Detailed Time Course of Decline in Serum C-Peptide Levels in Anti–Programmed Cell Death-1 Therapy–Induced Fulminant Type 1 Diabetes

Fulminant type 1 diabetes (FT1D) is a novel subtype of type 1 diabetes (T1D) and is characterized by a relatively low glycated hemoglobin (HbA_1c) level at onset, despite the abrupt occurrence of marked hyperglycemia with ketosis or ketoacidosis (1). Recent studies have shown that anti–programmed cell death (PD)-1/anti–PD-ligand 1 (PD-L1) therapy for malignancies leads to the development of T1D including FT1D (2–4). In typical FT1D, insulin secretion capacity is completely exhausted within ∼7 days after the onset (1); in some cases, a rapid loss of insulin secretion was observed during 2 days prior to onset of diabetic ketoacidosis (DKA) (5). Meanwhile, the detailed time course of insulin secretion capacity in anti–PD-1 therapy–induced FT1D has not yet been reported.

We present a case of an 82-year-old Japanese man with squamous cell cancer of the lung, who had received multiple courses of anti–PD-1 immunotherapy (pembrolizumab) every 3 weeks for ∼12 months. Written informed consent was obtained from the participant. On the day of the 16th course of anti–PD-1 therapy, his casual plasma glucose and HbA_1c levels were 106 mg/dL and 5.8% (40 mmol/mol), respectively. However, a blood test performed on the scheduled day of the 17th course revealed his casual plasma glucose was 432 mg/dL; thus, he was immediately hospitalized without the planned course.

His height was 158.5 cm; weight, 47.0 kg; and BMI, 18.7 kg/m². He had no history of glucose intolerance, no evidence of preceding acute viral infection, and no hyperglycemic symptoms before and during hospitalization. The initial laboratory examination revealed the following: negative urine ketones, casual plasma glucose level 319 mg/dL, HbA_1c level 6.1% (43 mmol/mol), casual serum C-peptide level 2.03 ng/mL, serum creatinine level 0.76 mg/dL, normal levels of serum exocrine pancreatic enzymes, negative islet-associated autoantibodies, arterial pH level 7.404, arterial HCO₃⁻ level 22.7 mmol/L, and serum total ketone body level 304.5 (normal range: ≤130) µmol/L. He carried a homozygote of HLA-DRB1*12:01 allele, which is not associated with autoimmune T1D. An abdominal ultrasonography revealed the absence of swelling in the pancreas. He had no diabetes complications. Intensive insulin therapy was started immediately after admission, and his plasma glucose levels finally improved with 10 units of insulin glargine and 11, 4, and 5 units of insulin aspart before breakfast, lunch, and dinner, respectively.

During hospitalization, fasting and casual serum C-peptide levels rapidly decreased as the days progressed without obvious hypoglycemia, and insulin secretion capacity was completely exhausted in 16 days, presumably not via reduced blood glucose stimuli but via β-cell damage (Fig. 1A and B), suggesting that the pace of reduction in serum C-peptide levels was milder than that in persons with typical FT1D. However, if each serum C-peptide level had been measured under adequate glycemic control, the time course might have shown a different pattern; this is a limitation of this study. After discharge, the patient’s serum C-peptide level remained below 0.01 ng/mL.

According to previous reports, the positivity rate of islet-associated autoantibodies may be relatively lower in Japanese individuals with checkpoint inhibitor–induced diabetes than in Caucasian individuals (4.76% vs. 56%, respectively) (3,6), which is consistent with the negative test results for islet-associated autoantibodies in this case. Moreover, a previous study has shown that anti-GAD antibodies (GADA)-positive T1D may develop in a shorter period of time after the initiation of anti–PD-1/anti–PD-L1 therapy than GADA-negative T1D (6). In this case, T1D developed over a long period of time (54 weeks) after the initiation of anti–PD-1 therapy, which potentially explains the negative test results for islet-associated autoantibodies.

Considering that anti–PD-1 therapy targets T-cell immunoregulatory molecules, a cellular immunity against islet-associated autoantigens may be involved in the development of T1D. However, any cellular immune responses against insulin molecules, including the main epitope of T cells in human T1D, could not be verified in this case. On this point, the HLA in our case, which was not associated with autoimmune T1D, might be related to the negative results, or the use of other islet-associated autoantigens might have obtained different results.

A previous study demonstrated that most patients with T1D caused by pembrolizumab presented with DKA at diagnosis, and all patients with DKA had elevated HbA_1c levels (6.85–10.70% [51–93 mmol/mol]), suggesting a certain period of preexisting hyperglycemic state before diagnosis (7). However, considering that our patient was diagnosed with T1D before the occurrence of hyperglycemic symptoms and DKA, the duration of the preexisting hyperglycemic state before diagnosis might have been relatively short, presumably leading to the low HbA_1c level (6.1% [43 mmol/mol]) at diagnosis.

We believe it could be clinically beneficial for oncologists, who have an opportunity to use anti–PD-1 drugs, to recognize the time course of β-cell exhaustion in this case.