ABSTRACT
Aim/Introduction
Although immune checkpoint inhibitor‐related type 1 diabetes mellitus (ICI‐T1DM) is a rare condition, it is of significant concern globally. We aimed to elucidate the precise incidence, risk factors, and impact of ICI‐T1DM on survival outcomes.
Materials and Methods
The study is a large retrospective cohort study, performed using the DeSC Japanese administrative claims database comprising 11 million patients. The database population is reportedly similar to the entire population of Japan. Patients administered ICI between 2014 and 2022 were enrolled in the study, including 21,121 patients. The risk factors for ICI‐T1DM development and their characteristics were evaluated by logistic regression analysis. Development of a new irAE after the day following the first administration of ICI was set as the study outcome.
Results
ICI‐T1DM was observed in 102 (0.48%) of the 21,121 patients after ICI initiation. PD‐(L)1 and CTLA‐4 combination therapy was associated with an increased risk of ICI‐T1DM compared with PD‐1 monotherapy (odds ratio [OR], 2.36; 95% confidence interval [CI], 1.21–4.58; P = 0.01). Patients with a prior diagnosis of diabetes mellitus (OR, 1.59; 95% CI, 1.03–2.46; P = 0.04) or hypothyroidism (OR, 2.48; 95% CI, 1.39–4.43; P < 0.01) also exhibited an increased risk of ICI‐T1DM. The Kaplan–Meier analysis revealed that patients with ICI‐T1DM showed higher survival rates than those without (log‐lank test, P < 0.01). Multivariable Cox regression analysis demonstrated that ICI‐T1DM development was associated with lower mortality (hazard ratio, 0.60; 95% CI, 0.37–0.99; P = 0.04).
Conclusions
Collectively, the results of this study demonstrate the precise incidence and risk factors of ICI‐T1DM. The development of ICI‐T1DM, like other irAEs, is associated with higher survival rates.
Keywords: Immune checkpoint inhibitors, Survival, Type 1 diabetes mellitus
Collectively, the results of this study demonstrate the precise incidence and risk factors of immune checkpoint inhibitor‐related type 1 diabetes mellitus (ICI‐T1DM). The development of ICI‐T1DM, like other immune‐related adverse events, is associated with higher survival rates.
INTRODUCTION
Endocrine immune‐related adverse events (irAEs) occur frequently in patients with cancer treated with immune checkpoint inhibitors (ICIs) 1 , 2 , 3 . Although ICI‐related thyroiditis and hypophysitis are relatively common, ICI‐related type 1 diabetes mellitus (ICI‐T1DM) is rare 1 with a reported incidence of 0.4–3.5% (~1%) 1 , 4 , 5 , 6 . However, most associated cohort studies have been based on relatively small sample sizes in single institutions. Therefore, the precise incidence of ICI‐T1DM is unknown. Moreover, ICI‐T1DM is almost exclusively associated with the use of antiprogrammed cell death 1 (PD‐1) or programmed cell death ligand 1 (PD‐L1) therapy. Although a few studies have reported that combining anticytotoxic T lymphocyte antigen 4 (CTLA‐4) with anti‐PD‐1 or anti‐PD‐L1 may increase the risk of ICI‐T1DM 4 , 7 , related data are limited.
ICI‐T1DM is characterized by rapid hyperglycemia onset, swift endogenous insulin deficiency progression, and sometimes diabetic ketoacidosis 3 , 4 , necessitating urgent diagnosis and insulin treatment 6 , 8 . Most patients require lifelong insulin therapy; however, the risk of developing ICI‐T1DM has not been comprehensively elucidated. In addition, high‐risk HLA alleles and haplotypes have been identified; however, they have not been sufficiently investigated in association with ICI initiation. Therefore, risk factors associated with clinical practice need to be identified 9 , 10 , 11 .
Recently, large case series studies have revealed that the development of several irAEs, including ICI‐related thyroiditis and hypophysitis, is associated with improved survival outcomes 12 , 13 . However, the relationship between ICI‐T1DM and survival outcome has not been fully clarified owing to its rarity. Therefore, we conducted a large retrospective cohort study based on the 11‐million person administrative claims database in Japan to elucidate the precise incidence, risk factors, and impact of ICI‐T1DM on survival.
METHODS
Study population
Patients administered ICIs were identified among the 11 million individuals in the Japanese administrative claims database (i.e., DeSC database) between April 2014 and February 2022. The ICIs included anti‐PD‐1 monotherapy (nivolumab, pembrolizumab), anti‐PD‐L1 monotherapy (atezolizumab, durvalumab, and avelumab), anti‐CTLA‐4 monotherapy (ipilimumab), and combination therapy. Combination therapy included simultaneous administration of drugs or the administration of multiple ICIs within the observation period. All patients were at least 18 years old; 48 patients diagnosed with T1DM prior to ICI administration and 16 without medical records after the first ICI administration were excluded (Figure 1). The constitution of the DeSC database population is reportedly similar to the entire population of Japan; therefore, the database can be used for representative medical epidemiology 14 . This study was approved by the Ethics Review Committee of Nara Medical University (approval number: 1123‐7); the requirement for informed consent was waived as all data were anonymized.
