Written Evidence Submitted by Genomics plc

(COG0003)

The importance of genomics in delivering a healthier nation and a future-fit healthcare system

Supplement to April 2019 submission (Annex A)

About Genomics plc

1. Genomics plc is a for-profit company spun out of the University of Oxford in 2014.  It is led by two of the world’s foremost human geneticists and statisticians, Professor Sir Peter Donnelly FRS (Founder and CEO, formerly Director of the Welcome Centre for Human Genetics, University of Oxford) and Professor Gil McVean FRS (Founder and CIO, formerly Director of the Big Data Institute, University of Oxford). Our company was founded with the vision of unlocking knowledge from population-scale genetics to make healthcare systems better for the individuals they serve.
2. Genomics plc specialises in genetics-enhanced risk prediction and its application in healthcare.  Our database of linked genotype-phenotype information is the largest and richest of its kind in the world. Our platform links over 14 million variants in the human genome to changes in 15,000 molecular, cellular, and physiological measurements and disease outcomes.  We apply best-in-class analytical tools to this resource to understand human biology and pathophysiology, to find new drug targets for serious diseases and to deliver precision health applications. 
3. Genomics plc is working in partnership with the NHS to introduce polygenic risk scores (PRS).  We are working with NECS to support GPs in identifying patients at increased risk of CVD earlier. We are also supplying polygenic risk scores for the Early Disease Detection Research Project (EDDRP) covering 5 million individuals.

Genomic Prevention and healthcare systems

4. Healthcare systems across the world are grappling with how to create a sustainable model which provides cost-effective care to a growing ageing population, often with multiple long-term conditions. The use of genomics can support this now and in the future.
5. The UK and the NHS already lead the world (in the form of the Genomics England 100,000 Genomes Project) in one major application of genomic information in healthcare, namely Genomic Medicine: the use of whole genome sequencing for rare diseases and cancers, to inform diagnosis and treatment of individuals who are already sick. 
6. We believe there is an opportunity for the UK and the NHS to lead in a second, much larger, area, namely Genomic Prevention: the harnessing of genetic information to improve the prediction, prevention, and treatment of common diseases including cancer.
7. Building on 15 years of ground-breaking advances in the understanding of human genomic variation and the genetic basis of common diseases, we are now able to use high throughput and cost effective technologies (which are up to 20-fold cheaper than the whole-genome sequencing which underpins the Genomics England use case in Genomic Medicine) to measure an individual’s genetic risk for many common diseases, including cardiovascular disease, common cancers, autoimmune disorders, and chronic conditions such as diabetes and COPD.
8. Polygenic risk scores (PRS) have been proven to give clinically valuable insights into individuals and populations, up to decades before the onset of symptoms.   They can provide useful and potentially actionable information on disease risk in apparently healthy individuals.  They therefore have the potential to be used within healthcare systems to predict and plan effective health services for chronic and life-limiting conditions as well as target prevention services more effectively.
9. PRS-generating tests can be delivered commercially through a direct-to-consumer model, or through a health service provider.  We would advocate the use of PRS in conjunction with the use of existing risk measurements (such as blood pressure and cholesterol levels).  Polygenic risk scores should, in general, not be considered as standalone diagnostic tools.

Principles for introducing genomic testing

10. We believe that introducing genomics to support population health should be underpinned by five principles:

○        Outcomes should be patient-focused;

○        Entities should work closely with, and collaborate with, healthcare system providers;

○        Decisions should be rooted in scientific and clinical evidence;

○        Clear patient consent processes and support packages must exist; and

○        The handling of genetic information must conform to the highest ethical and regulatory standards of information governance and privacy.

The role of genomics in transforming healthcare delivery

11. The NHS Long Term Plan, published in 2019, focuses on how to meet the healthcare needs of the country in the next decade through prevention, personalisation of care, and the eradication of health inequalities. 
12. Common diseases, such as cancer, cardiovascular disease, and diabetes exert an enormous burden on NHS services and population health management. If those at risk of such conditions could receive earlier warning this would support health systems in mitigating the impact through early intervention and preventive treatment as well as by empowering the individual to take proactive steps to improve their own health and wellbeing. 
13. Understanding the risk make-up of populations can support health systems to:

○        Deliver more person-centred care models and pathways;

○        Undertake effective stratification and management of population health needs at scale;

○        Adapt existing care pathways to deliver more preventative care; and

○        Tackle health inequalities in local areas.

