Case report: Strong GAD antibody positivity and type 1 diabetes-HLA-susceptible haplotype-DRB1*04:05-DQB1*04:01 in a Japanese patient with immune checkpoint inhibitor-induced type 1 diabetes

Immune checkpoint inhibitors (ICIs) are widely used in cancer treatment; however, they can lead to immune-related adverse events, including immune checkpoint inhibitor-induced type 1 diabetes mellitus (ICI-T1DM). While fulminant T1DM is common in East Asia, ICI-T1DM has predominantly been reported in Western countries. In this report, we present the case of a 66-year-old Japanese man with type 2 diabetes mellitus undergoing dialysis for diabetic nephropathy. The patient was diagnosed with left upper lobe lung cancer, and treatment with nivolumab and ipilimumab was initiated. After 48 days, the patient experienced impaired consciousness and difficulty moving. His blood glucose levels were 815 mg/dL, and metabolic acidosis was detected, leading to a diagnosis of diabetic ketoacidosis. The patient was subsequently treated with continuous intravenous insulin. However, his C-peptide levels rapidly depleted, and new-onset ICI-T1DM was diagnosed. Although most Japanese patients with ICI-T1DM test negative for glutamic acid decarboxylase (GAD) antibodies, this case exhibited a strong positivity. Thus, we reviewed the literature on 15 similar Japanese cases, revealing a mean HbA1c level at onset of 8.7% and a mean time from ICI administration to onset of 9.7 weeks, which was shorter than that in GAD-negative cases. Moreover, human leukocyte antigen typing revealed five cases of DRB1*04:05-DQB1*04:01, including the present case, and one case of DRB1*09:01-DQB1*03:03, both of which were susceptible to T1DM haplotypes. These findings suggest that GAD antibody positivity may be associated with acute onset and disease progression in some cases of Japanese patients with ICI-T1DM. Given that the prediction of new-onset ICI-T1DM is challenging, monitoring GAD antibody levels might be useful. However, further studies with large sample sizes and validation across different racial and ethnic populations are warranted.


Introduction
Immune checkpoint inhibitors (ICIs) have been widely used for the treatment of many cancers owing to their potent anti-tumor effects, achieved by inhibiting molecules such as cytotoxic Tlymphocyte antigen 4 (CTLA-4), programmed cell death 1 (PD-1), and programmed cell death ligand 1 (PD-L1) (1).However, immune-related adverse events following the use of ICIs have been increasingly reported (2,3).Fulminant type 1 diabetes mellitus (T1DM) is prevalent in East Asia, including Japan (4).ICI-induced T1DM (ICI-T1DM) has been widely reported in many Western countries, and its onset resembles that of fulminant T1DM (5).ICI-T1DM and acute T1DM are similar yet different conditions classified as "checkpoint inhibitor-associated autoimmune diabetes mellitus (CIADM)" (6), and ICI-T1DM has been recognized as a new category within T1DM.

