Low HDL cholesterol flags diabetes risk in Japanese men with prediabetes

High-density lipoprotein cholesterol, often called HDL-C, shows a strikingly different relationship with new-onset type 2 diabetes depending on a person’s starting glycaemic status, according to a population-based longitudinal cohort study published in BMC Endocrine Disorders. Among Japanese men with prediabetes, the researchers identified an L-shaped association between baseline HDL-C and incident diabetes, with a threshold concentration of 32.4 mg/dL. Below that threshold, each 10 mg/dL increment in HDL-C was associated with a nearly 62% lower hazard of developing type 2 diabetes, with a hazard ratio of 0.384 and a 95% confidence interval of 0.196 to 0.752. Above the threshold, no significant association was observed, with a hazard ratio of 0.947 and a 95% confidence interval of 0.862 to 1.042.

The study drew on the NAGALA cohort, enrolling 10,120 men between 5 January 2004 and 26 December 2015. Of these, 6,791 had normoglycemia and 3,329 had prediabetes at baseline. The researchers deliberately separated these two groups, because they suspected that pooling participants with different glycaemic status might explain why earlier studies of HDL-C and diabetes have produced inconsistent findings. During follow-up, 88 normoglycemic participants and 494 prediabetic participants developed type 2 diabetes. The median follow-up was 5.95 years for participants with normoglycemia and 4.33 years for those with prediabetes.

Statistically, the team used Cox proportional hazards models to examine the association between baseline HDL-C levels and incident type 2 diabetes. To evaluate whether the effect of HDL-C concentration had a threshold, they applied a two-piecewise linear regression model with a smoothing function. In both the crude model and the partially adjusted model, the risk of type 2 diabetes decreased significantly with increasing baseline HDL-C in both groups.

The threshold effect itself emerged only in the prediabetic group. In participants with normoglycemia, no significant threshold effect was detected, and the fully adjusted analysis found no significant association between baseline HDL-C and incident diabetes in that group either. This contrast between the two glycaemic states is the central finding of the study: an L-shaped relationship between baseline HDL-C concentration and the risk of incident type 2 diabetes was observed among Japanese men with prediabetes, while no significant association was detected among those with normoglycemia.

The authors frame their work as a response to a long-standing inconsistency in the literature. Many studies have indicated a negative correlation between HDL-C and the occurrence of diabetes, but some findings have disagreed. The study’s stated aim was to investigate the relationship between baseline HDL-C and incident type 2 diabetes separately among Japanese participants with normoglycemia and those with prediabetes, on the reasoning that the lack of this distinction in previous studies may partially account for the conflicting results.

Several limitations of scope should be kept in view when interpreting the results. The cohort consisted exclusively of Japanese men, so the findings cannot be assumed to extend to women or to other populations. The exposure was a single baseline HDL-C measurement rather than repeated assessments over time, and the observation that associations lost statistical significance after full adjustment indicates that the relationship may be partly explained by other measured confounders. The study reports associations from observational follow-up data, not evidence of a causal effect of raising HDL-C, and the piecewise model’s threshold estimate applies to the prediabetic subgroup within this cohort.

Within those bounds, the results suggest that among men who already have prediabetes, very low HDL-C concentrations mark a state of particularly elevated diabetes risk, and that the apparent protective gradient flattens out once HDL-C rises above roughly 32 mg/dL. For men with normal glucose tolerance, baseline HDL-C in this cohort did not significantly predict who would go on to develop type 2 diabetes. The study was published as open access research on 28 October 2026, and the version shared early is citable under a permanent DOI, subject to further editorial refinement before the final version of record appears.

Subject of Research: The association between baseline HDL cholesterol and incident type 2 diabetes in Japanese men with prediabetes versus normoglycemia.

Article Title: Association between high-density lipoprotein cholesterol and the risk of incident diabetes in the prediabetic and the normoglycemic Japanese men: a population-based longitudinal cohort study

Article References: Xuan, X., Kong, L., Hu, Q., Zhou, L., Zhu, H., Okamura, T., Hashimoto, Y., Obora, A., Kojima, T., Fukui, M., Hamaguchi, M., Liao, J., Zhang, J., Huang, S., He, S., & Xie, X. (2026). Association between high-density lipoprotein cholesterol and the risk of incident diabetes in the prediabetic and the normoglycemic Japanese men: a population-based longitudinal cohort study. BMC Endocrine Disorders. https://doi.org/10.1186/s12902-026-02699-9

Image Credits: AI Generated

DOI: 10.1186/s12902-026-02699-9

Keywords: HDL cholesterol, type 2 diabetes, prediabetes, normoglycemia, cohort study, NAGALA cohort, Japanese men, hazard ratio, threshold effect, epidemiology, BMC Endocrine Disorders, Association