Genetics and obesity: what FTO studies say about weight loss

Genes can influence body size without deciding your response to weight loss. FTO and polygenic score studies help separate risk, change, and maintenance.

日本語で読む（Japanese version）

You look through old family photographs and notice that your waist now resembles your mother’s. Back home, the bathroom scale seems to confirm a conclusion you reached before changing anything at dinner: weight gain runs in the family. The scale measures weight. Somewhere along the way, it has also become a reader of your family tree.

Genetic influences on obesity are real, but they do not automatically mean that changing food or activity cannot change weight. An analysis combining randomized trials in adults found no significant difference between FTO genotypes in response to weight loss interventions.[13] A possible explanation for why weight has increased is not, by itself, evidence that support will be pointless.

The useful questions are more specific: what predicts a higher body weight, what happens after a dietary change, and what happens when someone tries to maintain that change? Those are different stages with different outcomes. The related feature on weight stigma addresses self-criticism; here, the focus is on how far genetic research can answer practical questions about eating, movement, and continuing support.

Genetic influence is real, but it is not a weight forecast

A simple choice between genes and environment does not describe obesity well. A review of FTO research presents obesity as a multifactorial condition involving genetic susceptibility, environmental exposures, and behavior. FTO variants have been associated with body size, fat accumulation, and appetite regulation.[10] When people describe themselves as having a tendency to gain weight, the relevant biology may include appetite as well as appearance. That is a broader idea than a family resemblance in a photograph.

A genetic variant is a difference in part of the genetic information between people. Researchers compare groups carrying different variants and examine their body measurements or changes after an intervention.[10][13] The research question determines what a result means. A study of FTO may ask whether a particular genotype changes the response to an intervention. A study combining many variants may ask how well a score predicts differences in BMI, or body mass index.[8][13] Those questions can produce different answers without contradicting each other.

A polygenic score brings many genetic differences together into a numerical measure of their association with a trait. In a large study, researchers used genetic data from up to 5.1 million people to develop BMI scores. They created both ancestry-specific scores and a score combining information from multiple ancestry groups.[8] The scale of the dataset matters, but it does not remove the need to ask where the resulting prediction was tested.

The multi-ancestry score explained 17.6% of variation in BMI among UK participants of European ancestry. Its explanatory power was 16% among East Asian Americans and 2.2% among people in rural Uganda.[8] These percentages describe how much of the differences within the studied populations the score captured. They do not allocate a fixed percentage of an individual’s weight to genes. For readers in Tokyo with different family backgrounds, the population attached to a prediction is part of the finding, rather than a detail to skip.

The score’s explanatory power also needs to be distinguished from heritability. The 17.6% result is the proportion of BMI variation explained by that particular score in that population; it is not a measurement of heritability itself.[8] Nor does it say how much weight a person can lose. Before putting a percentage into your own story, identify its denominator and its outcome: differences in BMI across a group are not changes in weight after treatment.

The same study suggests a possible role for prediction in planning earlier support. Children with higher scores had faster BMI gain from age 2.5 years to adolescence and earlier adiposity rebound, the point at which body fat begins increasing again.[8] Adding the score to information available at birth increased the proportion of BMI variation explained at age 8 from 11% to 21%. Adding it to early-life BMI information improved prediction of BMI at age 18; with information at age 5, the explained proportion increased from 22% to 35%.[8] The age when information is collected and the age being predicted both shape the result.

These findings concern prediction over the life course. They cannot be substituted for evidence about how a child or adult will respond to support. A score may help researchers identify a pattern earlier while still leaving substantial uncertainty about the individual. Keeping that distinction visible prevents a useful research tool from becoming an imagined timetable for your body.

FTO risk variants do not mean an inability to lose weight

FTO is a gene repeatedly linked with obesity. The review discusses possible connections with energy intake, food preferences, and metabolic pathways.[10] But a plausible biological mechanism does not establish the amount of weight an adult with a particular genotype will lose after changing their diet. The question has to move from the development of higher weight to the response after an intervention begins.

An individual participant data meta-analysis addressed that question using 8 randomized controlled trials involving 9,563 adults with overweight or obesity.[13] Randomized trials allocate participants to intervention or comparison groups at random. An individual participant data analysis brings together information about each participant, rather than simply comparing published averages. This analysis used a common approach to adjustment across the contributing studies.[13] How the comparison was made is as relevant as the number of papers included.

