Presentation (CTI)

OBJECTIVES: Immune checkpoint inhibitors (ICIs) are associated with immune-mediated hyperglycemia. This study aimed to compare longitudinal blood glucose trajectories between PD-1 and PD-L1 inhibitor monotherapy and to identify independent risk factors for clinically significant hyperglycemia.
METHODS: A retrospective analysis was conducted using electronic medical records from a single institution (January 2019-December 2025). Adults (≥19 years) receiving ≥2 cycles of PD-1 or PD-L1 inhibitor monotherapy were included. All available random blood glucose (RBG) measurements during treatment were collected (baseline: most recent value within 4 weeks before initiation). Glucose trajectories were evaluated using a linear mixed-effects model (LMM) with a drug class × time interaction (random intercepts and slopes, REML); risk factors were identified by multivariable logistic regression. Both models were adjusted for age, sex, BMI, pre-existing diabetes, baseline glucose, and cumulative corticosteroid dose. Clinically significant hyperglycemia was defined as RBG >160 mg/dL (≥2 occasions), >250 mg/dL, or new antidiabetic medication (without diabetes); >250 mg/dL or new antidiabetic medication (with diabetes).
RESULTS: Of 415 patients (PD-1: n=276; PD-L1: n=139), 15,454 glucose measurements were analyzed. The LMM showed no significant difference in glucose trajectories between groups (interaction β=5.99 mg/dL per 100 days, 95% confidence interval [CI]: −4.48 to 16.46, p=0.264). Hyperglycemia was observed in 166 patients (40.0%; 38.8% vs. 42.4%, p=0.538). Three independent risk factors were identified: baseline glucose (adjusted odds ratio [aOR] 1.024/mg/dL, 95% CI 1.018 to 1.030, p<0.001), cumulative corticosteroid dose (aOR 1.31/log-unit, 95% CI 1.19 to 1.45, p<0.001), and pre-existing diabetes (aOR 1.88, 95% CI 1.14 to 3.11, p=0.014). ICI class was not a significant predictor (aOR 0.91, p=0.701; AUC=0.812).
CONCLUSIONS: After accounting for individual variability, glucose trajectories did not differ between PD-1 and PD-L1 inhibitor groups. Hyperglycemia risk was driven by patient-specific factors—elevated baseline glucose, corticosteroid exposure, and pre-existing diabetes—rather than ICI class, supporting individualized glucose monitoring regardless of drug class.