Refractory and/or relapsed (r/r) diffuse large B-cell lymphomas after treatment with two lines of systemic chemoimmunotherapy exhibit diversity in genetics, tissue biology, and pathology, as well as poor prognosis. Patient TCRαβ cells engineered with a CD19-specific chimeric antigen receptor (CAR) have shown promising clinical outcomes in r/r diffuse large B-cell lymphoma. The ZUMA-1 study, the JULIET study, and the TRANSCEND NHL 001 study of three prototype 19CAR-T cells have indicated an overall response rate of 52–82%, a complete response rate of 40–58%, and a 12-month progression-free survival of 33.2%–46.6%, with clinically manageable treatment related toxicity. At the 5-year follow-up, relapse was observed in approximately 57% of patients within 1 year. Understanding of the risk factors for non-response remains insufficient. In addition to intrinsic tumor resistance, such as aberrant apoptotic signaling, downregulation or loss of tumor-associated antigens (TAA), an immunosuppressive tumor microenvironment, and CAR-T cell exhaustion in vivo have been suggested to be important risk factors. Mechanisms underlying 19CAR-T cell exhaustion under chronic TAA exposure, and limited 19CAR-T cell trafficking and infiltration into the tumor mass have been reported. Moreover, tumor escape in the presence of low TAA density remains a challenge in 1928ζ CAR-T cell treatment. In this review, we provide an overview of modified modular CAR elements and their synergistic effects in controlling T-cell function. We then briefly discuss novel strategies against tumors with low TAA density, such as bispecific tandem or loop CAR recognition domains, the development of human leukocyte antigen-independent synthetic TCRαβ double-chain receptors integrated into the constant region of the TCRα chain, and armored CAR-T cells targeting the tumor microenvironment.