7760 Identifying Conventional and Novel Biomarkers to Predict Checkpoint Inhibitor Associated Autoimmune Diabetes

Wu L, Wentworth J, Liddle C, Carlino M, Brown D, Clifton-Bligh RJ, Long G, Sasson S, Tsang V, Menzies A, Gunton JE. Journal of the Endocrine Society, 8, no. Supplement_1, (Oct 05 2024), doi:10.1210/jendso/bvae163.803.

Abstract

Introduction: Checkpoint inhibitor associated autoimmune diabetes (CIADM) is a rare but highly morbid complication of immune checkpoint inhibitor (ICI) therapy. As indications for ICIs expand, the ability to predict CIADM has huge potential value and has yet to extensively explored.

Aims: To identify potential biomarkers for prediction of CIADM in patients commencing ICI therapy.

Methods: 14 patients with metastatic melanoma treated with ICI who subsequently developed CIADM were identified. 28 matched controls were identified (matched for ICI type, gender, cancer response and other immune related adverse events). Pre-treatment, on ICI and post CIADM serum and PBMCs were analysed. Serum was analysed for type 1 diabetes autoantibodies, C-peptide, glucose and an immune cytokine panel of TNFα, IL-2, IL-4, IL-6, IL-10, CXCL10, IL-1β, CCL2, IL-17A, CXCL8, TGF-B1, and IL-12p70. PBMCs were sorted using a BD Influx III cell sorter into 1000 CD8 cell subsets per sample. RNA was extracted and sequenced using a NovaSeq X with approximately 10 million 150bp paired end reads. RNA-Seq analysis was performed using edgeR.

Results: Pre-ICI-treatment anti-GAD had a positive predictive value for CIADM prediction of 63.6% and a negative predictive value of 82.1%. Anti-IA2, anti-IAA and anti-ZnT8 were not predictive. C-peptide fell rapidly from 1.8nmol/L on treatment to 0.18nmol/L post diagnosis for CIADM and remains normal in controls. There were no significant differences in circulating cytokines between CIADM patients and controls. Pathway enrichment analysis performed on differentially expressed genes in people who developed CIADM identified significant regulation in pathways for B cell receptor and interferon signalling, RAS signalling and IGF1R signalling. These changes were not observed in controls.

Conclusion: Anti-GAD has some predictive value for CIADM, but is incompletely sensitive. Further research is needed to prospectively test the predictive value of GAD Ab in combination with other potential markers such as pancreatic volumetry as we have previously published.