Diabetes duration modifies the association between serum osteocalcin and lumbar spine T-score in adults with type 2...

Type 2 diabetes mellitus (T2DM) is associated with increased fracture risk, but changes in bone mineral density (BMD) are site-specific and not always concordant with this risk. Osteocalcin (OCN) reflects bone turnover, yet its relationship with BMD in T2DM is heterogeneous. We examined whether diabetes duration modifies the association between OCN and BMD, with emphasis on the lumbar spine in community-dwelling older adults.

We performed a community-based cross-sectional analysis (2017-2018) of 170 adults with T2DM and 170 non-diabetic controls. Diabetes status and duration were ascertained from community health service center medical records and corroborated by participant/family report. Non-diabetic controls were used for descriptive between-group comparisons and control-only models without diabetes-duration terms; duration-interaction analyses were restricted to the T2DM group. For the duration-interaction analyses, 144 participants with T2DM had available duration data (mean age 65.0 ± 7.6 years; 31.9% male). For descriptive presentation, diabetes duration was grouped as ≤ 5, 6-10, and > 10 years. Lumbar spine osteoporosis was defined as lumbar spine (L1-L4) T-score ≤  - 2.5, using World Health Organization (WHO) criteria.

In participants with T2DM (n = 170), the mean lumbar spine T-score was lower than in non-diabetic controls (mean difference - 0.41; P = 0.015), whereas hip T-score was similar (P = 0.180). Within T2DM participants with available duration (n = 144), multivariable linear models showed a significant interaction between z-standardized ln(OCN) and diabetes duration on lumbar spine T-score (β_interaction = 0.049, 95% confidence interval (CI) 0.012-0.086; P = 0.009), indicating attenuation of the inverse OCN-BMD association with longer duration. Simple slopes were - 0.70 at 2 years, - 0.35 at 9 years, and 0.18 at 20 years. On the probability scale, adjusted marginal probabilities of lumbar osteoporosis across osteocalcin quartiles and duration strata are reported descriptively; the OCN × duration interaction term was not statistically significant in logistic models.

Older community-dwelling adults with T2DM showed lower lumbar spine T-scores compared with non-diabetic controls, while hip T-scores were similar. Within T2DM, we observed an OCN × duration interaction for lumbar spine T-score, with a stronger inverse association at shorter duration and attenuation at longer duration. These cross-sectional findings warrant confirmation in longitudinal studies with fracture outcomes.