“My dad had diabetes, so I suppose it’s only a matter of time before I get it too.”
It’s a common assumption, and an understandable one. Family history is one of the strongest risk factors for type 2 diabetes. In a large American study that followed families for decades, having one parent with type 2 diabetes roughly tripled a person’s risk, and having two affected parents increased it about sixfold [1].
However, families share diet, activity levels, sleep habits and socioeconomic circumstances as well as genes. So is it the genes that cause the diabetes or is it the shared (unhealthy) habits?
In my previous article I covered insulin resistance, ectopic fat and the legacy effect. This time I want to look at what is actually inherited, what those genes do, and how much of it is yours to change.
Different types, different genetics
All diabetes is defined by high blood glucose, and that is roughly where the similarities end.
Type 1 is autoimmune and accounts for roughly one in ten cases of diabetes in Australia [2]. The immune system destroys the beta cells that make insulin, most of the risk genes regulate immunity, and lifestyle plays no part.
Type 2, which accounts for about 85–90% of cases in Australia [2], develops when the pancreas can no longer make enough insulin to overcome the body’s resistance to it. Risk is spread across hundreds of genetic variants, each nudging the dial a little, and environment largely decides whether that risk is ever realised.
Then there is the one most people have never heard of. Monogenic diabetes, historically called MODY, comes from a single faulty gene and is usually passed down directly, so each child of an affected parent has a 50% chance of inheriting it. A 2025 Australian consensus statement estimates it accounts for 2–5% of all diabetes, and that more than 80% of people who have it are undiagnosed or mislabelled as type 1 or type 2 [3].
That matters because the gene often dictates the treatment: some forms need no treatment at all, and others respond to oral treatments in the place of insulin [3].
I suspect monogenic diabetes as the diagnosis when someone presents with diabetes before the age of 30, is lean, has negative pancreatic antibodies and a family tree with diabetes in every generation. Testing isn’t Medicare-funded and usually costs over $1,000, so it’s reserved for people whose picture fits [3]. When it does fit, one result can change treatment for a whole family.
There is no single type 2 diabetes gene
In 2024 researchers compared the DNA of more than 2.5 million people, about 428,000 of whom had type 2 diabetes, and found 1,289 places in the genome where a variant was more common in those with the disease [4]. Each variant shifts risk by a small amount, and most do almost nothing on their own. Their influence becomes more significant when there is a presence of multiple variants together, each adding its own risk, like weights on a scale.
Grouped by their effects on other metabolic traits, these variants fall into eight clusters, each tied to a particular organ or tissue [4]. Some act on the insulin-producing cells of the pancreas and govern how much insulin is made and released. Others work through body weight, through the liver and blood lipids, through blood vessels, through the hormone-producing cells of the gut, or through fat tissue itself. One cluster resembles a mild form of lipodystrophy, in which fat under the skin has limited storage capacity, so the overflow ends up in the liver and muscle. This is why the clusters matter. A list of 1,289 variants says little on its own; knowing that a person’s risk sits mainly in the pancreas, or mainly in fat storage, points to the organ under strain and to where prevention should be aimed.
This is why two people with identical HbA1c results can have arrived there by different roads. For one, the pancreas simply can’t make enough insulin. For another, the body can’t store fat safely.
And you don’t need a person’s genome to see this split. A Swedish study sorted nearly 9,000 newly diagnosed adults into subgroups using six ordinary clinic measurements, and the subgroups went on to different complications: the insulin-deficient group had the highest rate of eye disease, the insulin-resistant group a markedly higher rate of kidney disease [5]. Type 2 diabetes is several related conditions wearing one name and recognising which one a person has, is the first step toward precision treatment.
Why the pancreas matters
Insulin resistance alone doesn’t cause diabetes, and plenty of people live with it for years while the pancreas works overtime to keep glucose normal. Diabetes is diagnosed when the beta cells can no longer keep up.
How much capacity those cells have is partly inherited, which is why two people with similar insulin resistance can end up in very different places. The best-studied example is the gene, TCF7L2, first linked to type 2 diabetes in 2006 and still the strongest common genetic association we know of. One copy of the risk variant raised the odds of diabetes by about 45% in the original study; two copies more than doubled them [6]. The variant acts mainly by reducing insulin secretion.
What genes decide is how much beta-cell capacity you start with, but after that it’s wear and tear. Ageing, years of high glucose and fat building up in and around the pancreas all erode it, and two people who start with the same capacity can lose it at very different rates. Most of what sets that rate is metabolic health, which is the part we can treat.
Why are South Asian populations at greater risk?
People of South Asian ancestry tend to develop type 2 diabetes younger and at a lower BMI than people of European ancestry, and at the same BMI typically carry more abdominal and liver fat and less muscle [7].
