Prediction of response to GLP-1 receptor agonist therapy in Japanese patients with type 2 diabetes

BACKGROUND: Glucagon-like peptide-1 (GLP-1) receptor agonists can maintain good glycemic control in some diabetic. Here we compared the clinical characteristics and parameters reflecting glucose metabolism at the time of the initiation of GLP-1 receptor agonist therapy between patients who responded well to therapy and those who did not. METHODS: The records of 43 patients with type 2 diabetes who started receiving GLP-1 receptor agonist therapy during hospitalization were retrospectively reviewed. Glucagon stimulation tests were performed, and patients were started on liraglutide or exenatide therapy. Preprandial blood glucose levels were measured on days 2 and 3 of GLP-1 receptor agonist therapy. We used the Cox proportional hazard model to compare clinical parameters between responders (HbA1c level <8% at more than 3 months after the initiation of treatment) and non-responders (HbA1c level ≥8% at more than 3 months after the initiation of treatment or a switch to insulin therapy at any time). RESULTS: Twenty-six of the 43 patients were classified as non-responders. At baseline, mean HbA1c levels were 9.9% among responders and 9.7% among non-responders. Compared with treatment with only diet or metformin, the hazard ratio [HR] for non-response was 5.3 (95% confidence interval [CI]: 1.16-24.6, P = 0.03) for insulin therapy and 5.0 (95% CI: 1.13-22.16, P = 0.03) for sulfonylurea therapy. Compared with the lowest tertile, the HRs for non-response in the highest tertile were 3.1 (95% CI: 1.04-8.97, P = 0.04) for the mean preprandial blood glucose level on days 2 and 3 and 3.4 (95% CI: 1.05-11.01, P = 0.04) for the body mass index. The response was not significantly associated with the duration of diabetes or the glucagon stimulation test results. A receiver operating curve analysis showed that the mean preprandial blood glucose level had the highest area under the curve value (=0.72) for the prediction of non-responders. CONCLUSIONS: In patients with poorly controlled diabetes, the response to GLP-1 receptor agonist therapy was significantly associated with the treatment used before the initiation of therapy, the body mass index, and the mean preprandial blood glucose level during the 2 days after the initiation of therapy.