Health and genomics: what’s the score with polygenic scores?
Debate is growing about the use of a genetic/genomic approach called ‘polygenic scores’ to understand health and assess health risks. These scores are different to traditional genetic tests and can be used in relation to a vastly greater number of diseases and conditions. Advocates claim that these scores could revolutionise healthcare, and – in the UK context – help redefine the NHS. Critics retort that polygenic scores are of limited use, and that they are perilously easy to misconstrue.
A century ago, the British statistician and geneticist Ronald Fisher pioneered the idea that an inherited disease or characteristic can be influenced by many genes, each gene having only a small effect: the disease or characteristic is ‘polygenic’. But while our understanding of genetics and DNA advanced massively throughout the twentieth century, it was only practical to develop genetic tests for conditions affected principally by a single gene. By contrast, polygenic scores of the sort developed recently – using whole-genome sequencing, algorithms and genomic data from enormous numbers of people – allow us to look at and ascribe numbers to characteristics influenced by multiple gene variants. For example, an influential study last year produced polygenic scores for breast cancer, coronary heart disease, diabetes and inflammatory bowel disease by combing through a whopping 6.5million gene variants.
This certainly sounds impressive, but what does it mean in practice? The UK health minister, Matt Hancock, said in a speech at the Royal Society this year that he had discovered – via ‘a predictive polygenic risk test’ – that he was at elevated risk of prostate cancer, and that he would therefore seek medical advice and ‘make certain I don’t miss any screening appointments’. Prominent geneticists responded that what Hancock had received was in fact a polygenic score, that he had misconstrued it, that his risk of prostate cancer actually appeared normal, that there are no NHS ‘screening appointments’ for this particular cancer (for which screening is currently impractical), and that in seeking reassurance on the basis of his score he would in fact be wasting NHS time and resources. A cautionary tale, then.
But advocates still maintain that polygenic scores offer the only effective way to approach many common disorders that represent a great disease burden, and even sceptics concede that polygenic scores are useful in some narrow and specific contexts. Regardless of what the NHS does, these scores are now becoming widely available to the public via the private sector – for example, in ‘reports’ on diabetes and obesity risk that are available to buy from the company 23andMe. Also controversial is the fact that polygenic scores are necessarily contextualised by population. Since most of the people whose genomes have been studied so far have been white people of European ancestry, how useful can polygenic scores be for other ethnic groups?
So, what’s the score with polygenic scores? Do they offer useful information for patients and clinicians or could they lead to unnecessary anxiety and pointless medical intervention? Even if they do tell us something important, what are the ethical issues that arise when our future health outcomes can be gleaned from an analysis of our genes?