Longitudinal observation of insulin secretory ability before and after the onset of immune checkpoint inhibitor‐induced diabetes mellitus: A report of two cases

Abstract

Immune checkpoint inhibitor‐induced diabetes mellitus is a rare immune‐related adverse event. This report illustrates clinical data and insulin secretory ability before and after the onset of immune checkpoint inhibitor‐induced diabetes.

Immune checkpoint inhibitor‐induced diabetes mellitus is a rare immune‐related adverse event. This report illustrates clinical data and insulin secretory ability before and after the onset of immune checkpoint inhibitor‐induced diabetes.

1. INTRODUCTION

Immune checkpoint inhibitor‐induced diabetes mellitus is a rare immune‐related adverse event. This report illustrates clinical data and insulin secretory ability before and after the onset of immune checkpoint inhibitor‐induced diabetes.

The programmed cell death‐1 (PD‐1) receptor is expressed on the surface of activated T cells and regulatory T cells, and is an immune checkpoint molecule. PD‐1 interacts with two ligands, programmed cell death ligand‐1 (PD‐L1) and PD‐L2 within the tumor microenvironment.¹ Engagement of PD‐1 through PD‐L1 negatively regulates T cell–mediated immune responses. Immune checkpoint inhibitor (ICI) activates T‐cell immunity and suppresses tumor growth.² As the use of ICI therapy is expanding for treating many cancer types,³, ⁴ the increase in immune‐related adverse events (irAEs) has been obvious. It is important for cancer patients to pay careful attention to irAEs. The most common irAEs are rash, hypothyroidism, hepatitis, and gastroenteritis.⁴, ⁵ Although the percentage of ICI‐induced diabetes mellitus (ICI‐induced DM) is rare (<1%), it is a life‐threatening side effect without any means of early detection. Several studies reported that ICI‐induced DM had several characteristics, the first was rapid‐onset of severe hyperglycemia, the second was diabetic ketoacidosis (DKA) or very low to absent C‐peptide levels, and the third was persistent insulin dependence after the onset of diabetes.⁶, ⁷, ⁸ Most cases are caused by anti‐PD‐1 or anti‐PD‐L1 therapy, and ICI‐induced DM by anti‐cytotoxic T‐lymphocyte‐associated protein 4 (CTLA‐4) is rare. It has also been reported that the combination of anti‐PD‐1 antibody and CTLA‐4 inhibitor causes faster onset of ICI‐induced DM than anti‐PD‐1 antibody monotherapy.⁶, ⁷ In non‐obese diabetic mice, anti‐PD1 inhibitor administration rapidly induced diabetes, but anti‐CTLA‐4 inhibitors did not.⁹ Here, we report two cases in which insulin secretion was monitored before and after the onset of ICI‐induced DM in patients treated with nivolumab.

2. CASE REPORTS

2.1. Case 1

A 76‐year‐old woman with gastric cancer had undergone total gastrectomy and pancreatoduodenectomy after preoperative chemotherapy. Although adjuvant chemotherapy was given after surgery, the gastric cancer recurred 2 years after surgery. She then received nivolumab (3 mg/kg) as a third‐line therapy. She had no prior personal or family history of diabetes mellitus. At the time of nivolumab treatment, fasting blood glucose levels were 5.9 mmol/L and HbA1c 6.1%. Her body mass index (BMI) was 17.5 kg/m².

One hundred and 3 days after initial treatment with nivolumab, the patient was admitted to our hospital after 5 days of progressive general fatigue, thirst, and polyuria. She had no fever or other symptoms, suggesting she was free from a focal infection or systemic inflammation. On physical examination, she was alert. Laboratory data at the time of admission revealed hyperglycemia (blood glucose levels of 33.5 mmol/L) with ketosis (arterial blood pH 7.35, ketone body concentration 7620 μmol/L, the concentration of acetoacetate 1019 μmol/L, and the concentration of β‐hydroxybutyrate 5925 μmol/L). HbA1c was 9.0%, and serum C‐peptide levels were 0.05 nmol/L. Her pancreatic enzymes, thyroid hormones, and cholesterol levels were within the normal ranges. Abdominal computed tomography imaging showed porta hepatis and para‐aortic lymph node swelling, but no significant findings in the pancreas. Upon additional blood testing, she was not found to have serum anti‐glutamic acid decarboxylase, anti‐insulinoma‐associated antigen‐2, anti‐zinc transporter antibody 8, or anti‐islet cell antibodies. Her human leukocyte antigen typing was DRB1 *08: 02‐DQB1 *03: 02 and DRB1 *14: 54‐DQB1 *05: 03. She did not have the type 1 DM‐associated genotype. Acute‐onset ICI‐induced DM was diagnosed from her clinical course and laboratory data.

