Checkpoint inhibitors have an anticancer effect by removing a negative regulatory signal for T-cell activation from the tumor microenvironment (Fig. 1). They include cytotoxic T-cell–associated antigen (CTLA-4), programmed cell death protein-1 (PD-1), and programmed cell death ligand-1 (PDL-1) antibodies and are now being widely used for the treatment of different types of cancer. From the initial phases of checkpoint inhibitor use, there has been concern about the potential for the development of autoimmune disease as a result of T-cell activation. Subsequently, multiple autoimmune diseases were indeed observed as a result of these medications (1). Although both PD-1 and PDL-1 antibodies can precipitate type 1 diabetes in the nonobese diabetic mice model (2), only very recent reports have noted type 1 diabetes after PD-1 antibody use in humans (3,4). Here, we describe two older adults without diabetes receiving agents inhibiting the PD-1 pathway for resistant cancers who developed acute type 1 diabetes.
T-cell signaling and PD-1 pathway. T cells activate or deactivate through two signals. The major signal is always delivered through the T-cell receptor (TCR) after binding to the MHC. The second signal can be stimulatory (such as the binding between CD28 on the T cell and its ligand B7) or inhibitory (the binding between CTLA-4 or PD-1 and their ligands). If the second signal is stimulatory, phosphorylation of the downstream pathway is promoted, which results in interleukin-2 (IL-2) and B-cell lymphoma-extra large (Bcl-xl) production and T-cell activation. However, an inhibitory second signal results in deactivation by inhibiting the downstream pathway. Ag, antigen.
T-cell signaling and PD-1 pathway. T cells activate or deactivate through two signals. The major signal is always delivered through the T-cell receptor (TCR) after binding to the MHC. The second signal can be stimulatory (such as the binding between CD28 on the T cell and its ligand B7) or inhibitory (the binding between CTLA-4 or PD-1 and their ligands). If the second signal is stimulatory, phosphorylation of the downstream pathway is promoted, which results in interleukin-2 (IL-2) and B-cell lymphoma-extra large (Bcl-xl) production and T-cell activation. However, an inhibitory second signal results in deactivation by inhibiting the downstream pathway. Ag, antigen.
The first patient was a 70-year-old euglycemic male who was started on a PDL-1 antibody for advanced adenocarcinoma of the lung. After 15 weeks and five doses of the medication, he was noted to have plasma glucose of 512 mg/dL, and as he did not have any history of diabetes, he was started on metformin. Ten days later he presented with diabetic ketoacidosis (DKA) (Table 1). He remained insulin dependent and died of his advanced cancer 7 months later.
Patients’ characteristics at the time of presentation with DKA
| Patient 1 | Patient 2 | |
|---|---|---|
| BMI, kg/m2 | 23.2 | 15.1 |
| Plasma glucose, mmol/L | 22.83 | 41.77 |
| Anion gap (3–11) | 18 | 22 |
| Bicarbonate (22–32 mEq/L) | 15 | 7 |
| Arterial pH | Unavailable | 7.09 |
| A1C [4–6% (20–42 mmol/mol)] | 9.8 (84) | 9.4 (79) |
| Thyroid-stimulating hormone (0.4–5.0 μIU/mL) | 0.944 | 34.19 |
| Total triiodothyronine (73–178 ng/dL) | 126 | 26 |
| Free triiodothyronine (2.3–3.9 pg/mL) | — | 1.7 |
| Total thyroxine (4.8–10.8 μg/dL) | 13.2 | — |
| Free thyroxine (0.6–1.2 ng/dL) | — | 0.3 |
| Thyroid peroxidase antibody (0.0–8.9 IU/mL) | — | 83.3 |
| C-peptide (1.0–7.1 ng/mL) | 0.3 | <0.1 |
| GAD65 antibody | 0 (≤0.02 nmol/L)* | 465 (<142 WHO units) |
| Insulin autoantibody | 0 (0.0–0.02 nmol/L)* | 0.02 (<0.05 index) |
| IA2 autoantibody | — | 11 (<21 WHO units) |
| ZnT8 autoantibody | — | 0.00 (<0.033 index) |
| IA2β/phogrin | — | 0.001 (<0.0015 index) |
