Overcoming BCMA-negative or BCMA-low escape has been proposed as capable of reversing the resistance of malignant plasma cells to BCMA Si-CAR T-cell therapy ^[47]. However, after examining more than 200 patients treated with BCMA Si-CAR T-cell therapy in several trials ^{[8][48][49][50][51][52][53][54]}, BCMA-negative cells were detected in only two patients who relapsed. A BCMA-negative plasma cell population was present in one patient ^[50], while BCMA-negative and BCMA-positive plasma cells were present in the other patient ^[55]. By comparison, 10 patients relapsed with BCMA-positive expression or BCMA expression returning to the baseline level ^[53][56]. Therefore, evidence of relapse resulting from loss or down-regulation of BCMA expression derived from current clinical data is scarce, making the evidence of BCMA-negative or BCMA-low escape not as robust as that for CD19. 

Despite limited evidence on the failure of response due to BCMA escape among trials with Si-CAR T-cell therapy targeting BCMA, the combination of BCMA CAR and a second CAR is still being explored in MM  ^[11][12]. Given that the DOR of present BCMA Si-CAR T-cell therapy in MM is far from satisfactory, Bi-CAR T-cell therapy targeting other antigens together with BCMA might warrant further investigation. Although clinical efficacy, such as response and survival, has been reported to be irrelevant to BCMA expression ^[48][57], the data on the detailed expression pattern over time in responders who relapsed are limited. Meanwhile, it is unclear whether patients with reduced BCMA expression have been enrolled in current trials of Bi-CAR T-cell therapy. It may be worthwhile to design trials that include the tracking of BCMA expression and the related response in the individual patient during the clinical course, especially among patients treated with Bi-CAR T cells after relapse of BCMA Si-CAR T-cell therapy, to identify patients with reduced BCMA expression compared to baseline who could benefit from BCMA Bi-CAR T-cell therapy.

 Bi-CAR T-cell therapy is less likely to cause severe CRS and NT than Si-CAR T-cell therapy. There seems to be a difference in the safety profile with respect to the occurrence of CRS and NT between Bi-CAR T-cell therapy and Si-CAR T-cell therapy by simply looking at the numbers; however, considering the sample size, different clinical sites, and possible inadequate management of CRS and NT during early development of CAR T-cell therapy, it seems more investigations are needed to confirm this conclusion.

No matter how effective dual CAR strategies have been in preclinical models, it may only be considered a true success when it benefits patients in the clinic. Emerging clinical data in 2021 permit a fair comparison of different dual-targeting CAR T-cell therapies possible in patients (Table 4Table 1).

Table 41. Comparison of optimization process, transduction efficiencies, DOR, and OS among different dual-targeting CAR T-cell therapies (n > 10; ALL and NHL).

Abbreviations: ALL, acute lymphoblastic leukemia; CR, complete response; DOR, duration of response; EFS, event-free survival; f/u, follow-up; LFS, leukemia-free survival; mon, month(s); NA, not available; non-HSCT, no hematopoietic stem cell transplantation; OS, overall survival; PFS, progression-free survival; pts, patients; Ref., reference; RFS, relapse-free survival.

When comparing clinical outcomes on CD19/CD22 Loop Bi-CAR T-cell therapy with different locations of CD19 scFv and CD22 scFv on CAR in NHL subjects , the PFS in the trial on CAR T cells expressing Bi-CAR with CD19 scFv distal to 4-1BB was longer than the one with CD22 scFv distal to 4-1BB [34], both of which have similar OS. Of note, the transduction efficiency of the CD19/CD22 Loop CAR with CD19 scFv distal to the 4-1BB was lower than the lowest one of the CD19/CD22 Loop CAR with CD22 scFv distal to 4-1BB, indicating that optimization may still be needed. Transduction efficiencies of CD19/CD20 Tandem Bi-CAR T-cell product in patients with NHL dropped to nearly one-third of the those in vitro. However, the PFS and OS rates of patients given CD19/CD20 Tandem Bi-CAR T-cell therapy were higher than those with CD19/CD22 Loop Bi-CAR T-cell therapy, despite higher transduction efficiencies observed in the latter. Of course, caution is needed to interpret non-head-to-head studies, and the differences may be due to different targets in NHL. Overall, poor transduction efficiencies may not necessarily worsen clinical outcomes, though improving transduction efficiencies still matters in optimization.