The levels of HbA1c and serum C‐peptide levels were monitored before and after the onset of ICI‐induced DM. HbA1c increased from 6.1 to 9.0% during 103 days of her course. Fasting C‐peptide levels increased from 0.53 to 0.78 nmol/L, fasting blood glucose levels decreased from 5.9 to 5.2 mmol/L (Figure 1), 0–28 days after the start of nivolumab treatment, then serum C‐peptide levels decreased to 0.05 nmol/L at the onset of ICI‐induced DM. After admission, serum C‐peptide levels were depleted (Figure 1). She was initially treated with intravenous fluids and continuous insulin infusion, which was changed to subcutaneous multiple daily injections with a combination of rapid and long‐acting insulins.

FIGURE 1.

Longitudinal course of HbA1c, serum C‐peptide levels, and fasting blood glucose levels. HbA1c increased from 6.1 to 6.5%, the fasting C‐peptide levels increased from 0.53 to 0.78 nmol/L and fasting blood glucose levels decreased from 5.9 to 5.2 mmol/L, 0–28 days after the start of nivolumab (Nivo) treatment, then HbA1c increased to 9.0%, serum C‐peptide levels decreased to 0.05 nmol/L and fasting blood glucose levels increased to 33.5 mmol/L at the onset of immune checkpoint inhibitor‐induced diabetes. After admission, serum C‐peptide levels were depleted (Case 1). HbA1c increased from 5.2 to 5.6%, the fasting C‐peptide levels increased from 0.38 to 0.47 nmol/L and fasting blood glucose levels decreased from 6.7 to 6.2 mmol/L, 0–28 days after nivolumab treatment, then HbA1c increased to 9.5%, serum C‐peptide levels decreased to 0.13 nmol/L and fasting blood glucose levels increased to 36.3 mmol/L at the onset of immune checkpoint inhibitor‐induced diabetes (Case 2). X: at admission. *: random plasma glucose

2.2. Case 2

A 68‐year‐old woman with stage IVA gingival cancer had been receiving tegafur, gimeracil, oteracil potassium therapy, and neoadjuvant chemotherapy. Tegafur, gimeracil, oteracil potassium therapy, and radiation therapy were conducted after surgical resection of the hemimandible; however, gingival cancer spread to cervical lymph nodes and larynx, and she felt difficulty in swallowing. Then, she was treated with nivolumab (3 mg/kg) as a third‐line therapy. At the time of nivolumab treatment, fasting blood glucose levels were 6.7 mmol/L and HbA1c 5.2%. Her BMI was 16.8 kg/m². Three hundred days after the first nivolumab administration, she presented to our hospital complaining of persistent appetite loss and general fatigue for several days. She had acute airway obstruction and cardiopulmonary arrest due to mucous sputum filling in larynx when waiting for a medical examination. Even after cardiopulmonary resuscitation, hypoxic encephalopathy led to prolonged unconsciousness. Blood glucose levels were 36.3 mmol/L, and HbA1c was 9.5%. Serum C‐peptide levels were 0.13 nmol/L, and urinary C‐peptide level excretion was 11.9 μg/day. Her pancreatic enzymes were within the normal range. The ketone body measurements were not performed at onset. Abdominal computed tomography imaging showed no significant findings in the pancreas. Anti‐glutamic acid decarboxylase and anti‐IA‐2 antibodies were negative. Human leukocyte antigen typing showed DRB1*04:03‐DQB1*03:02, DRB1*15:01‐DQB1*06:02, and she did not have the type 1 DM‐associated genotype. We could not evaluate ketosis but diagnosed acute‐onset ICI‐induced DM from her clinical course and laboratory data.

Because she did not recover from the comatose state, a nasogastric tube was inserted for feeding. Multiple daily injections of long‐acting and ultra‐rapid insulin were used to control her blood glucose levels. Although nivolumab was effective for preventing cancer progression, this treatment was discontinued.

The time courses of HbA1c and serum C‐peptide levels were monitored before the onset of ICI‐induced DM. HbA1c increased from 5.2 to 9.5% in the period of 300 days. Fasting C‐peptide levels increased from 0.38 to 0.47 nmol/L, and fasting blood glucose levels decreased from 6.7 to 6.2 mmol/L, 0–28 days after nivolumab treatment, then serum C‐ peptide levels decreased to 0.13 nmol/L at the onset of ICI‐induced DM (Figure 1).

