Preliminary outcomes reported from three randomized controlled trials of CD19 CAR-T cell therapies in large B cell lymphoma

Main text

The first ever randomized controlled trials (RCTs) testing chimeric antigen receptor (CAR) T cell therapy against a standard of care (SOC) are to be reported at the American Society for Hematology Annual Meeting in December 2021. Based on the published abstracts, the control arms had similar outcomes between the three trials. However, results on the CAR-T cell arms differed, with two of the three trials reporting that they met their primary endpoint favoring CAR-T cell therapy.

Frontline treatment of diffuse large B cell lymphoma (LBCL) consists of chemoimmunotherapy (CIT) with anti-CD20 and anthracycline-containing chemotherapy, which cures 50%–80% of patients depending on baseline risk. In refractory or relapsing (R/R) patients, SOC (and the control arm in the three RCTs) comprises second-line CIT with anti-CD20 and platinum-containing chemotherapy. In patients responding to second-line CIT, autologous stem cell transplant (ASCT) is potentially curative. For patients who are refractory to second-line CIT, or who relapse after ASCT, CAR-T cell therapy is already approved for three products: axicabtagene ciloleucel (axi-cel), lisocabtagene maraleucel (liso-cel), and tisagenlecleucel (tisa-cel). Now, three separate RCTs have tested these three products if R/R after their frontline CIT: ZUMA-7 (axi-cel versus SOC),¹ TRANSFORM (liso-cel versus SOC),² and BELINDA (tisa-cel versus SOC)³ (Table 1).

Table 1.

Characteristics of randomized trials of CAR-T cell therapy in second line LBCL

RCT	ZUMA-7	TRANSFORM	BELINDA	Editorial comment
Product	axi-cel	liso-cel	tisa-cel
CAR target	CD19	CD19	CD19
Costimulation	CD28/CD3zeta	4-1BB/CD3zeta	4-1BB/CD3zeta
Vector	gamma retrovirus	lentivirus	lentivirus
T cell selection	No	yes	yes
CD4:CD8 selection	no	CD4:CD8 infused in a 1:1 ratio	no
Lymphodepletion	Flu 30 mg/m2 × 3 days	Flu 30 mg/m2 × 3 days	Flu 25 mg/m2 × 3 days	tisa-cel has the lowest lymphodepletion dose
	Cy 500 mg/m2 × 3 days	Cy 300 mg/m2 × 3 days	Cy 250 mg/m2 × 3 days
CAR-T dose	2 × 106 CAR-T/kg	1 × 108 CAR-T	0.6–6 × 108 CAR-T; median 2.9 × 108
Inclusion criteria	refractory or relapsed within 12 months of 1st line	refractory or relapsed within 12 months of 1st line	refractory or relapsed within 12 months of 1st line	similar inclusion populations
N	359	184	322	transform smaller than the others
Control arm				control arm outcomes similar across the three trials
Salvage regimen	2nd line CIT	2nd line CIT	2nd line CIT; 3rd line allowed	3rd line chemoimmunotherapy allowed on BELINDA
ASCT (%)	36	47	33	highest rate of ASCT on TRANSFORM
ORR (%)	50	48	43
CR (%)	32	39	28
EFS median (months)	2	2.3	3	events defined differently in each trial; control arm EFS similar
OS median (months)	35.1	16.4	–	differences in median OS between control arms of ZUMA-7 and TRANSFORM may be related to a durable OS of around 50% in both trials—longer follow-up needed
Crossover to CAR-T (%)	56	55	51	crossover built into the TRANSFORM trial, not ZUMA-7; similar crossover in all 3
CAR-T arm
Bridging regimen	steroids only (no chemo)	63%	83% 1 cycle = 36% 2 + cycles = 47%	effective bridging could improve CAR-T outcomes; ineffective bridging could worsen
CAR-T infused (%)	94	98	96
ORR (%)	83	86	46	efficacy outcomes worse with tisa-cel
CR (%)	65	66	28
EFS median (months)	8.3	10.1	3	events defined differently in each trial; tisa-cel EFS lowest
OS median	not reached	not reached	–
Grade 3 + cytokine release syndrome (%)	6	1	5
Grade 3 + ICANS (%)	21	4	3	higher frequency of severe ICANS with axi-cel
Comparison CAR versus SOC
Met primary endpoint?	yes—favor CAR-T	yes—favor CAR-T	no
	EFS p < 0.0001	EFS p < 0.0001
OS benefit?	interim p = 0.027	interim p = 0.026	N/A	threshold for significance more stringent than p < 0.05 for interim analyses