Despite a great deal of effort directed at optimizing the Bi-CAR T-cell product, the outcomes have not been able to outperform the simple strategy of the cocktail/sequential infusion of two Si-CAR T-cell products without relentless optimization. For ALL and NHL, the cocktail/sequential infusion of CD19/CD22 Si-CAR T-cell products achieved the longest median OS [16], not only providing convincing clinical evidence of dual-targeting CAR T-cell therapy to improve survival but also dwarfing other time- and cost-consuming trials from preclinic to clinic. It is time for different research groups to collaborate and share details on optimizing Bi-CAR structure and standardize clinical trials to compare different dual-targeting therapeutic strategies in a quest for the ideal construct to produce Bi-CAR. Meanwhile, the cocktail/sequential infusion of two Si-CAR T-cell products still merits clinical application to save the lives of patients with ALL and NHL if the commercialization of other dual CAR strategies requires additional time.

In terms of experience gained from the cocktail/sequential infusion of two Si-CAR T-cell products, the timing of the second infusion warrants further exploration. During the trial on the cocktail infusion of BCMA/CD19 Si-CAR T-cell products, CD19 Si-CAR T and BCMA Si-CAR T-cell product were infused on the same day ^[58]. During the cocktail/sequential infusion of CD19/CD22 Si-CAR T-cell products, CD22 Si-CAR T cells were infused one day before CD19 Si-CAR T cells ^[16]. Comparing the cocktail infusion of BCMA/CD19 Si-CAR T-cell products  ^[58] with BCMA/CD38 Tandem Bi-CAR T-cell product ^[59], the PFS of the former was much shorter than that of the latter. Again, the variance may be attributed to the difference in targets. However, it may be worthwhile to adjust the timing of the cocktail/sequential infusion of two Si-CAR T-cell products to standardize the comparison and investigate the influence of the timing of the infusion on the expansion of two different Si-CAR T-cell products in preclinical models.

4. Challenges and Perspectives

Advancing technologies have made Bi-CAR T-cell therapy readily available; however, three main limitations remain for Bi-CAR T-cell therapy: (1) Bi-CAR T-cell therapy does not address other proposed resistance mechanisms outside of target antigen loss; (2) evidence on the safety profile and in vivo activity of Bi-CAR T cells are insufficient; and (3) increased difficulty in manufacturing since the size of construct is bigger. The specific challenges within Bi-CAR T cell manufacturing are the complicated optimization process to find the suitable vectors for manufacturing, increased inconsistency in batch manufacture of viral vector, low transduction efficiency in Bi-CAR T cells, and high manufacturing failure rate due to the size of the bivalent and bicistronic vector.

In conclusion, dual-targeting CAR T-cell therapy has offered another hope for patients in the post era after the use of Si-CAR T-cell therapy. The clinical efficacy has been validated in trials on cocktail/sequential infusion of Si-CAR T-cell products and in a few trials of Bi-CAR T-cell therapy. However, an optimal Bi-CAR structure has not been established. The pooled safety profiles of Bi-CAR T-cell appear better than those of Si-CAR T-cell therapy, with a lower incidence of severe CRS and NT. No apparent effect of 1 + 1 > 2 in terms of DOR, OS, and PFS has been demonstrated in trials on Bi-CAR T-cell therapy, indicating that further optimization is needed. The optimization of the Bi-CAR should focus on finding the right targets for different indications, the appropriate spatial structure of two different scFvs, the suitable linker for scFvs, and the proper transduction efficiencies using patients’ T cells to enhance the efficacy and the persistence of Bi-CAR T cells in patients.

The lack of a magic bullet as Bi-CAR structure calls for collaboration of different research groups to develop solutions to benefit the global community. Models integrating clinical data with preclinical data to predict the optimal Bi-CAR may help design an ideal vector for Bi-CAR introduction. Meanwhile, more trials with Bi-CAR T-cell therapy in patients without prior exposure to Si-CAR T-cell therapy are also needed to compare the two types of CAR T-cell therapy.