In this single-center, prospective study, a total of 525 patients who were newly diagnosed with Philadelphia chromosome-negative B-ALL between January 2015 and September 2019 were included ( Figure 1 ). The inclusion criteria were (1) age of 1–18 years (2), achieved complete remission (CR) after induction chemotherapy (3), persistent positive MRD within three months from the start of treatment, and (4) achieved MRD negativity and the conversion of negative to positive MRD during consolidation chemotherapy. The exclusion criteria included (1) morphological relapse within two months of recurrent MRD and (2) severe heart, kidney, or liver disease. This study was approved by the Peking University People’s Hospital review board. All patients’ legal guardians provided written informed consent documents in accordance with the Declaration of Helsinki.

The lymphodepleting chemotherapy before CAR-T therapy included fludarabine (25 mg/m²/d on days –5 to –3) and cyclophosphamide (250 mg/m2/d on days –5 to –3). Multiple up-to-date CAR-T engineering technologies have been summarized (11). Anti-CD19 CAR-T cells constructed with a 4-1BB (79%) or CD28 (21%) costimulatory domain were generated via lentiviral vector from fresh leukapheresis material in this study.

For the CAR-T with a 4-1BB costimulatory domain, peripheral blood mononuclear cells (PBMCs) were collected from patients and stimulated with dynabeads coated with anti-CD3 and anti-CD28 mAbs (Thermo Fisher Scientific). Activated T cells were transduced with lentiviral vector encoding the anti-CD19 CAR construct consisting of CD19 recognition domain, transmembrane link domain, 4-1BB intracellular domain, CD3ζ intracellular domain. After lentiviral transduction, the 4-1BB CAR-T-19 cells were cultured in medium supplemented with 500 IU/ml IL-2 at 37°C/5% CO2 for approximately 5 to 11 days to obtain sufficient cells for infusion.

For the CAR-T with a CD28 costimulatory domain, peripheral blood mononuclear cells (PBMCs) were acquired through apheresis from the patients at relapses with sufficient lymphocyte counts, and the T cells were selected using CD3 magnetic beads. CD28 monoclonal antibodies were added for T cell activation in vitro. The activated T cells were transduced with the 4SCAR19 lentiviral vector encoding the CD19 CAR carrying a “safety switch”—iCasp9 for 3-5 days. The CAR-T cells were cultured in AIM-V (Invitrogen, San Diego, CA, USA) medium supplemented with IL-2, IL-7, and IL-15 at 37°C/5% CO2 for approximately 5–7 days to obtain sufficient cells for infusion.

Real-time quantitative polymerase chain reaction was used to quantify the level of the CAR gene (<100 copies/μg DNA was defined as negative). Flow cytometry (FCM) was performed to determine the transduction efficiency and the ratio of B cells in peripheral blood and bone marrow after CAR-T infusion. For patients who chose to be enrolled in the observation group after CAR-T infusion, the levels of the CAR gene and B cells were assessed every month, and maintenance chemotherapy was administered if CAR T cells were not detected in vivo and/or B cells were recovered.

The conditioning regimen for allo-HSCT was in accordance with previous reports (12, 13). Patients who received an HLA-mismatched HSCT received a regimen that included cytarabine (4 g/m2/day IV, days −10 and −9), busulfan (3.2 mg/kg/day IV, days −8 to −6), cyclophosphamide (1.8 g/m2/day IV, days −5 and −4), semustine (250 mg/m2 PO, day −3), and antithymocyte globulin (ATG) (2.5 mg/kg/day IV, days −5 to −2). Patients who received an HLA-identical HSCT were treated with a regimen identical to that of the patients who received an HLA-mismatched HSCT, but without ATG. All patients received acute graft-versus-host disease (aGVHD) prophylaxis consisting of cyclosporine A, mycophenolate mofetil, and a short-term methotrexate regimen.

The intensified chemotherapy regimens were in accordance with a previous report (14). These regimens included (1) an induction therapy (2), a consolidation therapy with two cycles of a re-induction block in between, and (3) a maintenance therapy. The induction and re-induction chemotherapy regimens consisted of vincristine, idarubicin, cyclophosphamide, and L-asparaginase. The consolidation chemotherapy regimens were comprised of methotrexate (MTX), vincristine, and peg-aspargase; and cytarabine (Ara-c), idarubicin, fosfamide, etoposide, and vincristine. The maintenance chemotherapy regimens included mercaptopurine (50 mg/m2/d PO, daily) and methotrexate (20 mg/m2/d IM, weekly). Patients in the chemotherapy bridge-to-allo-HSCT group received MTX (2–3 g/m2/d IV) and/or Ara-c (2 g/m2/d IV, days 1 to 3) based chemotherapy regimens.