All three trials chose as the control arm a population at high risk to have poor outcomes with SOC, including patients either refractory to frontline CIT or relapsed within 12 months. The results of the control arms of the three trials were remarkably similar, with overall response rates (ORRs) of 43%–50% and complete response (CR) of 28%–39%. This suggests that the patients enrolled across the three trials had similar risk characteristics. After second-line CIT, patients proceeded to ASCT based on investigator interpretation of response, which may vary from centralized review. The TRANSFORM trial had a higher rate of patients proceeding to ASCT at 47% versus 33% and 36% in the other two trials. Nonetheless, high-risk patients may relapse despite ASCT and the median overall survival (OS) of the control arm of the TRANSFORM trial was actually shorter than that of the ZUMA-7 control arm (16 versus 35 months), although medians may exaggerate small differences around 50%. Also affecting survival may be the frequency of crossover to subsequent CAR-T cell therapy in patients relapsing on the control arm. Despite differences in crossover strategy (crossover CAR-T cells were provided on TRANSFORM, while crossover on ZUMA-7 required commercial product or another clinical trial to obtain CAR-T cell therapy), the rates of crossover were similar and ranged from 51% to 56%. If perhaps 20% of patients on the control arms remain in remission after ASCT, this means that a majority of the relapsing 80% in the control arm crossed over to CAR-T.

A unique contributor to the outcomes of CAR-T cell therapy is manufacturing efficiency. “Bridging” therapy given while awaiting manufacturing differed across the three trials. On ZUMA-7, only corticosteroids were allowed as bridging. On TRANSFORM and BELINDA, CIT bridging was allowed, and 63% of patients on TRANSFORM received bridging, and 83% of patients on BELINDA. Furthermore, on BELINDA, multiple cycles of CIT were required in 47% of patients. The presumed reason for these differences is inferior manufacturing efficiency on BELINDA, but full data are needed to understand this better. Nonetheless, the proportion of patients receiving CAR-T cell therapy was high in all three trials, ranging from 94% to 98%. Bridging therapy that successfully reduces tumor burden is helpful to patients prior to CAR-T cell therapies, while for non-responders it is more likely to worsen outcomes. Further data on response to bridging may sharpen the comparison between ZUMA-7 and TRANSFORM. In terms of efficacy outcomes, axi-cel and liso-cel had high ORRs (83%/86%) and CR rates (65%/66%), while tisa-cel had much lower results at an ORR of 46% and CR rate of 28%. These differences translated into differences in EFS and, where reported, PFS. For LBCL, tisa-cel uses lower lymphodepletion intensity than axi-cel and liso-cel, and this may affect outcomes. Another notable difference is that CAR-T cell toxicities, such as cytokine release syndrome and neurotoxicity, were highest for axi-cel and low for liso-cel and tisa-cel. It will be important to know if this translates into non-relapse mortality differences across the three trials.

Comparison of CAR-T cell therapy to SOC: ZUMA-7 and TRANSFORM met their primary endpoint of event-free survival, while BELINDA did not. A careful evaluation of the definition of events, and reporting of progression-free survival may be further useful, but the reported effect size is impressive. A preliminary comparison showed possible OS improvements on ZUMA-7 and TRANSFORM; however, the significance threshold is more stringent than p < 0.05 at interim analysis. Longer follow-up is therefore needed for OS, as is an understanding of how patients are treated if they relapsed after CAR-T cell therapy.

In summary, the first ever RCTs of CAR-T cell therapy versus SOC are being reported. Having three trials totaling 865 patients, with slightly different designs, is extremely informative to understand the effect of CAR-T cell therapy in this setting. Full results are eagerly awaited, but the topline data suggest that axi-cel and liso-cel are effective in a group of patients who are at high risk with SOC therapy.