Predictive factors for the development of diabetes in cancer patients treated with phosphatidylinositol 3-kinase inhibitors

Abstract

PURPOSE: Targeted therapy using phosphatidylinositol 3-kinase (PI3K) inhibitors is used to treat cancer such as lymphoma. In animal studies, its use raised concern about alteration of glucose metabolism. To date, clinical data are inconclusive; therefore, we investigated the incidence and clinical manifestations of diabetes in cancer patients treated with PI3K inhibitors.

METHODS: In a retrospective review of diabetes-free patients with advanced solid tumors treated with PI3K inhibitor, we performed Cox regression to identify independent predictors for the development of diabetes.

RESULTS: Of 38 patients (mean age: 54.5 years, 23.7% female) having a mean duration of follow-up of 238.5 days who initiated PI3K inhibitors, 55.3% developed diabetes during treatment (mean 29.1 days); among these, 28.6% experienced remission of diabetes after discontinuing PI3K inhibitors (mean 72.1 days). Patients with incident diabetes had higher anti-hypertensive medication use, higher HbA1c levels and fasting glucose at baseline, and longer duration of PI3K inhibitor use (P = 0.024, P = 0.005, P = 0.008, and P = 0.023, respectively). Previous steroid use and lower baseline HbA1c level were significantly associated with development of diabetes (HR = 8.41, 95% CI 1.89-37.33; HR = 2.16, 95% CI 1.09-4.25, respectively). Patients whose diabetes remitted after discontinuing PI3K inhibitors were younger (P = 0.035) and had lower fasting glucose levels during PI3K inhibitor treatment (P = 0.001) compared to those non-remitters.

CONCLUSIONS: Previous steroid use and lower baseline HbA1c level may be important predictors for developing diabetes in patients with advanced solid tumors treated with PI3K inhibitors, warranting close observation and careful intervention.