*Geek Box: Classifying Biomarkers

*Geek Box: Causal Risk Factors vs. Systems Biomarkers

Studies investigating diet and health outcomes may use any number of measures to test effects of diet. Whether the outcome is composition of gut bacteria or blood cholesterol levels, it is important to distinguish between a physiological parameter that has been established as causative in a given disease process, or acts more as a “systems biomarker”, a term I first heard from Professor Chris Packard in a discussion about blood lipids and cardiovascular risk. So, what is the difference? We can delineate between the two as follows:

  • Independent risk factor: biomarkers in a causal pathway between the exposure and outcome;
  • Systems biomarkers: biomarker which provide indications of underlying cardio-metabolic processes, but are not causal independently.

This distinction is not academic. For example, LDL-cholesterol is an independent risk factor which is the causal pathway through which elevated cholesterol drives atherosclerosis. A systems biomarker, however, may not necessarily be causal of itself, but provides important additional granularity to the risk equation.

For example, high HDL-cholesterol is generally associated with lower risk for CVD, however, deliberately raising HDL-C does not reduce CVD risk, indicating that HDL-C is not directly causal of lower risk. But it remains an important systems biomarker; for example if two individuals had the same moderately elevated LDL-C levels, but one high and one low HDL-C, the individual with low HDL-C would likely be at higher CVD risk. Thus, HDL-C is this context is providing additional information to the risk assessment.

In the context of the present study, the question is whether TMAO is in the causal pathway driving cardio-metabolic disease processes, or is TMAO a biomarker for something else, perhaps underlying disease itself or the activity of the gut microbiota? And is diet in this causal chain? In addition to other lines of research to determine whether a risk factor has a causal role, a powerful tool in research design to look at potential independent causality is Mendelian randomisation studies [see the next Geek Box].