Testing a population for multiple genes at once will always be limited in its ability to accurately predict disease risk, experts are warning.
An analysis published in the British Medical Journal concludes that population-wide ‘polygenic scoring’ is inherently limited because many cases of disease occur among people who do not have high polygenic scores and many of those with high polygenic scores do not develop disease.
Polygenic scores also do not take into account the major environmental and behavioural factors that contribute to disease risk.
Researchers at The Institute of Cancer Research, London, the University of Oxford and UCL suggest that polygenic scores are only likely to modestly improve risk prediction for common diseases such as cancer.
They suggest enthusiasm for polygenic scoring should not overshadow efforts to address modifiable risk factors.
NHS genetic testing currently focuses on a small number of well-understood disease mutations which strongly raise the risk of disease – such as BRCA1 and BRCA2 in breast and ovarian cancer.
This paper discusses a different type of genetic test called polygenic scoring, which instead looks across thousands of variants in a person’s DNA, each of which will individually have only a small impact on risk, to gain a collective assessment of the genetic risk of disease.
Two recent Government reports show marked enthusiasm for polygenic scores in health care and assert that they offer a ‘step change’ in screening for disease.
The NHS is now partnering with the UK’s largest research programme, Our Future Health, to offer risk information based on polygenic scores to five million people in the UK population.
Such information is expected to inform clinical decision making including access to screening.
The authors argue that the risks and benefits of genetic testing must be carefully evaluated in large studies.
Their analysis concludes that initiatives which use polygenic scores to screen for people at risk of common illnesses such as cancer and heart disease will miss the majority of cases in the population.
When screening the population for a disease using polygenic scores, many have advocated for using a high polygenic score to target screening or preventative strategies for disease.
According to the researchers, doing so would miss most of the disease in the population and result in a sizeable number of healthy individuals undergoing invasive tests and therapies who will never go on to develop the disease.
In another example, NICE uses a 17 per cent lifetime risk of breast cancer as the threshold for deeming women at ‘moderate risk’.
But in a polygenic risk score study which also used this threshold, only 39 per cent of women who will go on to develop breast cancer have moderate or high-risk scores, meaning the majority of breast cancer cases are missed using these scores.
Meanwhile, 22 per cent of women who will not develop breast cancer will have a high polygenic risk score and will therefore be a ‘false positive’.
The scientists argue that since polygenic risk scores only indicate a small proportion of overall disease risk, that they should be presented in context, with the normal background disease risk communicated for comparison to patients and clinicians.
For example, women in the top five per cent of polygenic scores for breast cancer have a 19 per cent lifetime risk of developing the disease, compared with a population risk of 11.8 per cent.
For prostate cancer, those with a polygenic risk score in the top 5 per cent carry a 22 per cent risk compared with a 12.7 per cent population risk.
The differences become even more modest in less common cancers.
For example, people in the top five per cent of polygenic scores for ovarian cancer have a lifetime risk of 2.1 per cent, compared with a population risk of 1.6 per cent.
The authors outline various additional concerns with widespread use of polygenic scores to inform clinical decision making – including the cost and complexity of screening and diagnostic pathways, and the risk that they will expose large numbers of people with ‘high’ polygenic scores to often unnecessary follow-up tests.
Lead author Dr Amit Sud, Academic Clinical Lecturer in Genetics and Epidemiology at The Institute of Cancer Research, London, said: “There is a huge amount of enthusiasm about polygenic scores, and they do have the potential to improve our ability to predict who will or will not develop a disease, albeit rather modestly.
“But we argue that the benefits and harms surrounding the use of polygenic scores are carefully evaluated before they are widely implemented. Given the majority of disease in a population occurs in people who are not at high polygenic risk, these scores should not detract from effective population-wide screening and interventions to address modifiable and impactful risk factors like smoking and socioeconomic deprivation.”
Professor Anneke Lucassen, Professor of Genomic Medicine at the University of Oxford’s Nuffield Department of Medicine, said: “Polygenic scores offer really important insights in research, but using them in screening programmes – or clinical care – is often less predictive than people might expect. Our research is a reminder that polygenic scores only measure a small proportion of overall disease risk, and should not distract from efforts to address modifiable risk factors for diseases.”