14. 2020 has also seen a reboot of many healthcare service operations, with the COVID-19 Recovery Plan highlighting risk-based targeting strategies and the rapid adoption of digital solutions as key tactics. For the NHS, today represents a unique opportunity to adopt and implement innovative solutions that work to deliver on the Long-Term Plan, through key change-enabling organisations such as the Integrated Care Systems, the NHS England Regional Teams, and NHSX.

NHS partnership – Cardiovascular disease

15. Cardiovascular disease (CVD, which includes coronary disease and ischemic stroke) is the leading cause of death in the UK, with greater incidence among males. In the UK, GPs already routinely assess CVD risk using an established risk prediction tool called QRISK, with current guidelines recommending lipid-lowering treatments (statins) when 10-year CVD risk is above 10%. 
16. Genomics plc has developed the most powerful polygenic risk score available for coronary artery disease, and an integrated risk tool which combines this with QRISK. The integrated tool is significantly more powerful than QRISK alone.
17. CVD is highly correlated with other cardio-metabolic conditions including diabetes and hypertension. Over time, the inclusion of genetic, phenotypic, and contextual data would allow for a more targeted and integrated management of the highest-risk cohorts, taking account of not just the current clinical picture but also the risk of future aggravation and comorbidities.
18. Genomics plc is working in partnership with a Commissioning Support Unit in England to use PRS to support GPs and primary care in identifying patients at risk of CVD earlier. This pilot project aims to address the clinical and operational opportunities and questions around the inclusion of genetic information in the existing CVD risk tools used in primary care.

The science behind genomics and polygenic risk scores

19. For many common diseases, genetic risk provides the single strongest risk factor.  For most individuals and most diseases, this contribution to risk arises from the combination of tens or hundreds of thousands of small genetic changes which collectively affect many different genes and molecular processes.  Individually, these changes have limited predictive power.  But in combination, they can increase (or reduce) risk many-fold.  Since 2005, research initiatives across the world, many of them led by the UK (and specifically by the leaders of Genomics plc), have been steadily increasing our understanding of this, so-called, polygenic risk.  Today, for many diseases, and thanks in part to resources such as the UK Biobank, this knowledge has reached a level of accuracy and robustness that make it clinically relevant, over a number of common diseases.
20. In the short to medium term, we see the greatest benefits arising from: improving the speed, accuracy, and targeting of diagnostics; ensuring more personalised clinical management, within existing care pathways and guidelines; and empowering and enabling large-scale population health management in the NHS. Over time, we see significant potential in personalised therapeutic guidance, that is, using genome-wide scores to make much better - safer and more efficacious - choices of medications, surgery, and other interventions for individual patients.

Example case study – Breast cancer

21. Breast cancer is the second leading cause of cancer death in women worldwide[1]. While some women carry genetic variants that cause a high risk of cancer (e.g., BRCA1/2), these are rare and are present in fewer than 5% of cases. For an individual woman, a large number of other genetic variations, each with a small additional risk of cancer, can together be associated with a large increase in risk. These PRS can be used to stratify women based on their genetic risk. The top 3% of women by PRS - currently invisible to the system - have a 30% chance of developing breast cancer over their lifetime, equivalent to the ‘High Risk’ category of women generally associated with mutations like BRCA 1 & 2 in NICE guidelines. An actionable intervention would be to offer screening to women with very high PRS risk from an earlier age, as is currently the practice for other women in the currently defined High-Risk group[2].   Detecting cancers at an earlier stage is known to lead to better survival, and require less aggressive forms of treatment, benefiting both the women and their healthcare providers. These numbers could be further improved by integrating genetic factors with non-genetic risk factors. 
22. Figure 1 shows the evaluation of Genomics’ PRS for breast cancer in an independent validation set in UK Biobank.  This figure is reproduced from our April 2019 submission with the differences in the two figures also demonstrating the increasing power of our polygenic risk scores as our analytical capabilities grow.   