A panel of eight antibody combinations, which included cCD3, mCD3, CD2, CD5, CD7, CD10, CD19, CD20, CD34, CD38, CD45, CD58, CD99, CD123, and cTDT, were used for MRD detection. The standardized assays and quality controls were consistent with those of previous reports (15). Any MRD level was considered positive. MRD was assessed every month until MRD negativity, every two to three months during consolidation chemotherapy, and every six months during maintenance chemotherapy for patients in the chemotherapy group. MRD was assessed every month until one year of CAR-T therapy and every two to three months until two years of CAR-T therapy for patients in the CAR-T group. MRD was assessed at 1, 2, 3, 4.5, 6, 9, and 12 months post-HSCT and at six-month intervals thereafter for patients in the HSCT group.

CR was defined as the presence of <5% blasts in the bone marrow, an absolute neutrophil count of >1 × 10⁹/L, a platelet count of >100 × 109/L, and the absence of extramedullary disease. Recurrence of ≥5% bone marrow blasts and/or the development of extramedullary disease were defined as a relapse. Recurrent MRD was defined as two MRD-positive samples at an interval of one month in a patient who was previously MRD-negative. Non-relapse-related mortality (NRM), aGVHD, and chronic GVHD (cGVHD) were defined as previously described (12). Cytokine release syndrome (CRS) grading was based on the National Cancer Institute (NCI) consensus CRS scoring system.

The primary endpoint is LFS. The secondary endpoints are MRD negativity rate, overall survival (OS), and safety. LFS was measured from the time when CR was achieved; events of LFS included death in CR1 or relapse. MRD negativity was associated with an undetectable MRD by FCM. The event of OS was death at the date of the last follow-up. The patients’ characteristics were compared using the chi-square test or Fisher’s exact test for categorical variables and the Mann–Whitney rank test or Student’s t-test for continuous variables. The Kaplan-Meier method was used to analyze the LFS and OS. Comparisons between different LFS and OS probabilities were performed using the log-rank test. Multivariate analysis was performed using the Cox proportional hazards regression model. Statistical significance was set at P < 0.05. SPSS version 23.0 (SPSS Inc., Chicago, IL, USA) and R version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria) were used for data analysis.

A total of 525 patients diagnosed with Philadelphia chromosome-negative B-ALL were included in this study; 448 patients were excluded due to persistent negative MRD (n = 414), relapse within two months of recurrent MRD (n = 14), withdrawal due to personal reasons (n = 12), failure to achieve CR after induction chemotherapy (n = 6), and loss to follow-up (n = 2). Of the 77 patients with persistent/recurrent MRD who were screened and encouraged to receive CAR-T therapy, 43 patients were enrolled in the CAR-T group. The remaining patients were divided into the chemotherapy bridge-to-allo-HSCT group (n = 20) and the intensified chemotherapy group (n = 14) according to their personal willingness, economic background, and donor availability. After a month of CAR-T infusion, patients who failed to achieve MRD negativity received allo-HSCT (n = 4), and patients who achieved MRD negativity were divided into the bridge-to-allo-HSCT group (n = 31) and the observation group (n = 8). Tables 1 and 2 show the characteristics of the patients, and the baseline characteristics of the patients who received CAR-T therapy and those who did not were comparable.

In this study, 43 patients received CAR-T therapy. The median dose of infused CAR T cells was 3.85 × 10⁶ (0.45–8.45 × 10⁶)/kg. The viability of CAR-T pre-infusion was 93.4% (75.0-99.1%), transduction efficiency of CAR-T pre-infusion was 29.5% (8.7-83.8%). CAR T cells rapidly expanded during the first month in 42 (97.6%) patients. The mean CAR-T count at peak expansion was 303,238 (1,250–1,890,000), and the median time to peak expansion was 11.6 (7–17) days. One patient had unsatisfactory peak CAR-T counts (1,250 copies at peak expansion).

The median MRD level pre-lymphodepletion was 0.22% (0.01–2.86%), and the rate of patients who achieved MRD negativity after a month of CAR-T infusion was 90.7%. Patients who achieved MRD negativity and those who did not received comparable doses of CAR T cells (3.9 × 10⁶/kg vs. 5.0 × 10⁶/kg, P = 0.65), but the peak CAR-T counts of patients who achieved MRD negativity were significantly higher than those of patients who failed to achieve MRD negativity (362,350 copies vs. 70,323 copies, P = 0.02). Among the patients who failed to achieve MRD negativity after a month of CAR-T infusion (n = 4), one had unsatisfactory peak CAR-T counts, and another had undetectable CAR-T counts at 1 month after CAR T-cell infusion.

