The immune-related symptom web in immune checkpoint inhibitor therapy: a longitudinal clinical safety surveillance framework for immune-related adverse events

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Immune checkpoint inhibitors (ICIs) produce durable antitumor activity by releasing inhibitory immune pathways, but the same pharmacology can disrupt peripheral tolerance and generate immune-related adverse events (irAEs) that are delayed, recurrent, chronic, multisystem, and clinically nonspecific. Organ-specific guidelines are indispensable once toxicity is suspected; however, clinical surveillance commonly begins with an undifferentiated patient report or an objective abnormality rather than a confirmed organ diagnosis. This narrative review integrates clinical pharmacology, immunobiology, guidelines, pharmacovigilance principles, patient-reported outcome research, digital monitoring studies, and multidisciplinary implementation evidence. Focused, non-systematic searches of PubMed/MEDLINE and the Cochrane Library, supplemented by current guidelines and official regulatory sources, were finalized on 18 July 2026; representative strategies, an evidence-provenance map, and a structured comparison with existing approaches are provided in the Supplementary Material. We define an individual-level safety alert as a new symptom, functional change, laboratory abnormality, physiological change, imaging finding, or other clinically meaningful observation requiring contextual assessment. It is not equivalent to a formal population-level pharmacovigilance signal. The immune-related symptom web is a symptom-centered but not symptom-exclusive model that organizes exposure chronology, individual baseline, symptom and objective-data combinations, trajectories, competing diagnoses, corticosteroid tapering, discontinuation, rechallenge, and consequence-weighted urgency. The Signal-Attribution-Escalation-Longitudinal Management (SAELM) framework translates this information structure into an accountable workflow from alert capture to attribution, escalation, guideline-linked management, documentation, and reassessment. Its novelty lies in integrating and longitudinally organizing existing components rather than inventing each component. The Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) and electronic patient-reported outcome (ePRO) systems are positioned as capture and communication tools, not diagnostic or causal instruments. The models remain conceptual and hypothesis-generating. The most feasible first validation study is a prospective multicenter observational cohort with independent clinical adjudication; the primary safety outcome should be the false-negative proportion for severe or organ-threatening irAEs, accompanied by alert burden, response time, and potentially avoidable urgent escalation.