Despite the huge excitement that greeted the sequencing of the first human genome at the turn of the century, genetic testing isn’t routine. When will this change?
Historically, genetic testing was reserved for patients with known or suspected genetic disorders. Tests could identify single gene (“monogenic”) changes linked to conditions like sickle cell disease, haemophilia or cystic fibrosis which while rare individually are important collectively, totaling about 7,000 conditions affecting 1 in 10 people and now identifiable through single gene tests, gene panels, exome or whole genome sequencing.
Genetic testing for other indications has become more popular. “Personal genomics” satisfies people’s curiosity about their ancestry (were your forebears Vikings, or Zulu warriors?) and to explore DNA-based physical traits such as hair or eye colour. More importantly, genetic testing can improve the use of medications (the science known as “pharmacogenetics”) including those used to treat cancer (“precision oncology”).
Polygenic scores – genetic testing for the masses?
Polygenic scores (or Polygenic Risk Scores – PRS) could greatly expand the role of genetic testing. These tests target common health conditions like obesity, diabetes, coronary heart disease, high blood pressure, schizophrenia and others which do not, as a rule, result from single gene changes. A PRS sums the small effects of dozens, hundreds or thousands of genetic changes (“variants”) associated with the development of disease into a single numeric estimate of risk.
Advocates for polygenic scores say they enable more targeted therapeutic and preventative health interventions that will ultimately reduce disease burden and are ready for introduction to the clinic.
Challenges for polygenic score adoption
Problem: there is not enough genetic data from Africa and other global regions.
Genetic datasets do not accurately represent broader populations. For example, Genomics England and UK Biobank are pioneers in the field, but the BioBank’s percentage of non-European genomes is only about 5%. PRS scores have been poorly predictive in other ancestral groups. African genomes may be the most extreme example, not surprising because Africa is the continent with the most genetic diversity while being the least genetically explored.
Other challenges include:
- Gaps in professional understanding and interpretation of results and how to effectively communicate them to patients.
- Design and implementation of interventions after the return of results. Returning results is not beneficial without effective disease prevention or early detection strategies.
- Concerns about genetic determinism (the generally false belief that “DNA is destiny”; inevitability rather than probability), and the potential for stigma and discrimination (for example in limiting access to health, life or disability insurance).
- Quantitative improvements in risk prediction over existing risk models need to be large enough to justify potential costs and harms.
- Best practice for integration of PRSs into clinical programs has not yet emerged.
When will these problems be solved?
Maybe soon.
All of Us
The US National Institutes of Health (NIH) All of Us research programme will help.
All of Us is enrolling a diverse group of at least one million individuals “to accelerate biomedical research and improve human health”. A recent publication announced the release of nearly a quarter million whole-genome DNA sequences, plus more limited data on over 300,000 participants. Forty six percent are from racial and minority ethnic groups, and 77% from other groups historically under-represented in biomedical research, for reasons of race, ethnicity, age, geography, sexual orientation and gender identity, income, education, access to healthcare and/or disability. Combined with data from electronic health records, physical measurements, survey responses and wearables, All of Us is a great resource for research. It has already identified more a billion genetic variants, including more than 275 million previously unreported. These data are publicly available and accessible by researchers.
Diabetes – an important example of genetic diversity
Another recent publication used internationally diverse genomic data from 2,5 million individuals (39.7% not of European ancestry), including over 428,000 cases of Type 2 diabetes (T2D), to identify 611 sites of variation in the genome, of which 145 were previously unreported. Most people with diabetes are of non-white ethnicity, and South Asian, East Asian and African diabetes phenotypes (physical manifestations of disease) are markedly different from white populations. Studies investigating the genetics, phenotypes and pathophysiology of diabetes, mapped to treatment and long-term outcomes in these populations, are therefore desperately needed.
The eMERGE Network
The eMERGE (Electronic Medical Records and Genomics) Network is another US research project, addressing the challenges of adopting polygenic tests in mainstream practice. It focuses on the genetics of conditions that are common, well studied, heritable, and “actionable”, aiming to improve the “generalisability and portability” of genetic tests in diverse populations including African, Asian, and Hispanic ancestry groups. The Network has developed a “pipeline” for clinical PRS implementation, a framework for regulation, and a PRS clinical report. The pipeline uses data from 13,475 All of Us research participants to train and test the model and results from 25,000 subjects for risk stratification and to model harms and benefits over a patient’s lifetime.
eMERGE Study results
A recent report from eMERGE focuses on 10 conditions: asthma, atrial fibrillation, breast cancer; chronic kidney disease; coronary heart disease; hypercholesterolemia; obesity, prostate cancer; type 1 and type 2 diabetes.
