Potent Anti-leukemia Activities of Chimeric Antigen Receptor–Modified T Cells against CD19 in Chinese Patients with Relapsed/Refractory Acute Lymphocytic Leukemia

Patients with relapsed/refractory acute lymphocytic leukemia (R/R ALL) have a very poor prognosis under current therapeutic modalities (1, 2). In the past two decades, advances in chemotherapy drugs, targeted agents, and allogeneic hematopoietic stem cell transplantation (allo-HSCT) techniques have greatly improved the clinical outcomes for these patients. However, limited progresses have been made in reshaping the overall complexion of ALL treatment considering the crucial fact that the complete remission (CR) rate is relatively low, with a median overall survival (OS) time of 3–6 months and a 5-year OS rate of less than 10% for patients with primary refractory disease, a short duration of first remission (<12 months), relapse after allo-HSCT, or disease progression after multiple courses of therapy (1, 3). Allo-HSCT has been recognized as the only curative option for patients with R/R ALL (4), and salvage allo-HSCT is clinically available for these patients. However, according to the data from recent clinical trials, the high relapse rate and poor OS in these patients remain great challenges (5, 6). Subsequently, reinduction remission is an urgent issue to address to improve the clinical outcomes of patients with R/R ALL.

Chimeric antigen receptor–modified T cells against CD19 (CART19) have shown promise as a novel therapy for R/R ALL patients in clinical trials (7–12). High antileukemia efficacies of CART19s have been consistently reported by independent trials at different institutions (13–16). Among these trials, CART19s prepared by each institution differ in several respects, including CAR design, T-cell activation, and transduction methods. Costimulatory molecules such as CD28, 4-1BB, CD134, CD2, CD27, and ICOS are the integral CAR structural components, particularly CD28 and 4-1BB (17, 18). The Memorial Sloan Kettering Cancer Center (MSKCC, New York, NY) team used a CAR construct containing the costimulatory domain CD28/CD3-ζ via γ-retrovirus transfer, whereas investigators from CHOP/UPenn (University of Pennsylvania, Philadelphia, PA) used a CAR construct containing the costimulatory domain 4-1BB/CD3-ζ via lentivirus transfer (10, 12). According to current clinical trials, CD28/CD3-ζ–costimulated CART19s initially demonstrate potent effector functions, but the in vivo persistence of these cells seems inferior to that of 4-1BB/CD3-ζ–costimulated CART19s (13, 14).

Influence of CART19 doses, lymphodepleting chemotherapy regimens, and patient populations are major parameters evaluated in the clinical protocols (9–13). A broad range of CART19 doses, from 1 × 10⁵/kg to 1 × 10⁸/kg, were infused into patients with R/R ALL (13). Reports have suggested a correlation between the CART19 dose and the incidence as well as the severity of cytokine release syndrome (CRS; ref. 13). However, whether a higher dose is required for better efficacies remains uncertain. Updated lymphodepleting chemotherapies have been administered prior to CART19 infusion, including no chemotherapy, cyclophosphamide alone, and fludarabine combined with cyclophosphamide (13, 14); however, a well-recognized lymphodepleting regimen has not been acquired. By now, most clinical trials have been performed in Caucasians, Euro-Americans, Hispanic-Americans, and Asian-Americans (13, 14), and data from the systematic studies of Asian natives remain limited. To this end, validation trials are needed by confirming the safety and efficacy profiles of CART19 therapy in native Asian populations.

In the current study, we enrolled 15 consecutive patients with R/R ALL and administered individual lymphodepleting chemotherapy with 4-1BB/CD3-ζ–costimulated CART19s. We evaluated the efficacy and safety profiles to pursue an optimal CART-based therapeutic strategy in these Chinese patients.

This clinical trial was designed to assess the safety and efficacy of infusing autologous T cells modified to express the CD19-specific CAR/4-1BB/CD3-ζ into Chinese patients with R/R ALL (Chictr.org number, ChiCTR-OCC-15007008; Fig. 1A). The inclusion criteria were as follows: (i) age less than 60 years; (ii) relapsed or refractory CD19⁺ ALL; (iii) relapsed allo-HSCT without the evidence of graft versus host disease (GVHD) and not requiring immunosuppression therapy; and (iv) measurable disease and adequate performance status and organ function. Patients with central nervous system leukemia (CNSL) were ineligible. The protocol was approved by the First Affiliated Hospital, School of Medicine, Zhejiang University Institutional Review Board. All patients provided written, informed consent.

