Outcomes of tafasitamab and lenalidomide in large B‐cell lymphoma based on prior CD19‐directed CAR T exposure

The treatment landscape of relapsed and refractory (R/R) diffuse large B-cell lymphoma (DLBCL) has significantly improved with the development of CD19-directed chimeric antigen receptor T-cell therapy (CAR T).1, 2 Even so, approximately 50%–60% of patients treated with CAR T will ultimately experience disease progression.3 Tafasitamab (tafa) and loncastuximab tesirine (lonca) both target CD19 and are approved for the treatment of R/R DLBCL. Following the approval of CAR T in the second-line setting,4, 5 these agents may be used to treat DLBCL after CAR T failure. Notably, the L-MIND study of tafa with lenalidomide (tafa/len) excluded patients with prior CAR T, while the LOTIS-2 study for lonca enrolled only 13 post–CAR T patients.6, 7 Prior reports, including our own, have shown poor outcomes with CD19-directed therapies following CAR T, though it remains unclear whether this reflects the effects of prior CAR T itself or underlying high-risk disease features.8, 9 We aimed to evaluate the impact of prior CD19-directed therapy on subsequent CD19-based treatment, focusing on the impact of prior CD19 CAR T on tafa/len efficacy. We compared CAR T-exposed versus CAR T-naïve patients receiving tafa/len, and used propensity score matching to address baseline imbalances between CAR-exposed and CAR-naïve cohorts. We also explored CD19 expression patterns and outcomes with other CD19-directed sequences, including lonca and CAR T. This multicenter, retrospective cohort study included adults (≥18 years) with R/R DLBCL treated with tafa between August 2020 and August 2022 across 11 institutions. Eligible histologies included DLBCL-not otherwise specified (NOS), high-grade B-cell lymphoma (HGBCL), transformed indolent lymphomas, T-cell/histiocyte-rich large B-cell lymphoma, and primary mediastinal B-cell lymphoma. Responses to therapy were evaluated by each site using routine imaging and assessments. CD19 expression was locally assessed by immunohistochemistry or flow cytometry and defined as positive by local center criteria. We previously reported the overall outcomes of tafa/len.8 Subsequently, additional data were collected on the outcomes of other CD19-directed therapies aside from tafa. This study was approved by the institutional review boards of all participating institutions. Patients were categorized into CAR T-naïve and CAR T-exposed cohorts based on treatment exposure at the time of tafa/len initiation. Patients who had received prior CD19-directed therapy other than CAR T before tafa/len were excluded. Descriptive statistics were reported by cohort. Fisher's exact and Wilcoxon rank-sum tests were used to assess differences by cohort for categorical and continuous variables, respectively. The primary outcome was progression-free survival (PFS) from tafa initiation to progression or death. Overall survival (OS) was measured from tafa initiation to death or last follow-up. Kaplan–Meier methods were used to estimate survival, with log-rank tests and Cox regression models for comparisons with 95% confidence intervals calculated using Greenwood's method to estimate variance. Uni- and multivariable models are summarized with hazard ratios (HRs), 95% confidence intervals, and Wald P-values for each coefficient. The likelihood ratio test was used to assess the overall association for each variable in the adjusted models. Variable selection was performed using LASSO regularization on the univariable Cox predictors for PFS. CAR T-exposed patients were compared with a matched sub-cohort of CAR T-naive patients using propensity score matching (1:1, nearest neighbor, no replacement), accounting for age, Eastern Cooperative Oncology Group (ECOG) performance status, lactate dehydrogenase (LDH), and disease refractoriness—factors previously independently linked to PFS on multivariate analysis.8 Refractoriness was defined as less than complete response (CR) or progression within 6 months of last treatment. With 50 patients per group and 14-month median follow-up, the study was powered to detect a 1.5-month difference in median PFS (power 85%, α = 0.05). Analyses were conducted in R (survival and MatchIt packages). A total of 174 patients from 11 centers were included. Baseline patient characteristics at the time of tafa/len initiation are outlined in Table 1. In 41 patients with data on response to CAR T, the best overall response (BOR) to prior CAR T was 73% (30/41), and the CR rate was 51% (21/41). The median time from CAR T to tafa was 7.3 months (range 1–55 months, interquartile range [IQR] 3.6–14.8 months). In the overall cohorts, the complete response rate (CRR) was 19% (95% CI 9%–34%) in CAR T-exposed versus 20% (95% CI 13–28) in CAR T-naïve patients (P > 0.99). The overall response rate (ORR) in CAR T-exposed patients was 19% (95% CI 9%–34%) versus 36% (95% CI 27%–46%, P = 0.052) in CAR T-naïve patients. Median PFS was shorter with prior CAR T (1.7 vs. 2.8 months, P = 0.009; HR: 1.61 [95% CI 1.12, 2.30], Figure 1), and median OS was 5.0 versus 7.5 months (P = 0.25). At 6 months after tafa/len initiation, PFS was 28% (95% CI 21–37) in CAR T-naïve versus 14% (95% CI 7–28) in CAR T-exposed patients. Univariable analysis was performed in patients with prior CAR T exposure to identify factors associated with PFS and OS (Figures S1 and S2). Increased time from CAR T infusion to tafa/len initiation was associated with improved PFS (with each increase by 6 months, HR 0.78 [95% CI 0.62–0.94]). Median PFS was 1.4 (95% CI 1.0−2.5) versus 4.3 months (95% CI 2.0–NR), when tafa/len was given <12 versus ≥12 months after CAR T (P = 0.0059) (Figure 1C,D). Less than CR to CAR T was associated with inferior PFS (HR 2.86 [95% CI 1.43–5.87]), as was neurotoxicity with prior CAR T (HR 2.44 [95% CI 1.04–5.16]), international prognostic index (IPI) at tafa initiation (score 4–5 vs. 0–2, HR 3.38 [95% CI 1.36–9.88]) and LDH elevation (HR 2.62 [95% CI 1.27–6.03]). We next conducted a multivariable analysis evaluating factors that were significantly associated with PFS on univariable analysis. The time from CAR T to tafa/len was significantly associated with PFS (for each 6-month increase, HR 0.77 [95% Cl 0.61–0.98]; likelihood ratio test, P = 0.02). Neurotoxicity with prior CAR T was associated with worse PFS with borderline significance (HR 2.51 [95% Cl 1.01–6.23]; likelihood ratio test, P = 0.059). IPI 3 versus 1–2 was associated with inferior PFS (HR 3.56 [95% CI 1.08–11.79]), though overall IPI was not independently associated with PFS (likelihood ratio test, P = 0.13) (Figure S3). We used propensity matching to establish a cohort of 50 CAR T-naïve patients with similar characteristics to the 50 CAR T-exposed patients, using baseline characteristics associated with PFS after tafa/len therapy (Table S1).8 The ORR was 19% (9–34) in the CAR-exposed cohort versus 39% (25–54) in the CAR T-naïve cohort (P = 0.065). The CRR was 19% (9–34) for CAR T-exposed versus 20% (10–34) for CAR T-naïve. The median follow-up for the matched cohorts was 14 months. The median PFS in CAR T-exposed patients was 1.7 months, compared to 2.1 months in CAR T-naïve patients, not meeting prespecified significance (HR 1.44 [95% CI: 0.93, 2.23] (P = 0.096)) (Figure S4). The median OS was 5 months in the CAR T-exposed cohort versus 8 months in the CAR T-naïve cohort (P = 0.27). In the 50 CAR T-exposed patients, 31 had a biopsy reporting CD19 expression between CAR T and tafa/len. Methodology varied (flow cytometry in 11, immunohistochemistry in 10, both in 5, and unspecified in 5). Biopsies were done a median of 31 days (IQR 17–80) before tafa/len. Of these, 26 were CD19-positive, with ORR and CRR of 27% (7/26 CR, 0/26 PR). CD19 expression was associated with improved OS (HR 0.27 [95% CI 0.1–0.93]), but not PFS (HR 0.56 [95% CI 0.2–2.1]). Five patients had CD19-negative biopsies performed a median of 107 days after CAR T and 96 days before tafa/len. Among four patients with CD19-negative results within 1 year of tafa/len, ORR and CRR were 0%. One patient with a CD19-negative biopsy 3 months after CAR T and 4 years before tafa/len (with multiple intervening therapies) had a CR. Fifteen patients received tafa/len before CAR T, with a median interval between therapies of 2 months (IQR 1.4–3.2). The best ORR to prior tafa/len was 40% (6/15) and CRR 13% (2/15). Median follow-up was 17 months after CAR T. The CRR to CAR T after tafa/len was 40%, and ORR 60%. Median PFS was 4 months, and median OS was 10 months. All five patients who responded to tafa/len also responded to CAR T (ORR 100%, CRR 60%). Of 11 cases biopsied between therapies, 8 were CD19+. Among the three CD19− cases, one had partial response (PR) and two had PD to CAR T. Data were available on eight patients who received lonca after tafa/len, without prior CAR T. The ORR with lonca was 17% (one PR out of six evaluable) and CRR 0%. Median PFS was 