Part of that is genetic. A 2025 study of around 50,000 British adults of Pakistani and Bangladeshi ancestry broke genetic risk down by biological pathway. Those with high genetic risk for both reduced insulin production and impaired fat storage were diagnosed 8.2 years earlier, and at a BMI about 3 kg/m² lower, than those at low genetic risk. They also progressed faster to insulin and to small-vessel complications [8].
This is one reason I’m wary of BMI on its own. A DEXA scan estimates visceral fat and muscle mass in a way scales never will, provided the result is read alongside the rest of the clinical picture. Early-life nutrition, diet, physical activity and socioeconomic circumstances contribute as well [7].
Does high genetic risk mean diabetes is inevitable?
No, and intervention makes a meaningful difference. The Diabetes Prevention Program, a large American trial I covered in my previous article, took more than 3,000 adults with prediabetes and assigned some of them to a structured program of diet, exercise and weight loss. Over about three years, that group developed type 2 diabetes at a 58% lower rate than those given standard advice [9]. The obvious next question was whether that protection extended to people who had inherited a high genetic risk.
So the researchers went back to 823 people from the lifestyle group and gave each a genetic risk score built from 67 diabetes-related variants. Within the lifestyle group, those who met all three of the program’s targets (at least 7% weight loss, 150 minutes of moderate activity a week and a low dietary fat goal) developed less diabetes at every level of genetic risk. Among people in the lowest genetic-risk group, meeting the targets meant 1.8 fewer cases of diabetes for every 100 people followed for a year. Among those in the highest genetic-risk group, it meant 3.9 fewer [10].
What this secondary analysis shows is that inheriting genetic risk is no reason to give up and making lifestyle changes makes a significant difference to your risk.
Can genetic testing predict your risk?
A polygenic risk score adds up the effects of thousands of variants into a single estimate of inherited susceptibility. Since DNA doesn’t change, it can be measured at any age, potentially decades before glucose starts to drift.
The early results for polygenic risk scores are promising. A large UK study found that 3.5% of the population carried more than three times the average risk of type 2 diabetes, which is comparable to some monogenic conditions [11]. An Australian study of 12,000 people aged 70 and over found that adding a polygenic score to standard clinical risk factors modestly improved prediction of new diabetes over five years [12].
Genetic testing and polygenic scores have limitations however. These scores estimate relative risk, so three times a small number is still a small number, and age, waist circumference, HbA1c and blood pressure remain the backbone of any assessment. Accuracy varies by ancestry, because most of the foundational research was done in people of European descent, though newer multi-ancestry studies are closing the gap [4]. And knowing your risk doesn’t reliably change what you do; in a randomised trial, telling people their genetic risk of diabetes didn’t increase their enrolment in a prevention program compared with standard counselling [13].
That last one is important; a genetic test is only worth having if it changes what happens next.
The next decade
Three developments suggest genetics is moving from describing risk to shaping care:
- Advanced genetic scoring:
- A standard polygenic score adds everything into one number, so two people can have the same score for completely different reasons.
- The newer approach, used in the South Asian study above, keeps the parts separate: one score for the genes that affect insulin production, another for the genes that affect how fat is stored, and so on [8].
- That shows where a particular person’s pathology is.
- Someone whose risk sits mostly in the pancreas needs their beta cells protected, with early attention to glucose control.
- Someone whose risk sits in fat storage needs visceral and liver fat brought down aggressively.
- Still a research tool, though maturing quickly.
- Treatment matched to mechanism:
- The type 2 diabetes subgroups and genetic clusters carry different risks of kidney and cardiovascular disease [4, 5].
- Monogenic diabetes already shows what a genetic result can do to a prescription.
- The same approach could eventually reach a far larger population.
- A change in Australian law: Since 8 October 2026, life insurers have been barred from requesting or using predictive genetic test results when underwriting death, TPD, trauma or income protection cover on new or varied policies [14].
Nature vs nurture
In my own practice, a family history of type 2 diabetes changes four things:
- Test earlier and track the trend. HbA1c and fasting glucose are cheap, and a trajectory over several years tells you far more than any single result.
- Look past BMI. Waist circumference and, where appropriate, body composition add information weight alone can’t, especially for people of South Asian ancestry.
- Think about monogenic diabetes when several generations were diagnosed young, particularly if they were lean.
- Read any polygenic score together with your clinical risk factors.
Back to the question I started with. Is it the genes or the shared habits? Having gone through the research, I’d put it this way. Your genes decide how much insulin your pancreas can make and how much fat your body can store without harm. Your habits decide how hard those limits get tested, and how soon. That is why some people with diabetes on both sides of the family never develop it, and why the people who gained most from the Diabetes Prevention Program were the ones carrying the most genetic risk. In addition to this, genetic scores are getting better, and the insurance barrier that kept many Australians from being tested has largely gone. The polygenic scores are continuing to be researched and are likely to play a bigger role in diabetes management and even potential screening in the future.
If your father had diabetes, your risk is real and worth taking seriously. It is also, for most people, a risk that responds to what you do about it.