The single-chain fragment variable (scFv) sequence specific for CD19 was derived from Clone FMC63 (19). The 4-1BB costimulatory domain and CD3ζ signaling domain were generated as described previously (20–22). CART19-4-1BB vectors harboring anti-CD19 scFv and the human 4-1BB and CD3ζ signaling domains were cloned into a lentiviral backbone as described previously (22). Lentivirus was produced by transfecting 293T cells with CAR lentiviral vectors and viral packaging plasmids which were frozen at −80°C and thawed immediately before transduction. The lentivirus supernatant was harvested. CD3⁺ T cells were isolated and activated as described previously (23). The cells were then cultured in X-VIVO 15 medium (Lonza) containing 100 U/mL IL2 and transduced with lentivirus supernatant at high multiplicity of infection (MOI) from 5:1 to 10:1 within 24–48 hours. The CAR-transduced T cells were cultured for 11 days. Three days before administration, fresh culture media were replaced. After that, no manipulation was conducted to the cells until transportation for infusion. The transduction efficiency was evaluated by flow cytometry on day 5–7 after lentivirus transduction. The following anti-human antibodies were used: anti-hCD45 APC (BD Biosciences), anti-hCD3 FITC (BD Biosciences), biotin-labeled goat-anti-mouse IgG specific for F(ab′)2 fragment (Jackson Immuno-Research, cat # 115-065-072) and PE streptavidin (BD Biosciences). Data acquisition was performed using a CytoFLEX flow cytometer (Beckman Coulter).

Prior to CART19 infusion, FACS analysis of transduction efficiency and in vitro cytotoxicity assays of CART19s were performed for each patient as described in the Supplementary Materials. In addition, CART19 cultures were checked twice for possible contaminations by fungus, bacteria, mycoplasma, chlamydia, and endotoxin.

Peripheral blood mononuclear cells (PBMC) were obtained from patients by leukapheresis for CART19 preparation on day −11, and the first day of CART19s infusion was set as study day 0 (Fig. 1B). Patients were given a conditioning treatment for lymphodepletion. Fludarabine- and cyclophosphamide-based conditioning treatment varied according to the tumor burden in the bone marrow and peripheral blood (Fig. 1C). CART19s were transfused directly to patients in escalating doses over a period of 3 consecutive days without any premedication. Each day, CART19s were transported to hospital, washed, counted, checked for viability, and then prepared for administration to patients, who were then observed closely for at least 2 hours. CRS was graded according to a revised grading system (24). Other toxicities during and after therapy were assessed according to the NIH Common Terminology Criteria for Adverse Events Version 4.0 (http://ctep.cancer.gov/). Therapy responses were assessed by flow cytometry and morphologic analysis. When possible, patients were assessed by chimeric gene expression levels. The response type was defined as minimal residual disease (MRD) negative, complete response, complete response with incomplete count recovery, stable disease, and progressive disease as described in the Supplementary Materials.

Serial bone marrow and peripheral blood samples after CART19 infusion were collected in K₂EDTA BD vacutainer tubes (BD Biosciences). The persistence of CART19s from fresh peripheral blood and bone marrow in patients was determined by FACS. Circulating CART19 numbers per microliter were calculated on the basis of measured absolute CD3⁺ T lymphocyte counts. Simultaneously, CAR DNA copies were evaluated as another method of determining CART19s expansion and persistence. Genomic DNA was extracted using a QIAamp DNA Blood Mini Kit (Qiagen) from cryopreserved peripheral blood and bone marrow. CAR DNA copies were assessed by quantitative real-time PCR as described in the Supplementary Material.

The levels of cytokines IFNγ, TNFα, IL4, IL6, IL10, and IL17 in serum and CSF were measured in a multiplex format according to the manufacturer's instructions as described in the Supplementary Material.

Comparisons of continuous variables and risk factors that may influence variations in grade 3 CRS development were compared using the Mann–Whitney U test for two groups. Fisher exact test was used to evaluate the influence of categorical variables on grade 3 CRS between 2 groups. Correlations were calculated using a rank-based Spearman test. Overall survival (OS) and leukemia-free survival (LFS) probabilities were determined by the Kaplan–Meier method using all enrolled patients to determine OS and those with MRD-negative responses for LFS. All quoted P values are two sided, and P values less than 0.05 were considered statistically significant.