Figure 1: The evaluation of Genomics’ PRS in an independent validation set in UK Biobank (UKB).  For groups of individuals in UKB defined in terms of their PRS, the curves show the estimated cumulative incidence of breast cancer diagnosed in each group at particular ages.  Shadings show the 95% CI for the estimated cumulative disease incidence.  Individuals with high PRS have much higher incidence of disease, with a woman in the high-PRS group has the same risk of disease as a woman who is 10-15 years older.

Genomics plc

Oxford & Cambridge, UK

29 May 2020

Annex A: April 2019 submission

Science and Technology Committee inquiry on Commercial Genomics -Written evidence submitted by Genomics plc

Introduction

1. Genomics plc is a for-profit, commercial genomic analysis company, spun out of the University of Oxford in 2014 by four of the world’s leading geneticists. It has offices in Oxford and Cambridge, UK, and now employs over 50 scientists, software engineers, and others.

2. Genomics plc has developed a world leading data resource which links genetic variation in individuals to changes in health outcomes and other biological and molecular measurements. We apply sophisticated analytical tools to this resource to understand human biology and pathophysiology, to find new drug targets for serious diseases and to deliver precision health applications.

3. We welcome the opportunity to respond to certain aspects of the Committee’s inquiry[3], and would be happy to provide further assistance to the Committee if that would be useful.

Preamble - genetic information can be used in different ways

4. It is critical to understand that there are two, very different, areas of application for genetic information, and hence genetic testing, in healthcare:

●        The use of whole genome sequencing for rare diseases and cancer (as in the Genomics England 100,000 Genomes Project); and

●        The use of genome-wide genetic data to identify individuals at increased risk for certain common diseases.

5. There are three key differences between these two applications:

●        They involve different clinical conditions: rare diseases and cancer in the first case; and common diseases in the second case;

●        They require the collection of genetic information at different levels of detail, and very different cost: reading an individual’s entire genome (via ‘whole genome sequencing’) in the first case; and the collection of data on a fixed large set of genetic variants across the genome (usually via a ‘genotyping chip’ or ‘genotyping array’) in the second case. Currently whole genome sequencing costs £600-£1,000 per individual, whereas a genotyping array costs around £30 per individual; and

●        They typically apply differently to healthy and well individuals: The first category is applied to individuals who are already sick to inform diagnosis and treatment; the second to individuals who are typically well to facilitate disease prevention.

6. In our opinion it would be wrong for the Committee and others to overlook these significant conceptual differences when considering the types of health benefits and other services that could arise from the increasing use of both types of applications. The remainder of this preamble addresses these key differences in more detail, with a focus on the growing role of genomics in understanding, treating, and potentially preventing certain common diseases. 

Genomic testing for rare diseases and cancer

7. There is established clinical utility in the identification of specific genetic variants that contribute to the molecular basis of an individual’s disease in the context of rare ‘monogenic’ or Mendelian diseases (that is, diseases that occur as a result of changes in a specific area, often a gene, within the genome) and cancer.

8. Evidence from over 30 years of genetic testing in NHS molecular genetics laboratories and more recently through Genomics England’s 100,000 Genomes Project provide clear examples of the utility of genomic testing in this context.

9. The identification of disease-causing genetic variants underpinning rare monogenic disease can provide clarity of diagnosis and inform clinical management, therapeutic regimes and family planning.

10. More recently progress has been made, notably again through the 100,000 Genomes Project, in the detection of clinically relevant somatic mutations in cancer that can inform therapy selection and prognosis.

Most common diseases are polygenic

11. More than a decade of research has established that for all common diseases many thousands of genetic variants (single nucleotide polymorphisms, or ‘SNPs’) contribute to disease risk. Most of these individual risk-variants are common in the population, but each one typically has only a small effect on risk. The same is true for many other continuous traits in humans, including anthropometric, physiological, biochemical, and cognitive measurements.

12. This ‘polygenic’ basis for the majority of common diseases contrasts with many serious rare diseases and cancers, which are often ‘monogenic’ (as described above) in nature. For some common diseases (e.g. breast cancer) both types of genetics are in play: there are known genes where single changes can have substantial effects on risk (e.g. BRCA1 and BRCA2 for breast cancer), and also thousands of SNPs with individually small effects on risk.

