Abstract
Background: Chimeric antigen receptor (CAR) T-cell therapy has revolutionized treatment of relapsed/refractory large B-cell lymphoma (LBCL), but its administration is often complicated by cytokine release syndrome (CRS). Interleukin-6 (IL-6) is widely used to monitor CRS, though its clinical value diminishes after tocilizumab administration. We aimed to evaluate serum amyloid A (SAA), a dynamic acute-phase reactant, as a treatment-independent biomarker of inflammation and toxicity in CAR-T recipients. Methods: This retrospective study included 43 adults with LBCL treated with axicabtagene ciloleucel. SAA and other inflammatory markers were assessed from lymphodepletion through day +11 post-infusion. CRS and ICANS were graded per ASTCT criteria. Statistical analyses included Mann-Whitney U tests, Spearman's correlation, and ROC curve analysis to evaluate predictive performance. Results: SAA levels peaked at day +4 and normalized by day +11, displaying wave-like kinetics. Levels were significantly higher in patients with any-grade CRS at early timepoints but showed no association with ICANS. SAA correlated strongly with CRP, suPAR, sST2, fibrinogen, ferritin, procalcitonin, and IL-6. Compared to IL-6, SAA was more predictive of CRS at day +2 and +4, and unaffected by tocilizumab. Baseline SAA also correlated with the mEASIX score, suggesting linkage to endothelial stress. Non-responders at 3-month PET had higher baseline SAA than responders (196.0 vs. 17.7 mg/L, p = 0.036), with ROC analysis yielding an AUC of 0.74 and an optimal threshold of 79.8 mg/L. Conclusions: SAA is a robust and dynamic marker of systemic inflammation, with potential utility in both toxicity monitoring and response prediction in the CAR-T setting. Its independence from IL-6 modulation positions it as a promising biomarker for future integration into clinical algorithms.