CD19-specific CAR T Cells that Express a PD-1/CD28 Chimeric Switch-Receptor are Effective in Patients with PD-L1–positive B-Cell Lymphoma

Diffuse large B-cell lymphoma (DLBCL), a biologically and clinically heterogeneous cancer, is the most common subtype of lymphoma (1). Although anti-CD20 mAb, rituximab-based immunochemotherapy can improve clinical outcomes in DLBCL, a substantial number of patients suffer from primary refractory or relapse (R/R) disease after achieving complete remission (CR; refs. 2, 3). The prognosis for these patients is extremely poor, especially for those who have refractory disease (4).

The anti-CD19 chimeric antigen receptor (CAR) T-cell therapy has been demonstrated to induce remarkable clinical remissions in R/R B-cell lymphoma (5–9). The 2-year follow-up data from ZUMA-1 study show that axicabtagene ciloleucel, a first-in-class CAR T-cell therapy, can induce durable responses and a median overall survival (OS) of greater than 2 years in patients with R/R large B-cell lymphoma (10). Similar to the ZUMA-1 study, investigators from JULIET trial reported in their phase II trial of tisagenlecleucel, a second anti-CD19 CAR-T product, which treated 93 patients with R/R DLBCL. The overall response rate of patients who received tisagenlecleucel was 52%; 40% of the patients had CR and 12% had partial remission (PR). Durable responses were observed for up to 18.4 months after treatment (11). However, despite these encouraging clinical results, the existence of different immunosuppressive pathways, such PD-1/PD-L1 pathway, can restrict the full potential of CAR T-cell therapy (12). A multicenter retrospective study reported that among 76 patients with B-cell lymphoma, all three patients with PD-L1⁺ tumors were refractory to CAR T-cell (axicabtagene ciloleucel) therapy (13). In addition, a recent study showed that a patient with PD-L1⁺ DLBCL was refractory to anti-CD19-CAR T treatment, but following administration of PD-1–blocking antibody (pembrolizumab), the patient had a clinically significant antitumor response (14).

Recently, it was demonstrated that PD-1 blockade that may overcome PD-L1⁺ tumor immunosuppression significantly enhances anticancer activity of human CAR T cells against hematologic and solid tumors (15–18). In addition to the combination therapy with CAR T cells and immune checkpoint inhibitor, which has been shown to result in clinical responses and expansion of CAR T cells (14, 19), another strategy to block this pathway is delivery of PD-1 inhibitor by an “armored” CAR T cell that secrets an immune checkpoint blockade single-chain variable fragment (scFv; refs. 20–22). This strategy clearly works in clinically relevant syngeneic and xenogeneic mouse models of PD-L1⁺ hematologic and solid tumors (22). In addition, a novel PD-1/CD28 chimeric switch-receptor containing the extracellular domain of PD-1 fused to the transmembrane and cytoplasmic domain of the costimulatory molecule, CD28, not only engages its ligand, PD-L1, but transmits an activating signal (via the CD28 cytoplasmic domain) instead of the inhibitory signal normally transduced by the PD-1 cytoplasmic domain (23). Thus, T cells transduced with both an anti-CD19 CAR and a PD-1/CD28 chimeric switch-receptor are expected to have more potent antitumor efficacy, particularly for PD-L1⁺ B-cell lymphoma, compared with anti-CD19 CAR T cells.

In this study, we have generated a CD19-specific CAR T–expressing PD-1/CD28 chimeric switch-receptor, termed CD19-PD-1/CD28-CART. This CAR T has superior activity, compared with CD19-targeted second-generation CAR T, against PD-L1⁺ lymphoma cells in vitro and in a xenograft tumor model. We then conducted a phase Ib study to assess the antitumor efficacy, feasibility, and toxicity of CD19-PD-1/CD28-CAR T cells in patients with R/R PD-L1⁺ large B-cell lymphoma. We have shown, for the first time, the induction of durable clinical response in these patients, without significant toxicities.

