Back

Association of CD7⁺CXCR3⁺ CAR T cells with long-term remission in R/R DLBCL

Bartolini, R.; Trueb, L.; Daoudlarian, D.; Joo, V.; Noto, A.; Gentner, B.; Stadelmann, R.; Fenwick, C.; Perreau, M.; Coukos, G.; Pantaleo, G.; Arber, C.; Obeid, M.

2024-12-08 oncology
10.1101/2024.12.06.24318600 medRxiv
Show abstract

Background.CAR T-cell therapy is the standard of care for R/R DLBCL, but more than half of patients fail to achieve long-term remission. Identification of cellular biomarkers in CAR T- cell infusion products (IPs) that predict complete remission beyond six months may guide the development of strategies to improve outcomes. Methods.IPs from 13 R/R DLBCL patients were analyzed using a 39-marker mass cytometry panel, comparing cell populations between long-term responders (R) and non-responders (NR). Both unsupervised and supervised analyses were performed. Longitudinal blood samples were analyzed for 30 days to track CAR T-cell subpopulation dynamics. Results.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). The HR for PFS was 56.67 (95% CI: 7.3-439.3; P=0.0001). A subset of CD3+CXCR3+CD7+ CAR T-cells found in both CD4+ and CD8+ populations was significantly enriched in R and expressed higher levels of perforin, granzyme B, and NKG2D (restricted to CD8+). NR had more CXCR3+CD7+LAG3+ CAR T-cells. CD3, CD7, CXCR3, and NKG2D cell surface levels 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 levels were not different between groups. Conclusions.Increased frequency of CAR T-cells expressing CD7, CXCR3 and NKG2D in R versus LAG3 and CD71 in NR emerged as strong correlates of therapeutic outcome.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.