Background
Recent studies have observed an association between immune-related adverse events (irAE) and favorable clinical outcomes in the setting of cancer treatment with immune checkpoint inhibitors (ICI). However, results have been variable and inconclusive. Therefore, we have conducted a pan-cancer meta-analysis evaluating the relationship between irAEs and clinical outcomes.
Materials and Methods
The search included studies published in PubMed, Embase, and Web of Science from conception to 12.28.2019 as well as abstracts published in the ASCO and ESMO meetings from 2015 to 2019. Studies were included if ICI was used in advanced or metastatic cancer settings and excluded if data contained only combination therapy regimens or contained anti-CTLA-4. Raw data for overall response rate (ORR), hazard ratios (HR), number of patients (n), and p values for overall survival (OS) and progression-free survival (PFS) were extracted. Pooled sensitivity (SN), specificity (SP), positive (PPV) and negative predictive values (NPV), and odds ratios (ORs) were calculated using the 2 × 2 table and logit transformed proportions; and summary receiver operating curve (sROC) was generated using the bivariate approach for ORR. Pooled HRs were calculated using the means weighted by inverse of the variance for OS and PFS. Heterogeneity was assumed and random effects model was used throughout the analyses.
Results
Final analysis included 32 studies, among which ORR data were available in 15 studies, OS in 17, and PFS in 16. 17 studies evaluated non-small cell lung cancer (NSCLC), two studies melanoma, one study gastric cancer, three studies renal cell carcinoma (RCC), seven studies various cancer types, two studies urothelial carcinoma, and one study head and neck cancer (HNSCC). With respect to ORR, pooled SN, SP, PPV and NPV, and OR were 0.522 [0.423–0.619], 0.810 [0.771–0.844], 0.516 [0.413–0.618], 0.819 [0.764–0.864], and 4.59 [3.24–6.50], respectively. The area under the curve (AUC) derived from the sROC was 0.773. HR for OS and PFS were 0.47 [95% CI 0.37–0.60] and 0.46 [95% CI 0.37–0.56], respectively. Between-study publication bias was present for ORR, OS, and PFS; however, results remained significant after trim-fill analysis.
Conclusion
irAEs predict OR, OS, and PFS across different types of cancer and may represent useful biomarkers in the clinical setting.