Figure 1.
Study design. From a total of 21,185 patients receiving ICI in the Japanese administrative claims database, we excluded patients who were diagnosed with type 1 diabetes mellitus prior to ICI administration and those without medical records after the first ICI administration. Eventually, 21,121 patients with ICI administration were included in the analysis.
Definition of T1DM and comorbidities
Development of a new irAE after the day following the first administration of ICI was set as the study outcome. Diabetes was defined as a diagnosis of a disease code of diabetes and a prescription of diabetes medication. T1DM was defined as a disease code of T1DM and prescription of insulin with a self‐injection fee for T1DM, which is only applied for T1DM with intensive insulin therapy in the practice 15 . The validation analysis demonstrated that the sensitivity was 0.72 and specificity was 1.00 of T1DM based on this definition in the general population 16 . We defined the T1DM incidence date as the first day of instruction fee for T1DM. Hypothyroidism and Graves' disease were ascribed to patients treated with levothyroxine or antithyroid drugs (thiamazole and propylthiouracil), respectively. Adrenal insufficiency (primary or secondary) was defined as treatment with hydrocortisone. Diabetic ketoacidosis (DKA) was diagnosed in patients with the DKA disease code at the onset of ICI‐T1DM. Death was defined as a transcribed death in the claims data. Steroids administered prior to the start of ICI were injectable and oral glucocorticoids (Hydrocortisone, Prednisone, Methylprednisolone, Prednisolone, Dexamethasone, excluding applied, pasted, ophthalmic, and intraarticular injectable products) administered prior to the start of ICI during the observation period. Evaluation of patient characteristics included assessment of the updated Charlson Comorbidity Index (CCI) score 17 as a short‐term prognostic score.
Study analysis
The chi‐squared test was used for statistical analysis of independence and comparison of frequencies. Logistic regression analysis was performed to identify the risk factors for ICI‐T1DM development and their characteristics. Median overall survival (OS) was estimated using the Kaplan–Meier method. When the median OS was not reached, the 1‐year survival rate was applied for evaluation. The Cox model adjusted for the number of ICI doses and duration of ICI administration was used to assess the impact of ICI‐T1DM on survival. SPSS (IBM SPSS Statistics 28.0.1, USA) was employed for all statistical analyses. Statistical significance was defined as P < 0.05.
RESULTS
ICI‐T1DM incidence and characteristics
We identified 21,121 patients treated with ICI, with a median follow‐up period of 252 days (interquartile range [IQR], 112–499). ICI‐T1DM was diagnosed in 102 patients (0.48%; Table 1) with a median age of 73 years (IQR, 69–78), and 73.5% were males; no statistical differences were observed between the ICI‐T1DM and non‐ICI‐TIDM groups (P = 0.32, P = 0.81, respectively; Table 1). Additionally, 30.4% of patients with ICI‐T1DM had preexisting type 2 diabetes mellitus and 14.7% had preexisting hypothyroidism before ICI initiation. Regarding the type of ICIs among patients with ICI‐T1DM, anti‐PD‐1 monotherapy was the most common (77.5%), followed by anti‐PD‐(L)1 and anti‐CTLA‐4 combination therapy (12.7%). Regarding the type of tumors among patients with ICI‐T1DM, lung cancer was the most prevalent (36.3%), followed by gastric cancer (35.3%; Table 1). Moreover, 80.4% of patients with ICI‐T1DM were diagnosed within 1 year. In detail, 19.6% of patients were diagnosed with T1DM during the first 3 months after ICI initiation, 28.4% were during 4–6 months, 20.6% were during 7–9 months, 11.8% were during 10–12 months, 16.7% were during the second year, and 2.9% were during the third year. The median period for T1DM onset was 183 days (IQR, 96–278), and the latest was 909 days after ICI initiation (Figure S1); 27% of patients exhibited DKA at ICI‐T1DM onset. Following the onset of ICI‐T1DM, ICI treatment was resumed in 52.9% of the patients, while ICI treatment was discontinued in the others.
Table 1.
