Type 2 diabetes

Type 2 diabetes is a common condition of high blood sugar from reduced insulin action. This score sums common variants that shift your genetic risk.

What this type 2 diabetes score reads

This is PGS000125, assembled by Qi and colleagues for a 2017 Diabetes study of type 2 diabetes genetics in US Hispanic and Latino adults. It counts 80 variants that had reached genome-wide significance for type 2 diabetes, drawn from GWAS data covering 169,298 people with an unusually mixed makeup for a score of this era: about 41% European, 27% East Asian, 14% African, 12% South Asian and 6% Hispanic or Latino ancestry.

One detail is worth knowing. Every variant in this score carries the same weight, so it is a straight count of risk alleles rather than a weighted sum. That makes it simple and slightly blunt, since variants with genuinely larger effects, such as the TCF7L2 variant, count no more than the smallest.

How well it predicts type 2 diabetes

In the study's own Hispanic and Latino cohort (7,746 people, 2,499 with diabetes), each additional risk allele carried odds of 1.07, so the score separates groups rather than individuals. In a Middle Eastern cohort it reached an area under the curve of 0.646 for prediabetes and 0.761 for incident diabetes among people who already had prediabetes.

Read those last two numbers carefully: both models included age, sex and BMI alongside the score, and those three carry most of the discrimination. The genetic contribution on top of them is real but small.

How to read a high or low score

A high percentile means you carry more of these 80 risk alleles than most of the reference group. It is not a diagnosis, it is not a prediction of when, and it does not override the simple measurements that actually define diabetes: fasting glucose, HbA1c or an oral glucose tolerance test.

It is also not a verdict. In the Diabetes Prevention Program, an intensive lifestyle programme aiming for 7% weight loss and 150 minutes of activity a week cut progression from prediabetes to diabetes by 58% over about three years, and metformin cut it by 31%. That trial was not selected on genetics, and its effect sizes dwarf the spread this score describes.

What this score does not capture

Monogenic diabetes is invisible here. MODY, caused by single variants in genes such as HNF1A, HNF4A and GCK, accounts for a small share of diabetes but changes treatment substantially, and it is diagnosed by sequencing, not by a polygenic score. Diabetes diagnosed young, in a lean person, across several generations is the pattern that should prompt that question with a clinician.

Nor does it capture the things that dominate ordinary type 2 diabetes risk: body weight and where it sits, physical activity, diet, sleep, age, and gestational diabetes history.

Ancestry and accuracy

This score is better placed than most for cross-population use, since its source GWAS was closer to balanced than the European-heavy studies behind most scores, and it was validated in Hispanic and Latino and in Middle Eastern cohorts. That is a relative strength, not an exemption: percentiles still depend on which reference population you are compared against.

Variants in this score you can read on their own

This score includes variants that this site also explains individually, where you can see what each genotype means on its own: Fasting glucose / type 2 diabetes (MTNR1B) (rs10830963) · Type 2 diabetes protection (PPARG Pro12Ala) (rs1801282) · Type 2 diabetes risk (rs7903146).

How much of this score your file covers

All 80 variants in the published score are present in a standard consumer DNA file, so this reconstructs the published score in full.

Common questions

Can I prevent type 2 diabetes if my polygenic score is high? Genetic risk and lifestyle act largely independently, and the Diabetes Prevention Program showed a 58% reduction in progression from prediabetes to diabetes with intensive lifestyle change. A high score is a reason to keep an eye on fasting glucose or HbA1c, not a sentence.

Is a low score reassuring? Only mildly. The American College of Medical Genetics puts it directly: a low polygenic score does not rule out significant risk. Most people who develop type 2 diabetes do not have extreme genetic scores, and weight, activity and age carry more of the risk than these 80 variants.

What is the difference between this and the TCF7L2 result? TCF7L2 is the single largest common variant for type 2 diabetes and it has its own page here, where you can see what your genotype means. This score counts it as one of 80, so the two answer different questions: one variant in detail, or the whole common-variant tendency in aggregate.

Related

Asthma · Breast cancer (female) · Crohn's disease susceptibility · Fasting glucose / type 2 diabetes (MTNR1B) · HbA1c (blood sugar) · Longevity-associated variant (FOXO3) · Type 2 diabetes risk