3. DISCUSSION

We report two cases of acute‐onset ICI‐induced DM in which the longitudinal course of fasting C‐peptide levels and corresponding fasting blood glucose levels was assessed before and after the administration of nivolumab therapy. As far as we are aware, most of previous reports have measured blood glucose levels and serum C‐peptide levels after the onset of ICI‐induced DM. Few literatures have reported the longitudinal course of C‐peptide levels and corresponding fasting blood glucose levels before and after the onset of ICI‐induced DM. Our cases showed that insulin secretion increased 1 month after nivolumab administration and remained in the normal range until the onset of ICI‐induced DM. Furthermore, fasting plasma glucose level slightly decreased and the HbA1c level increased in both cases without any change in nutritional status before the onset. Nivolumab therapy also aggravates pre‐existing diabetes with incompletely deficient insulin secretion.¹⁰ It was suggested that mild glucose intolerance may have occurred while insulin secretion was maintained in the normal range before insulin secretion decreased markedly. The contribution of postprandial hyperglycemia is predominant in patients with well‐controlled HbA1c, over fasting hyperglycemia.¹¹ Postprandial glucose levels could not be assessed in our cases; however, we speculate that increased HbA1c and serum C‐peptide levels reflect elevated postprandial blood glucose levels.

Some studies have demonstrated that irAEs are associated with improved survival in melanoma or non‐small cell lung cancer patients treated with nivolumab.¹², ¹³ 33 isolated reports of ICI‐induced DM that included the effectiveness of anti‐PD‐1 antibody therapy can be found in the literature (Table 1). The overall response rate (complete plus partial response) was 52% (5 complete and 12 partial responses in 33 cases). ICI‐induced DM may be a predictive marker for durable clinical benefit in cancer patients, although the mechanisms underlying the association of ICI‐induced DM with the outcome of anti‐PD‐1 antibody therapy are unknown. In our cases, although no recurrence was observed and the response to anti‐PD‐1 antibody therapy was stable disease, anti‐PD‐1 antibody treatment was discontinued because of a patient or family's request.

TABLE 1.

Cases of immune checkpoint inhibitor‐induced diabetes mellitus with outcomes after anti‐PD‐1 antibody therapy