Genetic risk profiling in the healthy population

13. The idea of combining information across many common SNPs to predict disease risk has been around for more than 10 years, but it has only recently been possible to validate such methods and assess their potential clinical utility, largely thanks to the UK Biobank resource[4].
14. Outside of the contexts of rare disease and cancer, there is growing evidence that genetic variation observed in an individual’s genome can be used to quantify that individual’s risk of developing a range of common complex diseases.

15. Such approaches are typically referred to as Polygenic Risk Scores (PRS). They combine information from thousands of SNPs and can identify individuals with markedly higher (or lower) risk of developing a particular disease, on the basis of the DNA variants they have inherited.

Different genetic sequencing approaches

16. To identify the mutations which cause rare genetic diseases it is usually essential to sequence, or read, the entire genome of the individual (or at least the part that encodes genes: the ‘exome’).

17. In contrast, the many common variants needed to generate Polygenic Risk Scores can be measured with a different, and substantially cheaper, technology: genotyping arrays. These measure a predetermined set of about 1 million of the 3 billion positions in the human genome. From a single assay for one individual, one could calculate PRS for many diseases.

18. Currently whole genome sequencing costs £600-£1,000 per individual, whereas a genotyping array costs around £30 per individual.

The application of Polygenic Risk Scores

19. Polygenic Risk Scores should not be considered as stand-alone diagnosis tools. Rather, they could be used to complement existing risk measurements (e.g. blood pressure and cholesterol levels) to improve the identification of at-risk individuals.

20. In principle, Polygenic Risk Scores can be calculated for many diseases. The SNPs involved in calculating a PRS, and the weightings for them, will typically differ from disease to disease. Across a large set of individuals, for a particular trait, there will be a distribution of values for the PRS. Individuals with higher values of the score will be at higher risk of developing the disease on the basis of the common genetic variants they have inherited.

21. For any particular disease, a clinically meaningful increase in disease risk predicted by PRS typically will only apply to a few percent of individuals (although as for coronary artery disease, in a large population, this information can direct treatment to hundreds of thousands of individuals). Critically, while most individuals will have average risk for any particular disease, it is very likely that they will be at the extreme of genetic risk for at least one disease.

22. Early identification of these risks, through the availability of (relatively cheap) genotype data, could have a profound effect on individual and population health, and on health-related expenditure. The possibility of generating PRS for many common diseases at a population scale would identify individuals in the tail of the risk distribution for a subset of diseases, and present an exciting opportunity to optimise care, prevention and screening accordingly. For many diseases, elevated medical surveillance, targeted diagnostic screening combined with lifestyle changes or medical intervention for those individuals at high risk has the potential for disease prevention.

Specific examples of Polygenic Risk Scores

23. Women in the UK in the top 1% of the breast cancer PRS distribution (excluding BRCA genes) have a 30% lifetime risk of developing breast cancer. In this case it may well make sense to target screening preferentially at such individuals. For yet other diseases, including some mental health disorders, knowledge of increased genetic risk could be helpful in differential diagnosis, shortening the time between first seeing a doctor and getting the correct treatment. PRS are also likely to be helpful in guiding treatment choices, improving efficacy and reducing the risk of side effects.

24. Recent studies have shown that individuals in the top 5% of the coronary artery disease PRS distribution are at about threefold increased risk of developing the disease compared to the population average (Khera et al. 2018[5]). This effect is large enough to warrant clinical attention, and for many of these individuals there will be clinical interventions (and behavioural changes) which will reduce disease risk.

25. The figures below illustrate this risk stratification for breast cancer and coronary artery disease. These analyses were based on published genome-wide association studies (GWAS) for which the effectively anonymous summary statistics data were made available CARDIoGRAMplusC4D Consortium et al.[6] and Michailidou et al.[7].

26. In each case the team at Genomics plc evaluated performance in unrelated individuals in the UK Biobank cohort, matched on ancestry and sex (where appropriate) to the source studies.

27. In each of the figures, we include three indicative groups from UK Biobank, those individuals in the top 3%, median (ranging from 40-60%), and the bottom 3% of PRS.

Example 1 - Breast cancer

28. In breast cancer, the disease incidence in the three groups is very different. A 45-year-old woman in the high-risk group (red) has the same chance of having had breast cancer as a 55-year-old woman in the average group (blue) and a 75-year-old woman in the low-risk group (green). The chances of breast cancer by age 75 for a woman in the top group are around 25%, about three times higher than for one in the average group, and eight times higher than the low-risk group.