A total of 22 patients with pathologically confirmed CD19⁺ ALL between July 2015 and April 2016 were recruited for this trial. Six patients were not eligible for the CART19 clinical trial owing to abnormal liver function (2 patients), heart failure (1 patient), renal dysfunction (1 patient), and lung infection (2 patients). One patient developed a severe infection during CART19 generation. These 7 patients were excluded from receiving lymphodepletion chemotherapy and CART19s (Fig. 1A). Thus, 15 patients with CD19⁺ ALL aged between 7 and 57 years and with a median age of 32 years were enrolled in this trial (Supplementary Table S1). Of these patients, 4 had Ph⁺ ALL (ABL T315I mutation in 2 patients), 5 had prior allo-HSCT, and 8 had 2 or more times of relapse before receiving CART19 therapy. One patient (patient no. 3) had never achieved MRD-negative remission despite 5 courses of intensive chemotherapy and TKI-targeted therapy including the third-generation agent ponatinib. Of the 15 patients, 14 patients had detectable leukemic cells in bone marrow at the time of CART19 infusion, and 1 had extramedullary relapse in the testes while the bone marrow MRD status was negative. Four patients had previous CNSL, and 2 had previous extramedullary relapse in the testes. The median leukemia burden was 63.5% (ranging from 3% to 83%) of marrow blasts.

After 11 days of culture, cells were released for infusion. The median transduction efficiency of the final products was 33%, with a range of 5% to 50% (Table 1). In vitro cytotoxicity assays of CART19s showed robust CART19s activation and prominent antileukemia activities of CD19⁺ leukemia cells (Supplementary Fig. S1; Supplementary Fig. S2; Supplementary Table S2). A routine screening for fungus, bacteria, mycoplasma, chlamydia, and endotoxin was negative in CART19 cultures prior to infusion. All 15 patients received CART19s infusion at doses of 1.1 × 10⁶/kg to 9.8 × 10⁶/kg, with a median dose of 3.7 × 10⁶/kg, similar to the CART19 doses in previous studies (13, 14).

Therapy responses were evaluated in 13 patients on day 7–10 and 12 patients on day 30 after CART19s infusion. Excluding patient no. 14, who had a testicular relapse, of the 13 patients with bone marrow results, 6 were MRD-negative and displayed a complete response with incomplete count recovery, 5 obtained a complete response with incomplete count recovery but were MRD-positive, 1 patient (patient no. 10) had a stable disease, and 1 patient (patient no. 12) had a progressive disease 7–10 days after CART19 infusion. On day 30 after CART19 infusion, the 12 patients remained MRD-negative (Table 1). Patient no. 13, who had an extramedullary leukemia relapse in the testes, also obtained CR 1 month after CART19s infusion, which was confirmed by ultrasonography and pathology examination (Supplementary Fig. S3). These results demonstrate that CART19s induce CR rapidly and effectively in R/R ALL patients both in bone marrow and in extramedullary sites.

Distribution profiles of CART19s in peripheral blood were assessed by the means of FACS and qPCR assays at serial time points before and after CART19 infusion, respectively (Fig. 2A; Supplementary Fig. S4A). Of the 14 evaluable patients (patient no. 4 was not included), the median peak CART counts were 342/μL (95% CI, 140–532) and 96/μL (95% CI, 61.5–132.8) in the grade 3 CRS group and in the non-CRS or grade 1 or 2 CRS group, respectively (P = 0.002). The median peak CART DNA copies were 9.9 × 10⁵/μg (95% CI, 61.5×10⁶–132.8×10⁶) and 2.2 × 10⁵/μg (95% CI 1.5×10⁵–4.8×10⁵) in the grade 3 CRS group and in the non-CRS or grade 1 or 2 CRS group, respectively (P = 0.002 and 0.0047). During the 7–10 days after CART19s infusion, the median CAR DNA copies in bone marrow were higher in the grade 3 CRS group than in the non-CRS or grade 1 or 2 CRS group (2.5 × 10⁶/μg; 95% CI, 8.4 × 10⁵–8.3 × 10⁶ vs. 1.3 × 10⁵/μg; 95% CI, 5.9 × 10⁴–4.1 × 10⁵; P = 0.002), while the peak CART/CD3⁺ T-cell percentages in bone marrow were not significantly different between the two groups (Fig. 2B). In 4 patients, the CAR DNA copies were higher in bone marrow than in peripheral blood (Supplementary Fig. S4B). Moreover, CART19s were detectable in the blood and marrow for up to 7 months (Fig. 2A).

