The rapid expansion of immune checkpoint inhibitors, immune effector-cell and T-cell-engaging therapies, Antibody-Drug Conjugates (ADCs), and molecularly targeted agents has transformed both clinical research and standard oncology care. These advances have improved survival across many malignancies, yet they have also introduced a heterogeneous spectrum of adverse events that may be inflammatory, on-target, off-tumor, payload-related, or pharmacologic. Toxicities vary in onset and urgency, may emerge after therapy has stopped, and can affect virtually any organ system [1]. Safe treatment therefore increasingly depends on coordinated expertise that extends beyond the traditional scope of medical oncology.

Clinical practice guidelines have evolved in parallel. Recommendations from the American Society of Clinical Oncology (ASCO), National Comprehensive Cancer Network (NCCN), Society for Immunotherapy of Cancer (SITC), and European Society for Medical Oncology (ESMO) define toxicity grading, diagnostic evaluation, treatment holds, immunosuppression, and specialist consultation for many syndromes [2–6]. Cellular-therapy guidance similarly standardizes Cytokine Release Syndrome (CRS) and Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS) grading and management [7–9]. These documents establish what clinicians should do; they generally do not define who receives the first call, how quickly a specialist responds, who owns follow-up across care settings, or how an institution measures whether the recommended care was delivered. That operational gap is the focus of the framework described here.

The initiative covers the treatment strategies and representative agents shown in Table 1. The list is intentionally selective rather than exhaustive: the operational requirement is to map every newly adopted therapy to its sentinel toxicities, baseline and interval monitoring, red-flag symptoms, first-contact owner, and specialty escalation route before routine deployment.

Serious treatment toxicity often begins with nonspecific findings such as fatigue, rash, diarrhea, dyspnea, visual disturbance, weakness, hypotension, or an isolated laboratory abnormality. The same presentation may reflect infection, cancer progression, thrombosis, metabolic derangement, a concomitant medication, or treatment toxicity; empiric immunosuppression may be lifesaving in some syndromes but harmful when infection has not been considered. Combination regimens further complicate attribution: for example, checkpoint inhibitor plus VEGF-pathway inhibition can produce overlapping gastrointestinal, hepatic, endocrine, renal, cardiovascular, dermatologic, and fatigue syndromes. The system must therefore support simultaneous danger recognition, exclusion of mimics, agent-specific management, and early organ-specialist involvement rather than sequential ad hoc consultation.

At the University of Kansas Medical Center and the University of Kansas Cancer Center, these challenges prompted development of a multidisciplinary serious-adverse-event initiative. The initial clinical network integrates oncology with endocrinology, dermatology, and ophthalmology and is expanding through named pathways in neurology, pulmonology, gastroenterology, cardiology, rheumatology, nephrology, infectious diseases, and critical care. The program is not intended to replace the treating oncologist or organ-specific guidelines. It supplies an accountable operating layer: a common entry point, urgency rules, designated specialty champions, explicit handoffs, shared documentation, and a learning system that connects clinical care, education, and research.

Five linked components define the initiative. First, standardized triage scripts identify red-flag presentations and route patients to emergency, same-day, or expedited outpatient assessment. Second, named inpatient and outpatient owners preserve continuity through admission, steroid taper, specialty follow-up, and treatment-rechallenge decisions. Third, institution-specific pathways translate national and product-specific recommendations into order sets, referral instructions, and response-time expectations. Fourth, difficult, overlapping, refractory, recurrent, and rechallenge cases are reviewed in a multidisciplinary forum. Fifth, a structured registry captures treatments, organ systems, diagnostic certainty, management, and outcomes so that pathway performance can be audited and refined.

Figure 1 provides a therapy-agnostic front door for suspected toxicity. It is not a substitute for organ-specific management. Its purpose is to prevent delay while the diagnosis remains uncertain: danger is assessed first, the anticancer exposure and timing are reconciled, CTCAE grade and a working syndrome are assigned, mimics are evaluated in parallel, and a named clinician owns the next reassessment.