Literature	Clinical tumor response to anti‐PD−1 antibody	Islet‐related autoantibodies	Age/sex	Primary diagnosis	Anti‐PD−1 antibody	Occurrence in days after initiation of Nivo	Plasma Glucose at onset (mmol/L)	HbA1c at onset (%)	HLA
Gauci, M. L., et al. 201714	CR	Positive	73/m	Melanoma	Nivo	42	27.8	8.8	N/A
Lowe, J. R., et al. 201615	CR	N/A	54/m	Melanoma	Ipi, Nivo	133	N/A	N/A	N/A
Hofmann L., et al. ① 201616	CR	Negative	70/f	Melanoma	Nivo	42	N/A	N/A	N/A
Zezza, M., et al. ① 201917	CR	Positive	60/m	Melanoma	Ipi, Nivo	14	46	7.5	N/A
Hong, A. R., et al. ③ 202018	CR	Positive	78/f	Melanoma	Pembro	28	27.4	11.4	N/A
Takagi, T., et al. 201819	PR	N/A	72/m	Renal cell carcinoma	Nivo	182	N/A	N/A	N/A
Miyoshi, Y., et al.201620	PR	Negative	66/f	Melanoma	Nivo	121	29.5	7.3	DRB1*11:01‐DQB1*03:01 DRB1*13:02‐DQB1*06:04
Telo, G. H., et al. 201721	PR	Negative	51/m	Renal cell carcinoma	Ipi, Nivo	45	42.4	7.2	N/A
Hatakeyama, Y., et al. 201822	PR	Negative	60/m	Lung cancer	Nivo	491	23.1	9.1	N/A
Usui, Y., et al. ① 201723	PR	Positive	31/ m	Lung cancer	Nivo	13	41.2	6.4	DRB1*04:05‐DQB1*04:01
Usui, Y., et al. ② 201723	PR	Negative	62/f	Lung cancer	Nivo	42	13.7	6.5	DRB1*09:01‐DQB1*03:03
Chae, Y. K., et al. 201724	PR	Positive	76/m	Lung cancer	Pembro	28	34.2	N/A	N/A
Yamamoto, N., et al. 201925	PR	Negative	77/m	Renal cell carcinoma	Nivo	85	21.0	6.2	DRB1* 09, DRB1*12 DQB1* 03, DQB1* 03
Hong, A. R. et al. ① 202018	PR	Negative	76/m	Lung cancer	Pembro	77	27.4	10.4	N/A
Hong, A. R. et al. ④ 202018	PR	Negative	65/f	Billary cancer	Pembro	147	28.4	5.8	N/A
Yilmaz, M. 202026	PR	Negative	49/m	Renall cell carcinoma	Nivo	300	44.5	10.9	N/A
Wen, L., et al. 202027	PR	Negative	56/m	Hepatocellular carcinoma	sintilimab	168	22.2	7.8	DRB1* 12, DRB1*12 DQB1* 05, DQB1* 03
Sakaguchi, C., et al. 201928	SD	Negative	68/f	Melanoma	Nivo	420	17.4	8.2	DRB1*09:01
Matsuura, N., et al. 201829	SD	Positive	78/m	Lung cancer	Nivo	40	29.3	6.1	DRB1*0301‐DQB1*0803 DRB1*0601‐DQB1*1406
Capitao, R., et al. 201830	SD	Positive	74/f	Lung cancer	Nivo	25	58.8	8.7	DRB1*04
Araujo, M., et al. 201731	SD	Positive	73/f	Lung cancer	Nivo	11	>55.5	7.2	DRB1*0301‐DQB1*0201 DRB1*0401‐DQB1*03:02
Godwin, J. L., et al 201732	SD	Positive	34/m	Lung cancer	Nivo	28	41.0	7.1	DR9
Hofmann, L., et al. ② 201616	SD	Positive	78/f	Melanoma	Ipi, Nivo	18	N/A	N/A	N/A
Kapke, J., et al. 201733	SD	Positive	N/A	Head and neck cancer	Nivo	90	23.6	N/A	DRB1*08, DRB1*11 DQB1*03, DQB1*04
Tohi, Y., et al. 201934	SD	Negative	75/m	Ureteral cancer	Pembro	50	60.6	6.7	N/A
Hong, A. R. et al. ② 202018	SD	Negative	67/m	Urothelial cancer	Atezolizumab	189	29.4	9.8	N/A
Current study Case 1	SD	Negative	76/f	Gastric cancer	Nivo	107	33.5	9	DRB1*08:02‐DQB1*03:02 DRB1*14:54‐DQB1*05:03
Current study Case 2	SD	Negative	68/f	Head and neck cancer	Nivo	300	36.3	9.5	DRB1*04:03‐DQB1*03:02 DRB1*15:01‐DQB1*06:02
Li, W., et al. 202035	SD	Negative	73/m	Lung cancer	Anti‐PD−1 monoclonal antibody	221	51.0	7.6	N/A
Zaied, A. A., et al. 201836	PD	N/A	N/A/m	Renal cell carcinoma	Nivo	38	48.7	8.4	N/A
Li, L., et al. 201737	PD	Positive	63/m	Lung cancer	Nivo	27	32.9	7.2	N/A
Teramoto, Y., et al. 201738	PD	Negative	63/f	Melanoma	Nivo	189	36.7	8.9	N/A
Zezza, M. et al. ② 201917	PD	Positive	80/f	Melanoma	Ipi, Nivo	35	48.4	N/A	N/A

Abbreviations: CR, complete response; Ipi, ipilimumab; N/A, not available; Nivo, nivolumab; PD, progressive disease; Pembro, pembrolizumab; PR, partial response; SD, stable disease.

In a Japanese national survey of ICI‐induced DM, HbA1c and blood pH were higher, and fasting plasma glucose levels tended to be lower at the onset of ICI‐induced DM than conventional type 1 DM.³⁹ Baden M et al. reported that diabetic ketoacidosis was observed in 39% of ICI‐induced DM patients in Japan, while Akturk. HK et al⁴⁰ reported that diabetic ketoacidosis was observed 76% of ICI‐induced DM patients in their systematic review (Table 2). It might be one of the reasons for this difference that the higher positive rate of islet‐related autoantibodies in Akturk's report than Baden's report (presence of islet‐related autoantibody; 50.7% in Akturk's report vs 4.8% in Baden's report). Similar to the report of Baden, M. et al., HbA1c was as high as 9.0% or more, though diabetic ketoacidosis was not observed and islet‐related autoantibodies were negative in our cases. Recent research revealed that patients who had islet‐related autoantibodies were reported to develop ICI‐induced DM rapidly.⁶, ⁴⁰, ⁴¹ In Table 1, islet‐related autoantibodies were negative in 17 cases and positive in 13 cases. The duration to the onset of ICI‐induced DM was significantly longer in islet‐related autoantibody‐negative cases (the average duration between ICI initiation and DM development; 176 ± 134 days in negative group vs 31 ± 20 days in positive group, p = 0.000, Man‐Whitney test; Table 1). Our cases did not have islet‐related autoantibodies, and the duration to the onset of ICI‐induced DM was more than 3 months from the initial administration of nivolumab.