The study was an open-label and single-arm clinical study approved by the Ethics Committee of the First Affiliated Hospital, College of Medicine, Zhejiang University (Hangzhou, Zhejiang, P.R. China) and each participating institution, and was registered as ClinicalTrials.gov NCT03258047. All enrolled patients provided written informed consent in accordance with the Declaration of Helsinki. Patients between 18 and 75 years of age with R/R large B-cell lymphoma including DLBCL, transformed follicular lymphoma (tFL), and mantle cell lymphoma (MCL) were eligible. All patients had a performance status of 0–3, based on the Eastern Cooperative Oncology Group (ECOG) scale. The other inclusion and exclusion criteria for this study were the same as those for the ZUMA-1 study (8) and are available in the protocol. PD-L1 expression was evaluated by IHC using an anti–PD-L1 mAb (SP142, Abcam), and confirmed by the review committee with a central evaluation.

The PD1/CD28 switch-receptor was constructed as described previously by fusing a truncated extracellular PD-1 (AA1-155) domain with the transmembrane and cytoplasmic domains of CD28 (AA141-220) downstream of the mouse stem cell virus (MSCV) promoter (23). A second expression cassette under the regulation of a human EF-1α promoter expressed the anti-CD19 CAR, which was subcloned into pCDH-MSCV-MCS-EF1-copGFP-T2A-Puro lentiviral plasmid vector. The sequence of anti-CD19 CAR has been described previously (GenBank HM852952), except that cytoplasmic portions of CD28 were replaced with 4-1BB (CD137, AA 214-255, GenBank: U03397.1; refs. 5, 24).

Autologous peripheral blood mononuclear cells (PBMCs) were obtained via nonmobilized leukapheresis, and CD19-PD-1/CD28-CART was generated using lentivirus to transduce PBMCs that were treated with anti-CD3/CD28 antibody–coated beads (Gibco) to enrich for and activate T cells. Preparation of cell production is presented in detail in the Supplementary Table S1.

Multicolor flow cytometry, IHC, quantitative PCR (qPCR), cytokine assay, cytotoxic T-lymphocyte assay, and animal study were performed as described in the Supplementary Data.

Two independent pathologists (Dr. Jing Zhao and Prof. Zhaoming Wang from the First Affiliated Hosptial, College of Medicine, Zhejiang University) examined the stained slides in a blinded fashion. The staining percentages of tumor cells (from 0% to 100%) and the intensity (from 0 to 3: 0, negative; 1, weak positive; 2, median positive; and 3, strong positive) were used to calculate PD-L1 expression scores (H-scores) by multiplying the percentage of tumor cells by the staining intensity, as described previously (25).

Patients underwent leukapheresis to obtain PBMCs to manufacture CAR T cells. Before receiving CAR T-cell infusion, patients received cyclophosphamide 500 mg/m² and fludarabine 30 mg/m² daily on days −5 to −3. On day 0, patients were treated with a single intravenous infusion of CD19-PD-1/CD28-CART that was given at four dose levels, including 0.5, 1, 2, and 4 × 10⁶ CAR⁺ T cells/kg.

Response was assessed by PET-CT at month 3 according to the International Working Group Response Criteria for Malignant Lymphoma (26), and duration of response was evaluated by ultrasonic and CT every 3 months from month 3 to 12.

All patients underwent PET-CT scan before the onset of CAR T-cell therapy. Quantitative parameters were computed by a nuclear medicine physician (Prof. K. Zhao from the First affiliated Hospital, College of Medicine, Zhejiang University) blinded to patient outcome. Metabolic tumor volume (MTV) was defined as the tumor volume with 18F-2[18F]fluoro-2-deoxy-D-glucose uptake segmented above a threshold SUV of 2.5, which was calculated using the Syngo Volume-counting Program (Siemens Medical Solutions).

Adverse events (AEs) were assessed and toxicity was graded using the NCI Common Terminology Criteria for Adverse Events, v.4.03, and cytokine release syndrome (CRS) and CAR T-cell–associated neurotoxicity were assessed according to the American Society for Transplantation and Cellular Therapy grading systems (7).