Clinical characteristics of patients with and without ICI‐T1DM
| Total N = 21,121 | ICI‐DM | P value | ||
|---|---|---|---|---|
| Yes N = 102 | No N = 21,019 | |||
| Age, years, median | 73 | 73 | 73 | |
| IQR | 68–78 | 69–78 | 68–78 | |
| ≦70 | 8,045 (38.1) | 34 (33.3) | 8,011 (38.1) | 0.32 |
| >70 | 13,076 (61.9) | 68 (66.6) | 13,008 (61.9) | |
| Male | 15,742 (74.6) | 75 (73.5) | 15,671 (74.6) | 0.81 |
| ICI type | ||||
| Anti‐PD‐1 | 15,851 (75.0) | 79 (77.5) | 15,772 (75.0) | Reference |
| Anti‐PD‐L1 | 4,066 (19.3) | 9 (8.8) | 4,057 (19.3) | 0.02 |
| Anti‐CTLA‐4 | 42 (0.2) | 0 (0.0) | 42 (0.2) | 1.00 |
| Anti‐PD‐(L)1 + Anti‐CTLA‐4 | 678 (3.2) | 13 (12.7) | 665 (3.2) | <0.01 |
| Anti‐PD‐1 + Anti‐PD‐L1 | 684 (2.3) | 1 (1.0) | 483 (2.3) | 0.73 |
| Tumor type | ||||
| Lung | 11,433 (54.1) | 37 (36.3) | 11,396 (54.2) | Reference |
| Stomach | 4,746 (22.5) | 36 (35.3) | 4,710 (22.4) | <0.01 |
| Urothelial | 1,078 (5.1) | 8 (7.8) | 1,070 (5.1) | 0.05 |
| Melanoma | 581 (2.8) | 9 (8.8) | 572 (2.7) | <0.01 |
| Others | 3,200 (15.2) | 12 (11.8) | 3,188 (15.2) | 0.61 |
| Unknown + double | 83 (0.4) | 0 (0.0) | 83 (2.3) | 1.00 |
| Prior history before ICI | ||||
| Diabetes mellitus | 4,672 (22.1) | 31 (30.4) | 4,641 (22.1) | 0.04 |
| Hypothyroidism | 1,336 (6.3) | 15 (14.7) | 1,321 (6.3) | <0.01 |
| Graves' disease | 72 (0.3) | 0 (0.0) | 72 (0.3) | 1.00 |
| Steroid before ICI | 13,210 (62.5) | 50 (49.0) | 13,160 (62.6) | 0.01 |
| Charlson Comorbidity Index | ||||
| <6 | 9,537 (45.2) | 41 (40.2) | 9,496 (45.2) | 0.40 |
| ≧6 | 11,584 (54.8) | 61 (59.8) | 11,523 (54.8) | |
| After ICI | ||||
| Hypothyroidism | 1,757 (8.3) | 14 (13.7) | 1,743 (8.3) | 0.03 |
| Graves'disease | 16 (0.1) | 0 (0.0) | 16 (0.1) | 1.00 |
| Adrenal insufficiency | 778 (3.7) | 7 (6.9) | 771 (3.7) | 0.09 |
Univariate analysis was performed. The percentages of total, with, and without ICI‐T1DM represent the percentage of total (n = 21,121) and groups with ICI‐T1DM (n = 102) or without ICI‐T1DM (n = 21,019), respectively. Age denotes age at the time of first ICI administration. The median Charlson Comorbidity Index (updated score) was 6 on a 24‐point scale.
CTLA‐4, cytotoxic T lymphocyte antigen 4; ICI, immune checkpoint inhibitor; ICI‐T1DM, immune checkpoint inhibitor‐related type 1 diabetes mellitus; PD‐1, programmed cell death 1; PD‐L1, programmed cell death ligand 1.
Factors associated with ICI‐T1DM development
The univariate analysis demonstrated that the incidence of ICI‐T1DM was 3.8‐fold higher in patients receiving anti‐PD‐(L)1 and anti‐CTLA‐4 combination therapy (1.92%) than those with anti‐PD‐1 monotherapy (0.50%), followed by anti‐PD‐L1 monotherapy (0.22%), and anti‐PD‐1 and anti‐PD‐L1 therapy (0.21%). Meanwhile, no patients developed ICI‐T1DM following anti‐CTLA‐4 monotherapy (Figure S2). Multivariable analysis revealed that the odds ratio (OR) of ICI‐T1DM for the combination therapy was 2.36 (95% confidence interval [CI], 1.21–4.58; P = 0.01; Figure 2, Table S1). Patients with a prior diagnosis of diabetes mellitus (OR, 1.59; 95% CI, 1.03–2.46; P = 0.04) or hypothyroidism (OR, 2.48; 95% CI, 1.39–4.43; P < 0.01) also exhibited an increased risk of ICI‐T1DM. Age, sex, and CCI score were not significantly associated with the increased risk of developing ICI‐T1DM. Steroids administered prior to the start of ICI decreased the risk of ICI‐T1DM (OR, 0.60; 95% CI, 0.40–0.91; P = 0.02). ICI‐related hypothyroidism and adrenal insufficiency showed no effect (Table S1).
Figure 2.
Forest plot of factors associated with ICI‐T1DM risk (logistic regression analysis). Multivariable analysis was performed using age, sex, ICI type, tumor type, past medical history, steroid administration, prognostic score, and immune‐related adverse events as covariates. Lung cancer was the most common primary disease (54.1% of all study patients) and was therefore used as a reference for the tumor types. Anti‐PD‐(L)1 and anti‐CTLA‐4 combination therapy and a prior diagnosis of diabetes mellitus and hypothyroidism showed a higher risk of ICI‐T1DM. CI, confidence interval; CTLA‐4, cytotoxic T lymphocyte antigen 4; ICI, immune checkpoint inhibitor; ICI‐T1DM, immune checkpoint inhibitor‐related type 1 diabetes mellitus; PD‐1, programmed cell death 1; PD‐L1, programmed cell death ligand 1.
