Long-term follow-up of donor-derived CD7 CAR T-cell therapy in patients with T-cell acute lymphoblastic leukemia

Detailed study procedures for this single-center, single-cohort, open-label, phase I study (ChiCTR2000034762) have been reported (Data Supplement) [16]. Briefly, patients between the ages of 0–70 years who had CD7⁺ r/r T-ALL, Eastern Cooperative Oncology Group Performance Status (ECOG-PS) < 3, and no uncontrollable infections or organ failure were considered eligible. The trial followed the principles of the Declaration of Helsinki, and the protocol was approved by the Institutional Review Board at Beijing Gobroad Boren Hospital. All patients supplied written informed consent.

Patients who had received prior-SCT were infused with CD7 CAR T cells obtained from former SCT donors. Patients who had not undergone previous SCT received CD7 CAR T cells obtained from new donors who also provided stem cells for transplantation after CD7 CAR T-cell treatment. On days -5, -4, and -3 prior to infusion, patients who received prior-SCT donor cells underwent lymphodepletion with cyclophosphamide 250 mg/m²/day and fludarabine 30 mg/m²/day. Patients who received new donor cells underwent enhanced lymphodepletion with cyclophosphamide 30 mg/kg/day and fludarabine 30 mg/m²/day on the same days. CD7 CAR T cells were administered as a single intravenous infusion with a target dose of 1 × 10⁶ (± 30%) cells per kg body weight on day 0. Infusion of a lower dose of 5 × 10⁵ (± 30%) cells per kg body weight was allowed if the CAR T-cell product did not meet the target dose, and these data were also included in the safety and efficacy analyses.

Dose-limiting toxicities (DLTs) within 21 days and incidence of AEs within 30 days, or from day 30 until final visit, were primary endpoints. The ASTCT Consensus was used to grade CRS and neurotoxicity [17]. The EBMT consensus was used to grade GVHD [18]. Other AEs including infections and hematologic toxicities were graded with reference to CTCAE, Version 5.0. Plasma virus activation was routinely monitored by PCR. More details of AE assessment and management are in Additional file 1.

Overall response rate (ORR) (including complete remission [CR] or complete remission with incomplete blood count recovery [CRi]), as assessed by NCCN guidelines, version 1.2020 [4], progression-free survival (PFS), and overall survival (OS) were secondary endpoints. CAR T-cell pharmacokinetics in peripheral blood (PB) and cerebrospinal fluid (CSF) were also secondary endpoints. The survival and disease remission status continued to be followed after receiving SCT or other new anti-leukemia therapies, but documentation of AEs or pharmacokinetics was discontinued. Additional details about endpoints and assessments were as previously described (Data Supplement) [16].

The size of sample was determined based on clinical considerations. The dose was based on previous experience in CAR T-cell therapies in the center. No dose escalation was conducted due to limited capacity for manufacturing, and patient safe run-in still followed a modified 3 + 3 scheme. Time-to-event analysis was based on Kaplan–Meier method. Endpoints were analyzed in subgroups based on prior-SCT donors or new donors, as well as who received the target dose or a low dose. Additional details were in Additional file 1.

As previously reported, 20 participants with r/r T-ALL were enrolled between July 18, 2020 and December 21, 2020, and all patients (100%) received infusion of CD7 CAR T cells (Fig. 1) [16]. At data cutoff on December 20, 2022, the median follow-up time was 27.0 (range, 24.0–29.3) months. Baseline characteristics and early clinical outcomes (median 6.3-month follow-up) for all treated patients, including subgroup analyses of patients based on prior-SCT or on whether the target dose or low dose were administered, have been previously reported (Table 1) [16].

Short-term AEs that had been reported among the 20 treated patients (< 30 days post-infusion), including grade 1–2 CRS in eighteen patients (90%), grade 3–4 CRS in two patients (10%), grade 1–2 neurotoxicity in three patients (15%) and grade 1–2 GVHD in twelve patients (60%). Long-term (> 30 days post-infusion) safety analysis was mainly focused on SAEs of the 12 patients without SCT consolidation (Table 2, Additional file 1: Table S1). The eight patients who received new donor CAR T-cell infusion were either bridged to SCT (n = 7) or was off-study (n = 1) early after CAR T-cell infusion, and therefore stopped monitoring for AEs. Cytopenias, GVHD and infections were major long-term AEs. A total of six long-term SAEs other than cytopenias were observed, including two that have been previously reported [16].

Cytopenia, an anticipated side effect of lymphodepletion [19], occurred in all 20 patients within 30 days, of which 100% were grade 3 or higher, and 12 (60%) patients had grade 3 or worse cytopenias before enrollment. Among 12 patients who received prior-SCT donor derived CAR T-cell infusion, all had their cytopenias recovered to grade 2 or lower within 3 months following recombinant human granulocytes/macrophage colony-stimulating factor (rhGM-CSF) administration (n = 4) and intravenous infusion of CD34-positive stem cells without preconditioning (n = 1) [16]. Three patients had late-onset grade 3 cytopenias at 8, 12.5 and 13 months after CAR infusion, which were suspected to be related to the preceding infection events.

