NEW YORK – Results from the Phase III BELINDA study presented Tuesday at the American Society of Hematology's annual meeting have oncologists scratching their heads. Why did one anti-CD19 CAR T-cell therapy fail in the same disease setting where two others just succeeded?
After seeing results from the Phase III TRANSFORM and ZUMA-7 studies presented on Saturday and Sunday at this meeting, many were quick to declare that CAR T-cell therapy is now a viable second-line option for relapsed or refractory large B-cell lymphoma patients.
Both of the randomized studies pitted autologous CAR T-cell therapies against standard-of-care chemotherapy and stem cell transplant in the second-line setting for relapsed or refractory LBCL. In both cases, the CAR T-cell therapies — Gilead/Kite's axicabtagene ciloleucel (Yescarta; axi-cel) and Bristol Myers Squibb's lisocabtagene maraleucel (Breyanzi; liso-cel) — significantly improved event-free survival versus the standard treatment regimen.
While the TRANSFORM and ZUMA-7 studies had slight differences in terms of the patient population treated and were presented at different stages of data maturity, they were more similar than different in terms of their goals, treatment modalities, and disease setting. Therefore, it made sense that the two studies were both positive and showed that CAR T-cell therapy improved event-free survival.
But now a third randomized Phase III trial with the same goals, treatment modality, and disease setting has failed to demonstrate an event-free survival benefit with CAR T-cell therapy versus standard second-line treatment for LBCL.
In the Phase III BELINDA trial, presented Tuesday by Michael Bishop, director of the University of Chicago's cell therapy center, Novartis pitted its own autologous CAR T-cell therapy tisagenlecleucel (Kymriah; tisa-cel) against the same standard treatment as the other studies, chemotherapy followed by autologous stem cell transplant contingent on chemo response.
The trial randomized 322 patients to either arm, and after a median follow-up of 10 months, there was no difference in event-free survival, which was defined within BELINDA as disease progression, death by any cause, or stable disease after 12 weeks. In both arms, median event-free survival was three months.
The therapies' impact on secondary endpoints were fairly similar between the two arms, too. Twenty-eight percent of patients had a complete response in the tisa-cel and the standard treatment arms, and the overall response rates were 46 percent and 43 percent, respectively. Rates of toxicities were also similar with these regimens.
"All of us were very excited to see that the other two trials were positive," Bishop said in a presentation given to members of the media on Monday. "We were hoping that ours would be as well."
When asked to speculate on why the BELINDA study results were negative whereas the other second-line LBCL CAR T-cell studies were positive, Bishop admitted, "it's a tough question, a question that keeps me up every night."
The question is particularly vexing given the past several years of more-or-less comparable performance between Novartis' tisa-cel and Gilead/Kite's axi-cel in the third-line LBCL treatment setting. Both CAR T-cell therapies received US Food and Drug Administration regulatory approval in the later-line setting back in 2017 and 2018, based on single-arm studies demonstrating durable response rates for LBCL patients who had few, if any, other treatment options.
While there has never been a Phase III head-to-head comparison of axi-cel and tisa-cel, real-world analysis of the two drugs suggests they have similar benefit in later-line settings. For example, the Lysa study from the Descar-T Registry, presented Saturday during this meeting, showed that at least in the third-line real-world setting, patients on axi-cel or tisa-cel have comparable overall survival outcomes.
BMS's liso-cel was only approved this past year and was not included in the real-world survival comparison. But, like the other two, it demonstrated durable response rates in the third-line treatment setting.
At the meeting Bishop could not definitively explain why the benefit of tisa-cel should be on par with axi-cel and liso-cel in the third line but inferior in the second line. Still, he offered several possible explanations.
For one, bridging therapy, which is additional chemo administered prior to CAR T-cell infusion, was allowed in the BELINDA trial, and indeed, 35 percent of patients on tisa-cel received one cycle of bridging chemo, and 48 percent received more than one cycle of bridging chemo or a different regimen. In contrast, the ZUMA-7 trial allowed no bridging chemo, and the TRANSFORM trial allowed some, but not as much as BELINDA. Bridging chemo, experts have recognized, can confound trial results, or at least make it harder to distinguish the actual effects of the cell therapy independent of the chemo.
"There were significant differences between the three trials" in terms of allowing bridging chemo, Bishop recognized, "and ours was the most liberal in terms of how many and how much [bridging chemo] you could get." He further pointed out that many patients delayed tisa-cel infusions in BELINDA because they were receiving bridging chemo.
Indeed, in the BELINDA trial, the median time from randomization to infusion for patients treated on the tisa-cel arm was 52 days, which may have given patients' cancers more time to progress. Bishop noted that CAR T-cell infusion turnaround time in the US was slightly improved compared to non-US sites. In contrast, the median time to infusion for axi-cel in the ZUMA-7 study was 27 days from randomization.
"Our findings suggest the importance of preventing progressive disease prior to CAR T-cell infusion," Bishop said. "Shorter infusion times for this chemotherapy-refractory patient population could be critical to improve results."
Beyond the bridging chemo and the delayed infusion times, Bishop suggested that the way that BELINDA investigators defined event-free survival may have played into the differential results. This endpoint for the purposes of this trial comprised disease progression, death by any cause, or stable disease after 12 weeks. "The choosing of the 12-week mark, we think, was not necessarily fair," Bishop said during the press conference.
The other trials didn't put a time limit for stable disease in their event-free survival definitions. In ZUMA-7, event-free survival comprised "time from randomization to the earliest date of disease progression … commencement of new lymphoma therapy, or death from any cause." In TRANSFORM, researchers defined the endpoint as "time from randomization to death from any cause, progressive disease, failure to achieve complete response or partial response, or the start of new antineoplastic therapy due to efficacy concerns, whichever occurs first."
While this difference in defining event-free survival may not be the most important explanation for why BELINDA's results varied from ZUMA-7 and TRANSFORM, Bishop noted it could be a starting point for evaluating potential reasons.
Bishop further said that the different types of lymphodepleting chemotherapy regimens used to clear out patients' immune systems so their bodies accept the modified CAR T-cells may have also played a role.
Ultimately, ASH attendees were reminded on Tuesday that "the biology of [second-line LBCL] is significantly different" than other LBCL settings. According to Bishop, the fact that these patients didn't respond to their first-line treatment regimens is worth considering, since initial therapy for LBLC "can be highly successful."
"Tisa-cel has been highly successful in other lines in diffuse LBCL, but I think our trial design did not permit it to have ultimate efficacy," he concluded.
The BELINDA investigators are actively conducting additional analyses to determine what, if anything, may have affected outcomes in the trial.