Case description
A 66-year-old man with a history of diabetes since 43 years of age had been under treatment with antidiabetic agents, including alpha-glucosidase inhibitors, sulfonylureas, and thiazolidinediones.However, owing to inadequate control, he required the administration of a subcutaneous injection of long-acting insulin.At 47 years of age, his serum C-peptide concentration was 0.9 ng/ mL.At 59 years of age, he required hemodialysis for diabetic nephropathy.He was treated with 6 units of insulin glargine subcutaneously injected before bedtime and 0.3 mg of voglibose before each meal at another hospital.His fasting blood glucose ranged from 120 to 180 mg/dL, and his HbA1c levels ranged from 6.0 to 7.3%.
At 66 years of age, at our hospital, immunotherapy with nivolumab and ipilimumab was initiated for left upper lobe lung cancer (c-T2aN1M1a Stage 4A), with the initiation of immunotherapy defined as day 1.After starting immunotherapy, the patient experienced diarrhea and itching, which were considered side effects of immunotherapy.Accordingly, treatment with 10 mg/day of prednisolone was initiated concurrently with the third nivolumab administration on day 43.
On the fifth day of the third nivolumab administration (day 47), the patient began experiencing symptoms of fatigue, followed by difficulty moving the next morning, prompting hospitalization on that day.He had a history of hypertension and dyslipidemia but no relevant family history.His physical parameters were as follows: height, 164 cm; weight, 62.1 kg; body mass index, 23.1 kg/m 2 ; blood pressure, 155/137 mmHg; pulse, 93/min; body temperature, 36.8°C; and oxygen saturation, 95% (ambient air).The patient was unconscious; however, no obvious paralysis or neck rigidity was observed.The blood test results are presented in Table 1.His blood glucose level was 815 mg/dL, with HbA1c elevated to 8.7%.Plasma osmolality was 318 mOsm/L, and ketosis and metabolic acidosis were noted.His serum C-peptide level was 0.42 ng/mL, and the GAD antibody level exceeded 2000 U/mL.Chest radiography and computed tomography (Figure 1) confirmed known lung cancer, Abbreviations: ICI, immune checkpoint inhibitor; CTLA-4, cytotoxic Tlymphocyte antigen 4; PD-1, programmed cell death 1; PD-L1, programmed cell death ligand 1; irAE, immune-related adverse event; T1DM, type 1 diabetes mellitus; GAD, glutamic acid decarboxylase; HLA, human leukocyte antigen; ICI-T1DM, immune checkpoint inhibitor-induced type 1 diabetes mellitus; CIADM, checkpoint inhibitor-associated autoimmune diabetes mellitus; IA-2, insulinoma-associated antigen-2; ZnT8, zinc transporter 8. with no additional findings.Based on these findings, the patient was clinically diagnosed with diabetic ketoacidosis (DKA).

Yabuki
Upon admission, a central venous catheter was inserted.As the patient was on dialysis, the infusion was minimized, and treatment focused on a continuous intravenous insulin infusion, ranging from 0.2 to 5.0 U/h.By day 3, the patient's consciousness had gradually improved.As his blood glucose level decreased and his metabolic acidosis improved, continuous intravenous insulin infusion was terminated, and subcutaneous insulin injection was initiated (14-25 U/day).However, as blood glucose levels fluctuated, insulin injection levels were finely adjusted.On the 17th day of hospitalization (day 64), the patient was finally discharged with multiple daily injections of insulin aspart 6-6-4 and insulin glargine 0-0-0-9.His serum C-peptide levels decreased to 0.26 ng/mL and <0.03 ng/mL on days 63 and 94, respectively.Given his medical history and highly positive anti-GAD antibody test results, we diagnosed the patient with autoimmune T1DM caused by ICIs.HLA typing identified DRB1*04:05-DQB1*04:01, which is reportedly associated with T1DM susceptibility (Table 1).Following discharge, immunotherapy was resumed on day 85.Unfortunately, tumor growth persisted despite ICI treatment.Therefore, immunotherapy was discontinued six months postdischarge, and chemotherapy with carboplatin and paclitaxel was initiated as the second-line treatment.After completing six courses of treatment, tumor shrinkage was observed, and the patient is still under treatment and follow-up.The clinical course of the patient is summarized in Figure 2.

Discussion
In this report, we describe a case of strong GAD antibody positivity and T1DM-susceptible haplotype DRB1*04:05-DQB1*04: 01 in a Japanese patient with ICI-T1DM.This case presents several notable findings.First, although the prevalence of GAD antibody positivity in ICI-T1DM is lower in Japan than in Western countries, our patient exhibited a high GAD antibody titer, along with a remarkably short duration between the first ICIs and the acute onset of symptoms.Second, we identified the HLA haplotype as DRB1*04:05-DQB1*04:01, susceptible to T1DM haplotypes.Considering these findings, below we have discussed and reviewed the literature on GAD antibody-positive Japanese patients with ICI-T1DM and HLA haplotypes.

A brief review of ICI-T1DM
ICI-T1DM, also known as CIADM, has recently been recognized as a new category of T1DM (6).ICIs may inhibit PD-1 on the surface of T cells or PD-L1 on the surface of tumor cells, thereby activating T cells and enhancing tumor immunity.As PD-L1 is expressed in Langerhans cells, ICIs may block the PD-1 pathway, potentially activating self-reactive T cells to attack Langerhans cells (11,16).

