Efficacy and Safety of CAR-Modified T Cell Therapy in Patients with Relapsed or Refractory Multiple Myeloma: A Meta-Analysis of Prospective Clinical Trials

Background: In recent years, chimeric antigen receptor-modified T (CAR-T) cell therapy for B-cell leukemia and lymphoma has shown high clinical efficacy. Similar CAR-T clinical trials have also been carried out in patients with refractory/relapsed multiple myeloma (RRMM). However, no systematic review has evaluated the efficacy and safety of CAR-T cell therapy in RRMM. The purpose of this study was to fill this literature gap. Methods: Eligible studies were searched in PUBMED, EMBASE, the Cochrane Central Register of Controlled Trials (CENTRAL), CNKI, and WanFang from data inception to December 2019. For efficacy assessment, the overall response rate (ORR), minimal residual disease (MRD) negativity rate, strict complete response (sCR), complete response (CR), very good partial response (VGPR), and partial response (PR) were calculated. The incidence of any grade cytokine release syndrome (CRS) and grade ≥3 adverse events (AEs) were calculated for safety analysis. The effect estimates were then pooled using an inverse variance method. Results: Overall, 27 studies involving 497 patients were included in this meta-analysis. The pooled ORR and MRD negativity rate were 89% (95% Cl: 83–94%) and 81% (95% Cl: 67–91%), respectively. The pooled sCR, CR, VGPR, and PR were 14% (95% Cl: 5–27%), 13% (95% Cl: 4–26%), 23% (95% Cl: 14–33%), and 15% (95% Cl: 10–21%), respectively. Subgroup analyses of ORR by age, proportion of previous autologous stem cell transplantation (ASCT), and target selection of CAR-T cells revealed that age ≤ 55 years (≤55 years vs. > 55 years, p = 0.0081), prior ASCT ≤70% (≤70% vs. > 70%, p = 0.035), and bispecific CAR-T cells (dual B-cell maturation antigen (BCMA)/BCMA + CD19 vs specific BCMA, p = 0.0329) associated with higher ORR in patients. Subgroup analyses of remission depth by target selection suggested that more patients achieved a better response than VGPR with dual BCMA/BCMA + CD19 CAR-T cells compared to specific BCMA targeting (p = 0.0061). In terms of safety, the pooled incidence of any grade and grade ≥ 3 CRS was 76% (95% CL: 63–87%) and 11% (95% CL: 6–17%). The most common grade ≥ 3 AEs were hematologic toxic effects. Conclusion: In heavily treated patients, CAR-T therapy associates with promising responses and tolerable AEs, as well as CRS in RRMM. However, additional information regarding the durability of CAR-T cell therapy, as well as further randomized controlled trials, is needed.


INTRODUCTION
Multiple myeloma (MM) is the second most common hematological malignancy after non-Hodgkin's lymphoma. It is characterized by clonal evolution of malignant plasma cells (Lipe et al., 2016). During the past decades, autologous stem cell transplantation (ASCT) and the development of novel agents, such as proteasome inhibitors (PIs), immunomodulatory drugs (IMiDs), and monoclonal antibodies, have significantly prolonged patient survival. Although MM treatment options have gradually improved, relapsed and refractory diseases are common (Palumbo and Anderson, 2011;Rajkumar, 2011;Chim et al., 2018;Goldschmidt et al., 2019). It is, therefore necessary to develop innovative treatment strategies to achieve long-term remission for patients with relapsed/refractory MM.
Chimeric antigen receptor (CAR)-T cell therapy has shown the potential for inducing durable remission in certain hematologic malignancies (Makita et al., 2017;Mikkilineni and Kochenderfer, 2017;Neelapu et al., 2018). Meanwhile, anti-CD19 CAR-T-cell therapies reportedly offer promising efficacy in patients with leukemia or lymphoma. Based on previous successful results in B-cell neoplasms (Maude et al., 2014;Lee et al., 2015;Turtle et al., 2016a;Kochenderfer et al., 2017;Neelapu et al., 2017;Jain et al., 2018;Maude et al., 2018;Park et al., 2018), this approach has been licensed by the US Food and Drug Administration (FDA) for the treatment of relapsed or refractory acute lymphocytic leukemia (ALL), and diffuse large B-cell lymphoma (DLBCL). CAR-T cell therapy is defined as a novel immunotherapy that modifies T-cells with CAR, typically consisting of a target-recognition ectodomain, an anchored functional transmembrane domain, a hinge region, and signaling endodomains (Jensen and Riddell, 2015;Guedan et al., 2018). Selection of targets is the key to successful CAR-T therapy (Melchor et al., 2014). Currently, in the context of RRMM, targets used in clinical trials include the B-cell maturation antigen (BCMA), CD19, CD138, signaling lymphocytic activation molecule 7 (SLAM7), immunoglobulin light chains, and the fully human heavy-chain variable domain (FHVH) (Hajek et al., 2013;Lam et al., 2020).
Design and optimization of CAR-T therapy in RRMM has been a hot research area with several prospective clinical trials having been conducted to evaluate its efficacy and safety. However, there is a lack of quantitative and comprehensive statistical analyses on treatment outcome. Moreover, the factors contributing to CAR-T-cell therapy efficacy and safety in RRMM patients remain unclear. Therefore, a systematic review and meta-analysis on the efficacy and safety of the CAR-modified T cell therapy in RRMM patients were performed to offer an evidence-based reference for clinicians.