| HLA+ | — | DR3-DQ2/DR4-DQ8 |
| Islet-specific T cell^ | — | 7 positive protein bands |
| Patient 1 | Patient 2 | |
|---|---|---|
| BMI, kg/m2 | 23.2 | 15.1 |
| Plasma glucose, mmol/L | 22.83 | 41.77 |
| Anion gap (3–11) | 18 | 22 |
| Bicarbonate (22–32 mEq/L) | 15 | 7 |
| Arterial pH | Unavailable | 7.09 |
| A1C [4–6% (20–42 mmol/mol)] | 9.8 (84) | 9.4 (79) |
| Thyroid-stimulating hormone (0.4–5.0 μIU/mL) | 0.944 | 34.19 |
| Total triiodothyronine (73–178 ng/dL) | 126 | 26 |
| Free triiodothyronine (2.3–3.9 pg/mL) | — | 1.7 |
| Total thyroxine (4.8–10.8 μg/dL) | 13.2 | — |
| Free thyroxine (0.6–1.2 ng/dL) | — | 0.3 |
| Thyroid peroxidase antibody (0.0–8.9 IU/mL) | — | 83.3 |
| C-peptide (1.0–7.1 ng/mL) | 0.3 | <0.1 |
| GAD65 antibody | 0 (≤0.02 nmol/L)* | 465 (<142 WHO units) |
| Insulin autoantibody | 0 (0.0–0.02 nmol/L)* | 0.02 (<0.05 index) |
| IA2 autoantibody | — | 11 (<21 WHO units) |
| ZnT8 autoantibody | — | 0.00 (<0.033 index) |
| IA2β/phogrin | — | 0.001 (<0.0015 index) |
| HLA+ | — | DR3-DQ2/DR4-DQ8 |
| Islet-specific T cell^ | — | 7 positive protein bands |
Normal ranges given in parentheses where appropriate. WHO, World Health Organization.
Measured only at the clinical laboratory.
Genotyping at the HLA class II locus (IDDM1) was performed using the direct sequencing of separately amplified exons 2 of DRB1, DQA1, and DQB1.
The second patient was a 66-year-old female without any personal or family history of diabetes who was started on a PD-1 antibody for sarcomatoid squamous cell carcinoma of the jaw. Seven weeks after starting this therapy and after total of three doses, she presented with DKA. Detailed immunogenetic testing was consistent with autoimmune type 1 diabetes (Table 1).
We did not have the ability to study the first patient as completely as the second. His autoantibody status was negative for the two islet autoantibodies tested at a commercial laboratory, which has a lower sensitivity than research laboratories, but the very low C-peptide supports a type 1 diabetes diagnosis. The second patient’s autoimmune status or insulin secretory capacity prior to starting PD-1 therapy is unknown, but her normal plasma glucose values in the prior year, the timing of onset after the therapy, and the known effect of these agents to augment autoimmunity, all present a likely scenario for autoimmune diabetes development as a result of PD-1 pathway activation on T cells.
We conclude that anti–PD-1, and possibly anti–PDL-1, antibody use can result in a rapid progression of autoimmune diabetes in human subjects who have a high underlying genetic predisposition to type 1 diabetes, similar to what has been reported in a rodent model. Once diagnosed, autoimmune diabetes with severe insulin deficiency is a treatable disease, but the rapid presentation with DKA is important to recognize due to its morbidity and potential mortality. Physicians treating patients with this novel class of cancer therapy should be aware of this potential adverse event.
Article Information
Funding. I.B.H. has received grants from the Helmsley Charitable Trust. J.P.P. has received grants from the National Institutes of Health, which supports the T-cell studies.
Duality of Interest. I.B.H. has received grants from Sanofi and Novo Nordisk. He has also received consulting fees from Abbott Diabetes Care, Roche Laboratories, and Valeritas. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. M.M. and I.B.H. researched data, followed the second patient, and wrote the manuscript. K.D.E. and R.M. followed both patients and reviewed and edited the manuscript. B.M.B.-W. and W.A.H. performed immunologic and genetic testing and reviewed and edited the manuscript. J.P.P. reviewed and edited the manuscript. I.B.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