Unless otherwise noted, data are summarized as mean ± SEM. The unpaired Student t test was used to determine statistically significant differences between two groups. When three or more groups were compared, one-way ANOVA with Dunnett test for multiple comparisons was used. When multiple groups at multiple timepoints were compared, the unpaired Student t test or ANOVA for each timepoint was used. OS was calculated from the date of CAR T-cell infusion to death. Survival curves were compared using the log-rank test. ROC analysis was used to define the optimal cutoff of MTV and PD-L1 for survival prediction. Survival functions were calculated using Kaplan–Meier estimates and comparison between categories was made using the log-rank test. Univariate and multivariate analyses were performed using a Cox proportional hazards model. The reverse Kaplan–Meier method was used to estimate median follow-up time. Results were considered statistically significant with a two-sided P < 0.05. Statistical analyses were performed using Stata 11.0 for Windows (StataCorp LP).

The datasets supporting the conclusions of this article are included within the article and its Supplementary Data.

The study was approved by an independent Ethics Committee of the First Affiliated Hospital, College of Medicine, Zhejiang University (Hangzhou, Zhejiang, P.R. China), and conducted in accordance with the Declaration of Helsinki.

Schematic representations of the lentiviral CAR constructs used in this study are shown in Fig. 1A. The transduction rate of CD19-PD-1/CD28-CAR was similar to that of CD19-CAR (Fig. 1B). The PD-1/CD28 surface expression was demonstrated on T cells transduced with the CD19-PD-1/CD28-CAR vector (Fig. 1C). To evaluate the functional capacity of engineered T cells, the cytotoxic activity of CD19-CART, CD19-CART combined with anti–PD-1 antibody, and CD19-PD-1/CD28-CAR T cells was assessed by luciferase-based cytotoxicity assays at different effector/target (E/T) ratios. Both CAR T cells demonstrated dose-dependent cytotoxic effect when cocultured with the Raji cell line that was genetically modified to overexpress PD-L1 (Raji-PD-L1). However, CD19-PD-1/CD28-CAR T cells were significantly more effective than CD19-CART alone or combined with anti–PD-1 antibody in terms of cytolytic activity, respectively (Supplementary Fig. S1A and 1B). The cytokine production profiles were found to be different in the two types of CAR T cells. Upon antigen stimulation for 24 hours, we observed that CD19-PD-1/CD28-CAR T cells secreted significantly greater amounts of immunostimulatory cytokines, such as IL2, IFNγ, and TNFα, especially at higher E/T ratios in comparison with CD19-CAR T cells, which is consistent with a previous report (23). In contrast, CD19-CAR T cells had increased levels of IL4, IL6, and IL10 secretion (Fig. 1D). To assess whether the enhanced potency of CD19-PD-1/CD28-CAR T cells is attributed to cell intrinsic PD-1 blockade, CD28 triggering, or a combination of both, we compared the secretion of IL2 by CD19-PD-1/CD28-CART, CD19-CART alone, or when combined with anti–PD-1 antibody, anti-CD28 antibody, or a combination of both. The results showed that similar levels of IL2 were produced by CD19-PD-1/CD28-CART and CD19-CART combined with both anti–PD-1 antibody and anti-CD28 antibody, which were higher than that for CD19-CART and CD19-CART plus single antibody groups (Fig. 1E), suggesting that enhanced in vitro potency of CD19-PD1/CD28-CAR T is attributed to combination of cell intrinsic PD-1 blockade and CD28 triggering because PD1/CD28 chimera not only counteracts inhibitory costimulation, but also activates positive costimulatory pathways.