Twelve (60%) of the 20 treated patients developed GVHD within 30 days post-CAR T-cell infusion as previously reported. Late-onset GVHD occurred in 7 (58%) of 12 patients who did not receive SCT consolidation, including six (50%) mild and one (8%) severe cases. Seven patients experienced grade 1 or 2 skin GVHD, with a median onset of 87 days (range, 31–314), manifested as rash maculopapular mostly in the extremities and chest area (< 50% body surface) accompanied by desquamation and pruritus, which were alleviated after treatment with methylprednisolone and ruxolitinib. One of these patients concomitantly developed grade 4 intestinal GVHD at 245–265 days manifested as diarrhea (> 1000 ml/day) and abdominal pain and his symptoms alleviated to grade 1 after giving methylprednisolone, ruxolitinib and mycophenolate mofetil. Another one was a mild lung GVHD at 303–312 days featured by restrictive respiratory dysfunction, which was alleviated after treatment with methylprednisolone, ruxolitinib and cyclosporine. Interestingly, the patient who developed severe GVHD had obviously elevated levels of serum ferritin and lactate dehydrogenase (LDH) compared with those (n = 11) without severe GVHD (Additional file 1: Fig. S1). As previously reported, 14 patients received haploidentical donor CAR T cells, and six patients received matched sibling donor (MSD) or matched unrelated donor (MUD) CAR T cells. Patients receiving haploidentical donor cells and patients receiving MSD/MUD cells had a comparable GVHD incidence (9/14 [64%] vs 3/6 [50%] for early GVHD, and 5/8 [63%] vs 2/4 [50%] for late-onset GVHD) and persistence (median 2 days [range, 0–20] vs 1 days [range, 0–12] for early GVHD and median 213 days [range, 0–420] vs 88 days [range, 0–238] for late-onset GVHD) (Additional file 1: Table S2, Fig. S2).

Six (50%) of the 12 patients who did not receive SCT consolidation experienced infections of any grade during the study. Five patients (41.7%) had grade 3 or higher infections with a median onset time of 8.7 (range, 5.4–13.3) months post CAR T-cell infusion. One patient had been previously reported to succumb to pulmonary hemorrhage in the context of fungal pneumonia at 5.5 months. One patient had been previously reported to had mixed CMV and EBV infections at 5.4 months during treatment for his hematochezia in an outer hospital, and in this updated analysis he was recorded to finally die at 6.8 months. The other three severe infections were newly recorded, including a grade 3 CMV encephalitis at 11 months that was resolved after treatment with ganciclovir and anti-CMV immunoglobulin, a grade 5 pulmonary infection at 12.3 months during the immunosuppressant treatment for his intestinal GVHD, and a grade 5 pseudomonas aeruginosa pneumonia at 8.7 months (This patient initially controlled the infection at hospital, but he stopped treatment after discharge leading to disease exacerbation) (Additional file 1:Fig. S3). Of the seven patients with SCT consolidation after CAR T-cell infusion, one (14.3%) was recorded to have a severe infection.

Overall, non-relapse mortality occurred in five (25%) of 20 patients at a median time of 6.8 months (range, 2–12.3) after treatment, including four deaths caused by infections in patients without SCT consolidation and one death caused by engraftment syndrome in a patient after SCT consolidation.

The ORR in the treated population was previously reported as 95% (95% CI, 76.4–99.1) (n=19) at day 30; 85% (95% CI, 64.0–94.8) (n = 17) had CR at day 30; 5% (n = 1) achieved PR at day 30 and he reached CR at day 45; 5% (n = 1) had an objective response at day 30. Of 19 patients who responded, seven patients who received new-donor CAR T cells proceeded to SCT consolidation, two patients withdrew to take alternative therapies at day 55 and 271, and 10 patients did not receive further therapy and were followed for a median time of 27.0 (range, 24.0–29.0) months.

By data cutoff date, of 10 patients who did not receive further treatment, three were in remission status, three relapsed (two CD7-negative marrow disease [including a previously reported one], and one CD7-positive extramedullary disease), and four succumbed to infection; Of seven (37%) patients who underwent SCT consolidation, three maintained remission, three relapsed (two CD7-negative marrow disease, and one CD7-positive marrow disease), and one patient died of transplant-related complications (Fig. 2A–D). A total of six patients (33.3% of CR patients) had a relapse, with a median time of 6 (range, 4–11) months post infusion. Next-generation sequencing (NGS) revealed two frameshift and two missense mutations in CD7 gene in specimens from four CD7-negative relapse patients (Fig. 2E). One of them (E012) also performed CD7 sequencing on pretreatment tumor samples, but no mutation was found.

