Validation of prototype biomarkers to identify risk factors of inflammatory adverse events (iAEs) following idecabtagene vicleucel (ide-cel) infusion in patients with relapsed and refractory multiple myeloma (RRMM) in KarMMa-3.

person Sanhita Sengupta Bristol Myers Squibb, Princeton, NJ info_outline Sanhita Sengupta, Clara Amorosi, Mandeep Takhar, Yi Lin, Salomon Manier, Rachid C. Baz, Krishna Rangadhamarao Juluri, Allison Kaeding, Afshin Mashadi-Hossein, Julia Piasecki, Timothy Brandon Campbell, Shari Kaiser, Julie Rytlewski, Md Shamsuzzaman, Nathan Martin

Authors person Sanhita Sengupta Bristol Myers Squibb, Princeton, NJ info_outline Sanhita Sengupta, Clara Amorosi, Mandeep Takhar, Yi Lin, Salomon Manier, Rachid C. Baz, Krishna Rangadhamarao Juluri, Allison Kaeding, Afshin Mashadi-Hossein, Julia Piasecki, Timothy Brandon Campbell, Shari Kaiser, Julie Rytlewski, Md Shamsuzzaman, Nathan Martin Organizations Bristol Myers Squibb, Princeton, NJ, Division of Hematology, Mayo Clinic, Rochester, MN, Lille University Hospital, Lille, France, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL Abstract Disclosures Research Funding 2seventy bio and Celgene, a Bristol-Myers Squibb company Background: iAEs, such as cytokine release syndrome (CRS) and macrophage activation syndrome (MAS), can occur after infusion of chimeric antigen receptor T cell therapies such as ide-cel, but are typically low-grade and manageable with tocilizumab and corticosteroids. Although high-grade iAEs are rare with ide-cel, identification of patients at increased risk may improve management and remains an unmet need. Three composite biomarkers (multivariate models) of iAE risk were previously developed in a discovery cohort from the KarMMa and KarMMa-2 studies of ide-cel in RRMM (Mashadi-Hossein A, et al. J Clin Oncol 2023;41(16_suppl):e20005). Here, we validate these 3 prototype models for risk of developing high-grade iAEs after ide-cel infusion. Methods: Two models used pretreatment patient and routine clinical laboratory parameters; the third model also included change in 6 exploratory cytokines from baseline to 1 day following ide-cel infusion. To validate, each model was retrospectively applied to KarMMa-3 (NCT03651128) data in a blinded fashion, and sensitivity and specificity to identify grade ≥3 CRS and any-grade MAS were determined. Model performance thresholds used for sensitivity and specificity were 90% and 50%, respectively. Performance for identifying grade 2 CRS was also analyzed post hoc. Results: The first and simplest model comprised a manual algorithm with 7 features; it identified patients with grade ≥3 CRS or MAS with 85% sensitivity and 60% specificity. The second model comprised 19 routinely measured pretreatment features; it identified grade ≥3 CRS with 91% sensitivity and 59% specificity, and any-grade MAS with 60% sensitivity and 88% specificity. The third and most complex model comprised 7 pretreatment clinical features and 6 exploratory cytokines; it identified grade ≥3 CRS with 100% sensitivity and 51% specificity, and any-grade MAS with 100% sensitivity and 81% specificity. None of the models identified grade ≥2 CRS with passable sensitivity or specificity. Clustering analyses across all features showed that only a subset of grade 2 CRS cases was within the previously identified high-risk iAE cluster while the remaining grade 2 CRS cases were spread across other clusters. Conclusions: Two of the 3 composite biomarkers met validation criteria for grade ≥3 CRS risk. Criteria were not met for grade 2 CRS, suggesting these models are specific to higher-grade events and that grade 2 CRS represents a more heterogeneous group of patients. While high-grade iAEs are rare with ide-cel, these models may further optimize the robust benefit–risk profile that ide-cel has demonstrated by helping identify, prior to or shortly after infusion, patients most at risk for severe events. Clinical trial information: NCT03651128.

Clinical