Figure 1. Operational triage and co-management algorithm for suspected treatment-related toxicity. The workflow organizes access, danger screening, diagnostic evaluation, accountable follow-up, and feedback; organ-specific clinical care remains governed by current guidelines and product-specific recommendations [2–10].

This initiative is supported by an institutional environment that integrates clinical care, translational research, and early-phase drug development. The University of Kansas Cancer Center is an NCI-designated comprehensive cancer center operating as a matrix organization across multiple campuses and partner institutions [16]. Rapid adoption of novel therapies creates both the need and the opportunity to develop reusable toxicity pathways, connect community sites to subspecialty expertise, and study uncommon events across the cancer center network.

Prospective toxicity investigation is a central pillar. SWOG S2013 (I-CHECKIT; NCT04871542), available at our institution, is a multicenter observational study of patients receiving standard-of-care immune checkpoint inhibitor therapy [10,17]. In the completed checkpoint-inhibitor cohort reported in 2026, 2,020 eligible patients were evaluated: 13.3% experienced a grade 3 or higher nonhematologic immune-related Adverse Event (irAE) within one year, with a higher incidence after combination checkpoint blockade than single-agent therapy (27.4% vs 10.5%). Among grade 3 or higher events, hepatitis (21.2%), gastrointestinal disorders (19.3%), skin toxicity (11.9%), and respiratory toxicity (10.0%) were most frequent [10]. These prospective data underscore both the multisystem scope and the need to anticipate higher-risk regimens.

The institutional registry is designed to complement trial data by capturing real-world events across therapeutic classes, disease settings, and care locations. A minimum dataset includes anticancer agents and exposure dates; suspected organ system and CTCAE grade; alternative diagnoses considered; diagnostic tests and specialty input; treatment interruption, corticosteroid exposure, second-line immunosuppression, and supportive care; emergency department use, hospitalization, intensive care, readmission, time to improvement, persistent organ dysfunction, and treatment rechallenge. Standardized terminology and adjudication rules are needed because administrative codes alone may misclassify both toxicity and competing diagnoses [6].

Institutional experience with cellular therapy provides a mature model for this approach. Standardized CRS and ICANS grading, scheduled neurologic assessment, 24-hour escalation capability, pharmacy access to tocilizumab and other syndrome-directed therapy, infectious evaluation, and coordinated inpatient/outpatient follow-up define roles before a patient deteriorates [7–9]. The same design principles—predefined triggers, named responders, shared order sets, and prospective surveillance—can be adapted to checkpoint inhibitors, T-cell-engaging bispecific antibodies, ADCs, and targeted therapies without assuming that their mechanisms or treatments are interchangeable.

Although comparative evidence remains limited, published institutional programs demonstrate feasibility and signal potential benefit. A Johns Hopkins virtual multidisciplinary irAE team received 117 referrals involving 102 patients over 8 months; recommendations were delivered within 24 hours for all referrals, all surveyed referring clinicians used the recommendations, and 74% reported changing management [18]. An irAE tumor board model has similarly been used to streamline complex cases and identify recurrent service and educational needs [19].

At Massachusetts General Hospital, implementation of the multidisciplinary Severe Immunotherapy Complications Service was associated in a before-after analysis with lower irAE readmission (14.8% after vs 25.9% before; adjusted odds ratio 0.46, 95% CI 0.22–0.95) and a one-day reduction in median readmission length of stay; mortality, immunosuppressive use, and treatment discontinuation did not significantly differ [20]. These observational findings support evaluation but do not establish causality because secular changes and case-mix differences may remain.

MD Anderson's IOTOX initiative integrates standardized clinical practice, education, and clinical/translational research [21]. In a 126-patient pre/post study of hospitalized immune-mediated diarrhea and colitis, implementation of an inpatient algorithm and focused GI service increased GI consultation, shortened median time to clinical remission from 10 to 4 days, and was associated with fewer readmissions and recurrences [22]. Together, these programs provide benchmarks for response time, clinician adoption, readmission, length of stay, and organ-specific disease control.