ICI‐T1DM and survival
Among the 21,121 patients treated with ICIs, 4,728 (22.4%) died during the observation period, and 16 died after ICI‐T1DM diagnosis (Figure 3). Interestingly, patients with ICI‐T1DM exhibited superior OS compared with those without (P < 0.01, log‐rank test). One‐year survival rates were 89.9 ± 3.4% vs 76.7 ± 0.4% for patients with and without ICI‐T1DM, respectively (P < 0.01, chi‐squared test). Additionally, stratified analysis by age (≤70 years) and tumor type (lung cancer or not; because lung cancer was the most common primary disease and 54.1% of all study patients) demonstrated similar results (P < 0.01, log‐rank; Figure S3).
Figure 3.
Kaplan–Meier survival curve of patients with or without ICI‐T1DM. Patients with ICI‐T1DM showed superior overall survival compared with those without ICI‐T1DM (P < 0.01, log‐rank test) with 1‐year overall survivals of 89.9 ± 3.4% vs 76.7 ± 0.4%, respectively (P < 0.01; estimated value ± SE, with vs without ICI‐T1DM). ICI‐T1DM, immune checkpoint inhibitor‐related type 1 diabetes mellitus.
In addition, we performed multivariable analysis to avoid immortal time bias, which revealed that the development of ICI‐T1DM was independently associated with relatively low mortality (adjusted hazard ratio [aHR], 0.60; 95% CI, 0.37–0.99; P = 0.04) after adjustment for age, sex, ICI type, cancer type, prior diabetes mellitus, CCI score, ICI‐related hypothyroidism and adrenal insufficiency, and the number of ICI doses and duration of ICI administration (Figure 4, Table S2). Older age (HR 1.04; 95% CI, 1.04–1.05, P < 0.01) and higher CCI (HR 1.20; 95% CI, 1.13–1.28, P < 0.01) were associated with a higher mortality. ICI‐related hypothyroidism (HR 0.72; 95% CI, 0.65–0.81, P < 0.01) and adrenal insufficiency (HR 0.62; 95% CI, 0.51–0.74, P < 0.01) were associated with decreased mortality. Furthermore, steroid administration prior to ICI administration was also included as an adjustment factor in a sensitivity analysis, but the results were comparable (Table S3).
Figure 4.
Forest plot of factors associated with mortality risk (Cox regression analysis). Cox regression analysis adjusted for the number of ICI doses and duration of ICI administration was performed using age, sex, ICI type, tumor type, past medical history, prognostic score, and immune‐related adverse events as covariates. Lung cancer was the most common primary disease (54.1% of all study patients) and was therefore used as a reference for the tumor types. ICI‐related hypothyroidism and adrenal insufficiency showed a decreased mortality. In addition, the development of ICI‐T1DM showed a decreased mortality. CCI, Charlson Comorbidity Index; CI, confidence interval; ICI, immune checkpoint inhibitor; ICI‐T1DM, immune checkpoint inhibitor‐related type 1 diabetes mellitus.
DISCUSSION
In this study, we demonstrated the precise incidence of ICI‐T1DM (0.48%) in a large cohort study of more than 20,000 patients treated with ICI, representing one of the largest studies on ICI‐T1DM with a relatively unbiased population using the Japanese claims database. PD‐(L)1 and CTLA‐4 combination therapy and a prior diagnosis of diabetes or hypothyroidism were associated with an increased risk of ICI‐T1DM. Furthermore, the development of ICI‐T1DM was associated with superior OS.
In this study, the median age of patients with ICI‐T1DM was 73 years, with no difference compared to that of patients without ICI‐T1DM. As reported previously, age at ICI‐T1DM diagnosis was significantly higher than that in general T1DM 7 , 8 . This may be explained by the specific patient population that received ICI. Most studies have shown a male predominance at 55–64% 18 , 19 , consistent with our study. However, this likely reflects the patient population receiving ICI; notably, sex was not significantly associated with ICI‐T1DM risk in this study. Regarding the time of onset of ICI‐T1DM, it has been reported that the median period was 10 weeks (1–95 weeks) 7 and 155 days (range 13–504 days) 8 after the initiation of ICI, respectively. Several reports suggest a caution especially during the first 7 months after the initiation of ICI 20 , however, the present study demonstrated that only 48.0% of ICI‐T1DM was observed within the first 6 months, 80.4% was within the first year, and the latest case was observed more than 2 years after the initiation of ICI. In other words, careful monitoring of blood glucose levels is necessary, especially within the first year after initiation of ICI, but caution is needed for 3 years.