Among the patients who received CAR-T therapy, 23 (53.8%) had CRS of any grade, and six (13.9%) had severe CRS (grades 3 and 4). Neurological adverse events occurred in six (13.9%) patients, of which four experienced headache and confusion, one had seizures, and another had encephalopathy. No CAR-T-related mortality was observed.

All patients who failed to achieve MRD negativity after CAR-T therapy received allo-HSCT. Among the patients who achieved MRD negativity after CAR-T infusion (n = 39), 31 were bridged to allo-HSCT after CAR-T therapy, while eight were not ( Table 3 ). Among patients who bridged to allo-HSCT after CAR-T therapy, 26 of them received 4-1 BB CAR-T, CAR-T cell can be detected before allo-HSCT in 24 patients who received 4-1 BB CAR-T. Among patients who bridged to allo-HSCT after CAR-T therapy, 9 of them received CD28 CAR-T ( Table 4 ), the level of CAR-T cell after infusion was not monitored.

In patients who were not bridged to allo-HSCT after CAR-T therapy, the median dose of infused CAR T cells was 3.96 × 10⁶ (0.45–5.3 × 10⁶)/kg, the median persistence time of CAR T cells was 5.0 (2–12) months, and the median recovery time of B cells was 3.9 (1.5–7.5) months. B cells were not recovered in two patients until the last follow-up. One of them remained in CR after 23 months of CAR-T infusion, while the other patient with CD19 negativity relapsed.

In this study, 55 patients received allo-HSCT. A total of 35 patients received CAR-T therapy before allo-HSCT, and the median time from CAR-T therapy to allo-HSCT was 67 days. The other 20 patients received chemotherapy before allo-HSCT. Thirty (85.7%) patients in the CAR-T bridge-to-allo-HSCT group were MRD-negative pre-HSCT, while the other five patients were MRD-positive with a median MRD level of 0.17% (0.01–0.77%). Fifteen (75%) patients in the chemotherapy bridge-to-allo-HSCT group were MRD-negative pre-HSCT, while the other five patients were MRD-positive with a median MRD level of 0.44% (0.01–1.5%).

Of the patients who received allo-HSCT, five received allo-HSCT from matched sibling donors, while the others received haplo-HSCT. Patients achieved neutrophil engraftment at a median time of 13 (10–35) days, and all patients achieved platelet engraftment at a median time of 14 (7–58) days. The cumulative 100-day incidence of aGVHD grades II–IV and grades III–IV in the CAR-T bridge-to-allo-HSCT group were similar to those of the chemotherapy bridge-to-allo-HSCT group [24% (95% CI, 17–27%) vs. 23% (95% CI, 12–32%), P = 0.956; 8% (95% CI, 4–12%) vs. 6% (95% CI, 3–11%), P = 0.818]. The cumulative 3-year incidence of total cGVHD and severe cGVHD in the CAR-T bridge-to-allo-HSCT group were also similar to those of the chemotherapy bridge-to-allo-HSCT group [56% (95% CI, 38–65%) vs. 49% (95% CI, 39–55%), P = 0.687; 12% (95% CI, 6–19%) vs. 11% (95% CI, 5–15%), P = 0.918]. The cumulative 3-year incidence of NRM was 3% (95% CI, 1–6%).

Of the patients who received intensified chemotherapy without allo-HSCT, nine (64.2%) achieved MRD negativity. No serious treatment-related toxicity or TRM was observed.

Between January 1, 2015 and December 31, 2020, the median follow-up time for surviving patients was 44.0 (18.0–70.0) months. Of patients in the CAR-T group, 10 (23.2%) relapsed (four withdrew, three achieved second CR with allo-HSCT, two achieved second CR with CD22-CAR-T therapy, and one abandoned further treatment after no response to CD22-CAR-T therapy). Relapse occurred at a median time of 9.6 (4–17) months after CAR-T infusion. Nine (20.9%) patients experienced a CD19-positive relapse, while one (2.3%) patient experienced a CD19-negative relapse. Of the patients in the chemotherapy bridge-to-allo-HSCT group, six (30%) relapsed (four withdrew and two achieved second CR with allo-HSCT). Of the patients who received intensified chemotherapy, 10 (71.4%) relapsed (five withdrew, three achieved second CR with allo-HSCT, and two failed to achieve second CR with salvage chemotherapy). At the last follow-up, 17 (22.0%) patients died of relapse, and two (2.5%) patients died of transplant-related complications.

