9 January 2017 – Today’s “long read” in The Guardian is about sugar, perhaps the world’s most popular drug. It eases pain, seems to be addictive and shows every sign of causing long-term health problems. The article posits that perhaps it is time to quit sugar for good. There is something about the experience of consuming sugar and sweets, particularly during childhood, that readily invokes the comparison to a drug. This is of more than academic interest, because the response of entire populations to sugar has been effectively identical to that of children: once people are exposed, they consume as much sugar as they can easily procure. The primary barrier to more consumption – up to the point where populations become obese and diabetic – has tended to be availability and price. As the price of a pound of sugar has dropped over the centuries, the amount of sugar consumed has steadily, inexorably climbed. For the article click here.
The article was timely because I had spent the weekend catching-up on some recent studies published in the New England Journal of Medicine on “balanced lifestyles” and health and genetics as part of my artificial intelligence program at ETH in Zurich. I have spent some time on machine learning applications for the analysis of genome sequencing data sets.
Oh, you know the drill: how many times have you been to the doctor and was told that a “balanced lifestyle” is key for a healthy, disease-free, long life?
But how many times does genetics take the blame, and not your regimen? And do we really know whether genetic risk associated with complex diseases can be offset by behavioral changes?
Well, some recent studies bring us some answers. Herein a composite of the key points from several studies. If interested in the full studies, pop me an email. I have provided links to some terms with which you may not be familiar:
The idea that most human diseases are the result of a complex interplay between genes and environment is not new. Complex genetic disorders, those in which the effects of multiple genes are at play and clear-cut patterns of inheritance are difficult to pinpoint, are particularly influenced by lifestyle and environmental factors. Take heart disease as an example. While one’s own risk of cardiovascular disease increases by 3- to 5-fold if a person’s parent is diseased, factors such as diet, physical activity, tobacco use, environmental stress, and host factors such as age and sex also play a significant role in the risk of developing disease. The problem is that quantifying the contribution of each of these factors is tricky. One study was set up to try to answer the billion dollar question: can a healthy lifestyle compensate for a strong genetic risk of developing disease?
Genetic risk in complex diseases results from the combination of incremental effects of a large number of DNA sequence polymorphisms. In the case of coronary artery disease and myocardial infarct—a common and severe complication of cardiac disease—over 50 independent loci have been associated with risk of disease in genome-wide analyses. The aggregate of these risk genes provides has proved useful in defining genetic susceptibility to cardiovascular events. One study makes use of this strategy to stratify the ∼60,000 subjects in their prospective study into three genetic risk categories: low, intermediate, and high. Then they ask whether baseline healthy lifestyle choices impact the risk of developing coronary events and atherosclerosis. The definition of healthy lifestyle was drawn from the strategic goals of the American Heart Association to reduce the risk of cardiovascular in the general population and include no current smoking, no obesity, physical activity at least once weekly, and a healthy diet pattern. Importantly, the subjects in the study were previously enrolled in independent studies and represent three completely independent cohorts, what helps to power their analyses.
As one would expect, subjects with a high polygenic risk score were at much higher risk of developing coronary events than those at low genetic risk. What may come as a surprise is that further dividing subjects of different genetic risk groups according to their lifestyle choices shows that those who make healthy lifestyle choices present ∼45%–47% lower relative risk of developing coronary events. Even subjects at high genetic risk have their risk reduced when they make healthy lifestyle choices. The association also seems to hold true in men and women, and it does not seem to be under strong influence of ancestry. Moreover, not only the risk of adverse coronary events is reduced by adherence to healthy lifestyle but also coronary artery calcification, which is a prelude to more severe disease.
Does that mean that going to the gym and eating fresh fruit, vegetables, grains, nuts and fish (while avoiding saturated fat, red meat and sugar-sweetened beverages) can counteract completely the effect of genetic inheritance? Certainly not. A new study clearly shows that high genetic risk is a strong and independent risk factor that increases the risk of coronary events and atherosclerosis. What perhaps is unexpected is that genetics and healthy lifestyle choices seem to be independently contributing to the susceptibility of heart disease. While it is true that people with high genetic risk benefit the most from healthy choices, the positive effect is also seen in those with intermediate and low genetic risk. In other words, healthy lifestyle benefits everyone, and high genetic risk is not necessarily deterministic of disease outcome.
What do these findings mean from a basic biology perspective? Will we be able to get passed associations and begin to understand relationships of causality between healthy habits, genetics and disease? Will we be able to move beyond generic prescriptions of diet and exercise and tailor one’s choices to his or her genetics? Recent data have already showed that the response to food, for instance, is highly individualized when it comes to metabolism and glucose response, speaking to the need a personalized approaches for nutrition and medicine. Further, can the risk to other complex diseases be modified by behavioral changes?
While I eagerly wait for answers to some of these questions, perhaps it would be wise to stick with carrots.
Coming in Part 2:
The field of machine learning, which aims to develop computer algorithms that improve with experience, holds promise to enable computers to assist humans in the analysis of large, complex data sets. It has had a critical role in the analysis of genome sequencing data sets, including the annotation of sequence elements and epigenetic, proteomic or metabolomic data. But there recurrent challenges.
Genetic algorithms are fascinating and they do a good job of introducing the “cost function” or “error function” – concepts both important and common to most other ML algorithms. My legal technology readers will quickly get the drift.
Genetic algorithms are inspired by nature and evolution, which is seriously cool to me. It’s no surprise, either, that artificial neural networks are also modeled from biology: evolution is the best general-purpose learning algorithm we’ve experienced, and the brain is the best general-purpose problem solver we know.
In Part 2 I will give you an overview and try to explain some genetic algorithm basics … and where this all fits in with the health issue I posed earlier.