lifestyle changes and genetic risk for CAD

A recent study looked at the relative effects of genetic risk and healthy lifestyle in the development of coronary artery disease (see cad genetic risk dec CAD with lifestyle change NEJM2016 in dropbox, or DOI: 10.1056/NEJMoa1605086​). Details:
-- 3 prospective cohorts were followed: the Atherosclerosis Risk In Communities (ARIC, with 7814 white people between the ages of 45 and 64), the Woman's Genome Health Study (WGHS, with 21222 white female health professionals), and the Malmo Diet and Cancer Study (MDCS, with 22389 Swedish people aged 44 to 73 and free from prevalent cardiac disease). Also included were 4260 people the cross-sectional BioImage Study who had genetic/risk factor data and had coronary artery calcium (CAC) scores
-- they evaluated these people for up to 50 single–nucleotide polymorphisms (SNPs) known to be associated with coronary artery disease, and then derived a polygenic risk score based on the number of risk alleles at each SNP, multiplied by the sum of the literature-based clinical effect size.
-- they also assessed 4 lifestyle behaviors: no current smoking, BMI <30, physical activity at least once a week, and a healthy diet pattern (consisting of increased amounts of fruit, nuts, vegetables, whole grains, fish, dairy, as well as reduced amounts of refined grains, processed meats, red meats, sugar sweetened beverages, trans fats, and sodium)
-- a favorable lifestyle was defined as at least 3 of the 4 healthy lifestyle factors, with an intermediate lifestyle being 2 of these factors.
Results:
-- 1230 cardiac events were observed in the ARIC cohort over 18.8 years, 971 in the WGHS cohort over 20.5 years, and 2902 in the MDCS cohort over 19.4 years
-- a risk gradient was noted across quintiles of genetic risk, pretty consistently among the studies, with a hazards ratio of 1.91 comparing the top vs. bottom quintile of risk scores, reflecting a 91% higher attributable genetic risk, controlling for age, sex, self-reported education level, and analysis of ancestry when available
-- Levels of LDL cholesterol were only modestly increased across categories of genetic risk, and genetic risk scores were independent of other cardiometabolic risk factors as well as the 10- year predicted cardiovascular risk
-- A family history of coronary disease was an imperfect surrogate for genotype-defined risk [perhaps part of the issue: several older studies i have seen have found pretty dramatic discordance between elicited family history of heart disease and actual known heart disease events in the family members, up to 35-40%; the Framingham study found a 17% discordance.]
-- Each of the lifestyle factors were associated individually with decreased coronary risk: non-smoking with a 44% decreased risk, BMI <30 with a 34% decrease risk, regular physical exercise with a 12% decreased risk, and a healthy diet with a 9% decreased risk
-- people with unfavorable lifestyles (<2 of the 4) had higher rates of hypertension and diabetes, higher BMI, and less favorable lipids; overall they had an adjusted hazard ratio of cardiac disease of approximately 2 in each of the three prospective cohorts (ie, twice the risk).
-- Within each genetic risk category, lifestyle factors were strong predictors of coronary events.
    -- Adherence to a favorable lifestyle (at least 3 of the 4 factors) vs. an unfavorable lifestyle (<2) was associated with a 45% lower relative risk in the group at low genetic risk, a 47% lower relative risk among those when intermediate genetic risk and 46% among those at a higher genetic risk. [ie, the relative risk was equivalently lower in each genetic subgroup with more favorable lifestyle]
    -- among people at high genetic risk, in the 3 prospective cohorts the 10-year coronary event rates were also approximately twice as high in those with unfavorable vs. favorable lifestyles.
-- Further analysis showed that those with high genetic risk but healthy lifestyles had about the same number of cardiac events as those with low genetic risk but unhealthy lifestyles in each of the 3 prospective cohorts, though this was partially explained by differences in traditional risk factors
-- there were limited data for the black population and generally less well-validated genetic loci for coronary disease, although evaluating the black cohort in the ARIC study yielded similar findings
-- in the BioImage study, both genetic and lifestyle factors were associated with higher CAC scores, and within each genetic group there was a significant trend toward decreased CAC scores in those with a healthier lifestyle

Commentary:
-- this study validates that both genetic risk as well as lifestyle are continuously related to cardiac events, and that the genetic risk is largely independent of traditional cardiac risk factors
-- but, this study really undercuts genetic determinism. In fact, there was the same relative decrease in cardiac risk in each of the genetic subgroups (high vs. low genetic risk) with improved lifestyle. And, in terms of absolute risk, those at the highest genetic risk had much more benefit from a healthy lifestyle change than those of lower genetic risk.
-- One general concern I have, tangentially related to the above, is with the availability of genetic analyses for the general population (e.g., 23andme, which advertises "discover what your 23 pairs of chromosomes say about you", and costs only $199). My concerns are in part that we don't really know how to interpret many of these genetic findings prospectively, but also that this analysis may reinforce a sense of genetic determinism in the population.
-- It was also a little surprising that the benefits for the individual lifestyle markers were what they were. We do know from prior studies that smoking cessation is the single most beneficial individual intervention in decreasing cardiac risk, as also suggested in this study which found that current non-smoking was associated with the lowest cardiac risk . What was most surprising to me was that the benefit of exercise and healthy diet were not so profound: I suspect that the quantification of diet and exercise was likely not that accurately represented as a bimodal, all-or-none issue (e.g., unclear what the cutpoint was for defining a healthy vs. unhealthy diet, and using one day of exercise a week without even quantifying the amount/type of exercise makes these lifestyle markers hard to interpret, especially as compared to smoking or BMI which have clear bimodal cut-points)
So, I think this study provides some important information in dealing with patients. I have certainly seen many patients who feel that they are destined to die from heart disease because of their family history. This study provides a tool to speak even more convincingly with patients, in that those at higher genetic risk actually achieve the most benefit by improved lifestyle.