Median PFS and duration of response (DOR) were respectively 11.0 (range, 6.9–12.5) months and 10.5 (range, 6.4–12.0) months among the 19 responders, and median OS was 18.3 (range, 12.5–20.8) months. The 2-year PFS rate of the 19 responders and OS rate of all 20 treated patients were 36.8% (95% CI, 13.8–59.8%) and 42.3% (95% CI, 18.8–65.8%), respectively (Fig. 3A, B and Additional file 1: Fig. S4A). Post hoc analyses of PFS, DOR, and OS were conducted comparing patients according to whether they received SCT consolidation. For patients without SCT consolidation, 2-year PFS and OS rates were 31.8% and 35%, respectively, and the median PFS and OS were 11.0 (range, 6.9–12.5) months and 18.3 (range, 8.8–20.8) months, respectively. For patients with SCT consolidation, 2-year PFS and OS rates were 42.9% and 58%, respectively, and median PFS was 9.1 (range, 5.8–9.1) months. Median OS was not reached by study endpoint in the subgroup of patients who received SCT consolidation (Fig. 3C, D and Additional file 1: Fig. S4B). Statistical comparison was not conducted owing to the small sample size of the subgroups.

The short-term kinetics of CAR T-cell infusion was reported previously. CD7 CAR T cells reached peak levels in blood between days 7–14 after treatment, with a median concentration of 83.70 (range, 5.45–1,300.00) cells/μl. One patient who was discharged from the study early and seven patients who received SCT consolidation did not have long-term monitoring with CAR T cells. Among 12 patients without SCT consolidation, CAR T cells could be detectable by flow cytometry in 100% (seven of seven evaluable patients) and 50% (two of four evaluable patients) at 6 and 12 months post-CAR T-cell infusion (Fig. 4A). CD7 CAR T cells were still detectable by flow cytometry in two patients before their CD7⁻ relapses and undetectable by flow cytometry in a patient before his CD7⁺ relapse. All evaluable patients had CAR transgenes detectable by quantitative polymerase chain reaction (PCR) at the time of the last assessments, and the median duration of CAR T-cell persistence at flow cytometric level in the 12 patients without SCT consolidation was 255 (range, 30–682) days (Fig. 4B). Of note, peak CAR T-cell counts did not significantly differ between patient subgroups according to long-term remission or relapse. The incidence of late-onset severe infection, GVHD or cytopenias was not significantly associated with a higher peak CAR T-cell counts (Fig. 4C–F). There was no difference in the persistence of CAR T cells between patients receiving haploidentical and MSD/MUD cells (Fig. 4G).

In all treated patients, non-CAR CD7⁺ T and NK cells were rapidly depleted within 15 days of CD7 CAR T-cell infusion, which was accompanied by an elevated count of CD7⁻ T and NK cells (Fig. 5A–F and Additional file 1: Fig. S5). Patients who received prior-SCT donor CAR T cells had complete chimerism status in peripheral blood T cells after infusion, whereas patients with new donor CAR T cells had complete or mixed chimerism after infusion (Additional file 1: Table S3). T and NK cells were monitored with long-term in the 12 patients who did not receive SCT consolidation. CD7⁺ T and NK cells remained undetectable in all patients until the last visit, excepting one patient who had recovery of CD7⁺ T and NK cells 25.6 months post-infusion following loss of flow-cytometry-detectable CD7 CAR T cells at 22.7 months. The number of CD7⁻ T,  total T and NK cells progressively increased, and by the last visit, total T-cell counts recovered to normal levels in 7 (58%) of 12 patients at a median time of 1.9 (range, 0.4–4.3) months (Fig. 5C). Total NK-cell counts recovered to normal levels in 6 (50%) of 12 patients at a median time of 5.1 (range, 1.9–21.4) months (Fig. 5F). Of the five patients with severe infections, two had T-cell recovery and one had NK-cell recovery, whereas of the 7 patients without severe infection, five had T-cell recovery and five had NK-cell recovery (Fig. 5G, H). All patients (100%) exhibited a low CD4⁺/CD8⁺ T-cell ratio within 1 month, and seven patients (55%) recovered to normal CD4⁺/CD8⁺ T-cell ratio before last visit (Additional file 1: Fig. S6). B-cell recovered in seven patients (> 5% of lymphocytes) between days 25 and 100 as previously reported [16]. In this follow-up study, one additional patient had B-cell recovery at 17 months, and the other four patients remained < 5% B cells among lymphocytes until the last visit, which was suspected to be related to the presence of GVHD.

Our previous short-term analysis showed a lack of naïve subpopulations of non-CAR T cells early after CD7 CAR T-cell infusion, but patients had a partially preserved response to viral and fungal antigen stimulation [16]. Long-term monitoring of T-cell phenotype and function in 2 patients showed that the central memory T-cell subpopulation gradually increased, and low levels of naïve and stem-cell-memory T-cell subpopulations were detectable in 1 of these patients after 15 months (Fig. 5I, Additional file 1: Fig. S7A). TCR diversity in patients after CD7 CAR T-cell infusion remained lower compared to healthy donors (Fig. 5J, Additional file 1: Fig. S7B). However, the response of T cells post-infusion to CMV and EBV antigen stimulation was in a trend of elevation, suggesting certain degree of protective function against these viruses. (Fig. 5K, Additional file 1: Fig. S7C).