Cancer and Cell Therapy · us
Lung cancer CAR-T response tracked for over a year, with CNS relapse highlighting limits of efficacy
A2 Bio has updated two early-stage trials, providing preliminary signals for logic-gated CAR-T therapy in solid tumors. However, the lung cancer patient experienced a central nervous system relapse, and the complete response in anal cancer remains a single case. The durability and applicability of efficacy remain to be confirmed.
Enabling immune cells to recognize cancer cells while sparing normal tissue is a challenge for CAR-T therapy in solid tumors. On November 3, A2 Biotherapeutics announced updates from two Phase 1/2 trials. The research team continued to classify one lung cancer patient's complete response as ongoing after more than 12 months of follow-up. However, a central nervous system relapse documented in an earlier announcement means this clinical course cannot simply be described as “continuously cancer-free throughout the body.”
The patient with non-small cell lung cancer who received A2B694 had KRAS G12V and STK11 mutations and had experienced disease progression after carboplatin, pemetrexed, and pembrolizumab. According to the company's May announcement, the patient achieved a complete response on day 90 after infusion, confirmed by central review on day 180. However, an isolated central nervous system relapse occurred on day 243, while a complete response outside the central nervous system was maintained on day 284. The November update provided no further details on the management or subsequent course of that relapse. Therefore, “more than a year of follow-up” cannot be interpreted as having had no relapse during that period, much less as a cure.
A2B694 uses the patient's own T cells, equipped through the Tmod platform with two receptor systems: one recognizes mesothelin (MSLN) and activates an attack, while the other recognizes HLA-A*02 and inhibits the attack. The trial selects patients whose tumors have lost HLA-A*02 while their normal cells retain this marker, seeking to use this difference to protect normal tissue. This selectivity is a design goal, and it also means the results cannot be directly applied to all patients with lung cancer.
Beyond the lung cancer case, overall responses remain limited. As of September 1, among 13 evaluable patients who received A2B694 in EVEREST-2, 1 had a complete response, 8 had stable disease, and 4 had progressive disease. The trial also enrolled 3 patients who received A2B543, which incorporates an IL-12 enhancement. The best response in both of the 2 evaluable patients with pancreatic cancer was stable disease. There is currently insufficient evidence to determine whether the enhancement can improve efficacy.
The other trial, DENALI-1, uses donor-derived A2B395, which targets EGFR. As of August 21, 11 patients had received treatment, and 9 were evaluable for efficacy: 1 had a complete response, 1 had a partial response, 5 had stable disease, and 2 had progressive disease. The patient with anal cancer who achieved a complete response did so on day 28 after infusion, after five chemotherapy regimens and two courses of radiotherapy. A partial response was confirmed on day 180 in the patient with head and neck cancer. These two cases had already appeared in the September announcement; the November data added no new cases of objective response.
Safety must also be considered in the context of each trial. EVEREST-2 has had no dose-limiting toxicities to date, but there was one case of grade 3 immune effector cell-associated neurotoxicity syndrome (ICANS) in the A2B694 group. The September announcement for DENALI-1 reported no ICANS, graft-versus-host disease, or fatal adverse events, and both cases of grade 1 or grade 2 cytokine release syndrome had resolved. These small datasets involving different products and different patients are insufficient to compare the safety of the two.
These company-announced results, prepared for presentation at the SITC annual meeting, provide early evidence of activity for logic-gated CAR-T in solid tumors. However, both trials are small, lack control groups, and include multiple cancer types. A deep response in a single patient cannot establish whether most patients can benefit. More complete records of relapse sites and subsequent treatment, longer follow-up, and results from more patients with the same cancer type are needed to determine whether this recognition mechanism can deliver reproducible clinical benefit.