AbstractBackground and AimsChimeric antigen receptor engineered T cells (CARTs) for HCC and other solid tumors are not as effective as they are for blood cancers. CARTs may lose function inside tumors due to persistent antigen engagement. The aims of this study are to develop low‐affinity monoclonal antibodies (mAbs) and low‐avidity CARTs for HCC and to test the hypothesis that low‐avidity CARTs can resist exhaustion and maintain functions in solid tumors, generating durable antitumor effects.Methods and ResultsNew human glypican‐3 (hGPC3) mAbs were developed from immunized mice. We obtained three hGPC3‐specific mAbs that stained HCC tumors, but not the adjacent normal liver tissues. One of them, 8F8, bound an epitope close to that of GC33, the frequently used high‐affinity mAb, but with approximately 17‐fold lower affinity. We then compared the 8F8 CARTs to GC33 CARTs for their in vitro function and in vivo antitumor effects. In vitro, low‐avidity 8F8 CARTs killed both hGPC3high and hGPC3low HCC tumor cells to the same extent as high‐avidity GC33 CARTs. 8F8 CARTs expanded and persisted to a greater extent than GC33 CARTs, resulting in durable responses against HCC xenografts. Importantly, compared with GC33 CARTs, there were 5‐fold more of 8F8‐BBz CARTs in the tumor mass for a longer period of time. Remarkably, the tumor‐infiltrating 8F8 CARTs were less exhausted and apoptotic, and more functional than GC33 CARTs.ConclusionThe low‐avidity 8F8‐BBz CART resists exhaustion and apoptosis inside tumor lesions, demonstrating a greater therapeutic potential than high‐avidity CARTs.