29. Currently in the NHS, breast cancer screening is based solely on age, with women being offered screening when they turn 50. Adding PRS information to the current screening system could direct screens to those most at risk while reducing unnecessary screens for low risk individuals. In addition to targeting screening more effectively, Polygenic Risk Scores affect the interpretation of screening results. A positive mammogram result for a woman in the top group is much less likely to be a false positive than one for a woman in the bottom group. 

Example 2 - Coronary artery disease

30. For coronary artery disease a similar pattern is observed, with very different risk profiles across the genetically defined groups. A 50-year-old man in the high-risk (red) group has the same chance of having developed heart disease as a typical 60-year-old (blue) and the same chance as a 70-year-old in the low-risk group (green). Almost a third of the men in the red group will have developed heart disease by the time they are 75.

31. While further work is necessary, high-risk individuals could potentially be prescribed cholesterol-lowering statin medication. Deploying these medicines earlier to such individuals could potentially reduce coronary disease and its consequences such as myocardial infarction. At least some of the individuals in the high-risk group may also be more likely to make lifestyle adjustments to reduce their risk, although further empirical studies to assess this will be helpful.

32. With this context in mind, the remainder of our submission will address some of the specific questions raised by the Committee.

---

Health or other benefits that consumers can derive from using commercially available genomic testing

32. Genome sequencing-based approaches currently being delivered by those such as Genomics England and Genomic Laboratory Hubs have clear benefits in the context of rare disease and cancer. This area will continue to be important. However we believe it is likely that, in time, the largest impact of genomics in medicine will relate to common disease risk prediction and prevention.
33. The approach for common disease risk prediction using Polygenic Risk Scores from genome-wide data is distinct, both scientifically and logistically, from approaches currently used in rare disease and cancer.
34. Polygenic Risk Scores should not be considered as stand-alone diagnosis tools. They can provide useful and potentially actionable information on disease risk in apparently healthy individuals. Accordingly there are opportunities for PRS-generating tests to be delivered commercially via a direct-to-consumer model, or through a health service provider. We would advocate the use of PRS in addition to the use of existing risk measurements (e.g. blood pressure and cholesterol levels) to improve the identification of at-risk individuals.
35. For the individual, knowledge of which diseases they are at elevated risk of has potential benefits for health planning with the aim of disease prevention. This could include taking steps such as elevated medical surveillance, targeted diagnostic screening, combined with lifestyle changes, medical intervention, or both.

The industrial strategy opportunity for genomics within the UK biotechnology sector, and how the Government could support UK growth (including for exports)

36. The UK has extraordinary strength in genetics. UK-based researchers have made leading contributions to the major research efforts in this area over the past two decades, such as the Human Genome Project, the Wellcome Trust Case Control Consortium, and the International HapMap Project. It is important to note that UK-based researchers have also led global efforts aimed at understanding the contribution of genetic variation to common diseases.
37. The application of genomics to healthcare has potentially huge commercial value, not just in the UK but across a global market. Given its existing expertise, the UK has the potential to be a world leader in this space. Some current and future UK-based SMEs could become global players, and in so doing create massive benefits for individuals’ health, health care systems, and the economy as a whole.
38. The Government’s backing for the translation of genetic research, particularly in the case of the ongoing support for UK SMEs, is welcomed.
39. The Government could do more to support routine sharing of data from genetic research, particularly when such research has been funded by taxpayers. It is possible for robustly de-identified or effectively anonymous summary-level data to be used very effectively by other researchers e.g. to help generate Polygenic Risk Scores. Providing adequate funding to create and maintain resources to ensure that such data can be freely and consistently shared, would be of great benefit to the conduct of genetic research in the UK and beyond.