Polygenic score results for persons of African or African-American ancestry along with those of European, Asian and Hispanic/Latino origin are compared in the figure below:
- Scores from all population groups are predictive; differences are not large except in a few instances (e.g., Type 2 diabetes).
- Between 2% (e.g., Type 2 diabetes, chronic kidney disease) and 10% (e.g., prostate cancer) of the population are categorized as high risk. ‘High-PRS threshold’ is the percentile cutoff for a specific condition above which a high-risk result is reported for a condition, based on statistically significant differences between ‘high-risk’ and ‘not high-risk’ groups.
- Odds ratios (square dots) range mostly between 2 and 4, representing the higher numerical likelihood that an individual will have the condition, compared to those with a score below the specified threshold. 95% confidence intervals (CIs) are shown in the whiskers.
- The ‘Number of SNPs’ is the number of genetic variants included in each score.
- ‘Age ranges for return’ indicate the participant ages at which a PRS is calculated for a given condition.
Figure 1. Results from the eMERGE study “Selection, optimization and validation of ten chronic disease polygenic risk scores for clinical implementation in diverse US populations”. Source: https://www.nature.com/articles/s41591-024-02796-z/figures/2
Key: AFIB, atrial fibrillation; BC, breast cancer; CKD, chronic kidney disease; CHD, coronary heart disease; HC, hypercholesterolemia; PC, prostate cancer; T1D, type 1 diabetes; T2D, type 2 diabetes
The eMERGE Study report
How can polygenic risk scores be communicated? The eMERGE PRS report adopts a simple approach (Figure 2) which is compared to an example from a commercial lab, Allelica (Figure. 3) below. The report states only that the PRS score (for asthma) is “high”. This implies, on average, a two to three-fold increase in genetic risk. The Allelica report in contrast offers a more detailed graphic showing where an individual lies on the distribution for polygenic risk.
Figure 2. eMERGE study PRS report. Source: https://www.nature.com/articles/s41591-024-02796-z
Figure 3. PRS for breast cancer risk as reported by the company Allelica (https://eu.allelica.com).
Overall risk, absolute risk, relative risk are key issues
The eMERGE study has a welcome focus on accuracy across diverse populations. However, accurately reporting relative genetic risk is of limited clinical value unless genetic risk is combined with a measure of absolute lifetime risk. Why? A twofold difference in risk for a rare condition (say 1 in 10,000) has a very different impact on clinical decision making than the same increase in lifetime risk for a common condition, such as breast cancer, which about 1 in 9. To be useful in practice, a PRS must be combined with a clinical risk of disease prevalence in the population.
Image source: https://emerge-network.org/genomic-risk/
The typical separation between control and PRS-based at risk populations is not huge but still significant. For example, reclassification of about 9-10% of intermediate clinical risk to a higher level of risk can occur. Individuals at the bottom 10% of risk do not show enough predicted risk reduction to be actionable for example to recommend a reduction of preventive screening; even PRS advocates have been cautious to suggest such a move.
This study (Figure 4) suggests that genetic risk of cardiovascular disease rises in a linear fashion as clinical risk rises, and that combined risk could be estimated by simple multiplication of clinical with genetic risk. This would be very convenient if true across a range of different clinical conditions.
Figure 4. A cardiovascular risk score (SCORE2) is multiplied by relative genetic risk from a PRS score to refine the absolute risk of cardiovascular disease. About 10% of individuals are reclassified from intermediate to high risk, with a two-fold higher predicted incidence of disease.
Image source: https://academic.oup.com/eurheartj/advance-article/doi/10.1093/eurheartj/ehae048/7637416
Genetic testing in SA
In SA, the genetic testing environment was dominated until recently by the US company Invitae which offered much lower prices for a range of genetic tests along with high reported levels of support and quality. Local laboratories, including commercial labs and NHLS, found it difficult to compete.
The Invitae story came to a head last year, with the company declaring bankruptcy, and knowing glances from SA laboratories who believed the company’s dominance of the local market was economically unsustainable at the prices charged. The possibility that testing was being run as a loss leader to gain access to African genomes was rumoured.
Discovery Health’s population genetics project, announced 8 years ago, but never launched could have made a major contribution to genetics research and clinical application in the local context. That’s an interesting story for another day.
Conclusions
PRS is coming nearer.
Research from All of Us and the eMERGE Network takes us forward in the implementation of PRS-based risk assessment that will make accurate genetic risk assessment possible globally and in Africa. This must happen in combination with other risk estimates from single gene (“monogenic”) testing, family history, and standard clinical scores.