Biomarkers for Predicting CAR-T Cell Therapy Outcomes in B-Cell Acute Lymphoblastic Leukemia: A Systematic Review

衍昊 柯^1,2芬 周^1*

• ¹Department of Hematology and Oncology, Wuhan Children’s Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China
• ²Huazhong University of Science and Technology, Wuhan, China

Chimeric antigen receptor (CAR) T-cell therapy has revolutionized the treatment of relapsed/refractory B-cell acute lymphoblastic leukemia (B-ALL), yet challenges such as cytokine release syndrome (CRS), neurotoxicity (ICANS), and variable long-term efficacy persist. This systematic review evaluates the role of biomarkers in predicting CAR-T therapy outcomes, toxicity risks, and guiding personalized strategies. Following PRISMA guidelines, 33 studies (n=2,095 patients) published between 2018–2024 were analyzed through PubMed, Web of Science, and Embase. Key findings identified tumor burden and minimal residual disease (MRD) as dual-predictive biomarkers. High tumor burden (≥40% blasts) correlated with reduced complete remission (87% vs. 100%) and increased CRS/ICANS risks, while MRD negativity (NGS threshold <10-⁶) predicted superior 2-year event-free survival (68% vs. 23%). CAR-T functional parameters, including PD-1/LAG-3 expression (>5.2% in CD4+ cells) and peak expansion kinetics, linked efficacy to toxicity trade-offs. Genetic biomarkers (IKZF1 mutations, complex karyotypes) and biochemical indicators (m-EASIX >6.2, ferritin ≥10,000 ng/mL) further stratified risks. Unidirectional efficacy biomarkers included T-cell subsets (e.g., CD8+ naive T cells) and B-cell aplasia, while IL-6 dynamics specifically predicted CRS severity. Despite promising insights, heterogeneity in toxicity grading systems, inconsistent biomarker thresholds, and retrospective study designs limit clinical standardization. Future directions emphasize cytoreductive bridging therapies, biomarker-guided combinatorial approaches (e.g., MDM2 inhibitors for TP53 mutations), and multicenter validation of integrated predictive models.