The extent to which currently available genomic sequencing and interpretation can provide accurate and unambiguous health results, for healthy and ill sections of the population

40. In the context of rare disease, the application of current genome sequencing approaches provide molecular diagnosis for 20-50% of individuals.
41. In contrast, Polygenic Risk Scores have the potential for wide impact across the entire population. Individuals with particularly high PRS are at substantially elevated risk of disease. We illustrate this using the examples of coronary artery disease and breast cancer in the preamble.
42. For any particular disease, a clinically meaningful increase in disease risk predicted by PRS typically will only apply to a few percent of individuals. However, for many common diseases, such as coronary artery disease, in a large population this information can direct treatment in respect of hundreds of thousands of individuals. Critically, while most individuals will have average risk for any particular disease, it is very likely that they will be at the extreme of genetic risk for at least one disease.
43. Early identification of these risks, through the availability of (relatively cheap) genome-wide genetic information, could have a profound effect on individual and population health, and on health-related expenditure. The possibility of generating PRS for many common diseases at a population scale would identify individuals in the tail of the risk distribution for a subset of diseases, and present an exciting opportunity to optimise care, prevention, and screening accordingly.
44. It should be noted that due to the fact that most studies of the genetic basis of common complex disease have been undertaken in individuals of European ancestry, Polygenic Risk Scores are best calibrated to define disease risk in individuals of recent European ancestry. It is necessary to address this disparity both with studies of the genetic basis of common diseases in other non-European populations and also to evaluate the validity and calibration of translating risk estimates between populations.

The counselling or other support offered for those receiving, or considering asking for, commercial genomic test results, and whether this is to the standard required.

45. In our view, the conceptual differences between the two applications of genomics addressed in the preamble should be reflected when considering counselling and support for recipients of results.
46. The NHS workforce is well-versed in delivering information on disease risk in the context of predictive tests specifically conveying the results risk calculations e.g. explaining cholesterol levels or blood pressure measurements in relation to the risk of cardiovascular disease. Disease risk emerging from Polygenic Risk Scores follows a very similar paradigm for these tests.
47. Crucially, unlike tests for monogenic rare diseases, predictive tests based on polygenic factors of common diseases are not deterministic in terms of either the individual or their family members.
48. Providers of genetic test results in any context must have a comprehensive understanding and ability to accurately convey concepts of disease risk and the consequence of the nature of the disease risk. In our opinion, it is not evident that communication about disease risk emanating from Polygenic Risk Scores would require specific counselling or support beyond that which is in place for the communication of disease risk from other disease risk factors.

The potential benefits and risks for the NHS that arise from the increasing availability of commercial genomic testing

49. Determination of disease risk from genomic information via Polygenic Risk Scores has clear utility both for the individual and healthcare systems as a whole.
50. One risk in the short term for the NHS is that individuals will be motivated to find out their genetic risk from direct-to-consumer companies offering PRS information but will turn to the NHS for help, advice, and interventions to modify this risk.
51. As discussed above, a significant potential benefit to the NHS, especially if these types of tests become routine in healthcare, is the opportunity to optimise care, screening, and even disease prevention for those individuals shown to be at high risk for particular common diseases.  

The regulations or standards that commercial genomic tests are currently subject to, and if any new or strengthened regulations or standards should be introduced to mitigate any perceived risks associated with commercial genomic testing.

52. We are not aware of any different regulatory treatment of genetic tests that assess the monogenic basis of rare diseases and cancers, as compared to tests that could assess the polygenic basis of common diseases.
53. In our view, the provision of genetic information relating to risks of developing common diseases, using approaches such as Polygenic Risk Scores, ought to be regulated in the same way as predictive tests for other risk factors (e.g. tests for cholesterol, blood pressure etc). These predictive tests are unlike those for monogenic rare diseases, which tend to offer significant information that is determinative for individuals and their families. In the case of tests for monogenic rare diseases there is a clear need for appropriate, informed communication (e.g. involving genetic counsellors) when results are provided. We would not support any regulatory requirement for genetic counselling alongside the communication of disease risk information arising from Polygenic Risk Scores: the information communicated in those cases would refer to individually small effects on risk for individuals and their families. 
54. More generally, we believe that regulation in this area ought to reflect the context in which information from genetic tests is provided i.e. whether it is done in a medical or non-medical setting. We would support the continuing application of general consumer protection regulations to any genetic tests - of whatever type - made available directly to consumers. We would also support appropriate (i.e. not disproportionate) regulation or standards designed to ensure that providers of genetic tests - of whatever type - were transparent about the scientific validity and credibility of their tests.  

Genomics plc

Oxford & Cambridge, UK

26 April 2019

(May 2020)