We also observed a rise in division number and absolute cell number of CD19-PD-1/CD28-CAR T cells (Supplementary Fig. S2A and S2B). Upregulation of NKG2D and CD27, a costimulatory receptor involved in the generation of T-cell memory, may contribute to the enhanced cytolytic capacity of CAR T cells (27, 28). Both of these molecules were shown to be upregulated in CD19-PD-1/CD28-CAR T cells, but not in CD19-CAR T cells and CD19-CAR T cells plus anti–PD-1 antibody, when challenged with mitomycin C–treated Raji-PD-L1 cells (Supplementary Fig. S2C). Notably, CD19-CAR T cells exhibited a senescence immunophenotype, as evidenced by high expression of KLRG1 (ref. 29; Supplementary Fig. S2D). In addition, we found that CD19-PD-1/CD28-CART cocultured with Raji-PD-L1 cells at E/T ratios of 5 and 10 for 72 and 96 hours, respectively, showed much lower expression levels of TIM3/LAG3 double-positive cells than CD19-CART (Supplementary Fig. S2E), indicating that PD-1/CD28 switch-receptor reduced the exhaustion of our CD19 CART incorporating the 4-1BB costimulatory domain. To detect whether the exhaustion phenotype correlated with the expression of CD69, a costimulatory molecule for T-cell activation and proliferation (30), we quantitated the percentages of the CD69⁺LAG-3⁻/Tim-3⁻ and CD69⁻LAG-3⁺/Tim-3⁺ CAR T cells by flow cytometric analysis. Results showed that the exhaustion phenotype negatively correlated with CD69 expression. CD19-PD-1/CD28-CART represented a high portion of CD69⁺LAG-3⁻/Tim-3⁻ T-cell subtypes (Supplementary Fig. S2F). We also examined the differentiation phenotype of two kinds of CAR T cells that were cocultured with Raji-PD-L1 for 24, 48, 72, and 96 hours. Flow cytometric analysis demonstrated a less differentiated phenotype in CD19-PD-1/CD28-CART, but terminally differentiated T-cell subsets in CD19-CAR T cells (Supplementary Fig. S2G).

We next compared the in vivo antilymphoma efficacy of CAR T-cell therapies by inoculating NSG mice with Raji-PD-L1 cells intravenously. We observed that a single treatment with anti–PD-1 antibody failed to control tumor progression. In contrast, CD19-CAR T cells alone and in combination with anti–PD-1 antibody significantly inhibited tumor growth. We confirmed that the combinational therapy, but not CAR T cells, resulted in significant suppression of tumor progression by IVIS imaging (Fig. 2A). Importantly, CD19-PD-1/CD28-CAR T cells showed significantly higher antitumor function in PD-L1⁺ tumor–bearing mice as dynamically determined by IVIS imaging and spleen size at the endpoints, compared with an infusion of CD19-CAR T cells or CD19-CART combined with anti–PD-1 antibody (Fig. 2A and B). The minimum tumor burden quantified by real-time photon emission was also confirmed in the mice treated with CD19-PD-1/CD28-CAR T cells (Fig. 2C). We further evaluated the biodistribution of CD3⁺CAR⁺ CAR T cells and CD19⁺ tumor cells in the spleen of NSG mice. The results showed that the tumor cells could be detected in the spleen of mice with xenograft of Raji-Luc cells. Approximately 23.2% and 84.3% CD19⁺ lymphoma cells were quantitated in the spleen tissue, from each of the two control mice. In comparison, only 0.3%, 4.0%, and 0.3% CD19⁺ cells were detected in the spleen from each of the three mice treated with CD19-CART cells (Supplementary Fig. S3). Moreover, we also confirmed the same level of TIM3 and LAG3 expression in CAR T cells that were obtained from bone marrow of mice treated with CD19-PD-1/CD28-CART, CD19-CART, or CD19-CART plus anti–PD-1 antibody, respectively (Supplementary Fig. S2H). However, CD19-PD-1/CD28-CAR T cells expressed higher levels of T-cell activations marker, CD69, than does other groups (Supplementary Fig. S2I).

At the data cutoff (20 November 2019), 18 patients were enrolled in the study. One patient became ineligible due to central nervous system lymphoma before leukapheresis. Seventeen patients received conditioning chemotherapy and subsequent CD19-PD-1/CD28-CAR T cells at doses ranging from 0.5 to 4.0 × 10⁶ CAR⁺ T cells/kg, but did not receive any bridging therapy. Median age was 55 years (range, 24–70 years). Our cohort consisted of 13 DLBCL cases, predominantly non-germinal center B-cell (GCB) subtype (76.9%; 10/13) compared with GCB subtype (23.1%; 3/13) by Hans algorithm (31). Four (30.8%) of 13 DLBCL cases had double expressor B-cell lymphoma and five (38.5%) had high-grade B-cell lymphoma, including one (7.6%) with triple expressor and four (30.8%) with high-grade B-cell lymphoma otherwise specified. Two patients had MCL and 2 had tFL. Overall PD-L1 positivity was 15 of 17 (88.2%) and the mean (range) of PD-L1 scores was 49.4 (0–165; Table 1).