Methods
In performing this study, we abided by the standards set by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Knobloch et al., 2011).

Literature Search
We searched PUBMED, EMBASE, the Cochrane Central Register of Controlled Trials (CENTRAL), CNKI, and WanFang from inception of the study to December 20, 2019 without any language restriction. We combined Medical Subject Headings (MeSH) terms and free-text terms regarding "CAR" and "myeloma" to search for potentially eligible studies.

Inclusion and Exclusion Criteria
We included clinical trials (phase 1 and phase 2 single arm trials) involving patients with relapsed or refractory MM receiving CAR-T cell therapy. Qualified studies reported at least one of the following variables: efficacy outcomes (overall response rate, ORR), strict complete response (sCR), complete response (CR), very good partial response (VGPR), partial response (PR), minimal residual disease (MRD) negativity rate, and safety outcomes (any grade cytokine syndrome, CRS), grade ≥ 3 AEs (anemia, neutropenia, lymphopenia, thrombocytopenia), and grade ≥ 3 CAR-T-related encephalopathy syndrome (CRES). No restrictions on sample size or length of follow-up were imposed.

Study Qualitative Assessment
The Methodological Index for Non-randomized Studies (MINORS) was adopted to assess the methodological quality of the inclusive studies. MINORS contained 12 items, eight of which were specified for non-comparative studies (Slim et al., 2003;Cullis et al., 2020). The eight items included: study aims, consecutive patient inclusion criteria, prospective pooling of data, endpoint consistent with the study aim, unbiased evaluation of endpoints, follow-up period, loss to follow-up less than 5%, and prospective calculation of the sample size. The items were scored 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate).

Data Extraction
Two investigators independently reviewed and extracted the following information: study characteristics (first author, publication year, ClinicalTrials.gov number, research design), patient characteristics (the group number, age, median time from diagnosis, prior lines of treatment, high-risk cytogenetics, previous ASCT, anti-CD38 monoclonal antibodies exposed, extramedullary-disease), intervention (CAR-T cell dose, target selection, costimulatory domain, conditioning regimen), and outcomes of interest (treatment response, adverse events (AEs)). Discrepancies were settled by discussion or by adjudication by a third reviewer.

Statistical Analysis
We used the Metaprop module in the R-3.4.3 statistical software package to analyze therapeutic efficacy and safety. The effect estimates were pooled using an inverse variance method. Heterogeneity among studies was evaluated by the chi-squared test (χ2 test) and I-squared test (I 2 test). In case of potential heterogeneity (I 2 > 50%), analysis was conducted using the random-effect model; otherwise, the fixed-effect model was employed. Subgroup analysis by age (≤55 vs. >55 years), proportion of high-risk cytogenetics (≤50% vs. >50%), proportion of previous ASCT (≤70% vs. >70%), conditioning regimen (cyclophosphamide plus fludarabine vs cyclophosphamide only), target selection for CAR-T therapy (specific BCMA vs. dual BCMA/BCMA + CD19 vs BCMA + others), costimulatory domain (4-1BB vs. CD28 vs. CD28 + OX40) was performed to explore the sources of heterogeneity. P values < 0.05 were considered statistically significant. Sensitivity analysis was aimed at estimating the effect with removal of the largest sample size among all studies. Li et al. (2018) 28 Cohen et al. (2019) Damian (2018)

