Triage for Early Recognition of Immune-Related Adverse Events in Patients Receiving Immune Checkpoint Inhibitors: A Multicenter Prospective Diagnostic Accuracy Study

Background: 

Immune checkpoint inhibitors (ICIs) have transformed treatment across multiple solid tumors, but immune-related adverse events (irAEs) occur in approximately 40% of treated patients across all grades and require prompt recognition to limit morbidity. In settings with a high patient-to-oncologist ratio, nurse-led triage may extend safe monitoring capacity, but prospective validation against a physician reference assessment has been lacking in Indian cancer centers. 

 Objective: 

To evaluate the diagnostic accuracy of a structured Nurse-Led irAE Triage Checklist (NLITC) for identifying clinically significant irAEs (Common Terminology Criteria for Adverse Events [CTCAE] v5.0 grade ≥ 2) among patients receiving ICI therapy at two tertiary oncology centers in India. 

 Methods:

In this prospective, two-center diagnostic accuracy study, consecutive adult patients receiving ICI monotherapy or combination therapy were assessed at infusion visits and through a 24-hour nurse-staffed helpline using the NLITC. Each nurse-triage assessment was compared with a blinded oncologist reference assessment; actionable encounters underwent same-day physician assessment, whereas non-actionable encounters underwent assessment at the next scheduled infusion visit. Sensitivity, specificity, predictive values, likelihood ratios, and Cohen's kappa were calculated. 

 Results:

Of 262 enrolled patients, 258 completed at least one paired assessment (612 total encounters). A clinically significant irAE was confirmed by the reference standard in 94 of 258 patients (36.4%). Nurse-led triage showed a sensitivity of 94.7% (95% CI, 88.1–97.7) and specificity of 90.2% (95% CI, 84.7–93.9), with positive and negative predictive values of 84.8% and 96.7%, respectively. Agreement in actionable/non-actionable classification was high (κ = 0.83; 95% CI, 0.74–0.90). Median time from symptom report to completed nurse triage was 24 minutes (IQR, 15–40). 

 Conclusion:

In this prospective two-center Indian cohort, a structured nurse-led triage checklist demonstrated high sensitivity and strong agreement with oncologist assessment for identifying clinically significant irAEs. Because the study used asymmetric verification timing and included repeated encounters, the findings should be considered hypothesis-generating. Further multicenter validation with symmetric reference-standard verification and analyses that account for repeated encounters is warranted.