Several reviews have reported the incidence of ICI‐T1DM; however, most original research studies have been based on a relatively small cohort with sample heterogeneity 4 , 5 , 6 . The largest report using USA‐based insurance claim data reported an ICI‐T1DM incidence of 0.86% in 30,337 patients treated with ICIs with a median follow‐up of 308 days. However, this study used insurance claim data, and the definition of T1DM was only based on the ICI‐10 code, while it lacked information regarding severity, which may have caused overestimation of the diagnosis 7 . Therefore, we applied a more clinically strict definition of T1DM as a patient who was diagnosed with T1DM and prescribed insulin, with a self‐injection fee. Based on this definition and a relatively large cohort, we reported the incidence of ICI‐T1DM as 0.48%.
Globally, the incidence and prevalence of T1DM vary substantially. The incidence of T1DM is relatively high in Caucasians and it is the highest in Finland (>60 cases per 100,000 people each year) 20 , by contrast, it is approximately 2 cases per 100,000 people each year even in childhood in Japan 21 . On the other hand, among the types of T1DM, the incidence of fulminant T1DM is higher in Asians than in Caucasians 22 . These differences are mainly explained by genetic background including HLA haplotypes 23 , 24 , 25 . Interestingly, the incidence of ICI‐T1DM is 0.4–3.5% in various countries as reported previously, compared with 0.48% in the present study, suggesting that regional and racial differences in the incidence of ICI‐T1DM are much smaller than those of T1DM. Importantly, the pathophysiological conditions of ICI‐T1DM are different between Western countries and Asia. Wherein half of the patients with ICI‐T1DM in Western countries are positive for anti‐GAD antibodies 26 and these patients show rapid presentation, whereas patients without them show a delayed onset of disease 27 . On the contrary, most patients in Asia including Japan, these antibodies are negative 8 and 50.0% of these patients fulfilled the criteria of fulminant T1DM in Japan 26 . These data suggest that the relatively increased incidence of ICI‐T1DM in Asian may be explained by the increased type of fulminant onset in ICI‐T1DM that is rarely observed in Caucasians.
Anti‐PD‐(L)1 therapy was almost exclusively associated with ICI‐T1DM development, whereas anti‐PD‐(L)1 and anti‐CTLA‐4 combination therapy increased ICI‐T1DM risk. In contrast, anti‐CTLA‐4 monotherapy was rarely associated with ICI‐T1DM risk 4 , 28 . In this study, ICI‐T1DM incidence was the highest in anti‐PD‐(L)1 and anti‐CTLA‐4 combination therapy, followed by anti‐PD‐1 monotherapy, anti‐PD‐L1 monotherapy, and anti‐PD‐1 and anti‐PD‐L1 therapy; ICI‐TIDM was not observed in any patients following anti‐CTLA‐4 monotherapy. The results of multivariate analysis showed a significantly increased risk for ICI‐T1DM in anti‐PD‐(L)1 and anti‐CTLA‐4 combination therapy with an odds of 2.36 (1.21–4.58) compared with the anti‐PD‐1 therapy, which was consistent with previous reports 7 , 29 . Mechanistically, upregulation of PD‐L1 in pancreatic β‐cells during the early stages of T1DM has been suggested 30 , 31 , 32 . Therefore, during combination therapy, the addition of anti‐CTLA‐4 may boost the anti‐PD‐L1 therapeutic effect.
We demonstrated that a history of hypothyroidism or diabetes mellitus was a risk factor for developing ICI‐T1DM. Similarly, a previous report revealed that 17% of patients with ICI‐T1DM had comorbid Hashimoto's disease 33 . Indeed, nearly 40% of ICI‐T1DMs reportedly have other concomitant endocrine irAEs 5 , 8 . In the present study, among patients with ICI‐T1DM, 13.7% had ICI‐related hypothyroidism and 6.9% had ICI‐related adrenal insufficiency. Therefore, these common autoimmune predispositions for endocrine organs, including genetic background, may also be associated with ICI‐T1DM.
We have identified preexisting diabetes mellitus (type 2 diabetes and other types of diabetes) as a risk factor for ICI‐T1DM, as described in previous studies 6 , 7 . Certain patients diagnosed with T2DM potentially have autoimmune diabetes mellitus with anti‐GAD antibodies, which may predispose ICI‐T1DM 34 . In addition, β‐cell damage associated with glucose toxicity may increase immunogenicity and provoke such immune responses.