Paired measurements of CAR DNA copies in bone marrow and peripheral blood on day 7 to 10 after CART19s infusion and the percentage of leukemia cells after fludarabine and cyclophosphamide (FC) chemotherapy showed strong correlations with the CAR DNA copy count (CAR DNA copies in bone marrow vs. peripheral blood, Spearman r = 0.956, P < 0.001; CAR DNA copies in bone marrow vs. percentages of leukemia cells after FC chemotherapy, Spearman r = 0.699, P = 0.011; CAR DNA copies in peripheral blood vs. percentages of leukemia cells after FC chemotherapy, Spearman r = 0.755, P = 0.005; Supplementary Fig. S4C). Paired measurements of peak CAR DNA copies in bone marrow or peak CART19 counts after CART19 infusion and CART19 dose showed no correlations (peak CAR DNA copies in bone marrow vs. CART19 dose, Spearman r = 0.248, P = 0.438; peak CART19 count in peripheral blood vs. CART19 dose, Spearman r = 0.172, P = 0.594; Supplementary Fig. S4C).

Of the 12 patients with a median follow-up of 142 days (ranging from 30 to 281 days), 6 patients maintained CR. Six patients with CR at 1 month subsequently relapsed: 2 with CD19 (+) blasts (patients no. 1 and 10, Supplementary Fig. S5A), 2 with both CD19(−) and CD19(+) blasts (patients no. 1 and 3, Supplementary Fig. S5B), and 2 with CNS relapse while maintaining CR in bone marrow (patients nos. 5 and 11). Two patients with only CNS relapse received therapeutic intrathecal injection with cytarabine and methotrexate before achieving CR again. Patient nos. 2 and 3 underwent salvage haploidentical allo-HSCT. Patient nos. 7 and 13 were MRD-positive but still maintained CR in bone marrow and underwent a second allo-HSCT. All 4 patients maintained CR after allo-HSCT (Fig. 3A).

Of the enrolled 15 patients, 6 patients (40%) died: 3 (20%) owing to severe infection (2 died during the pancytopenia period, and 1 died during the follow-up period) and 3 (20%) owing to disease progression or relapse. At 150 days, the OS rate was 65.5%. The LFS rate was 37.8%. The cumulative incidence of relapse in bone marrow or the CNS was 54.5%, whereas the cumulative incidence of relapse in bone marrow only was 36.3% (Fig. 3B and C). The nonrelapse mortality (NRM) rate was 7.7% (Supplementary Fig. S6). The median interval of LFS was 143 days (95% CI, 24–262 days).

In this trial, 10 of 15 patients (66.7%) experienced CRS. Of these patients, three had grade 1 CRS, 1 had grade 2 CRS, and 6 had grade 3 CRS, as shown in Table 1 and Fig. 4A. No grade 4 or 5 CRS occurred in this trial. CRS mostly occurred within a median of 2.5 days after infusion (range 1–10 days) and lasted for a mean of 5.9 days (range 2–9 days). The syndrome was fully reversible in all patients and was well managed with supportive care alone (n = 4), supportive care plus the anti-IL6 receptor mAb tocilizumab (n = 3), supportive care plus tocilizumab and corticosteroids (n = 2), and supportive care plus corticosteroids (n = 1; Table 1).

All 10 patients experienced pyrexia. Patients with grade 3 CRS had pyrexia of higher temperatures that lasted for longer periods than those with grade 1 or 2 CRS (Fig. 4B).

Serum levels of the cytokines IL2, IL6, IL10, IFNγ, IL4, TNFα, and IL17 as well as CRP, D-dimer, and ferritin were evaluated. Serum levels of IL6, IFNγ, IL10, CRP, D-dimer, and ferritin were elevated during CRS (Supplementary Fig. S7A), and the peak levels differed significantly between the grade 3 CRS group and the non-CRS or grade 1 or 2 CRS groups (Fig. 4C; Supplementary Fig. S7B). IL6 is one of the most important biomarkers for CRS. Paired peak serum levels of IL6, CRP, ferritin, and D-dimer showed strong correlations (IL6 vs. CRP, Spearman r = 0.617, P = 0.014; IL6 vs. ferritin, Spearman r = 0.574, P = 0.028; IL6 vs. D-dimer, Spearman r = 0.789, P < 0.01; Supplementary Fig. S7C). Serum IL2, IL4, TNFα, and IL17 levels were not associated with CRS (data not shown).

