WGS data from around one million adults. The key novelty of our approach lies in the starting point: clinical studies begin with patients or families with symptoms and then attempt to make a genetic diagnosis (a 'phenotype-first' approach); population studies begin with genetic variants and then attempt to determine whether individuals have a specific condition (a 'genotype-first' approach).

The former generally produces high estimates of penetrance, which are important for counselling patients in genetics clinics but don't necessarily apply to a screening scenario, where most individuals tested don't have – and won't develop – the condition. In our studies, although all three conditions were considered to have high penetrance clinically, we found incomplete penetrance in population cohorts. For example, in some cases, more than half of adults identified with a pathogenic variant had no recorded symptoms of the condition. If identified through screening, without appropriate validation, incompletely penetrant variants could result in substantial over-diagnosis and over-treatment. Similar findings have been reported by other groups, underlining the importance of population-based studies when evaluating conditions for inclusion in genomic screening programmes.

We are also investigating penetrance in several autosomal recessive conditions, in which pathogenic variants must be present in both copies of a gene to cause a condition. So far, the picture looks rather different for autosomal dominant conditions, and the potential for overdiagnosis can be substantially reduced by careful variant curation. Several conditions, including spinal muscular atrophy, severe combined immunodeficiency syndrome and primary ciliary dyskinesia, appear to have high penetrance in UK Biobank. However, this depends on applying appropriate variant-detection algorithms and quality-control procedures to ensure that individuals carry true pathogenic variants in both copies of the relevant gene.

Beyond detailed gene-disease associations, we also raised the idea of revisiting penetrance in an era of genomic screening and considering what counts as a relevant phenotype. Is the outcome a disease diagnosis, a related phenotype or a quantifiable biomarker? These seemingly minor distinctions can make a huge difference to penetrance estimates and could ultimately affect how the success of genomic screening programmes is assessed.

Another important part of our work explores the ethical aspects of genomic NBS, including public and patient attitudes. This is an enormous topic in itself, to which we cannot hope to do justice in this short article. We performed a systemic mapping of stakeholder perspectives to understand which perspectives have been represented in the literature and to identify gaps. We observed a somewhat limited range of perspectives, with mothers and people with prior experience of a genetic condition in the family relatively frequently represented.

In light of our findings on incomplete penetrance, we also identified a worrying trend: participants often appeared to assume that genomic findings offered a high degree of predictive certainty. Understanding and communicating uncertainty, both in genomics and in screening more generally, is therefore crucial to decision-making around genomic NBS and maintaining public confidence.

Genomic NBS pilot projects continue apace across the globe, and have extraordinary potential to improve the early detection and treatment of many genetic conditions. However, context matters. Large-scale population studies are needed both to reduce risks of newborn genome screening and to ensure that we design the best possible screening programmes using the technologies at our disposal.

Funding: Project grant from the Medical Research Council [MR/X021351/1] to investigate scientific and ethical approaches to newborn genome screening.