A new CAR takes a test drive in DLBCL

In this issue of Blood, Roddie et al report results from the phase 1 ALEXANDER trial using a novel bicistronic anti-CD19 and anti-CD22 dual-targeted autologous chimeric antigen receptor (CAR) T-cell product, AUTO3, in combination with pembrolizumab, as a third-line or later treatment in patients with relapsed/refractory large B-cell lymphomas (LBCLs).¹ Currently available CAR T cells directed at the single CD19 antigen produce durable remissions in ≈40% of patients with relapsed/refractory LBCL as third-line or later therapy,^2,3 so most patients will progress despite access to this major treatment advance. Targeting a second tumor antigen with the same CAR T cell represents a novel and appealing strategy to try to overcome a mechanism of resistance and enhance efficacy.

The AUTO3 construct and ALEXANDER study design seek to address 2 dominant mechanisms of resistance against CAR T-cell therapy in LBCL: antigen escape and immune escape. Antigen escape refers to downregulation or loss of the target antigen, likely resulting from selective pressure, whereas immune escape refers to CAR T-cell exhaustion with loss of a functional immune response against the antigen-bearing tumor cells. Antigen loss appears to represent only a minority of anti-CD19 CAR T-cell failures, likely up to one-third of LBCL relapses.⁴ Most relapsing patients, however, still express the target antigen and still have detectable CAR T cells in the peripheral blood, pointing to immune escape as a dominant mechanism of resistance.

So how did the novel AUTO3 construct plus pembrolizumab perform? Sixty-two AUTO3 products were manufactured, and 52 patients were treated. Ten patients did not receive their product infusion primarily due to progressive disease or death before product availability. Patients were heavily pretreated with a median of 3 prior lines of therapy, 36.5% had double-hit/triple-hit lymphoma, and 71% were refractory to their last line of therapy. Among 52 treated patients, AUTO3 was generally well tolerated. Cytokine release syndrome (CRS) occurred in 36.5% of patients and was severe in only 1 patient (1.9%) with grade 3 CRS; no grade 4 or 5 CRS events were reported. The incidence of neurologic toxicities was low at 7.7% overall and was severe in 2 patients (3.8%). The low incidence of CAR T-cell–associated toxicity allowed exploration of outpatient dosing of AUTO3, which appeared feasible.

Among 52 treated patients, two-thirds of patients responded to AUTO3, with a complete response rate of 48.9%. The median duration of response and progression-free survival were 8.3 and 3.3 months, respectively. Despite the encouraging response rates, durability was disappointing, with only 25.8% of treated patients remaining progression free at 12 months. Only elevated lactate dehydrogenase and high tumor burden, as measured by sum of the perpendicular diameters, were significantly associated with lower rates of complete response on univariate analyses of patient- and disease-specific variables, which is consistent with findings in other CAR T-cell studies in this patient population.

Did use of a dual-targeted CAR T-cell product along with pembrolizumab prevent either antigen or immune escape? It is not clear that it did. The rate of recurrence does not appear lower than what we would expect with a single targeted anti-CD19 CAR T cell; in fact, the durability appears shorter.^2,3 Furthermore, CD19 and CD22 expression was low or undetectable in 2 and 7 of the 13 relapses, respectively, so antigen loss still occurred. More important, CD22 expression was low or undetectable in most patients at baseline, raising the question of whether CD22 represents an optimal target in relapsed/refractory LBCL. Immune escape was still a common mechanism of treatment failure despite inclusion of pembrolizumab, which is also the case with relapses after anti-CD19 CAR T cells administered without immune checkpoint inhibition. Of 33 relapsing patients, 15 had detectable CAR T cells at the time or recurrence, and increased programmed death ligand 1 (PD-L1) expression after AUTO3 treatment was observed in only 3 of 13 relapsing patients. There were also no differences in outcome or pharmacokinetics observed between the 2 pembrolizumab schedules, either a single dose administered on day −1 or 3 doses given 21 days apart beginning on day 15. These results suggest the pembrolizumab may have added little to the success of the treatment program, but because only 3 patients were treated without pembrolizumab and they were all treated at the lowest AUTO3 dose level, the study does not provide a sufficient control population to specifically address the value of incorporating the immune checkpoint inhibitor. Incorporation of a programmed cell death protein 1 (PD-1) inhibitor, regardless of dose and schedule in this study, therefore appears insufficient to prevent CAR T-cell exhaustion in large B-cell lymphomas. Induction of T-cell exhaustion is multifactorial, however, and so PD-1 may not be the critical player in relapsed LBCL. PD-L1 and PD-L2 are uncommonly expressed in diffuse LBCL (DLBCL), and as observed in the ALEXANDER study, may not be broadly induced by CAR T-cell therapy itself. Treatment of patients with DLBCL relapsing after anti-CD19 with PD-1 inhibitors has also shown disappointing results,⁵ suggesting the presence of alternate mechanisms of resistance in most patients. Multiple other immune checkpoints may also contribute to CAR T-cell exhaustion, so targeting a single checkpoint will likely not be the answer for all patients. Furthermore, numerous additional genomic factors in both the tumor cells and the microenvironment have been associated with CAR T-cell resistance,^6,7 and vary patient to patient. Ultimately, we will need to understand the discrete mechanisms of CAR T-cell resistance in a given patient, and personalize therapy accordingly to overcome that resistance at relapse, or ideally prevent it in the first place. The ALEXANDER study is a step in the right direction in seeking to understand and overcome CAR T-cell resistance so as to optimize the success of this transformative therapy in relapsed/refractory LBCL.

Conflict-of-interest disclosure: J.S.A. reports consulting for AbbVie, Astra-Zeneca, BeiGene, Bluebird Bio, BMS, Celgene, Epizyme, Incyte, Kymera, Genmab, Genentech, Ono Pharma, Mustang Bio, MorphoSys, Regeneron, Century Therapeutics, Kite Pharma, Lilly, Janssen, Takeda, Caribou Biosciences, Interius, and Cellectar.