Development of a novel risk stratification model for immune-related adverse events for patients with advanced melanoma and non-small cell lung cancer treated with immune checkpoint inhibitors.
Bibliographic record
Abstract
2649 Background: Immune checkpoint inhibitors (ICI) transformed treatment paradigm across cancers. There remain few reliable, clinically accessible predictors of ICI-induced immune related adverse events (irAEs). We derive a novel risk stratification model for irAEs using baseline patient, tumor and treatment variables in a large cohort of patients with advanced melanoma (AM) or non-small cell lung cancer (NSCLC) treated with ICI. Methods: We conducted a multi-centre retrospective observational cohort study of consecutive patients with AM or NSCLC receiving ≥ 1 cycle of single-agent or combination ICI, in any line, 2015 - 2023, in Alberta, Canada. Clinically significant irAEs, defined as those requiring treatment delay or systemic steroids/steroid sparing agents, were identified as outcome of interest. The association between irAEs, overall survival (OS), and time to next treatment (TTNT) was assessed with Cox Proportional Hazards regression. Stepwise logistic regression was used to select and weight baseline variables associated with development of irAEs to derive a predictive risk score. Model validation was carried out on 500 iterations of bootstrapped samples. Harrel’s C-index was calculated and internally validated to ascertain the model’s discriminatory performance. Results: 1,292 total patients were included, 519 (218/489 [44.6%] AM and 301/803 [37.5%] NSCLC) developing a clinically significant irAE. Using a subset of 801 patients with available baseline characteristics, the following variables were identified and weighted for creation of risk model (risk score attributed): tumor type (NSCLC) (+1), age >60 (+1), ECOG ≥1 (-1), BMI ≥ 25 (+1), ICI after first line (-1), combination ICI (+4), >10 cycles of ICI (+2), albumin level < LLN (+2), adrenal metastasis (+1), multiple sites of metastasis (-1). Patients were stratified into 3 irAE risk groups based on combined score: low (n = 230, risk score ≤ 0), intermediate (n = 412, risk score 1-2), high (n = 159, risk score ≥3). The risk model performed well with an optimism-corrected c-index of 0.707 in internal validation, and strong association with odds of irAE occurrence (Table). The development of irAE was associated with an improvement in OS (HR 0.48, 95% CI 0.41-0.44, p<0.001), with a median OS of 34.3 (95% CI 28.8-39.6) months compared to 12.0 (95% CI 10.6-14.3) months for those who did not develop an irAE. Similar robust stratification was also seen with TTNT. Conclusions: We presented and internally validated, a simple risk stratification tool that utilizes readily available baseline patient, tumor, and treatment characteristics to robustly stratify risk of irAE development. [Table: see text]
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".