Several studies have reported that endocrine irAEs, such as thyroiditis 35 , 36 and hypophysitis 37 , 38 , are associated with more optimized survival and a preferable response to the tumor. In contrast, the association between ICI‐T1DM and prognosis has been controversial, with certain reports showing favorable responses and prognosis in a small‐scale analysis 28 , 29 , and another study reporting no association with mortality 7 . Meanwhile, the current study clearly demonstrated that the development of ICI‐T1DM was associated with improved mortality according to Kaplan–Meier and multivariate analyses adjusted for the number of ICI doses and duration of ICI administration. The mechanisms underlying the association of ICI‐T1DM with better OS are still unknown. Recently, it has been reported that at least a part of ICI‐related hypophysitis is caused as an immune‐mediated paraneoplastic syndrome and defined as paraneoplastic autoimmune hypophysitis, in which an ectopic pituitary antigen expression in the tumor evoked autoimmunity against pituitary‐specific antigens, resulting in hypophysitis and exhibiting the injury of specific anterior pituitary cells by cytotoxic T cells 39 , 40 . In addition, the development of vitiligo as irAE was associated with better response in patients with melanoma, suggesting that melanoma cells and normal melanocytes may have shared antigens 41 . These data suggest that a cross‐reactivity or molecular mimicry between the antigen in the tumor and target tissues caused the irAE; therefore, the effectiveness of ICI to the tumor can be associated with the onset of the irAE 2 . Although specific antigens have not been identified, it is suggested that there may be a common antigen between pancreatic beta cells and tumor cells.
The relatively optimized survival of patients with ICI‐T1DM shown in this study suggests the significance of clinical management, considering the effect of optimized diabetic control on patient prognosis. In addition, ICI was resumed in only 52.9% of patients. Considering that ICI‐T1DM is associated with a favorable ICI effect, continued ICI‐T1DM treatment is recommended.
This study had several limitations. The data depended on claims data codes for diseases, drugs, and instructional charges and lack of laboratory values; therefore, the detailed analysis of pathological conditions and clinical course such as the presence of pancreatitis and circulating anti‐GAD antibodies at the onset of ICI‐DM, is difficult. In particular, lack of the information on C‐peptide and anti‐GAD antibodies resulted in difficulty of the diagnosis of T1DM; however, we used the prescription of insulin with a self‐injection fee for T1DM, which is strictly applied for T1DM for the definition. Indeed, it has shown that thesensitivity was 0.72 and specificity was 1.00 based on this definition in the general population 16 . In our definition, patients with relatively slow progressive and mild T1DM may be underestimated, and this definition in ICI‐DM, which is predominantly elderly, has not been validated. Nontheless, using this definition, it is considered that clinically relevant T1DM based on the insulin‐dependent status is appropriately included. Additionally, the use of a sufficient sample size and relatively less‐biased data allowed for precise determination of the incidence and association with survival. However, even with this sample size, due to the low incidence of ICI‐T1DM, subgroup analysis by tumor type has not been performed, and further studies are warranted.
In summary, we demonstrate the precise incidence and risk factors of ICI‐T1DM using a large administrative claims database. The development of ICI‐T1DM is associated with superior survival in patients with cancer treated with ICI.
DISCLOSURE
FK received speaker fees from Dainippon Sumitomo, Sanofi, and Kyowa Kirin. YN received consultation fees from Novo Nordisk. SO received speaker fees from Ono; Mitsubishi Tanabe; Dainippon Sumitomo; Eli Lilly, Japan; Takeda; AstraZeneca; Novartis Pharmaceuticals; Novo Nordisk; Mochida Pharmaceutical; Kyowa Kirin; and Terumo. YT received consultant fees from Novo Nordisk, Otsuka, and Recordati and speaker fees from Novo Nordisk, Sumitomo Dainippon, Eli Lilly, Ono, Novartis, Nippon Boehringer Ingelheim, AstraZeneca, and Kyowa Kirin. The other authors declare that they have no conflicts of interest.
Approval of the research protocol: This study was approved by the Ethics Review Committee of Nara Medical University (approval number: 1123‐7).
Informed consent: The requirement for informed consent was waived as all data were anonymized.
Registry and the registration no. of the study/trial: N/A.
Animal studies: N/A.
FUNDING
This study was supported by Grants‐in‐Aid for Scientific Research (21K10474, 24K13531 to FK; 22H03355, 23K24613, 24K22341 to YN; and 21K10451 to SO), the Ministry of Health, Labor and Welfare (Hypothalamo‐hypophyseal Disorders and Endocrine Syndrome with Sexual Differentiation and Maturation grant to YT), and Nara Medical University Center for Diversity and Inclusion (grant to FK).
ROLE OF THE FUNDER
None.
Supporting information
Table S1. Logistic regression analysis of factors associated with ICI‐T1DM risk.
Table S2. Multivariable Cox regression analysis of mortality.
Table S3. Multivariate Cox regression analysis of mortality (sensitivity analysis including steroid administration prior to ICI administration).
Figure S1. Time from ICI initiation to ICI‐T1DM onset (n = 102).
Figure S2. Incidence of ICI‐T1DM by ICI types.
Figure S3. Kaplan–Meier survival curve of patients with or without ICI‐T1DM. (A) Stratified by age (≤70, group 1; >70, group 2). (B) Stratification by cancer type (LK, group 1; other than LK, group 2).
ACKNOWLEDGMENTS
The DeSC database was provided by DeSC Healthcare, Inc. under their academic research support program.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from DeSC Healthcare Inc. (Tokyo, Japan) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are, however, available from the corresponding author upon reasonable request and with permission of DeSC Healthcare Inc. (Tokyo, Japan).
