Chimeric antigen receptor (CAR) T-cell therapy is the standard of care for relapsed or refractory (R/R) diffuse large B-cell lymphoma (DLBCL). However, more than half of patients fail to achieve durable remission. Identifying predictive biomarkers within the CAR T-cell infusion product (IP) may guide strategies to improve clinical outcomes.
This single-center observational study conducted at Lausanne University Hospital (CHUV), Switzerland, analyzed IPs from 13 patients with R/R DLBCL who underwent standard-of-care CAR T-cell therapy. A 39-marker mass cytometry panel was used to compare phenotypic and functional markers between long-term responders (R) and non-responders (NR). Unsupervised and supervised analytic approaches were applied to IP data, and longitudinal peripheral blood samples were collected over 30 days post-infusion to track CAR T-cell subpopulation dynamics.
At a median follow-up of 13·5 months, median progression-free survival (PFS) was 13·3 months (95% CI 9·7-24·3) in R (n=8) versus 3·5 months (95% CI 0·5-5·4) in NR (n=5) (hazard ratio 56·67 [95% CI 7·3-439·3]; p=0·0001). A CD3 ⁺ CXCR3 ⁺ CD7 ⁺ CAR T-cell subpopulation-found in both CD4 ⁺ and CD8 ⁺ compartments-was significantly enriched in R. These cells showed increased expression of perforin, granzyme B, and NKG2D (restricted to CD8 ⁺ cells). In contrast, NR had a higher frequency of CXCR3 ⁺ CD7 ⁺ LAG3 ⁺ CAR T-cells. Surface expression of CD3, CD7, CXCR3, and NKG2D were higher in R, whereas LAG3, Ki67, and CD71 were elevated in NR. A predictive cut-off ratio of CD3 ⁺ CXCR3 ⁺ CD7 ⁺ LAG3 ⁺ CAR ⁺ T-cells <0·83 and CD3 ⁺ CXCR3 ⁺ CD7 ⁺ NKG2D ⁺ CAR ⁺ T-cells >1·034 yielded a predictive accuracy of 0·92. Serum CXCL9 and CXCL10 concentrations did not differ between groups.
Enrichment of CD7 ⁺ CXCR3 ⁺ CAR T-cells alongside elevated NKG2D expression in R, in contrast to higher LAG3 and CD71 in NR, emerged as potentially robust correlates of therapeutic outcome. Although derived from a small, hypothesis-generating cohort, these findings suggest that targeted analysis of IP composition may inform the development of biomarker-driven strategies to optimize CAR T-cell products and improve the likelihood of durable remission in R/R DLBCL.