Analysis of Early Apheresis Immunophenotype, Batch Data Variables and Clinical Outcomes for Optimization of CAR-T-Cell Therapy in Aggressive B-Cell Lymphomas

Introduction: A plethora of clinical factors is known to influence long-term success of anti-CD19 CAR-T-cell therapy in patients (pts) with relapsed/refractory large B-cell lymphoma (LBCL). CAR-T-cell functionality in pts with fewer treatments (Tx) before apheresis (APH) (Neelapu, Dickinson et al. 2022) might be superior, however, the impact of the APH material composition on outcome is less well investigated.

Before the availability of CAR-T in 2^nd line Tx of LBCL in 12/2021, we hypothesized that early APH (APH before 2^nd relapse = EA) in pts with 1^st early relapse (ER) might have a positive influence on final product (FP) quality and long-term success of CAR-T therapy. Only tisagenlecleucel (tisa-cel) is an approved anti-CD19 CAR-T therapy to be produced from frozen APH material, allowing this EA strategy. This study aimed at the characterization of tisa-cel pts APH and FP batch data (BD) to investigate potential effects of EA product quality on overall success rate associated with pts outcome and prior Tx.

Methods: We included LBCL pts treated with tisa-cel, who had EA or standard APH (after confirmed 2^nd relapse = SA) for CAR-T therapy production or at least had EA, but did not receive CAR-T therapy between 7/18 and 12/22 with follow-up until 12/23. Immunophenotype (IP) (192 IP marker measured) and BD for tisa-cel production were obtained from Novartis Pharma Gmbh. Overall survival (OS) and progression-free survival (PFS) were calculated by Kaplan Meier estimates. OS and PFS probabilities were compared by log-rank test. Principle component analysis (PCA) was used to reduce the number of dimensions of IP and BD to obtain potentially correlated variables. Correlation analysis of IP and BD evaluated potential correlation of those parameters with respect to the quality of APH and CAR-T therapy FP. Student’s t-test was used on IP and BD to compare data for EA and SA pts.

Results: We identified 6 pts with EA before CAR-T therapy (median age 57.5 years [32-79], 100 % male , IPI > 2 in 2 pts [33 %], median Tx lines of 2 [2-4]. Three (50 %) pts had primary refractory disease (no response/relapse within 6 months (mos) of 1^st line Tx =PREF) and 3 (50 %) pts had an ER within 1 year after 1^st line.

10 pts had EA without CAR-T therapy(median age 58 years [29-73], 5 female pts (50 %)., IPI at initial diagnosis > 2 in 7 (70 %) pts, 4 (40 %) pts were PREF to 1^st line Tx, 4 pts had ER and 2 pts late relapse (LR) > 1 year after 1^st line Tx).

Twenty-eight pts had SA (median age 53.5 years [18-80], 12 female [43%] pts, median Tx lines before CAR therapy 2 [2-6], IPI > 2 before CAR therapy in 10 pts [36 %)]). Five (18 %) pts were PREF to 1^st line Tx, 13 (46 %) had ER 1 year after 1^st line Tx and 10 (36 %) pts had LR > 1 year after 1^st line Tx.

All 34 pts treated with tisa-cel had a median PFS (mPFS) of 2.7 mos (1-year PFS 20.6 %) and median OS (mOS) of 9.1 mos (1-year OS 40.1%). In EA and SA pts, mPFS was similar with 2.4 mos [95% CI: 0.35 to 2.16] and 2.7 mos [95% CI: 0.46 to 2.81], respectively, (mOS was 4.7 mos [95% CI: 0.19 – 1.26] and 9.5 mos [95% CI: 0.78-5.14]). PFS and OS of EA pts was 16.7 % after 6 mos (1-year PFS/OS 0 %). For SA pts, PFS and OS were 28.6 % after 6 mos (1-year PFS 25%) and 79 % after 6 mos (1-year OS 47 %) respectively.

PCA and correlation analysis of IP data of APH and CAR-T therapy FP showed trends for an adverse impact of T-cell exhaustion markers (PD-1, LAG-3, TIM-3) on FP quality. For APH and CAR-T FP batch data, a potential correlation between cell expansion during CAR-T production and cell seeding strategy was shown.

Characterization analysis of IP for EA and SA pts showed significant differences in the composition of APH and CAR-T FP. PD1+ cytotoxic T-cells were more frequent in APH products of SA compared to EA (p = 0.010). In contrast, CAR+ cytotoxic T-cells and T-helper-cells with central memory T-cell subtype were more frequent in FP of EA batches (p = 0.008 for cytotoxic T-cells and p = 0.017 for T-helper-cells) suggesting more favourable product characteristics of EA material.

Conclusion: With the limitation of small pts numbers and a possible selection bias of high risk patients, the EA strategy did not lead to improved outcome in high-risk pts with multiply relapsed LBCL as compared to SA. Thus, our data support CAR-T therapy at 1^st relapse of LBCL. However, characterization of IP and BD suggest more favourable characteristics of EA material further highlighting the impact of prior therapies before APH.