Risk factors associated with CRS were analyzed. Univariate analysis showed that the MRD after the conditioning regimen and the number of previous relapses were two factors associated with a high risk of grade 3 CRS (P = 0.026 and 0.036, respectively; Table 2). That is, multiply relapse and the high tumor burden after the conditioning regimen were 2 factors associated with grade 3 CRS. Other risk factors including age, gender, previous therapy (chemotherapy or allo-HSCT), CART19 dose, and MRD before the FC-conditioning regimen were not associated with the risk of grade 3 CRS (Table 2). Multivariate analysis was not performed because of the small number of patients. Our results imply that the number of previous relapse and cases of MRD after the conditioning regimen might be novel predictors for grade 3 CRS.

Reversible neurotoxicities were observed in 5 patients (patients 2, 5, 8, 13, and 15). Patient 2 developed confusion and gait disturbances 7 days after CART infusion and dramatically improved after tocilizumab (4 mg/kg) therapy on day 9 after CART infusion. Patient 5 experienced headache and vomiting, recurrent right-sided facial and limb paresis, blurred vision and defective visual field with decreased myodynamia, high blood pressure, papilledema, and positive Babinski and Kernig signs since day 3, as we reported previously (22). The symptoms and signs were relieved by methylprednisolone and mannitol treatment. Patient no. 8 developed confusion followed by amnesia during CRS and recovered quickly once CRS was controlled. Patient no. 13 developed epilepsy once after 2 doses of tocilizumab (8 mg/kg) therapy and CRS recovery. Patient no. 15 developed transient muscle clonus in limbs at the time of CRS recovery after 2 doses of tocilizumab (8 mg/kg) therapy.

Three patients (nos. 2, 8, and 15) did not undergo cerebrospinal fluid (CSF) examination when neurotoxic symptoms occurred owing to pancytopenia. In patient no. 5, cytokine levels were much higher in CSF than in serum, with a serum IFNγ concentration of 152 pg/mL versus a CSF concentration of 2,977 pg/mL, and a serum IL6 concentration of 46 pg/mL versus a CSF concentration of 8,475 pg/mL. Furthermore, qPCR analysis showed 3032,265 CAR copies/μg DNA in CSF versus 988,747 CAR copies/μg DNA in peripheral blood (22). In patient no. 14, no white blood cells (WBC) were found in the CSF, but the CSF IL6 level (935 pg/mL) was higher than that in serum (132 pg/mL).

The occurrence, management, and outcomes in other specific organs are listed in Supplementary Table S3.

In the current study, we utilized a CART19 strategy to treat Chinese R/R ALL patients. The clinical output of CART19 therapy in enrolled patients was encouraging. This trial demonstrated that CART19 therapy was a feasible approach for the majority of enrolled Chinese patients with R/R ALL, with an induction of MRD-negative CR at 1 month. As progressing refractory disease is inevitable in patients not eligible for allo-HSCT, our data revealed that these patients would have a new opportunity to achieve CR quickly by choosing CART19 therapy as a transitional modality, increasing the opportunities for further therapies including, but not limited to, allo-HSCT. This study also established a comprehensive evaluation protocol covering multiple systems to monitor CART19-associated toxicities.