GAD antibody positivity in acute-onset Japanese ICI-T1DM
In our case, our patient exhibited a high GAD antibody titer with a considerably short symptom onset duration from the first ICIs.The positivity rate of islet-related antibodies in ICI-T1DM is reportedly approximately 26-56%, which is lower than that in T1DM (5, 10-13, 23, 38).Additionally, Qiu et al. reported a positivity rate of 45.7% and 9.5% in islet-related antibodies in Caucasians and Asians with ICI-T1DM, respectively (5).Notably, the positivity rate of islet-related antibodies in ICI-T1DM in Asians is low (5,9,24).To further elucidate these characteristics, we searched for Japanese patients with GAD antibody-positive ICI-T1DM.
To the best of our knowledge, so far, 15 cases have been reported, as summarized in .The mean age of the patients was 65.1.The HbA1c level at the time of ICI-T1DM onset was 8.7%, and the duration of ICI-T1DM onset after the first ICIs  Clinical course.On the 48th day after the start of immunotherapy, the blood glucose level spiked; thereafter, the blood glucose levels fluctuated.The HbA1c and C-peptide levels increased and decreased rapidly, respectively.Continuous intravenous insulin was administered at 0.2 to 5.0 U/h from day 48 to day 50.FBS, fasting blood sugar; HbA1c, glycated hemoglobin; Ipi, ipilimumab; nivolumab, nivolumab.6 weeks in islet antibody-negative and -positive ICI-T1DM cases, respectively.Therefore, the onset duration in the reviewed patients with ICI-T1DM, including the current case, was short, consistent with previous reports (5,(10)(11)(12)(13)23).In our current case, the onset duration was 47 days (6.7 weeks), which is considerably short.The prevalence of DKA is reportedly high in cases positive for islet-related antibodies (10,13).Our review indicated a DKA prevalence of 38.9%, similar to that reported by Baden et al. (9).The reasons behind the differences reported in the clinical courses between GAD-positive and GAD-negative cases in ICI-T1DM remain unclear.In T1DM, anti-islet autoantibodies are generated as a result of b-cell destruction by T cells (39), and GAD antibody titers do not always reflect disease progression in T1DM (39).However, our results suggest that GAD antibodies may serve as surrogate markers of disease progression in some ICI-T1DM cases.
Predicting new-onset ICI-T1DM is difficult, as demonstrated in our case.In a recent retrospective cohort study, Akturk et al. reported that blood glucose monitoring at every ICI infusion visit failed to detect new-onset ICI-T1DM (40).They proposed that continuous glucose monitoring or daily self-monitoring may be useful for some high-risk patients (40).Although we acknowledge their suggestion, in our case, despite daily self-monitoring, early detection of new-onset ICI-T1DM was not achieved.Initially, we attributed the elevated glucose levels to prednisolone, delaying the recognition of acute glucose elevation and subsequent unconsciousness.Therefore, identifying high-risk patients could facilitate early intervention, such as initiating early insulin injections or increasing the dose, to prevent worsening metabolic disorders related to elevated glucose levels.
The onset of ICI-T1DM is unpredictable, suggesting that monitoring GAD antibody titers could serve as a valuable tool for detecting new-onset ICI-T1DM.However, there are currently no clear criteria for the timing of its measurement.One of the reasons for this is that the pattern of change in GAD antibody titer varies, with some cases reported to have been positive before the start of immunotherapy (41,42), others to have become positive after the start of immunotherapy (43), and still others to have increased antibody titers after the start of immunotherapy (29, 30).Considering this issue, we propose that anti-GAD antibodies should be measured before immunotherapy, and positive cases should be considered high-risk patients and periodically monitored after the start of immunotherapy to confirm that the titer does not increase rapidly.Another issue is that this assessment plan cannot be applied to cases that become positive for GAD antibodies after the start of immunotherapy.As mentioned above, the average period from the start of immunotherapy to the onset of disease in Japanese ICI-T1DM patients with positive GAD antibodies is 9.7 weeks (n=15).In this regard, measuring GAD antibodies once, for example, a few weeks after the start of immunotherapy, may be useful.While it is difficult to suggest a timing for GAD antibody measurement at present, this is a clinically important issue, and further studies are needed in this regard.
In addition, although islet antigen-2 (IA-2) and zinc transporter 8 (ZnT8) antibodies have been reported to be surrogate markers of pancreatic beta cell destruction (39), their titers were not high in our review (Table 2).Wu et al. reported that multiple autoantibodies are less frequently detected in ICI-T1DM than in T1DM (13), possibly indicating low positivity rates for IA-2 and ZnT8 antibodies in patients with ICI-T1DM.Therefore, anti-islet antibodies, such as GAD or IA-2 antibodies, may play different roles in T1DM and ICI-T1DM, and future studies should focus on elucidating their roles.The association between HLA haplotypes and ICI-T1DM remains unclear.However, the HLA haplotype of our patient was identified as DRB1*04:05-DQB1*04:01, a T1DM-susceptible haplotype.
Second, an association between GAD antibody positivity and HLA haplotypes has been reported.(49), which is inconsistent with the observations from our review.These findings suggest that susceptible HLA-related haplotypes differ in Japanese patients with ICI-T1DM positive for GAD antibodies, warranting further studies with larger sample sizes.

