Introduction Chimeric antigen receptor T-cell therapy (CART) leads to similar response rates in trials evaluating treatment as 2nd (2L) vs 3rd line (3L) and beyond in patients (pts) with relapsed/refractory (R/R) aggressive B-cell non-Hodgkin lymphoma (ZUMA-7 and TRANSFORM for 2L vs ZUMA-1 and TRANSCEND for 3L and beyond). Whether earlier vs later use of CART improves survival has not been established. We performed a multicenter retrospective analysis evaluating survival outcomes with CART according to line of therapy, specifically 2L vs 3L vs 4th line and beyond (4L+) to inform better practice. Methods Adult pts with R/R de novo diffuse large B-cell lymphoma (DLBCL) or transformed follicular lymphoma (tFL) treated with CD19-directed CART from 2015-2024 across 13 US centers were identified. Baseline characteristics compared between cohorts included: age, sex, histology, IPI, ECOG, LDH at CART collection, double hit (DHL) and double expressor status (DEL), prior autologous stem cell transplant (autoHCT), lines of therapy pre-CART, bridging therapy (BT), receipt of CART < 12 months from diagnosis, primary refractory disease (PRD) defined as refractory or relapse < 12 mo from frontline therapy, CART on clinical trial, CART construct, and toxicity. Median progression free (mPFS) and overall survival (mOS) were estimated via Kaplan-Meier. Cox regression was used to assess impact of clinical variables on survival. A p-value <0.05 was considered significant. Results 713 pts were included (n=142 [2L], 281 [3L], 290 [4L+]). Demographic and clinical features were similar between groups including CART construct and prevalence of pts with DHL/DEL or PRD. Exceptions included: more 2L pts received CART < 12 mo from diagnosis (51.4 [2L], 40.1 [3L], 26.6% [4L+], p<0.001), and more 3L pts received autoHCT prior to CART (7 [2L], 18 [3L], 7% [4L+], p<0.001); 4L+ line pts had longer lasting neurotoxicity (4.5 [2L], 2 [3L], 6 days [4L+], p=0.01); 2L pts trended towards more tFL histology (26 [2L], 19 [3L], 17% [4L+], p=0.07) and use of BT (54 [2L], 49 [3L], 43% [4L+], p=0.07). Median follow up from CART infusion was 25.1 mo. CART response rates were similar for 2L, 3L, and 4L+ (ORR 70 [2L], 67 [3L], 64% [4L+], p=0.6; CR 59 [2L], 51 [3L], 48% [4L+], p=0.1). From CART infusion, three-year PFS and OS were similar in 2L vs 3L but shorter in 4L+ pts (PFS 39.5 [2L], 35.9 [3L], 28.1% [4L+], p<0.001; OS 47.4 [2L], 51.8 [3L], 37.7% [4L+], p<0.001), as was mPFS and mOS (mPFS 8.2 [2L], 10.3 [3L], 4.4 [4L+] mo, p<0.001; mOS 29.7 [2L], 36.8 [3L], 20.5 [4L+], p<0.001). In PRD pts, mPFS and mOS were similar for 2L vs 3L but shorter in 4L+ pts (mPFS 7.2 [2L], 6.0 [3L], 3.0 mo [4L+], p=0.04; mOS 27.4 [2L], 31.6 [3L], 14.9 mo [4L+], p=0.04). In pts receiving BT, mPFS and mOS were similar for 2L vs 3L but shorter in 4L+ pts (mPFS 7.8 [2L], 7.4 [3L], 3.0 mo [4L+], p<0.001; mOS 53.6 [2L], 25.7 [3L], 14.5 mo [4L+], p=0.003). 2L pts who received BT had similar mPFS and mOS to 3L pts who did not receive BT (mPFS 7.8 [2L with BT], 12.5 mo [3L without BT], p=0.7; mOS 53.6 [2L with BT], 69.4 mo [3L without BT], p=0.4). After CART failure, CART line of therapy did not impact mPFS with next line therapy (3.8 [2L], 2.7 [3L], 2.3 mo [4L+], p=0.3) nor mOS (14.6 [2L], 11.3 [3L], 9.9 mo [4L+], p=0.4). On Cox regression, pts with PRD, irrespective of line of CART, were more likely to have a PFS/OS event (PFS: HR 1.7 [1.2-2.7], p=0.02; OS: HR 1.7 [1.2-2.7], p=0.02), as were pts who received CART < 12 mo from diagnosis (PFS: HR 2.3 [1.5-3.3], p<0.001; OS: HR 2.3 [1.5-3.3], p<0.001). Age, LDH at time of CART, and DHL/DEL status did not impact survival. When comparing the interaction between CART line (2L and 3L only) and PRD or DHL, 3L pts with PRD were more likely to have a PFS/OS event than 2L (PFS: HR 2.5 [1.7-3.7], p<0.001; OS: HR 2.5 [1.7-3.7], p<0.001); 2L and 3L pts with DHL had similar risk for PFS/OS events (PFS: HR 1.2 [0.9-1.6], p=0.2; OS: HR 1.2 [0.9-1.6], p=0.2). Conclusions This multicenter analysis is the first to compare outcomes in DLBCL/tFL pts according to CART line of therapy and demonstrates that survival with 2L or 3L CART is similar but superior to CART 4L+, irrespective of use of BT. The 2L cohort was enriched with pts who received CART < 12 mo from diagnosis, a characteristic associated with poor survival, and suggests the inclusion of more pts with aggressive disease in our 2L cohort. When comparing 2L vs 3L CART, PRD pts may have improved survival with earlier use of CART.