1 month, and median OS was 2 months. Nine patients received lonca after both tafa/len and CAR T; ORR was 25% (2 PRs), median PFS 2.4 months, and median OS 3.2 months. CD19 expression was assessed in 24 post-tafa biopsies (median 18 days after infusion); 54% were CD19-positive. Notably, all samples collected >21 days after tafa (6/6) were CD19-positive, compared to 39% (7/18) of those obtained within 21 days. In this retrospective multicenter cohort evaluating tafa/len in DLBCL, prior CAR T exposure was associated with inferior PFS, though this association was not detected after propensity score matching to account for potential confounders. Patients with DLBCL progression after CAR T have a poor prognosis.10, 11 In this setting, where treatment options are limited, understanding the effectiveness of subsequent CD19-directed therapies is essential. Use of consecutive CD19-directed therapies raises theoretical concerns, as targeting the same antigen may lead to loss or decreased expression of CD19, and reduced efficacy of the next CD19-directed therapy.10, 12 In this study, when accounting for potential confounders, we were not able to identify worse efficacy with tafa/len related to prior CD19 CAR T. Analogously, in a large retrospective study, the LYSA group observed no significant difference in outcomes between tafa/len and non-CD19-directed therapies.13 Together, these findings suggest that CD19-directed therapies can be considered in the post–CAR T setting. However, in both studies, the overall efficacy of tafa/len was modest, regardless of prior CAR T exposure. Alternative therapies, such as CD20-directed bispecific antibodies (BsAbs), which have demonstrated relatively higher response rates in cohorts of heavily pretreated or refractory DLBCL, may be preferred in the post–CAR T setting.14-16 In our analysis, a longer time between CAR T and tafa/len was associated with longer PFS, potentially reflecting better disease biology or immune recovery. Similar findings were observed in the LYSA study.13 Surprisingly, neurotoxicity with prior CAR T was associated with worse PFS, possibly due to small sample size (only nine patients with prior neurotoxicity), aggressive disease biology, dysregulated immune responses, or other unmeasured high-risk characteristics. Among patients with post–CAR T biopsies, 84% retained CD19 expression. Only one of five CD19-negative cases (whose biopsy was performed 4 years before tafa/len) responded to tafa/len, suggesting that CD19 expression should be considered before tafa/len use. After tafa/len, CD19 expression was negative in 46% of patients; however, all biopsies performed more than 3 weeks after the last tafa dose were CD19-positive. Instead of antigen loss, tafa may mask the CD19 epitope or drive transient downregulation, as has been hypothesized in other cases.17, 18 Biopsies more than 3 weeks after tafa treatment may yield more useful CD19 results. Limitations to this study include its retrospective design and limited generalizability due to academic site enrollment. While the propensity score matched comparison was powered to detect a difference in median PFS, a more modest difference may not be detected with smaller sample sizes. Additionally, despite attempts to account for risk factors, unmeasured confounders within the matched cohorts may contribute to bias. Finally, patients with CD19 loss may have been less likely to receive tafa/len, biasing the population of patients receiving tafa/len after CAR T. In summary, while overall outcomes were poor, when accounting for other high-risk factors, prior CAR T exposure was not independently associated with inferior tafa/len efficacy. These findings suggest that CD19-directed therapy may be considered in select patients after CAR T failure; however, with modest efficacy, combinations with other active agents may be preferred. Currently, tafa/len combined with the BsAb epcoritamab (NCT07030699), and lonca with the BsAb glofitamab (NCT04970901), are under evaluation in R/R DLBCL, including after CAR T failure. Overall, there remains a significant need for more effective therapies in this high-risk population. Mwanasha H. Merrill: Writing—original draft; validation; visualization; writing—review and editing; formal analysis; data curation; project administration. Robert A. Redd: Writing—original draft; writing—review and editing; methodology; formal analysis. Nicholas Lambert: Writing—review and editing; project administration. Paolo F. Caimi: