Background

Mendelian Randomization (MR) uses genetic polymorphisms to determine if phenotypes, such as serum triglycerides (TG), have causal effects on disease endpoints. The underlying principal is that genetic variants are inherited randomly relative to other variables influencing a disease.Data on causal relationships for diet studies are limited because of difficulty randomizing people to diets for long enough periods to detect effects on diseases. Cohort studies of diets are susceptible to confounding, a problem that has led virtually all of medicine to adopt randomized controlled trials (RCT) as the gold standard for making treatment decisions.

Methods

This review combines results from MR with a review of published meta-analyses of RCT trials of low carbohydrate and ketogenic (LCK) versus high carbohydrate (HC) diet effects on TG. MR uses genetic variants in either specific genes or in genetic risk scores as randomizing variables to sort people into groups with variable levels of traits that may cause disease.

Results

Meta-analyses of RCT reproducibly show that LCK diets significantly decrease serum TG and increase HDL cholesterol. MR for alleles of lipoprotein lipase (LPL) showed that they cause changes in serum TG and cause cardiovascular disease (CVD). LPL alleles had no effect on LDL cholesterol.MR for alleles of low-density lipoprotein receptor (LDLR) show they influence serum cholesterol and CVD. Effects of LPL and LRLR alleles on CVD were equally large when normalized for apolipoprotein B levels.In contrast, associations from cohorts show much smaller effects of TG than LDL on CVD.

Conclusions

Results from previous studies show that LCK diets improve several independent CVD risk factors including TG. MR also demonstrates that serum TG independently effects CVD risk using cohort data that are easier to obtain than RCT.MR may provide a general method to make diet recommendations when RCTs are not possible.