Regressions of outcomes on genes, when controlling for parental genes, yield causality as much as could be desired: which of the parental genes an individual gets is randomized at the molecular level. And controlling for what genes the parents have (which form the pool from which all but mutations come from in the child) controls for the effect of the parental genes on how the parents do their child-rearing.

Often, though, we are not satisfied with knowing causality from genes because, when it comes to interventions, the only thing causality from genes says directly is what one might be able to get from genetic engineering interventions.

It is often the case that a set of genes G cause both outcome X and outcome Y. This is called a genetic correlation. Correlation does not imply causation and genetic correlation between X and Y does not necessarily imply either that X causes Y or that Y causes X. And a genetic correlation between X and Y does not necessarily imply (X causes Y OR Y causes X). G could cause X entirely without going through Y and separately cause Y entirely without going through X. As a crucial example, G could cause Z, then Z cause X entirely without going through Y, and separately Z cause Y entirely without going through X.

Even though genetic correlation between X and Y does not imply causation, it should make us wonder whether X causes Y or Y causes X. Correlations of any substantial magnitude always raise the question of causation even though they don’t prove causation.

The most surprising finding is a clear positive genetic correlation between fraction of calories derived from protein (controlling for overall calories) and obesity. The next most surprising finding is a negative genetic correlation of fraction of calories from sugar and waist circumference and the closely related waist-to-hip ratio. The authors give a careful discussion of the possible causal explanations. Here is their discussion, with headings added for the two different passages:

Protein

The genetic correlations we find between protein and obesity, waist-hip ratio, fasting insulin, type 2 diabetes, HDL cholesterol, and heart disease, together with the association we find between the BMI-increasing FTO allele and increased protein intake, point to an intriguing hypothesis: relative protein intake may play a role in the etiology of metabolic dysfunction. This hypothesis coincides with a growing (but often overlooked [71]) body of evidence that links protein intake to obesity and insulin resistance [72,73,74,75,76,77,78,79,80]. There is some related evidence from randomized trials with infants, which found a causal relationship between high-protein baby formula and infant body fat [81]. While the underlying biological mechanisms are unclear, high consumption of protein or certain types of amino acids (i.e., building blocks of protein) is known to induce insulin resistance [82,83,84], rapamycin signaling [77], and growth factor signaling [85], which might increase metabolic dysfunction and early mortality risk. Indeed, a recent phenotypic meta-analysis of prospective observational studies (pooled N = 154,344) found that low carbohydrate diets, which restrict carbohydrate in favor of increased animal protein or fat intake, were robustly associated with increased mortality [86].

We caution, however, that the strong and consistent links between protein and poor health outcomes might also be consistent with alternative explanations. Causation could run in the reverse direction: overweight individuals may have higher protein needs or use high-protein diets as a weight-loss strategy. The associations might also be caused by other, unmeasured variables such as unhealthy lifestyle factors or co-consumed ingredients.

Sugar

These correlations may suggest that dietary sugar, beyond its energy content, does not have negative health effects [87,88,89,90], contrary to some popular beliefs (e.g., [91]). Another possibility is that exercise offsets negative metabolic effects of high sugar intake [92, 93]. Those with a higher predisposition to be physically active may tend to consume more sugar, as sugar is a metabolically convenient source of energy during exercise [94] and may enhance endurance [95]. If so, the positive genetic correlation between sugar and physical activity might partially explain the lack of genetic correlations between sugar and poor health.

I continue to think that sugar is causally bad and the animal protein is causally bad, in line with many previous posts here. But these genetic correlation results add nuance to that view. In particular:

• Based on the authors’ discussions, their may be more mechanisms for protein to be a problem than the high insulin index for many protein-rich foods and cancer-feeding aspects of protein (and especially of animal protein) that I have emphasized.

• Other healthy behaviors—especially physical activity—leading to more sugar consumption may obscure negative effects of sugar.

But all of these ideas are only possibilities. Genetic correlations are only one clue in figuring out diet and health relationships.

For annotated links to other posts on diet and health, see:

• Miles Kimball on Diet and Health: A Reader's Guide