Patients enrolled in this study obtained a high CR rate in bone marrow which confirmed a favorable outcome for CART19 therapy in R/R ALL. Our results also demonstrated the prominent therapeutic efficacy of CART9s in Chinese patients with R/R ALL. Recently, evidences for the efficacy of CART19 therapy in other patient populations for R/R ALL have been reported. Early in 2013, Grupp and colleagues reported that 2 children with R/R ALL achieved CR after CART19 infusion (25). Brentjens and colleagues reported a 100% CR rate in 5 patients with R/R ALL and an 88% CR rate in 16 adult patients with R/R ALL (10, 26). Then, in 2014, a 90% CR rate in 30 pediatric and adult patients with R/R ALL was reported by Grupp and colleagues (12). Other groups have reported similar results (13, 14). Lee and colleagues reported a 66.7% CR rate in an NCI intent-to-treat analysis of children and young adults with ALL (9). Most of patients in these reports were not native Asian. In the current study of Chinese patients, the CR rate was comparable or even superior to those in these previous studies. CART19 expansion and persistence was another indicator for evaluating efficacy. In a previous study, a shorter persistence duration of 1 to 6 months was reported following CART19s containing CD28/CD3-ζ infusion compared with that following CART19s containing 4-1BB/CD3-ζ infusion in adults with B-ALL, as assessed by FACS and qPCR (13, 14). Consistent with these results, robust expansion and long-term persistence of 4-1BB/CD3-ζ costimulated CART19s were observed in our trial with Chinese patients. Moreover, we observed the efficacy of CART19s in extramedullary relapse in the CNS and testes. This study is the first to report antileukemia activities of CART19s in testicular relapse. Trafficking of CART19s to several sites including the liver, lymph nodes, CNS, and bone marrow has been demonstrated previously (27). In this study, we noted that the CART19 count in peripheral blood was not significantly different from that in bone marrow. We also observed that CART19s migrated to tumor sites and exerted antitumor effects in bone marrow, CNS, and testes. These results presented that CART19s were transported to leukemia sites and eradicated leukemia cells effectively and quickly in Chinese patients with R/R ALL.

In this study, we applied an FC-based lymphodepletion strategy before CART19 infusion. The preinfusion chemotherapy facilitates CART19 engraftment by eliminating immune cells potentially competing for homeostatic cytokines (28). In addition, the chemotherapy exhibits direct antileukemia activities. However, a lymphodepletion regimen is not a standard option and has varied widely in different trials. Recent studies have demonstrated that lymphodepletion chemotherapy is associated with a better prognosis. Zhang and colleagues reported that the 6-month EFS for patients administered a lymphodepletion regimen before CART19 infusion was 94.6%, compared with 54.5% in patients without lymphodepletion (P < 0.001; ref. 29). Turtle and colleagues observed that the addition of Flu to Cy lymphodepletion improved the persistence of CARTs and EFS time (30).We exploited FC chemotherapy before CART19s infusion in this study after considering the following disease-related factors: (i) relapsed or refractory leukemia cells in patients with high tumor burden usually have a high proliferation profile, which leads to a high risk of hyperleukocytosis; (ii) preparation of CARTs is a time-consuming procedure, and the allocated time for CART generation is approximately 10 to 14 days; (iii) during the period of CART generation, patients are at a high risk of hyperleukocytosis, which may cause severe morbidity and mortality by inducing leukostasis and tumor lysis syndrome (31). Thus, the rationale for chemotherapy before CART19s in these patients is confirmed. We subsequently established a suitable FC-based regimen by considering the de facto tumor burden or leukemia cell proliferation activity in individual patients. In our study, no patient developed hyperleukocytosis-associated complications during CART19 preparation, despite the high tumor burden in most patients.

Serum levels of IL6, IFNγ, and IL10 were simultaneously increased during CRS, suggesting that these are causative cytokines. These results are consistent with those of previous studies (32–36). In addition, we observed that serum ferritin, CRP, and D-dimer levels correlated with the severity of CRS and declined in response to tocilizumab or corticosteroids, implying that serum CRP, D-dimer, and ferritin levels might be CRS biomarkers for CRS diagnosis and grade. Severe grade 4 or fatal grade 5 CRS has been reported in other clinical trials. None of the 15 patients in this trial had grade 4 or 5 CRS, which may largely be attributable to the patients' immunologic backgrounds, their overall status, and suitable intervention with tocilizumab or corticosteroid during the CART therapy. Previous reports have suggested that the severity of CRS may be associated with the disease burden at enrollment (before the lymphodepletion regimen; refs. 9, 10, 13). In our study, the tumor burden was evaluated at 2 time points: the time of enrollment and when the FC lymphodepletion regimen was completed. We observed that the tumor burdens at the time of enrollment were not associated with the risk of grade 3 CRS; however, tumor burdens at the end of the FC regimen were statistically associated with grade 3 CRS, indicating that tumor burden after FC regimen is a more precise risk factor associated with grade 3 CRS. Interestingly, the number of previous relapses was another possible factor associated with grade 3 CRS. Multiply relapsed ALL usually indicates a high risk of grade 3 CRS. Leukemia relapse indicates aberrant selection for and emergence of increasingly malignant clones during progression and therapy (37). Leukemia cells with multiple relapse experiences exhibit greater chemotherapy resistance and immune escape, and the eradication of these leukemia cells by indicated therapies is limited (38, 39). Consequently, CART19s should be more rigorously activated to target these cells and simultaneously result in a more severe CRS. Our data demonstrate that, in addition to the tumor burden, the biological features of leukemia cell were associated with severe CRS. These results may further the understanding of post-CART CRS pathogenesis and more data from clinical trials are warranted.