Changes in BMI, body weight, and waist circumference after dietary, physical activity, or drug-based weight loss interventions were not significantly different between FTO genotypes. Analyses by intervention type or length, sex, age, and other participant or study characteristics did not provide evidence that this conclusion changed.[13] No specific body-weight difference or confidence interval was reported for the overall comparison.[13]

No significant difference does not mean everyone lost exactly the same amount. It means the combined analysis did not find evidence for dividing intervention response according to the FTO genotype studied.[13] There can be variation between individuals even when a proposed genetic explanation does not distinguish the groups. The researchers assessed several body measurements, including waist circumference, so the finding was not limited to a single reading on a scale.

Another meta-analysis reported a small difference favoring one genotype. Across 10 studies involving 6,951 adults, people with the AA genotype lost an average of 0.44 kg more than those with TT. The 95% confidence interval was 0.09 to 0.79 kg. For TA compared with TT, the difference was 0.18 kg, with a confidence interval of −0.09 to 0.45 kg, and was not statistically significant.[3] AA, TA, and TT are the genotype labels used in these comparisons. A genotype associated with susceptibility to obesity was not necessarily associated with less weight loss.

For the AA–TT comparison, read the average difference of 0.44 kg alongside the 0.09–0.79 kg confidence interval.[3] It is also a small average difference, rather than a personal prediction. The authors said that clinical applications required further investigation.[3] A precise decimal does not make a genetic test a reliable forecast of the result of your next dietary change.

The same paper combined TA and AA in a comparison of 14 studies with 7,700 participants. It described 0.20 kg greater weight loss than in TT, reported a confidence interval of −0.43 to 0.04 kg, and gave a p value of 0.10; the result was not significant.[3]

A 2024 meta-analysis combined 30 studies with 46,976 adults with overweight or obesity. It found weight reduction after diet and exercise interventions among carriers of the FTO risk allele. The standardized mean difference was −0.619, with a 95% confidence interval of −1.137 to −0.100.[1] A standardized mean difference puts results onto a common comparison scale; it is not a number of kilograms to expect personally.

Agreement across the included studies also varied. For TA versus TT, the between-study heterogeneity measure was high at 91.12%. For AA versus TT, the standardized mean difference was −0.148, with a 95% confidence interval of −0.282 to −0.014, and heterogeneity was 24.96%.[1] The genotype comparison therefore affected both the estimated difference and the consistency of the evidence. An average can conceal quite different patterns across studies.

These analyses do not fully agree on every small genotype difference or its direction. Their common message is that adults carrying FTO risk variants can lose weight after interventions involving food and activity.[1][3][13] Whether an intervention can produce change and whether a genetic variant predicts a small difference in that change are separate questions. Reading them separately gives a clearer account than labeling a gene either helpful or harmful for dieting.

When findings disagree, examine the people and the comparison

There is less favorable evidence too. A pilot study followed 18 adults with overweight or obesity who had type 2 diabetes or impaired blood glucose regulation. Participants carrying the FTO A allele lost less weight than TT carriers and had a smaller BMI reduction from baseline to 12 months.[2] The weight difference was not reported in kilograms.[2] The result belongs in the discussion even though it points in a different direction from the larger analyses.

A small study in people with particular health conditions cannot establish that the same outcome will occur in all adults. The authors describe these findings as preliminary.[2] That does not make the observations irrelevant. It means the sample, health status, and intervention belong beside the conclusion. If the result sounds personally relevant, the first task is to identify how closely its participants and support resemble your own circumstances.

The comparisons also differ across the evidence. The individual participant data analysis tested genotype differences in intervention response across randomized trials.[13] The pilot study examined a particular nutritional and lifestyle program in a small clinical group.[2] Other analyses pooled studies using different genotype groupings.[1][3] They are not interchangeable experiments. Selecting whichever result offers the preferred answer would conceal the very uncertainty the research is trying to resolve.

A useful way to read these papers is to keep the outcome visible throughout: body weight, BMI, and waist circumference are related measurements, but the reported change is attached to one of them. Likewise, a within-group reduction and a difference between genotype groups answer different questions. The larger body of evidence supports continuing to consider intervention, while leaving room for variation in response and for further investigation of particular settings.[1][2][3][13]

Numbers: genetics and weight loss

9,563 adults: an analysis of 8 randomized controlled trials found no significant FTO genotype difference in weight loss response.[13]

0.44 kg: the average additional loss for AA versus TT in another adult meta-analysis, with a 95% confidence interval of 0.09 to 0.79 kg.[3]

17.6%: the share of BMI variation explained by a polygenic score among European-ancestry participants, rather than a personal allocation of body weight.[8]

223 participants: a dietary intervention trial compared adults with high and low genetic scores to examine short-term response.[11]

0.20–0.28 cm: the equivalent waist difference at average weight loss in a genetic score analysis; the authors considered it too small to be clinically significant.[18]