March 20, 2019

expert reaction to Matt Hancock’s speech on genetic testing

Reactions to Matt Hancock’s speech on genetic testing

Prof Nicholas Wald, Professor of Environmental and Preventative Medicine, Queen Mary University of London (QMUL), said:

“Whole genome analysis has considerable potential in medical practice particularly in diagnosis. Its role in screening and the prediction of disease, however, is not established. Demonstrating associations between genetic variants and disease is not enough to show whether screening, based on these genetic variants, is worthwhile. The contribution of a person’s risk of a disease arising from a combination of many genetic variants is relatively small, even if the associations with the disease are large from a causal perspective. Knowing that a person has a 50% higher risk of a particular disorder or a 30% lower risk is not sufficient to assess its value as a predictive test, ie its value in medical screening. This is determined by a test’s discriminatory efficiency and needs to be judged in combination with other criteria, such as the availability of an effective and acceptable remedy. Polygenic risk scores are poor predictors of disease and it is, therefore, misleading to refer to them as “predictive polygenic risk tests”. Most affected cases occur in people without an elevated risk score (ie missed) and are potentially denied the opportunity to access preventive action while many unaffected people have an elevated risk score (ie are false-positive) and are needlessly worried.

Robert Old and I recently wrote a Commentary on this in Genetics in Medicinetitled “The illusion of polygenic disease risk prediction”.¹ The application of whole genome analysis in screening and disease prediction needs further thought and careful consideration before being applied in routine practice.“

1​Wald NJ, Old R. The illusion of polygenic disease risk prediction. Genet Med 2019; https://doi.org/10.1038/s41436-018-0418-5

Prof Ewan Birney, Director, European Bioinformatics Institute (EMBL-EBI), said:

“Matt Hancock talked at the Royal Society on the use of his own polygenic risk score for prostrate cancer, indicating the difference from rare disease genetics, which often has a clear cut diagnosis to this more complex change in risk which, for example, might alter his screening profile for prostate cancer. He also emphasised the need to understand this information with the help of a trained medical professional.

“Using genetic risk scores to inform public health screening strategies and provide more detailed information to patients has been an area of active research for over five years worldwide. This research is now close to the first health practice trials, but many aspects need to be tested carefully. These include the response by individuals, the practical operation of health professionals and integration into standard risk screening. The UK, along with other countries, eg. Finland, is in the leading group of countries in exploring this. A key issue is how to handle the diverse cosmopolitan population in the UK with many individuals having complex recent genetic ancestry. The importance of equitable access to health care from the NHS means that these complex ancestries have to be tackled; this is challenging but feasible. The roll out of responsible use of genetic risk scores require careful integration from genetic research to health care practice. The UK is a world leader in this area.”

Prof David FitzPatrick, MRC Human Genetics Unit, University of Edinburgh, said:

“The predictive power of the genome is evidenced by how similar identical twins usually are in appearance and physiology.  Excepting genetic variants of the highest impact which cause rare and ultra rare genetic disease our ability to recognise the elements of the genome that are predictive is strictly limited. Even more difficult is knowledge of how these can be modified by other genomic variants.” 

Professor Diana Eccles, Dean of Medicine and Professor of Cancer Genetics, University of Southampton, said:

“Firstly 15% risk of prostate cancer compared to a UK population lifetime prostate cancer risk of 18% for men born after 1960 (CRUK data) sounds like a pretty modest risk. There are many additional risk factors for prostate cancer including obesity – deprivation is not a risk factor for prostate cancer diagnosis (similar to breast cancer) but the high risk genetic factors for prostate cancer are very uncommon.

“Prostate cancer is not one disease – many are indolent and early detection means knowing about for longer when it may never impact on life expectancy. Unless the intervention actually prolong life (unproven for current approaches to prostate cancer for early detection).

“More research is needed. Risk stratified screening is a really good idea but only if it works – we can’t just assume it will.”

Professor Anneke Lucassen, Professor of Clinical Genetics, University of Southampton and Chair of the British Society for Genetic Medicine, said:

“I welcome Secretary of State support for genomics and his thoughtful consideration of the ethical issues it can raise.

“I think it is unfortunate that his polygenic risk score has been portrayed as a strongly predictive genetic test. The claim that his life has been saved by this test is somewhat over egging the pudding. The Department of Health and Social Care is polling whether he has a duty to disclose this result to his siblings which begs the Q what is there to disclose? Polygenic risk scores are not inherited en bloc and, as my colleague Professor Diana Eccles points out, his lifetime risk is barely increased from the background population risk.”

Professor Peter Donnelly FRS, Founder and CEO, Genomics plc, said:

 “Genomic data and analysis has the potential to transform healthcare by identifying people most at risk of certain diseases, allowing for targeted prevention measures and earlier detection and treatment.

“Through the 100,000 Genomes Project, the NHS has led the world in use of genetics for rare diseases and cancer. The next step is to harness the power of genomic analysis, through the use of polygenic risk scores, to more effectively manage common diseases. Doing so can save lives and money by making health systems more proactive and efficient.

“The nature of genetic data means that the population the data comes from can be important. With current approaches, polygenic risk scores tend to be most accurate when they are applied to the same population as the individuals in the genetic study from which they were derived, typically those of European ancestry. A major focus at Genomics plc is to assess the methods carefully in people with ancestry from other parts of the world, and where necessary to improve the methods, and to take advantage of genetic studies underway in different ancestry groups, to produce tools that are equally effective for everyone.”