During the first 4 weeks, all AEs were graded and reported. Seventeen (100%) patients had one or more AEs, with maximum grade 3 and 4 events observed in seven (41.18%) and six (35.29%) patients, respectively. There were no grade 5 events (Table 2). The most common AEs of grade ≥3 were leukopenia (88.23%), thrombocytopenia (35.29%), anemia (23.53%), and pyrexia (23.53%). Fifteen patients (88.24%) had mild symptoms of CRS, such as pyrexia, tachycardia, and dizziness. However, seven patients (41.17%) had hypotension and one patient (5.88%) had hypoxia, which responded to supportive care, including fluid and oxygen supplementation (Table 3). No patient received tocilizumab or glucocorticoids. One patient, patient 17, died from lymphoma progression on day 30 after infusion. She did not experience any severe CRS or CAR T–related neurotoxicity.

Ten of 17 (58.8%) patients achieved an objective response evaluated at 3 months, with seven of 17 (41.2%) patients achieving CR, including one of two patients with MCL, and one of two with tFL (Fig. 3A). Figure 3B shows PET-CT images of the seven patients who achieved CR evaluated on day 90. Patient 14 with double-expressor DLBCL achieved CR after CAR T-cell therapy; however, progressive disease (PD) was noted at 5-month and proven as CD19⁺ relapse by biopsy. Currently, nine of 10 patients with CR or PR have ongoing responses. Patient 9 had advanced triple-expression lymphoma with International Prognostic Index (IPI) score of 3 and was refractory to four lines of immunochemotherapy followed by ibrutinib. He received 1 × 10⁶ CAR⁺ T cells/kg and achieved persistent CR. In post hoc analyses, the estimated proportion of patients with progression-free survival (PFS) at 18 months was 80% [95% confidence interval (CI), 0.41–0.95] among those with CR or PR at 3 months (Fig. 3C). The median OS was not reached, and median PFS survival was 5 months (Fig. 3D).

The tumor burden has previously been associated with long-term survival of patients with acute lymphoblastic leukemia (ALL) treated with CD19-CAR T cells (32). In this study, the impact of MTV measured at baseline on prognosis was analyzed. In the whole population, the mean MTV was 256 cm³ (median, 148 cm³; 25th–75th percentiles, 91–408 cm³). We found a significantly longer PFS and OS in patients with lower MTV compared to those with higher MTV. ROC analysis demonstrated that the optimal MTV cut-off value was 291 cm³ for predicting both PFS and OS (Fig. 4A and B). AUC ROC was 0.83 (95% CI, 0.61–1.00) for PFS and 0.78 (95% CI, 0.52–1.00) for OS. The 291 cm³ cut-off value had a sensitivity and a specificity of 77.8% and 100% for PFS and 75.0% and 88.9% for OS, respectively (Fig. 4C and D). Kaplan–Meier curves demonstrated that MTV using this cutoff, was a strong prognostic factor of both PFS and OS. A high MTV (MTV ≥ 291 cm³) was predictive of both PFS and OS at univariate level, and also at multivariate level when tumor burden was adjusted for baseline IPI (Table 1).

Because PD-1/CD28 chimeric switch-receptor has been demonstrated to engage PD-L1 and transmit an activating signal instead of the inhibitory signal, leading to enhanced antitumor effects (23), we attempted to identify the relationship between PD-L1 tumor expression and the severity of AEs, as well as long-term outcomes. There was no statistically significant correlation between PD-L1 expression and severity of AEs. However, an H-score of ≥10 was found to be predictive of OS at univariate level and at multivariate level according to baseline IPI (P = 0.019; Fig. 4E). Sensitivity and specificity of the H-score cutoff were 88.9% and 62.5% for OS, respectively (Fig. 4F).

