Letter to the Editor: “Checkpoint Inhibitor-Associated Autoimmune Diabetes Is Distinct From Type 1 Diabetes”

We read with great interest the article by Tsang et al (1) titled “Checkpoint Inhibitor-Associated Autoimmune Diabetes Is Distinct From Type 1 Diabetes,” which is a retrospective cohort study of 538 melanoma patients treated with antiprogrammed cell death-1 (PD-1) either alone or in combination with ipilimumab, of which 10 patients (1.9%) developed new onset insulin-requiring diabetes. These patients had a decrease in c-peptide from baseline and glycemic variability, suggesting a rapid development of insulin deficiency. This article is written on a very pertinent topic with various strengths, including performing continuous glucose monitoring, including a comparator group of type 1 diabetes (T1DM) patients with varying disease duration, testing for c-peptide in 6 of 10 patients at baseline, and HLA class II loci testing, thus adding significantly to the limited available literature on this topic.

However, there are some important questions that remain unanswered either due to the sample size or because they have not been considered in this study. The authors mention that to date there are no large published series that have examined the characteristics of checkpoint inhibitor-associated diabetes mellitus (CIADM). However, they did not cite pertinent cohort studies by Stamatouli et al (2) or our study (3) (which evaluated the largest cohort of CI-treated patients with various malignancies to identify CIADM), which were also included in recent reviews (4–6). The follow-up duration after the initiation of the checkpoint inhibitor to identify the occurrence of CIADM was also unclear. In our retrospective cohort study of 1444 patients with different malignancies treated with a CI, of which 1163 were treated with a PD-1 inhibitor, 12 (0.8%) developed new onset insulin-dependent DM and 9 (0.6%) had a worsening of pre-existing type 2 DM (T2DM), with DM occurring most frequently with pembrolizumab (2.2%) compared to nivolumab (1%) and ipilimumab (0%) (3).

In the study by Tsang et al (1), 1 out of 10 patients had T2DM, but the presence of T2DM or prediabetes in those without CIADM is not mentioned. In our study, 9 of 96 patients (9.3%) with pre-existing T2DM had a worsening of glycemic control defined as initiation or intensification of insulin therapy after starting a PD-1 inhibitor. Similar to other cohort studies (2, 3) on this topic, the current study also lacks testing of autoantibodies at baseline, thus failing to provide a definitive role of autoantibodies as a marker for CIADM. In this study, 2 of 10 patients (20%) had a positive autoantibody at the time of diagnosis, which is slightly lower than the 50–70% reported in other series. Also, the small sample size of this study precludes a definite conclusion of association of HLA with CIADM.

Overall, the presentation and course of CIADM shares some similarities with T1DM, but it appears to cause more rapid beta cell destruction. The occurrence of acute hyperglycemia or diabetic ketoacidosis in most patients in this study (1) as well as others (2, 3) highlights the importance of a heightened suspicion for the development of diabetes in patients receiving PD-1/PD-L1 inhibitors specifically and developing better risk prediction.

Additional Information:

Disclosure Summary: The authors have nothing to disclose.

References