Factors associated with outcomes after a second CD19-targeted CAR T-cell infusion for refractory B-cell malignancies

Abstract

CD19-targeted chimeric antigen receptor-engineered (CD19 CAR) T-cell therapy has shown significant efficacy for relapsed or refractory (R/R) B-cell malignancies. Yet, CD19 CAR T cells fail to induce durable responses in most patients. Second infusions of CD19 CAR T cells (CART2) have been considered as a possible approach to improve outcomes. We analyzed data from 44 patients with R/R B-cell malignancies (acute lymphoblastic leukemia [ALL], n = 14; chronic lymphocytic leukemia [CLL], n = 9; non-Hodgkin lymphoma [NHL], n = 21) who received CART2 on a phase 1/2 trial (NCT01865617) at our institution. Despite a CART2 dose increase in 82% of patients, we observed a low incidence of severe toxicity after CART2 (grade ≥3 cytokine release syndrome, 9%; grade ≥3 neurotoxicity, 11%). After CART2, complete response (CR) was achieved in 22% of CLL, 19% of NHL, and 21% of ALL patients. The median durations of response after CART2 in CLL, NHL, and ALL patients were 33, 6, and 4 months, respectively. Addition of fludarabine to cyclophosphamide-based lymphodepletion before the first CAR T-cell infusion (CART1) and an increase in the CART2 dose compared with CART1 were independently associated with higher overall response rates and longer progression-free survival after CART2. We observed durable CAR T-cell persistence after CART2 in patients who received cyclophosphamide and fludarabine (Cy-Flu) lymphodepletion before CART1 and a higher CART2 compared with CART1 cell dose. The identification of 2 modifiable pretreatment factors independently associated with better outcomes after CART2 suggests strategies to improve in vivo CAR T-cell kinetics and responses after repeat CAR T-cell infusions, and has implications for the design of trials of novel CAR T-cell products after failure of prior CAR T-cell immunotherapies.

Graphical abstract

Figure 1.

Toxicity and efficacy after CART2. (A) CRS grade according to the Lee et al 2014 consensus criteria (left). Neurotoxicity grade according to the CTCAE 4.03 grading scale (right). (B) Response rates by disease type. CTCAE, Common Terminology Criteria for Adverse Events; SD, stable disease.

Figure 2.

Outcomes after CART2. (A) OS and PFS probabilities after CART2 in ALL (left), CLL (middle), and NHL (right). (B) OS stratified by day 28 response after CART2. (C) PFS in CR/PR patients (n = 17) after CART2 (best response). Solid lines, Kaplan-Meier estimates; dashed lines, 95% CIs. For the analysis of PFS, an event was defined as no response, relapse or disease progression, or death. Death was the event for analysis of OS. Patients who did not have an event were censored at the date of last follow-up. Data were not censored at the time of initiation of new therapy, including allogeneic HCT, in the absence of disease progression. P values per the log-rank test.

Figure 3.

In vivo CAR T-cell expansion and persistence. (A-B) CD8⁺ (A) and CD4⁺ (B) CAR T-cell peak expansion (cells/µL) by flow cytometry after CART1 and CART2. Gray lines connect paired values between identical patients. (C,E) CAR T-cell peak expansion by flow cytometry categorized by best response (C) and pre-CART1 lymphodepletion (E). CAR T-cell persistence by qPCR (CAR transgene copies/µg of DNA) categorized by best response (E) and pre-CART1 lymphodepletion (F). Dashed lines, limit of detection. (A-C,E) Bold horizontal lines, median; box, IQR; vertical lines, quartiles ± 1.5 × IQR. (D,F) Bold lines, polynomial regression lines using the LOESS method, shaded areas, 95% CI of the LOESS estimates.

Figure 4.

Outcomes and CAR T-cell persistence in NHL and CLL patients stratified according to pre-CART1 lymphodepletion and CART2 dose relative to CART1. (A) Response rates, (B) PFS, and (C) CAR T-cell persistence by qPCR in NHL and CLL patients according to pre-CART1 lymphodepletion and CART2 dose relative to CART1. Solid lines, Kaplan-Meier estimates; dashed lines; 95% CIs. P values per the log-rank test. Two patients in the pre-CART1 Cy-Flu CART2 > CART1 dose received a CART1 dose of 2 × 10⁵/kg and a CART2 dose of 2 × 10⁶/kg. (C) Bold lines, polynomial regression lines using the LOESS method, shaded areas, 95% CI of the LOESS estimates.