We assessed the serum levels of six cytokines within 30 days after CAR T treatment. The fold increase over the preinfusion treatment level was calculated for each cytokine in each patient. About half of the cases experienced slightly increased levels of IL2, IL4, IL6, and TNFα, whereas no significant elevation of IL10 and IFNγ levels was observed (Fig. 5A). Similar results were observed for serum C-reactive protein (CRP) levels (Fig. 5B). Patient 14 had significantly higher levels of IL10 and IFNγ at 12 days after CAR T infusion with 2 × 10⁶ cells/kg. However, he only experienced a grade 1 CRS, which resolved spontaneously. The phenotypic characterization of the patients’ apheresed PBMCs and CD19-PD-1/CD28-CAR T cells was also analyzed on the basis of CCR7 and CD45RA expression. CD19-PD-1/CD28-CAR T-cell product, compared with the apheresed T cells, had higher proportions of cells with naïve and Tcm phenotypes, and lower proportions of cells with Tem and terminally differentiated effector T cells (TEMRA) phenotypes, suggesting a less differentiated phenotype (Fig. 5C). The presence of CAR T cells in patients’ peripheral blood was monitored by qPCR. Peak expansion of CAR T cells occurred 2 weeks after cell therapy, and the peak CAR T-cell levels of patients who achieved CR were higher than that of patients with PR and PD (Fig. 6A). CAR⁺ cells were still observed in five of 10 (50%) patients with response at 6 months (Fig. 6A). Slightly decreased serum globulin level was observed during first 90 days after infusion, compared with the pretreatment levels, but no statistical difference was seen. No difference was observed among groups of patients with CR, PR, and PD (Fig. 6B). Six months after treatment with CAR T cells, circulating normal B cells were followed and measured in a subset of patients with CR or PR by flow cytometry. Ongoing B-cell aplasia was seen in three patients, but recovered well in patients with CR within 16 months (Supplementary Table S2).

To our knowledge, this is the first multicenter study of treatment with CD19-CAR T cells expressing PD-1/CD28 chimeric switch-receptor in R/R large B-cell lymphoma. The data demonstrated a favorable safety profile and durable clinical responses. All 17 patients before enrollment in this study had received heavy immunochemotherapies and new agents, such as ibrutinib and lenalidomide, thus were challenging to treat (4). Importantly, of the 17 patients, 15 (88.2%) were PD-L1⁺ in tumor cells.

PD-L1 has been shown to be heterogeneously expressed in 11%–61.1% of DLBCL samples with PD-L1⁺ cutoff >5% or >10%, and PD-L1 expression in lymphoma cells has been associated with poor OS (33, 34). PD-L1 binds to the inhibitory receptor PD-1 on T cells, resulting T-cell exhaustion (22). A recent meta-analysis showed that patients with PD-L1 overexpression in R/R lymphoma benefited more from anti–PD-1 therapy (35). In large B-cell lymphoma, PD-L1 expression in the tumor cells has been shown to be a potential mechanism of resistance to CD19-CAR T-cell therapy (13, 36). PD-1/CD28 switch-receptor, an engineered receptor that will transmit an activating signal instead of the inhibitory signal by binding of PD-L1 (37), has been documented to boost CAR T-cell antitumor activity with CAR T cells infiltration, decrease in susceptibility to tumor-induced hypofunction, and attenuation of inhibitory receptor expression (23). Here, we have demonstrated enhanced activation and efficacy of CD19-PD-1/CD28-CART in a preclinical model of PD-L1–overexpressing Raji cells (Fig. 1). Consistent with earlier results (23), PD-1/CD28 switch-receptor augmented CD19-CAR T cells antitumor function to a greater degree than anti–PD-1 antibody (Fig. 2). In this clinical study, 10 of 17 (58.8%) patients achieved an objective response within 3 months of the CAR T treatment, with seven of 17 (41.2%) achieving a CR (Fig. 3). The response rate of the DLBCL was seven of 13. Using an H-score methodology, we found that patients with H-score ≥10 had a trend toward longer OS (Fig. 4). Taken together, our data suggest that CD19-PD-1/CD28-CART has the potential to improve outcomes in patients with PD-L1⁺ large B-cell lymphoma who had a R/R disease.