Vivienne Parry, Head of Engagement Genomics England, said:

“Our current focus is those with rare disease or cancer and we are only just beginning to think about risk scoring for healthy people which will need a great deal of work before it is ready for widespread NHS use.  But 25% of our rare disease participants are from ethnic minority backgrounds so any work we do will be based on a representative sample.”

*Please note Prof Curtis’ comment is in response to an article in the Times (print) today*

Professor David Curtis, UCL Genetics Institute & Barts and the London School of Medicine and Dentistry, said:

“The quotes from Matt Hancock provide a really striking illustration of some of the things that can go wrong with providing people with polygenic scores which are supposed to indicate risk. It is obvious that he has massively misinterpreted the meaning of the score he has been given. As a result of his misunderstanding, he first suffered unnecessary anxiety and then took up valuable medical time having to be counselled about the reality of the situation. Now he is going to waste even more of the NHS’s scarce resources by booking a completely unnecessary appointment with his GP to discuss a course of action to address a problem which essentially does not exist.

“As a health secretary, he displays a quite astonishing level of ignorance about the NHS. He says he is going to make certain that he does not miss any screening appointments. That should be easy, because there is no such thing as a screening appointment for prostate cancer. We don’t do them because they don’t work, they’re a waste of time and money, they cause unnecessary anxiety to patients and unnecessary work for health professionals.

“His claim that this test may have saved his life is frankly ridiculous. It really demonstrates an astonishing and worrying degree of innumeracy and lack of comprehension of health issues. The genetic test said his risk was 15% and the average risk is 11%. For all practical purposes these are basically the same number – we wouldn’t see any difference between an 11% risk and a 15% risk. 15% is not high risk. 50% would be high risk. 15% is average risk. And bear in mind also that some of these commercial tests have very doubtful validity. There’s no regulation of them. And even the ones which are valid are not necessarily particularly accurate so that the difference between 11% and 15% may be no more than the margin of error in the estimate. The claim about the test possibly having saved his life also overlooks the fact that far more people get prostate cancer than actually die of it. When the disease is local the cure rate is about 100%. And even when people are left with the disease it may just be something that they live with for a long time while they actually end up dying from something completely different.

“We have a major problem in the UK that there is no regulation at all of these tests and basically any company can tell anybody anything they like about their risk of an illness using the flimsiest data or none. There really needs to be a review of the regulation around this.

“Professor Stokes-Lampard is absolutely correct in her criticisms of these tests. I fully support her comments that the tests can pick up things which are unimportant or dubious and that they can cause unnecessary distress, as well as causing extra unnecessary work for already over-burdened GPs.

“Matt Hancock has provided a really bad example of how genetic testing issues should be dealt with. In fact, this account provides the perfect argument for why testing based on polygenic risk scores should not be introduced into clinical practice.”

Professor David Curtis, UCL Genetics Institute & Barts and the London School of Medicine and Dentistry, said:

“The quotes from Matt Hancock provide a really striking illustration of some of the things that can go wrong with providing people with polygenic scores which are supposed to indicate risk. It is obvious that he has massively misinterpreted the meaning of the score he has been given. As a result of his misunderstanding, he first suffered unnecessary anxiety and then took up valuable medical time having to be counselled about the reality of the situation. Now he is going to waste even more of the NHS’s scarce resources by booking a completely unnecessary appointment with his GP to discuss a “course of action” to address a problem which essentially does not exist.

“As a health secretary, he displays a quite astonishing level of ignorance about the NHS. He says he is going to make certain that he does not miss any screening appointments. That should be easy, because there is no such thing as a screening appointment for prostate cancer. We don’t do them because they don’t work, they’re a waste of time and money, they cause unnecessary anxiety to patients and unnecessary work for health professionals.

“His claim that this test may have saved his life is frankly ridiculous. It really demonstrates an astonishing and worrying degree of innumeracy and lack of comprehension of health issues. The genetic test said his risk was 15% and the average risk is 11%. For all practical purposes these are basically the same number – we wouldn’t see any difference between an 11% risk and a 15% risk. 15% is not “high risk”. 50% would be high risk. 15% is average risk. And bear in mind also that some of these commercial tests have very doubtful validity. There’s no regulation of them. And even the ones which are valid are not necessarily particularly accurate so that the difference between 11% and 15% may be no more than the margin of error in the estimate. The claim about the test possibly having saved his life also overlooks the fact that far more people get prostate cancer than actually die of it. When the disease is local the cure rate is about 100%. And even when people are left with the disease it may just be something that they live with for a long time while they actually end up dying from something completely different.

“We have a major problem in the UK that there is no regulation at all of these tests and basically any company can tell anybody anything they like about their risk of an illness using the flimsiest data or none. There really needs to be a review of the regulation around this.

“Professor Stokes-Lampard is absolutely correct in her criticisms of these tests. I fully support her comments that the tests can pick up things which are unimportant or dubious and that they can cause unnecessary distress, as well as causing extra unnecessary work for already over-burdened GPs.

“Matt Hancock has provided a really bad example of how genetic testing issues should be dealt with. In fact, this account provides the perfect argument for why testing based on polygenic risk scores should not be introduced into clinical practice.”

Declared interests

Prof Ewan Birney: “I am a non executive director of genomics England.”

Professor Peter Donnelly: Genomics plc is an Oxford University spin-out which develops predictive tests, or polygenic risk scores, of the kind discussed in Matt Hancock’s speech; Peter Donnelly is former director of the Wellcome Centre for Human Genetics.

None others received.