TABLE 2.

Clinical characteristics of immune checkpoint inhibitor‐induced diabetes mellitus

	Akturk. H. K., et al, 201940 n = 71	Baden, M., et al, 2019a 39 n = 22
Age, years (SD)	61.7 (12.2)	63 (12)
Gender (Female; %)	45	41
immune checkpoint inhibitor (nivolumab; %)	53.5	100
Duration to the onset of immune checkpoint inhibitor‐induced diabetes mellitus, days (SD)	83.5 (88.5)	121(21)
Diabetic ketoacidosis (%)	76	39
Plasma glucose, mmol/L (SD)	33.4 (11.5)	34.2 (13.8)
HbA1c, % (SD)	7.8 (2.2)	8.1 (1.3)
Presence of islet‐related autoantibody (%)	50.7	4.8

^a all patients were Japanese.

There is no standard version of case definitions for ICI‐induced DM. ICI‐induced DM often characterized by the presence of diabetic ketoacidosis or very low to absent C‐peptide levels and insulin dependence after the onset of diabetes.⁶, ⁷, ⁸ Baden, M. et al³⁹ suggested that C‐peptide levels of acute‐onset type 1 diabetes were 0.15 (0.07–0.23) nmol/L, while those of fulminant type 1 diabetes were 0.07 ( ± 0.06) nmol/L in Japanese patients with ICI treatment. Tsang V. H. M. et al⁸ reported that C‐peptide levels were 0.41 nmol/L several months after diagnosis of ICI‐induced DM. However, specific cutoff values for C‐peptide levels in ICI‐induced DM have not been determined.⁷ In the current cases, ICI‐induced DM was diagnosed by the clinical symptoms and laboratory data. Hyperglycemia symptoms such as general fatigue, thirst, and polyuria appeared for several days in both cases. The serum C‐peptide levels were inappropriately low (case1: 0.05 nmol/L and case2: 0.13 nmol/L), and both patients were insulin dependent for over 1 year after the onset of ICI‐induced DM. In case 2, although the course of insulin secretion after the onset could not be assessed, ICI‐induced DM was diagnosed because the serum C‐peptide levels at the onset had already markedly decreased and the patient was insulin dependent.

As ICI‐induced DM is a rare, but life‐threatening side effect of nivolumab, close monitoring of blood glucose levels is recommended for the early diagnosis of ICI‐induced DM. Pathogenic mechanisms underlying the development of ICI‐induced DM and predictive markers for selection of patients who have increased risk of the onset of ICI‐induced DM have not been elucidated. Monitoring clinical data including insulin secretory ability might help in early detection of ICI‐induced DM and understanding clinical features of the development of DM during ICI treatment. Our two cases showed that insulin secretion was maintained in the normal range and temporarily increased before the onset of ICI‐induced DM. Accumulating cases of ICI‐induced DM can be helpful to identify predictive markers for early diagnosis and prevent this side effect.

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

AUTHOR CONTRIBUTIONS

All authors contributed significantly. Noriko Fujiwara, Mayu Watanabe and Akihiro Katayama contributed the design of the work. Noriko Fujiwara and Mayu Watanabe wrote the manuscript in consultation with Akihiro Katayama, Yohei Noda, Jun Eguchi, Hitomi Kataoka, Syunsuke Kagawa, and Jun Wada.

ETHICAL APPROVAL

This manuscript has not been published or presented elsewhere in part or in entirety and is not under consideration by another journal. This study was conducted with the approval of the Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences and Okayama University Hospital, Ethics Committee (No 1704–009). We obtained verbal and written informed consent were obtained from each patient.

DISCLOSURE

Jun Wada receives speaker honoraria from Astra Zeneca, Daiichi Sankyo, MSD, Novartis, Tanabe Mitsubishi, Taisho Toyama and receives grant support from Baxter, Chugai, Dainippon Sumitomo, Ono, Teijin.

Fujiwara N, Watanabe M, Katayama A, et al. Longitudinal observation of insulin secretory ability before and after the onset of immune checkpoint inhibitor‐induced diabetes mellitus: A report of two cases. Clin Case Rep. 2021;9:e04574. 10.1002/ccr3.4574

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.