Strengths and limitations
In this study, we reviewed Japanese patients with GAD antibody-positive ICI-T1DM and their associated HLA haplotypes.Our study highlights several important findings.First, GAD antibody positivity may be associated with acute onset and disease progression in some cases of Japanese patients with ICI-T1DM.Second, as the prediction of new-onset ICI-T1DM is challenging, monitoring GAD antibody levels might be useful.Third, DRB1*04:05-DQB1*04:01 is a potentially important HLA haplotype of GAD antibody-positive ICI-T1DM in Japanese patients.These findings provide important insights into guiding clinicians in treating patients undergoing ICI therapy.
This study has several strengths.First, as only a few cases of GAD antibody-positive Japanese ICI-DM have been reported, the characteristics of this condition remain unclear.To the best of our knowledge, this is the first review focusing on GAD antibodypositive Japanese patients with ICI-T1DM.Recently, various rare onsets and related cases of ICI-related T1DM have been reported (50)(51)(52), and our study adds valuable inputs for understanding the pathogenesis and management of this condition.Second, we have discussed the HLA haplotype of GAD-positive antibodies in a small number of Japanese ICI-T1DM cases, warranting future investigations with larger sample sizes and validation across different racial and ethnic groups.
However, this study has some limitations.First, we could not measure GAD antibody titers before ICI treatment.Second, data on islet-associated autoantibodies other than GAD antibodies are limited.Third, we could not determine the HLA haplotypes of DPA1 and DPB1 owing to phase ambiguity.Fourth, although we carefully searched the literature, some reports did not contain detailed information, and we could not draw detailed clinical manifestations in some cases.Finally, no individual GAD-positive cases were identified in several observational studies.

Conclusions
In summary, we present a case of strongly positive GAD antibody and T1DM-susceptible haplotype DRB1*04:05-DQB1*04:01 detected in a Japanese patient with ICI-T1DM.As the high-GAD antibody titer ICI-T1DM may be acute and have a short onset from the first ICIs, clinicians should carefully observe patients.However, our results should be validated in studies with large sample sizes and across different racial and ethnic groups.

FIGURE 1 Contrast
FIGURE 1Contrast-enhanced computed tomography on admission.Contrastenhanced computed tomography showing left upper lobe lung cancer (arrowheads).

TABLE 1
Patient laboratory data on admission.

TABLE 2
(9)iew of anti-GAD antibody-positive Japanese ICI-T1DM cases.DKA was observed in 6 cases, 3 cases had type 2 DM, and 3 cases had slowly progressive insulin-dependent DM.Baden et al. reported that the mean onset duration from the first ICI was 22.1 weeks in 22 Japanese ICI-T1DM cases, with only one patient testing positive for GAD antibodies(9).Similarly, Wu et al. reported a mean onset duration from the first ICI of 21.5 and 15.