Characteristics of patients and anti-CD7 CAR-T infused products
From April 15, 2021 to January 13, 2022, 11 patients were enrolled in this trial. Expect for one patient who achieved complete remission (CR) before CAR-T cell infusion, 10 were treated (Fig. 1). Seven patients (70%) were male, and the median age was 32 (range, 16–69) years (Table 1). All enrolled patients had CD7+ T-cell malignancies including adult T-ALL, T-cell lymphoblastic lymphoma, angioimmunoblastic T-cell lymphoma and mycosis fungoides. The median lines of previous therapies were 4 (range, 2–10). One patient received an excision of mediastinal tumor, 7 patients received allogeneic SCT, and one received donor lymphocyte infusion. Five patients (50%) had ALL including one with extramedullary disease (EMD) involving the nasopharynx and clavicle. Four patients were diagnosed with lymphoma grading II-IV according to Ann Arbor stage. Two of them had BM infiltrations and four had extranodular lymphoma including mediastinum, lung, chest wall, gastrointestinal tract, spleen, and subcutaneous tissue. Patient 1 had IIIA mycosis fungoides staging by TNMB (tumor, node, metastasis, blood) classification (28). The median blasts in BM were 15% (range, 0–53%) in all patients. Three patients had high-risk mutations including RUNX1 (29, 30), HOX11L2 (31), and KMT2D (32).
Based on the patient’s tumor burden, pancytopenia status, the preference and donor availability, PBMCs were collected from the patients (n = 5) or donors (n = 5) for CAR-T cell production. The characteristics of infused CAR-T cell products are presented in Supplemental Table 1. The median transfection efficiency was 46.3% (range, 40.2%-81.7%) and the median CD4+/CD8+ T cell ratio was 6.6 (range, 1.5–55.9). All cell products were frozen due to long-distance transportation and were resuscitated within 15 minutes before infusion. Patient 6 and patient 7 received chidamide, a histone deacetylase inhibitor, as bridging therapy after aphesis and before lymphodepletion.
Safety
Eight patients (80%) experienced CRS, with grade 1–2 in 5 patients (50%) and grade 3 in one patient (10%) (Table 2). The median time of onset and duration of CRS was 10 (range, 7–15) days and 4 (range, 2–9) days, respectively. Both tocilizumab and corticosteroids were administrated for 5 patients. Considering the serum levels of interleukin-6, 3 patients received dexamethasone alone (Supplement Table 2) (33). Immune effector cell-associated neurotoxicity syndrome was not observed among these participants. Two patients were diagnosed with hemophagocytic lymphohistiocytosis (HLH). Patient 3 recovered from HLH after intravenous steroids. Steroids and etoposide failed to control Epstein-Barr virus (EBV)-related HLH of patient 10, and the patient died from fungal pneumonia on day 63.
GVHD of grade 1–2 occurred in two patients. Patient 4 who received allogeneic CAR-T cells developed acute GVHD presented as diarrhea and maculopapular rash. Patient 2 had previously received allogeneic SCT. She was treated with autologous CAR-T cells and developed chronic GVHD characterized as skin desquamation and pigmentation. GVHDs of these patients were well controlled by tacrolimus and methylprednisolone.
Pancytopenia was generally observed in these patients which might correlate to patients’ marrow reserves, lymphodepletion or CAR-T therapy (Supplemental Table 3). Five patients (50%) had grade ≥ 3 cytopenia before lymphodepletion. All patients developed grade 4 lymphopenia, neutropenia and leukopenia after infusion. Persistent grade 3–4 lymphopenia and neutropenia over one month occurred in 2 patient and 3 patients, respectively. Secondary appearance of grade 4 lymphopenia and neutropenia were observed in patient 2 and 10. Patient 4 developed grade 4 lymphopenia during day 52 to day 75. Seven patients (70%) suffered grade ≥ 3 thrombocytopenia. Grade ≥ 3 anemia occurred in 7 patients (70%). A significant secondary decrease of platelets was found in patients 4, 8 and 9 (Supplemental Fig. 1). Patient 9 recovered to an absolute platelet count of at least 50,000 per cubic millimeter by day 28 but plunged to 21,000 per cubic millimeter on month 2. The patient died from a sudden intracerebral hemorrhage on day 81 and the pathogenesis was unclarified.
Six patients experienced 9 infections (Supplement Table 2). Five occurred within one month after infusion, with 3 bacterial infections, 1 fungal infection and 1 viral infection. Three patients suffered from cytomegalovirus (CMV) or EBV activation, manifested with cough, fever or diarrhea. Patient 4 developed viral pneumonia associated with EBV infection on day 62 and died from respiratory failure on day 75.