The disease relapse after CART19 therapy has been observed in previous studies (9, 13). During follow-up, in this limited-scale study, we observed that disease relapse remained the main obstacle for CART19s therapy. The following factors might be causative for relapse. (i) our patients had a high tumor burden, with a median leukemia burden of 63.5%; (ii) the enrolled patients were more refractory. Most patients had several experiences of relapse or primary refractory. Therefore, the genetic background and mutations in leukemia stem cells were unlikely to be eradicated completely; (iii) most patients had extramedullary relapses. Extramedullary leukemia can reside in a specialized stromal compartment that differs from the bone marrow microenvironment and facilitates tumor growth directly via paracrine secretion of growth factors and the provision of nutrients to permit escape from CART19 attack and bone marrow relapse. Various treatment strategies have been applied for relapse after CART19 therapy, but consensus on an effective option has not been obtained. Several studies have attempted CARTs infusion to enhance antileukemia effects in those who had no CART persistence or low levels of CARTs, but the efficacy was less adequate. In our study, we observed that haploidentical HSCT after CART19 therapy remarkably reduced the relapse rate and improved OS. Haploidentical HSCT appears to be a promising strategy with a theoretically high donor availability of almost 100%. On the basis of our previous results, haploidentical HSCT is expected to trigger a more potent graft versus leukemia effect compared with HLA-identical transplants (40). Results from our another study showed that haploidentical HSCT improved outcomes in patients with high-risk leukemia (41). In this study, 4 patients who had received CART19 therapy then underwent haploidentical HSCT, and no transplant-related deaths occurred, indicating that haploidentical HSCT is a favorable option for patients at high risk of relapse after CART19 therapy.

In conclusion, this study demonstrated that CART19s had potent antileukemic activities and improved clinical outcomes in Chinese patients with R/R ALL. Our data suggest that CART19s could provide a new therapeutic approach for patients with R/R ALL. CART19 therapy might be an effective transitional modality to bridge a brief remission with further interventional strategies, such as haploidentical HSCT and even a second allo-HSCT, to further improve overall clinical outcomes in patients at high risk of relapse after CART19s therapy.

L. Xiao has ownership interest in Innovative Cellular Therapeutics CO., LTD. No potential conflicts of interest were disclosed by the other authors.

Conception and design: Y. Hu, Z. Wu, L. Xiao, H. Huang

Development of methodology: Y. Hu, Z. Wu, Y. Luo, J. Shi, W. Ni, A. Jin, H. Zhang, Z. Cai, L. Xiao, H. Huang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Hu, Z. Wu, Y. Luo, J. Shi, J. Yu, C. Pu, G. Wei, J. Sun, J. Tu, J. Wang, A. Jin, H. Zhang, Z. Cai, L. Xiao, H. Huang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Hu, Z. Wu, Y. Luo, J. Shi, J. Yu, C. Pu, Z. Liang, G. Wei, Q. Cui, J. Jiang, J. Xie, W. Ni, J. Tu, J. Wang, A. Jin, H. Zhang, L. Xiao, H. Huang

Writing, review, and/or revision of the manuscript: Y. Hu, Z. Wu, Y. Luo, J. Shi, J. Yu, G. Wei, Q. Cui, J. Jiang, J. Xie, H. Huang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Hu, Z. Wu, Y. Tan, L. Xiao, H. Huang

Study supervision: Z. Wu, L. Xiao, H. Huang

This work was supported by grants from the 973 Program (2015CB964900), the Zhejiang Provincial Natural Science Foundation of China (LY14H080002), the Natural Science Foundation of China (81230014, 81470341, 81520108002, 81500157), Zhejiang Medical Technology & Education Foundation of China (2014KYA064, 2014KYA066), and the Key Project of Science and Technology Department of Zhejiang Province (2015C03G2010091).

This work was supported by grants from the 973 Program, Innovative Cellular Therapeutics Co., Ltd., Zhejiang Provincial Natural Science Foundation of China, the Natural Science Foundation of China, Zhejiang Medical Technology & Education, and Key Project of Science and Technology Department of Zhejiang Province.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.