Structured education is being consolidated into a recurring multidisciplinary symposium on serious adverse events from modern anticancer therapy. The program will combine new-agent and guideline updates with case-based sessions, organ-specific breakout discussions, nursing and Clinical Nurse Coordinator (CNC) triage training, simulation of time-critical syndromes, patient/caregiver perspectives, and review of institutional performance data. Shorter case conferences and faculty updates between symposia will translate lessons into revised scripts and pathways. This model parallels the educational and operational integration of the MD Anderson IOTOX initiative while remaining tailored to local resources and referral geography [21].

The ophthalmology pathway illustrates how a general framework becomes an actionable local process. KU Ophthalmology designated two triage technicians to receive acute oncology messages through a single telephone route (913-588-6600, option 3). Oncology teams leave a structured message identifying the anticancer agent, timing, symptoms, visual acuity when available, and callback contact; messages received during business hours are reviewed before 4 PM. Abrupt vision loss, severe ocular pain, photophobia with marked vision change, field loss, or suspected retinal, optic-nerve, or orbital emergency bypasses the routine queue for emergency assessment. This pathway is relevant to checkpoint-inhibitor uveitis and neuro-ophthalmic events, ADC-associated ocular-surface and corneal toxicity, and MEK-inhibitor retinopathy, for which prompt ophthalmic characterization may prevent unnecessary treatment discontinuation or irreversible injury [12,14].

The KU initiative is in an implementation and prospective-evaluation phase. We do not yet have a mature, analytically validated institutional cohort that can support a claim of improved outcomes; anecdotal feedback is therefore not presented as evidence of effectiveness. Evaluation will distinguish implementation outcomes from clinical outcomes and use prespecified definitions (Table 2) [23].

A pragmatic evaluation can begin with a 12-month pre-implementation baseline followed by prospective surveillance and segmented interrupted-time-series or risk-adjusted pre/post analysis. Rare fatal syndromes should be reported descriptively and pooled across sites when possible. Pathway changes, data definitions, and contextual factors should be documented in accordance with SQUIRE 2.0 principles so that positive and negative findings are interpretable [24]. Electronic patient-reported symptom monitoring may supplement, but should not replace, clear human review and escalation [25].

Implementation and service outcomes should be reported separately from clinical outcomes. Syndrome-specific definitions and response-time targets should be version controlled as guidance evolves.

Implementation barriers are predictable. Specialty capacity is finite; a new centralized service may create referrals without reducing work elsewhere. Diagnostic uncertainty can produce both delayed treatment and over-immunosuppression. Guidelines and product labels evolve faster than static order sets. Inpatient and outpatient teams may disagree about ownership, and community clinicians may lack direct access to tertiary subspecialists. Finally, documentation burden and rare-event denominators make outcome measurement difficult.

Mitigation requires tiered urgency criteria, named alternates for specialty champions, concise EHR templates, pharmacy and nursing participation, scheduled pathway review, and protected administrative/data support. Infectious disease and critical-care input should be incorporated when immunosuppression or organ failure is likely. Governance should specify who may activate a pathway, who can recommend treatment interruption, how disagreement is resolved, and when a case returns to the treating oncologist.

Important limitations remain. This article describes a single-center framework at an early stage, and the workflow has not yet been shown to improve KU patient outcomes. The external studies cited are largely observational and focused on checkpoint-inhibitor toxicity; their findings may not transfer to ADCs, cellular therapy, bispecific antibodies, or targeted agents. The representative drug/toxicity table cannot replace current prescribing information. Resource needs, reimbursement, and referral geography will vary, and faster specialist access could increase low-value consultation if triage criteria are not audited.

As anticancer therapy becomes more mechanistically diverse, safe delivery will depend not only on drug efficacy and published toxicity recommendations but also on reliable systems for recognition, escalation, co-management, and learning. The proposed KU framework converts guidance into an operational model with a common triage architecture, therapy-and organ-specific pathways, accountable handoffs, multidisciplinary education, and prospective measurement. Its effectiveness remains to be established locally; publishing prespecified implementation and clinical outcomes, including negative findings and balancing measures, will be essential. Transferability to other settings should not be inferred until the model has been evaluated prospectively and its resource requirements and contextual dependencies are described.