Study Quality
All studies illustrated the aim of the study. Their endpoint was appropriate to the aim of the study and data were prospectively collected. In most studies (approximately 80%) consecutive patients were enrolled, an unbiased evaluation of endpoints was performed, and loss to follow-up did not exceed 5%. Twenty-six studies (96%) did not prospectively calculate the sample size. In general, the overall rating was high, and the overall quality of the selected studies was adequate ( Table 3).

Sensitivity Analysis
Sensitivity analysis showed that after removal of the largest sample size among all studies, the pooled ORR did not change significantly. Moreover, the results of the meta-analysis were stable (Table 7).

DISCUSSION
In the last decade, CAR-T therapies have been extensively developed for the advancement of individualized clinical cancer immunotherapy. This meta-analysis, which examined 27 prospective studies involving 497 patients, has demonstrated that CAR-T therapy offered promising outcomes with a tolerable safety profile in RRMM patients.
Our meta-analysis suggests that CAR-T cell therapy could address the negative effects associated with high-risk cytogenetics (≤50% vs. > 50% 84.21% vs. 82.17%) and exhibited a higher efficacy against MM resistant to previous therapies including IMiDs, PIs, anti-CD38 monoclonal antibody, and ASCT. Notably, patients who did not receive prior ASCT achieved a better response, suggesting that ASCT is an irreplaceable component of RRMM patient treatment.
CAR-T cell-based therapies mechanistically differ from all other MM treatment modalities. CAR-T cells can be optimized to specifically kill tumor cells, or reshape the tumor microenvironment by releasing soluble factors capable of regulating the function of matrix or immune cells (Fujiwara, 2014;Maus et al., 2014;Park et al., 2016). Hence, they represent a powerful tool for targeting multiple constituents of the tumor ecological system (Ye et al., 2018). When stimulated by primary MM cells, anti-BCMA-CAR-transduced T cells produce IFN-c and kill them. In fact, serum from patients receiving BCMAspecific CAR-T cells kill target cells that express BCMA in vitro through complement-mediated lysis and antibody-dependent cytotoxicity (Bellucci et al., 2005). Some studies also suggest that earlier CAR-T intervention, at a stage when T cells may be intrinsically "fitter," may be particularly effective (Kay et al., 2001;Dhodapkar et al., 2003;Suen et al., 2016). Based on these arguments, deciding whether CAR-T therapy should be administered early is challenging, particularly for patients with unfavorable cytogenetics.
Additionally, the efficacy appeared to be independent of conditioning scheme, as the combination of cyclophosphamide/fludarabine (Cy-Flu) appears to produce CAR-T cell dynamics similar to that of cyclophosphamide alone. This differed from the CD19-specific CAR-T cell-based therapy in relapsed/refractory B cell non-Hodgkin's lymphoma, where Cy/Flu lymphodepletion resulted in higher response rates (50% CR, 72% ORR) compare to those elicited by the Cy-based lymphodepletion without Flu (8% CR, 50% ORR) (Turtle et al., 2016b). Our research demonstrates that the normal expansion and activity of CAR-T cells in MM may not require exhaustive lymphatic depletion, as patients with MM may have intrinsically "fitter" T cell reserves compared to patients with B cell non-  Hodgkin's lymphoma. Therefore, a single CAR-T conditioning protocol may be applied in future patient management. Previous studies have suggested that specific product features, including the design of engineered costimulation, may impact therapeutic efficacy (Long et al., 2015;Zhao et al., 2015). In contrast, our present study showed that a similar overall response rate (ORR) was elicited by different costimulatory domains (4-1BB, CD28, and CD28 plus OX40), which may indicate that the Frontiers in Pharmacology | www.frontiersin.org December 2020 | Volume 11 | Article 544754 9 small patient samples sizes, as well as the diverse differences in study designs, including the inclusion criteria, broad range of efficacious doses, treatment schedule, and lymphodepletion regimen, preclude drawing definitive conclusions. Notably, the production of CAR-T cells depends, to a large extent, on numerous manual, open-process procedures, and cell culture media to reach a clinical therapeutic dosage (Sadelain, 2009;Sadelain et al., 2013). These characteristics may limit the application of this approach to large-scale, multicenter clinical trials. Therefore, studies are needed to streamline and optimize the production process. Moreover, additional steps should be standardized to maximize the process consistency (Roberts et al., 2018).
