Fragility fractures are an important hallmark of aging and an increasingly recognized complication of Type 2 diabetes (T2D). T2D individuals have been found to exhibit an increased fracture risk despite elevated bone mineral density (BMD) by dual x-ray absorptiometry (DXA). However, BMD and FRAX-scores tend to underestimate fracture risk in T2D. New, reliable biomarkers are therefore needed. MicroRNAs (miRNAs) are secreted into the circulation from cells of various tissues proportional to local disease severity. Serum miRNA-classifiers were recently found to discriminate T2D women with and without prevalent fragility fractures with high specificity and sensitivity (AUC > 0.90). However, the association of circulating miRNAs with incident fractures in T2D has not been examined yet.

In 168 T2D postmenopausal women in the AGES-Reykjavik cohort, miRNAs were extracted from baseline serum and a panel of 10 circulating miRNAs known to be involved in diabetic bone disease and aging was quantified by qPCR and Ct-values extracted. Unadjusted and adjusted Cox proportional hazard models assessed the associations between serum miRNAs and incident fragility fracture. Additionally, Receiver operating curve (ROC) analyses were performed.

Of the included 168 T2D postmenopausal women who were on average 77.2 ± 5.6 years old, 70 experienced at least one incident fragility fracture during the mean follow-up of 5.8 ± 2.7 years. We found that 3 serum miRNAs were significantly associated with incident diabetic fragility fracture:​ while low expression of miR-19b-1-5p was associated with significantly lower risk of incident fragility fracture (HR 0.84 (95% CI:​ 0.71–0.99, p = 0.0323)), low expression of miR-203a and miR-31-5p was each significantly associated with a higher risk of incident fragility fracture per unit increase in Ct-value (miR-203a:​ HR 1.29 (95% CI:​ 1.12–1.49), p = 0.0004, miR-31-5p HR 1.27 (95% CI:​ 1.06–1.52), p = 0.009). Hazard ratios of the latter two miRNAs remained significant after adjustments for age, body mass index (BMI), areal bone mineral density (aBMD), clinical FRAX or FRAXaBMD. Women with miR-203a and miR-31-5p serum levels in the lowest expression quartiles exhibited a 2.4–3.4-fold larger fracture risk than women with miR-31-5p and miR-203a serum expressions in the highest expression quartile (0.002 ≤ p ≤ 0.039). Women with both miR-203a and miR-31-5p serum levels below the median had a significantly increased fracture risk (Unadjusted HR 3.26 (95% CI:​ 1.57–6.78, p = 0.001) compared to those with both expression levels above the median, stable to adjustments. We next built a diabetic fragility signature consisting of the 3 miRNAs that showed the largest associations with incident fracture (miR-203a, miR-31-5p, miR-19b-1-5p). This 3-miRNA signature showed with an AUC of 0.722 comparable diagnostic accuracy in identifying incident fractures to any of the clinical parameters such as aBMD, Clinical FRAX or FRAXaBMD alone. When the 3 miRNAs were combined with aBMD, this combined 4-feature signature performed with an AUC of 0.756 (95% CI:​ 0.680, 0.823) significantly better than aBMD alone (AUC 0.666, 95% CI:​ 0.585, 0.741) (p = 0.009).

Our data indicate that specific serum microRNAs including senescent miR-31-5p are associated with incident fragility fracture in older diabetic women and can significantly improve fracture risk prediction in diabetics when combined with aBMD measurements of the femoral neck.