Writing—review and editing; project administration. Priyanka Pullarkat: Writing—review and editing; project administration. Richard C. Godby: Writing—review and editing; project administration. David A. Bond: Writing—review and editing; project administration. Graham T. Wehmeyer: Writing—review and editing; project administration. Jason Romancik: Writing—review and editing; project administration. Behzad Amoozgar: Writing—review and editing; project administration. Lori Leslie: Writing—review and editing; project administration. Loretta J. Nastoupil: Writing—review and editing; project administration. Jennifer L. Crombie: Writing—review and editing; project administration. Jeremy S. Abramson: Writing—review and editing; project administration. Arushi Khurana: Writing—review and editing; project administration. Grzegorz S. Nowakowski: Writing—review and editing; project administration. Kami Maddocks: Writing—review and editing; project administration. Sarah C. Rutherford: Writing—review and editing; project administration. Brad Kahl: Writing—review and editing; project administration. Michelle Okwali: Writing—review and editing; project administration. Michael J. Buege: Writing—review and editing; project administration. Connie L. Batlevi: Writing—review and editing; project administration. Philippe Armand: Writing—review and editing; project administration. Gilles Salles: Writing—review and editing; project administration. David A. Qualls: Writing—original draft; writing—review and editing; visualization; validation; formal analysis; project administration; data curation; supervision. D.A.B. has received research funding from Incyte, BMS, Kite/Gilead, Novartis, Nurix Therapeutics, Accutar, GenMab, and AstraZeneca, and has consulted for Kite/Gilead, Nurix Therapeutics, ADC Therapeutics, SeaGen, and Novartis. C.L.B. is currently employed by and holds equity in Genentech/Roche. K.M. consults for AbbVie, ADC Therapeutics, AstraZeneca/Acerta, BMS, Caribou, Genentech, Genmab, Gilead/Kite, Incyte, Janssen, Lilly, Morphosys, and Pharmacyclics. G.S. consults for AbbVie, BeiGene, BMS, Debiopharm, Epizyme, Genentech/Roche, Genmab, Innate Pharma, Incyte, Ipsen, Janssen, Kite/Gilead, Merck, Modex, Molecular Partners, Nurix, Orna Therapeutics, Pfizer, and Treeline, and has received research funding from AbbVie, Genentech, Genmab, Janssen, Ipsen, and Nurix. L.L. has consulting/honoraria from AbbVie, ADC Therapeutics, AstraZeneca, BeiGene, Celgene/BMS, Eli Lily Epizyme, Genmab, Gilead/Kite, Janssen, Merck, Pharmacyclics, and Seagen. P.F.C. has consulting/honoraria from BMS, Novartis, Synthekine, SOBI, and Genentech, and research funding from AbbVie, ADC Therapeutics, Genmab, and Recordati Rare Diseases. J.S.A. has served as a consultant for AbbVie, ADC Therapeutics, AstraZeneca, BeiGene, BMS, Cellectar, Caribou Biosciences, Gilead, Interius, Janssen, Lilly, Novartis, Roche, Seagen, and Takeda, and has received research support (to institution) from BMS, Cellectis, Merck, Mustang Bio, Regeneron, and Seagen. P.A. has served as a consultant for Merck, BMS, ADC Therapeutics, Tessa, Genmab, Enterome, Regeneron, Epizyme, AstraZeneca, Genentech, Xencor, and ATB Therapeutics; has received research funding (institutional) from Merck, BMS, Adaptive, Genentech, and Kite; and honoraria from Merck. B.K. has received consulting from AbbVie, BeiGene, BMS, Genentech, Roche, GSK, Lilly, Pfizer, and Merck, and research support from BeiGene, Roche, and Lilly. S.C.R. has received consulting from AbbVie, ADC Therapeutics, BMS, Genmab, Incyte, Karyopharm, Kite, and Pfizer/Seagen DSMB: Karyopharm, and research support from Constellation, Genentech, and Karyopharm. J.L.C. has received consulting from Genentech/Roche, AbbVie, Genmab, Regeneron, ADC Therapeutics, Lilly, and Novartis. L.J.N. has received honorarium for participation in advisory boards from AbbVie, AstraZeneca, BMS, Daiichi Sankyo, Genentech, Genmab, Gilead/Kite, Incyte, Janssen, Merck, Novartis, Regeneron, and Takeda, and has received research support from BMS, Caribou Biosciences, Daiichi Sankyo, Genentech, Genmab, Gilead/Kite, Incyte, Janssen, Merck, Novartis, and Takeda. D.A.Q has received consulting/honoraria from Genmab, Kite Pharma, and ADC Therapeutics. The remaining authors declare no competing financial interests. None. The data that support the findings of this study are available from the corresponding author upon reasonable request. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.