Why nutrition studies often clash — and how genetics may help

Australians are increasingly encouraged to make dietary choices that reduce their risk of chronic diseases such as cardiovascular disease and type 2 diabetes. As the National Health and Medical Research Council revises the Australian Dietary Guidelines for the first time since 2013, a key challenge remains: how do we know whether foods associated with better health are truly causing those benefits?

Much of nutrition research relies on observational studies, which compare what people eat with their health outcomes. These studies have generated important insights, but they cannot fully account for the many ways people differ from one another. Individuals who consume more vegetables, for example, may also exercise more, smoke less, have higher incomes, or engage more frequently with preventive health care.

This makes it difficult to distinguish causation from correlation. Is a food directly improving health, or is it simply a marker of a healthier lifestyle?

Using genetics as a natural experiment

Randomised controlled trials are often considered the strongest way to establish if a specific food causes a disease, but they can be difficult to conduct in nutrition research. Long-term dietary interventions can be expensive and difficult to maintain, while chronic diseases often take years to develop.

One increasingly popular alternative is Mendelian randomization, a method that uses genetic variants as natural experiments. Because genetic variants are assigned at conception and generally remain unchanged throughout life, they are less susceptible to many of the confounding factors that affect observational studies.

Mendelian randomization has transformed several areas of medicine and helped establish causal relationships between risk factors and disease. For example, it confirmed that LDL (“bad”) cholesterol causes coronary heart disease while challenging the long-held belief that simply raising HDL (“good”) cholesterol would reduce risk, a conclusion later supported by clinical trials. However, applying the same approach to diet has proven particularly challenging.

The problem with studying diet

The reliability of Mendelian randomization depends heavily on the genetic variants used as proxies for the behaviour being studied. Ideally, these variants should influence the dietary behaviour of interest and little else.

However, food choices are shaped by a complex interplay of biological, behavioural, and social factors. As a result, many genetic variants associated with dietary behaviours are also associated with education, income, body weight, and health status. This makes it difficult to determine whether observed associations reflect the effects of diet itself or these other factors.

Identifying genetic variants that are specific to food preferences and dietary choices has therefore become one of the major challenges in using Mendelian randomization to study diet and health.

A new way to study diet: the biology of taste and smell

A recent study published in BMC Medicine explored whether genes involved in taste and smell could provide more reliable genetic proxies for dietary research.

Rather than searching across the entire genome for variants associated with dietary behaviours, researchers focused on genes involved in taste and smell. This idea is based on a simple premise: people experience the taste and smell of foods differently, and these sensory differences help shape food preferences and dietary choices.

Humans possess hundreds of taste and olfactory receptor genes that influence how foods taste and smell. Variations in these genes can affect food preferences and dietary choices, making them attractive candidates for studying the health effects of diet.

Using data from approximately 160 000 participants in the UK Biobank, researchers identified numerous associations between receptor genes and food preferences. Several findings were replicated in an independent cohort, providing support for their biological relevance.

One example involved onion preference. A variant in the olfactory receptor gene OR2T6 showed a strong association with liking onions, suggesting that differences in how people perceive food aromas may contribute to individual differences in food preferences. Unlike many previously used diet-related variants, this variant showed little evidence of association with common socioeconomic or lifestyle factors.

Using this receptor variant as a genetic proxy for onion preference, the researchers found evidence suggesting that greater onion consumption may reduce blood pressure and lower the risk of type 2 diabetes. While these findings require further confirmation, they illustrate how biologically informed genetic approaches may help improve the quality of evidence in nutrition research.

Building a stronger evidence base for dietary guidelines

The importance of the study lies not in onions themselves, but in the broader principle. By selecting genetic variants based on biological knowledge rather than statistical association alone, researchers may be able to improve our ability to identify foods that genuinely influence health.

As Australia develops its next generation of dietary guidelines, methods that strengthen causal evidence will become increasingly important. Human genetics will not replace clinical trials or traditional epidemiology, but it may provide a valuable additional tool for understanding the health effects of foods. Ultimately, the goal is to ensure that nutrition recommendations and public health policy are guided by the most reliable evidence possible.

Dr Daniel Liang-Dar Hwang is a Senior Research Fellow at the University of Queensland specialising in sensory nutrition and human genetics.