Efficacy And Long-term Outcomes
At a median follow-up of 6 (range, 2.7–14) months, the best overall response rate (ORR) was 70% (Fig. 2A). Among 7 responders, 3 patient received autologous cells and 4 were treated allogeneic therapies. Within a month post-infusion, 7 patients with BM infiltration (100%) achieved MRD-negative CR. Four patients with high-risk mutations (100%) achieved molecular remission. Two patients with EMD or extranodular infiltrations responded to CAR-T therapies, including 1 CR (Fig. 2B) and 1 partial remission (PR) (Fig. 2C). Patients 6 and 7 with progressively bulky lymphoma withdrew from the trial on day 14 and 25, respectively. Shrunken lymphadenopathy was observed in patient 10 on day 49, but unfortunately, the reduction did not reach the criteria of PR.
None of patients received SCT after CAR-T cell therapies. After autologous CAR-T cell therapies, patient 1 with mycosis fungoides achieved PR on day 65, relapsed on day 180 and attained CR followed by 11 courses of low-dose local radiotherapy; patients 2 and 3 suffered CD7+ relapses on day 103 and 140, respectively. After allogeneic CAR-T cell therapies, patient 8 with KMT2D and HOX11L2 mutations had a CD7− relapse at month 3; patient 5 with T-ALL maintained MRD-negative remission for 9 months; patients 4 and 9 died during remission as previously described.
Kinetics Of Anti-cd7 Car-t Cells And Serum Cytokines
We adopted two methods to detect in vivo CAR-T cells. During treatment, FCM is more convenient to trace the expansion of CAR-T cells. However, PCR has higher sensitivity for tracking a trickle of CAR-T cells during follow-up. The median time of peak expansion measured by FCM was 14 (range, 7–23) days after infusion. The median levels of peak expansion were 409.0 (range,11.4–8640.0) per µl measured by FCM and 7.95 × 104 (range, 2.88 × 102 -1.75 × 105) copies per microgram genomic DNA measured by ddPCR, respectively (Fig. 3A, 3B). Limited by the small sample size, peak CAR copies were not significantly correlated with cell sources, disease subtypes, tumor burden and dose of infused cells, but they were associated with the best efficacy (P = 0.02) (Fig. 3C, Supplemental Fig. 2). Four patients (57.1%) had a relatively high level of CAR-T copies detected by ddPCR at month 2, among whom 3 received allogeneic CAR-T cells and 1 received autologous products.
Sixteen serum biomarkers were detected after infusions, among which 10 rise at different levels in a group of patients (Fig. 4A). Unfortunately, the data was insufficient for statistical analysis. Serum interleukin-6, interleukin-10, interferon-γ showed significant summits in patients 2, 5, 7, 8 and 9 (Fig. 4B). These summits clung to the occurrence of CRS-related symptoms and were prior to CAR-T cell peak amplification (Supplemental Fig. 3).
The absolute count of T cells had a transient decline after CAR-T infusion followed by a dramatic rise due to CAR-T cell proliferation (Supplemental Fig. 4). Accompanied by CAR-T amplification, CD7+ T cells were rapidly eliminated and CD7− cell subsets remained alive (Fig. 4D). T cells reached a median count of 401.82 (range, 46.47-1998.40) per µl a month post-infusion.
Single-cell Rna Sequencing
Patients 1 and 5 achieved durable remission after receiving CAR-T cell therapies. To further explore the immune reconstitution after CAR-T cell infusion, we performed single-cell transcriptomic sequencing of PBMCs derived from the two patients. Sample 1 was from patient 1 at 12 months after autologous CAR-T treatment, and sample 2 from patient 5 at 9 months after allogeneic CAR-T therapy. After performing quality control, we included 8627 cells from sample 1 and 12455 cells from sample 2 for further analysis. UMAP analysis revealed 22 clusters in both sample 1 and 2 (Supplemental Fig. 5–6). Sample 2 contained a higher proportion of T cells compared with sample 1 (79.40% vs 55.04%). Among T cells, CD8+ T cells accounted for 74.27% in sample 2. A predominant cluster of NK cell accounted for 10.46% in sample 1 (Fig. 5A). Notably, CD7 expression differed significantly between the two samples, especially higher expression on NK cells from sample 1, but almost absent on sample 2 (Fig. 5B). We analyzed differentially expressed genes in CD4+ T, CD8+ T, and NK cells, and found that only CD4+ T in sample 1 highly expressed FOS, which was enriched in the TNF signaling pathway (Supplemental Fig. 7–8). CD7 and FCER1G were upregulated in NK cells in sample 1, but were not enriched to any pathway.
We further divided T cells into 16 clusters via Principal Component Analysis and found that 93.9% of T cells from sample 2 expressed T cell effector function-related genes (Supplemental Fig. 9–10). Patient 1 had higher levels of memory T, γδ T and cytotoxic T cells with high granzyme B compared with patient 5 (Fig. 5C-D). Immune checkpoint, TIGIT, was widely expressed on effector T cells of patient 1 but limited to exhausted T cells of patient 5 (Fig. 5D). Patient 5 had relatively less regulatory T cells compared with patient 1, and myeloid-derived suppressor cells and tumor-associated macrophages were not detected.