The initial success of the CD19-targeted CAR-T cell therapy in B-cell malignancy emphasizes that selecting the optimal surface target antigens is critical for efficient CAR-T cell therapeutics. However, first-rank surface antigens remain to be identified in MM. Nevertheless, several alternative antigens have been used in CAR-T cell therapy against MM (Bolli et al., 2014;Tai et al., 2016). In our study, the BCMA, dual BCMA, CD196, CD38, TACI, and FHVH were considered. The results show that LCAR-B38M and combined CD19/BCMA exhibit higher overall response rates and deeper responses compared to specific BCMA. In the design of LCAR-B38M, the antigen recognition portion consists of two camel antibody heavy chains against two BCMA epitopes. This structure may enhance the antigen recognition specifically as well as the affinity of CAR-T cells for antigen, resulting in a stronger anti-MM effect (Shah et al., 2020). In terms of immunophenotype, the dominant clones of most myeloma patients are similar to the most differentiated normal plasma cell subset: CD38 + CD138 + CD19 − . A few MM clone subsets with poorly differentiated plasma cell phenotypes (CD138lo/-or CD19 + ), or a B cell phenotype (CD138-CD19 + CD20 + ) can also be found in patients. Moreover, according to a clinical trial and in vitro study using immunodeficient mice, poorly differentiated components in MM clones are also involved in disease pathogenesis. In addition, CD19 was found to be expressed on only a small proportion of myeloma cells (Bagg et al., 1989;Paiva et al., 2017;Garfall et al., 2018;Nerreter et al., 2019). Hence, the combination of CD19 and BCMA may tackle MM pathogenesis more effectively and result in enhanced antitumor effects.
Although our study included some patients without an MRD status reported, the high rate of pooled MRD negativity in patients (81%, 67%-91%) was inspiring. In contrast, a recent study exploring the effects of daratumumab plus pomalidomide-dexamethasone for RRMM showed that 35% and 29% of the patients could be assessed as MRD negative at a threshold of 10 −4 and 10 −5 nucleated cells, respectively (Chari et al., 2017). Meanwhile, previous studies showed that the MRD status was one of the most relevant independent prognostic factors in MM. Compared with patients achieving CR who are MRD positive, patients who are MRD negative may have longer overall, and progression-free survival (PFS) (Paiva et al., 2015;Kumar et al., 2016;Munshi et al., 2017). Despite the high response rate, it remains unknown whether CAR-T cells have the potential to induce long-lasting remission in RRMM, as observed with the CD19 CAR-T cells in B-cell malignancy. Longer follow-ups for patients who exhibit a response and are MRD negative will be required to address this question.
CRS was determined to be primarily of grade 1 or 2. The reported incidence of grade 3 or higher with CD19-directed CAR-T cells was 46% with tisagenlecleucel and 13% with axicabtagene ciloleucel (Neelapu et al., 2017;Maude et al., 2018), which is higher than our results (11%). The overall occurrence of grade three or four neurologic toxic events was also low (8%). Generally, the safety profile was tolerable and manageable.
In conclusion, in an era in which numerous novel agents for MM are emerging, CAR-T cells demonstrate a high overall response and a good remission rate in heavily treated patients (Miguel et al., 2013;Lonial et al., 2016;Chen et al., 2018). However, further information regarding the durability of the CAR-T cell-based therapy is needed. Owing to the lack of control groups and the small sample sizes of the examined studies, our results require confirmation by randomized controlled trials. Finally, as continuous development of MM therapeutic agents is underway, the optimization of timing, sequensce, and combination with other therapies will be crucial to obtain adequate responses and substantially increase patient survival (Trudel et al., 2018;Kumar et al., 2019;Parrondo et al., 2020).

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
XX collected, analyzed the data, and wrote the article. QH, YO, and WW collected the data, helped in subgroup analysis and prepared the figures and tables. YW and QH designed research, provided the plan and modified the manuscript. All authors read and approved the final manuscript.