Patients who received CAR-T therapy (n = 43) had a higher 3-year LFS [77.8% (95% CI, 65.6–89.9%) vs. 51.1% (95% CI, 33.8–68.3), P = 0.033, Figure 2A ] and a trend of higher OS [86.0% (95% CI, 93.4–75.6%) vs. 62.6% (95% CI, 45.5–76.5%), P = 0.059, Figure 2B ] than those who did not (n = 34). Patients in the CAR-T group (n = 43) also had a trend of higher 3-year LFS than those in the chemotherapy bridge-to-allo-HSCT group (n = 20) [77.8% (95% CI, 64.8–90.7%) vs. 68.7% (95% CI, 47.7–89.6%), P = 0.575] and had a significantly higher 3-year LFS than those in the intensified chemotherapy group (n = 14) [77.8% (95% CI, 64.8–90.7%) vs. 28.6% (95% CI, 4.9–52.3%), P = 0.001) ( Figure 2C ). The 3-year OS of patients in the CAR-T group (n = 43) tented to be higher than that in the chemotherapy bridge-to-allo-HSCT group (n = 20) [86.0% (95% CI, 75.6–96.3%) vs. 73.3% (95% CI, 52.9–93.6%), P = 0.470] and was significantly higher than that in the intensified chemotherapy group (n = 14) [86.0% (95% CI, 75.6–96.3%) vs. 49.0% (95% CI, 22.3–75.6%), P = 0.010] ( Figure 2D ).

Multivariate Cox regression modeling showed that MRD ≥1% at any checking point and non-CAR-T therapy were independent risk factors associated with inferior LFS in all patients ( Table 5 ).

In patients who received CAR-T therapy (n = 43), the LFS and OS of patients who were bridged to allo-HSCT after CAR-T infusion (n = 35) were comparable with those of patients who were not (n = 8) [75.0% (95% CI, 59.9–90.0%)vs. 75.0% (95% CI, 45.0–104.9%), P = 0.765, Figure 2E ; 85.2% (95% CI, 73.2–97.1%) vs. 75.0% (95% CI, 45.0–104.9%), P = 0.470, Figure 2F ]. MRD ≥1% at any checking point pre-CAR-T therapy (n = 15) tended to lower the LFS of the CAR-T group, but the trend was not statistically significant [65.5% (95% CI, 40.8–90.1%) vs. 82.9% (95% CI, 67.6–98.1%), P = 0.236], indicating that the negative impact of high-level MRD can be abrogated by CAR-T therapy to some extent. Multivariate Cox regression modeling showed that high-risk cytogenetics was an independent risk factor associated with inferior OS in patients who received CAR-T therapy ( Table 5 ).

In patients who received allo-HSCT (n = 55), the 3-year LFS and OS of patients who received CAR-T therapy pre-HSCT (n = 35) were higher than those of patients who did not (n = 20) [75.0% (95% CI, 59.7–90.2%) vs. 68.7% (95% CI, 47.7–89.6%), P = 0.586; 85.6% (95% CI, 69.9–97.3%) vs. 73.3% (95% CI, 52.9–93.6%), P = 0.382]. MRD ≥1% at any checking point pre-HSCT (n = 16) significantly lowered the LFS and OS of the allo-HSCT group [46.1% (95% CI, 20.2–71.9%) vs. 85.5% (95% CI, 73.7–97.2%), P = 0.006; 61.4% (95% CI, 33.9–88.8%) vs. 88.7% (95% CI, 78.3–99.0), P = 0.045]. The 3-year LFS of patients with cGVHD (n = 29) [85.9% (95% CI, 71.0–100.7%)] was higher than that of patients without cGVHD (n = 26) [61.4% (95% CI, 41.9–80.8%)] (P = 0.045). Multivariate Cox regression modeling revealed that MRD ≥1% and not having cGVHD were independent risk factors associated with inferior LFS, and MRD ≥1% was also an independent risk factor associated with inferior OS.

In patients who received intensified chemotherapy (n = 14), the 3-year LFS of patients with recurrent MRD (n = 9) was lower than that of patients with persistent MRD (n = 5) [0.0% vs. 40.0% [95% CI, 9.0–69.0%], P = 0.350). MRD ≥1% at any checking point (n = 4) tended to lower the LFS (0.0% vs. 40.0% [95% CI, 9.6–70.3], P = 0.125).