Cancer — 7

As you may recall, a polygenic risk score can be inexpensively obtained from a gene chip (array) that assesses over a million common genomic variants that have been quantitatively validated to be associated with risk of a particular cancer, in terms of a score. While these were initially based on European ancestry cohorts, their accuracy and clinical utility have been markedly improved with diverse ancestry and large population data. That’s why some health systems have begun clinical implementation, despite some skepticism among researchers who continue to call for randomized trials to determine whether people having these scores saves lives. My view is that such trials are not likely to get done, and even if there was a funding commitment, it would take many years of follow-up. The principal goal of a polygenic risk score is to identify high-risk individuals so that they can be followed closely. Even if some people wind up getting surveillance measures that prove to be unnecessary (a false positive), that’s better than the one-size-fits-all current approach. Polygenic risk scores can miss high-risk cases, which is why it is so important to have multiple layers of orthogonal (complementary, additive) data, as shown in figure 6.4. The Mass General Brigham health system in Boston has begun providing polygenic risk scores for its patients. In published results from more than thirty-six thousand participants of diverse ancestry, 8.6 percent were found to be at high risk for breast cancer, 5.4 percent at high risk for colon cancer, and 13 percent at increased risk for prostate cancer. All were at more than double the risk of the general population, with some variation by ancestry. Polygenic risk score data from Finland’s FinnGen databank of over 450,000 individuals has indicated when colon cancer screening should be initiated, with the highest risk (top 1%) at age forty-nine years and the lowest risk (20%) at age sixty-seven years. That would save most people several colonoscopies! Similarly, partitioning by low polygenic risk score among more than twenty-five thousand women in the Healthy Nevada project indicated the potential to defer mammography by five to ten years, for those with a more than 60 percent reduction of risk.

The case for routine whole genome sequencing to determine risk of cancer keeps getting stronger. Whole genome sequencing of our three billion letters can now be done for about $200. We’ve known about well over a hundred cancer predisposition genes for over a decade that would all be detected by sequencing. In Iceland, nearly 58,000 people underwent whole genome sequencing, more than 108,000 people had a gene chip, and 4 percent were found to have actionable disease-causing gene mutations, nearly half of which were accounted for by cancer gene mutations (such as BRCA1, BRCA2, PALB2, MSH6, PMS2) for high risk of breast, ovarian, prostate, pancreatic, or colon cancer. Similar frequencies of these cancer gene mutations were found in the UK Biobank and other cohorts. The Icelandic data is unique because it has data from death certificates, and it shows carriers of cancer-causing variants died three years sooner. In the Icelandic population, there is a founder mutation in BRCA2, which was linked to a seven-year lifespan reduction, and the overall analysis was performed without including this founder mutation. As the accompanying editorial pointed out,