The significance of robust in vivo expansion and persistence of CAR T cells is still controversial in the literature. It has been associated with remission in B-cell malignancies in some reports, but has shown no significant prognostic effect in the others (32, 38, 39). In this study, we observed that patients with CR displayed a higher degree of CAR T-cell expansion and extended duration of CAR T-cell persistence in vivo than those patients with PD. However, no statistical difference of absolute number of CAR⁺ cells in blood between these two groups of patients was observed (Fig. 6). The observation suggests that duration of CAR T cells in vivo plays a more important role in achieving superior clinical efficacy than copy-number change. Previously, the less differentiated T-cell subsets, such as T memory stem cells, had been shown in mouse models to have superior proliferative capacity, prolonged persistence, and higher antitumor activity than the more differentiated T-cell subsets (40). We also observed that CD19-PD-1/CD28-CART acquired a less differentiated phenotype, while CD19-CART, when challenged with Raji-PD-L1 cells in vitro, had a phenotype in-line with more differentiated T cells, including effector memory T cells and terminally differentiated effector T cells (Fig. 5).

The prognostic role of tumor burden has been recently highlighted by Park and colleagues (32), in a phase I trial involving 53 patients with ALL treated with CD19-CAR T cells. However, such a relationship has yet to be explored in B-cell lymphoma. In our study, the baseline MTV appears as a new tool to better predict response to CD19-PD-1/CD28-CART therapy (Fig. 4). Gauthier and colleagues (41) also showed that for R/R DLBCL, increased tumor burden, measured by the sum of the products of the diameters, was adversely associated with PFS and OS. On the contrary, a recent report has provided evidence indicating that the baseline MTV did not differ significantly between patients with response and those without response (41). Difference in prognostic value of disease burden in patients with lymphoma observed among investigators might result from the small number of patients analyzed. Thus, studies in a large number of patients are necessary to confirm this finding.

In our study, treatment with CD19-PD-1/CD28-CAR T cells was well tolerated. Toxicity was mild, especially with CRS (Table 2). One possible reason is that CD19-PD-1/CD28-CART generated less IL6, a proinflammatory cytokine that plays a key role in CRS pathophysiology (42, 43), than CD19-CART when exposed to PD-L1⁺ Raji cells. This finding is also compatible with the feature of clinical course of the patients revealing lower levels of serum cytokines, including IL6 and IFNγ. Another possible mechanism is that PD-1/CD28 chimeric switch-receptor may offer a potential way to deliver CAR T cells with potent activation launched specifically within the immunosuppressive tumor microenvironment (23). The patients showed low serum levels of cytokine (Fig. 4), whereas CD-19-PD-1/CD28-CART produced higher levels of cytokine compared with CD19-CART in vitro (Fig. 1). These data may suggest persistent CAR T-cell function within tumor microenvironment.

In summary, this first-in-human study demonstrates the tolerability, safety, and encouraging efficacy of CD19-PD-1/CD28-CART in PD-L1⁺ large B-cell lymphoma. While promising, the data are limited because of the small number of patients enrolled in this study. Further study through a large-scale randomized controlled, multicenter clinical trial should be conducted.

No disclosures were reported.

H. Liu: Formal analysis, project administration. W. Lei: Project administration. C. Zhang: Software. C. Yang: Project administration. J. Wei: Project administration. Q. Guo: Project administration. X. Guo: Project administration. Z. Chen: Project administration. Y. Lu: Project administration. K.H. Young: Supervision, investigation, writing-review and editing. Z. Lu: Supervision, writing-original draft, writing-review and editing. W. Qian: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, methodology, writing-original draft, writing-review and editing.

The research was supported by National Natural Science Foundation of China (grant Nos. 81830006, 81670178, and 81800188) and funds from Science Technology Department of Zhejiang Province (grant No. 2018C03016-1).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.