REFERENCES
- 1. Barroso‐Sousa R, Barry WT, Garrido‐Castro AC, et al. Incidence of endocrine dysfunction following the use of different immune checkpoint inhibitor regimens: A systematic review and meta‐analysis. JAMA Oncol 2018; 4: 173–182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Postow MA, Sidlow R, Hellmann MD. Immune‐related adverse events associated with immune checkpoint blockade. N Engl J Med 2018; 378: 158–168. [DOI] [PubMed] [Google Scholar]
- 3. Chang LS, Barroso‐Sousa R, Tolaney SM, et al. Endocrine toxicity of cancer immunotherapy targeting immune checkpoints. Endocr Rev 2019; 40: 17–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Cho YK, Jung CH. Immune‐checkpoint inhibitors‐induced type 1 diabetes mellitus: From its molecular mechanisms to clinical practice. Diabetes Metab J 2023; 47: 757–766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Wang Y, Zhou S, Yang F, et al. Treatment‐related adverse events of PD‐1 and PD‐L1 inhibitors in clinical trials: A systematic review and meta‐analysis. JAMA Oncol 2019; 5: 1008–1019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Stamatouli AM, Quandt Z, Perdigoto AL, et al. Collateral damage: Insulin‐dependent diabetes induced with checkpoint inhibitors. Diabetes 2018; 67: 1471–1480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Chen X, Affinati AH, Lee Y, et al. Immune checkpoint inhibitors and risk of type 1 diabetes. Diabetes Care 2022; 45: 1170–1176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Tachibana M, Imagawa A. Type 1 diabetes related to immune checkpoint inhibitors. Best Pract Res Clin Endocrinol Metab 2022; 36: 101657. [DOI] [PubMed] [Google Scholar]
- 9. Tsang VHM, McGrath RT, Clifton‐Bligh RJ, et al. Checkpoint inhibitor‐associated autoimmune diabetes is distinct from type 1 diabetes. J Clin Endocrinol Metab 2019; 104: 5499–5506. [DOI] [PubMed] [Google Scholar]
- 10. Inaba H, Morita S, Kosugi D, et al. Amino acid polymorphisms in human histocompatibility leukocyte antigen class II and proinsulin epitope have impacts on type 1 diabetes mellitus induced by immune‐checkpoint inhibitors. Front Immunol 2023; 14: 1165004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Inaba H, Kaido Y, Ito S, et al. Human leukocyte antigens and biomarkers in type 1 diabetes mellitus induced by immune‐checkpoint inhibitors. Endocrinol Metab 2022; 37: 84–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Shankar B, Zhang J, Naqash AR, et al. Multisystem immune‐related adverse events associated with immune checkpoint inhibitors for treatment of non‐small cell lung cancer. JAMA Oncol 2020; 6: 1952–1956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Das S, Johnson DB. Immune‐related adverse events and anti‐tumor efficacy of immune checkpoint inhibitors. J Immunother Cancer 2019; 7: 306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Okada A, Yasunaga H. Prevalence of noncommunicable diseases in Japan using a newly developed administrative claims database covering Young, middle‐aged, and elderly people. JMA J 2022; 5: 190–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Nishioka Y, Noda T, Okada S, et al. Association between influenza and the incidence rate of new‐onset type 1 diabetes in Japan. J Diabetes Investig 2021; 12: 1797–1804. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Nishioka Y, Noda T, Okada S, et al. Incidence and seasonality of type 1 diabetes: A population‐based 3‐year cohort study using the National Database in Japan. BMJ Open Diabetes Res Care 2020; 8: e001262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Quan H, Li B, Couris CM, et al. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol 2011; 173: 676–682. [DOI] [PubMed] [Google Scholar]
- 18. Tan MH, Iyengar R, Mizokami‐Stout K, et al. Spectrum of immune checkpoint inhibitors‐induced endocrinopathies in cancer patients: A scoping review of case reports. Clin Diabetes Endocrinol 2019; 5: 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Akturk HK, Kahramangil D, Sarwal A, et al. Immune checkpoint inhibitor‐induced type 1 diabetes: A systematic review and meta‐analysis. Diabet Med 2019; 36: 1075–1081. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Atkinson MA, Eisenbarth GS, Michels AW. Type 1 diabetes. Lancet 2014; 383: 69–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Onda Y, Sugihara S, Ogata T, et al. Incidence and prevalence of childhood‐onset type 1 diabetes in Japan: The T1D study. Diabet Med 2017; 34: 909–915. [DOI] [PubMed] [Google Scholar]
- 22. Imagawa A, Hanafusa T. Fulminant type 1 diabetes‐east and west. J Clin Endocrinol Metab 2023; 108: e1473–e1478. [DOI] [PubMed] [Google Scholar]
