Earlier this week Carrie Arnold wrote an article for Mosaic about the anxiety and confusion uncertain genetic test results can provoke. I wrote a short accompanying piece and over the next few weeks I will be expanding on some of the issues and discussion points the article raised. This week I will focus on the urgent need for all of us to reset our baseline assumptions about genetic variation.
Genetic variation knowledge first came from studying genetic diseases
For 20 years, gene tests were performed if it had been decided that a genetic diagnosis was likely, through specialist medical evaluations. We obtained interesting and useful information from these tests, but they gave us a highly skewed view of genetic variation, because we only had data from a small number of genes in exceptional individuals with rare genetic conditions. But it was all we had. So, inevitably and appropriately, we used the information to make our best guess at what the overall landscape of genetic variation in humans looked like. A pervasive assumption from this time was that protein-altering genetic variation was rare and had a high chance of impacting human health.
We now have direct data on human genetic variation
One of the most important outcomes of cheap, fast DNA sequencing has been our ability to analyse many thousands of genomes, including from healthy individuals. Rather than having to guess the nature and structure of human genetic variation, we can now simply observe and describe it. Major initiatives such as the 1000 genomes project and the ExAC dataset of >60,000 individuals have given us unparalleled insights into our genetic makeup. They have also given us the opportunity to review how good our predictions were. We got some things right. But many of our previous assumptions about genetic variation have turned out to be naive, incomplete or incorrect.
Rare genetic variation is common
One of the most startling and important discoveries has been the recognition that we all have many thousands of genetic variants. This has been a universal, uncontroversial finding from all studies. For example, my group performed exome sequencing to analyse all 20,000 genes in 1000 individuals from the UK general population. This dataset, which is called the ICR1000 UK exome series, is available to other researchers and there is an open-access paper that describes our findings.
We found that, on average, each person had 22,000 gene variants, of which 160 were very rare, not present in anyone else in the study. Among these were, on average, 90 that altered the protein (nonsynonymous variant) and 6 of the most severe types of variant, which stop the gene from working at all (protein-truncating variants). This might seem surprising, but we each have two copies of every gene, so even if a genetic variant stops one copy working, the second copy can often cover for it.
Most rare variants do not cause disease
If this study had been done in individuals with disease many would have assumed there were causal links between the protein-altering variants and the disease in question. Indeed, when we first presented the results, colleagues would often suggest that the people with rare truncating variants must have undiagnosed genetic diseases, because it was so ingrained that these variant types cause disease.Of course, as soon as you realise that would apply to every single person, you also realise that the old baseline assumptions about rare variants must be wrong. Clearly, rare variants, including those that stop genes from working, are the norm. Individually they are rare, but collectively they are common. We all have them. And therefore, overall they must mostly be harmless. Otherwise all of us would be suffering from genetic diseases.
Time to reboot
It is easy to understand the message of these many studies. Disease-causing variants are rare, but most rare variants do not cause disease. In the same way that bananas are yellow but most yellow objects are not bananas.
This new understanding should have a profound impact on how we handle rare variants identified in genetic tests. But, unfortunately, our management of rare genetic variants has not yet caught up with our new knowledge. We need to reboot how we think about their likely impact, how we communicate about them, and how we use them to influence patient care.
Next week I will expand on the reasons for the lag in implementation, and what we might do about it.
Image: Photo by byronv2 on Flickr, CC-BY-NC