- 23. Steck AK, Rewers MJ. Genetics of type 1 diabetes. Clin Chem 2011; 57: 176–185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Tsutsumi C, Imagawa A, Ikegami H, et al. Class II HLA genotype in fulminant type 1 diabetes: A nationwide survey with reference to glutamic acid decarboxylase antibodies. J Diabetes Investig 2012; 3: 62–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Liu YC, Liu H, Zhao SL, et al. Clinical and HLA genotype analysis of immune checkpoint inhibitor‐associated diabetes mellitus: A single‐center case series from China. Front Immunol 2023; 14: 1164120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Baden MY, Imagawa A, Abiru N, et al. Characteristics and clinical course of type 1 diabetes mellitus related to anti‐programmed cell death‐1 therapy. Diabetol Int 2019; 10: 58–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ruiz‐Esteves KN, Shank KR, Deutsch AJ, et al. Identification of immune checkpoint inhibitor‐induced diabetes. JAMA Oncol 2024; 10: 1409–1416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Quandt Z, Young A, Perdigoto AL, et al. Autoimmune endocrinopathies: An emerging complication of immune checkpoint inhibitors. Annu Rev Med 2021; 72: 313–330. [DOI] [PubMed] [Google Scholar]
- 29. Muniz TP, Araujo DV, Savage KJ, et al. CANDIED: A pan‐Canadian cohort of immune checkpoint inhibitor‐induced insulin‐dependent diabetes mellitus. Cancers 2021; 14: 89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Colli ML, Hill JLE, Marroquí L, et al. PDL1 is expressed in the islets of people with type 1 diabetes and is up‐regulated by interferons‐α and‐γ via IRF1 induction. EBioMedicine 2018; 36: 367–375. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Ansari MJ, Salama AD, Chitnis T, et al. The programmed death‐1 (PD‐1) pathway regulates autoimmune diabetes in nonobese diabetic (NOD) mice. J Exp Med 2003; 198: 63–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Martinov T, Spanier JA, Pauken KE, et al. PD‐1 pathway‐mediated regulation of islet‐specific CD4(+) T cell subsets in autoimmune diabetes. Immunoendocrinology 2016; 3: e1164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Kotwal A, Haddox C, Block M, et al. Immune checkpoint inhibitors: An emerging cause of insulin‐dependent diabetes. BMJ Open Diabetes Res Care 2019; 7: e000591. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Akturk HK, Alkanani A, Zhao Z, et al. PD‐1 inhibitor immune‐related adverse events in patients with preexisting endocrine autoimmunity. J Clin Endocrinol Metab 2018; 103: 3589–3592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Muir CA, Clifton‐Bligh RJ, Long GV, et al. Thyroid immune‐related adverse events following immune checkpoint inhibitor treatment. J Clin Endocrinol Metab 2021; 106: e3704–e3713. [DOI] [PubMed] [Google Scholar]
- 36. Osorio JC, Ni A, Chaft JE, et al. Antibody‐mediated thyroid dysfunction during T‐cell checkpoint blockade in patients with non‐small‐cell lung cancer. Ann Oncol 2017; 28: 583–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Faje AT, Sullivan R, Lawrence D, et al. Ipilimumab‐induced hypophysitis: A detailed longitudinal analysis in a large cohort of patients with metastatic melanoma. J Clin Endocrinol Metab 2014; 99: 4078–4085. [DOI] [PubMed] [Google Scholar]
- 38. Kobayashi T, Iwama S, Yasuda Y, et al. Pituitary dysfunction induced by immune checkpoint inhibitors is associated with better overall survival in both malignant melanoma and non‐small cell lung carcinoma: A prospective study. J Immunother Cancer 2020; 8: e000779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Takahashi Y. Paraneoplastic autoimmune hypophysitis: A novel form of paraneoplastic endocrine syndrome. Endocr J 2023; 70: 559–565. [DOI] [PubMed] [Google Scholar]
- 40. Kanie K, Iguchi G, Bando H, et al. Mechanistic insights into immune checkpoint inhibitor‐related hypophysitis: A form of paraneoplastic syndrome. Cancer Immunol Immunother 2021; 70: 3669–3677. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Teulings HE, Limpens J, Jansen SN, et al. Vitiligo‐like depigmentation in patients with stage III‐IV melanoma receiving immunotherapy and its association with survival: A systematic review and meta‐analysis. J Clin Oncol 2015; 33: 773–781. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Logistic regression analysis of factors associated with ICI‐T1DM risk.
Table S2. Multivariable Cox regression analysis of mortality.
Table S3. Multivariate Cox regression analysis of mortality (sensitivity analysis including steroid administration prior to ICI administration).
Figure S1. Time from ICI initiation to ICI‐T1DM onset (n = 102).
Figure S2. Incidence of ICI‐T1DM by ICI types.
Figure S3. Kaplan–Meier survival curve of patients with or without ICI‐T1DM. (A) Stratified by age (≤70, group 1; >70, group 2). (B) Stratification by cancer type (LK, group 1; other than LK, group 2).
Data Availability Statement
The data that support the findings of this study are available from DeSC Healthcare Inc. (Tokyo, Japan) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are, however, available from the corresponding author upon reasonable request and with permission of DeSC Healthcare Inc. (Tokyo, Japan).