ET019003 cells simultaneously expressed fully human ET190L1-TCRs and scFv/CD28 co-stimulatory molecules (Fig. 1A). ET019003 TCR-T cells maintained comparable antitumor potency but released less inflammatory cytokines in vitro and in vivo compared with ET190L1-CAR-T cells (Fig. 1B–F). ET019003 cell manufacturing was successful for all the eight treated patients (6 at 2 × 106/kg, and 2 at 4 × 106/kg). Three patients received a second infusion, the cryopreserved ET019003 cells of the first production was used for patient 5, and a second production was used for patient 2 and 7 (Additional file 1: Table S1). The mean transduction efficiency was 59.52% (range, 44.90–75.90%), with 61.38% of TCR+CD4+ T cells and 34.93% of TCR+CD8+ T cells (n = 10). The mean turnaround time from leukapheresis to infusion was 14 (range, 11–25) days.

Molecular design and preclinical evaluation of ET019003. A Schematic structure of ET019003 compared with natural γ/δ TCR and ET019L1-CAR. L, light; H, heavy; scFv, single-chain fragment variable. B Cytotoxicity of ET019003 and ET019L1-CAR-T cells on CD19+ Raji and Nalm-6 cells (n = 5). C Production of interleukin (IL)-2, granulocyte macrophage colony-stimulating factor (GM-CSF), tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ) in coculture of ET019L1-CAR-T and ET019003 cells with Raji and Nalm-6 cells (n = 5). D Luciferase live imaging of Raji xenograft mice on different timepoint after receiving mock T cells, ET019L1-CAR-T cells, and ET019003 cells. E Kaplan–Meier survival plot of Raji xenograft mice. F Serum IL-2, IL-6, GM-CSF, TNF-α, and IFN-γ in Raji xenograft mice on day 7 after infusion of ET019L1-CAR-T and ET019003 cells (n = 6). The unpaired t test was used in Fig. A–C and F, and the log-rank test was used in Fig. E. Ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001

As of August 20, 2022, the median follow-up time after infusion was 34 (range, 6–38) months. Patient enrollment was ceased in June 2020. Of the 11 patients with DLBCL screened, ten patients were enrolled, and eight patients received ET019003. Two patients died from disease progression and discontinued the study, one died before leukapheresis and another before infusion (Additional file 1: Fig. S1). Patients ranged from 33 to 71 years of age and had received a median of 4 (range, 2–8) prior lines of treatment, and three (37.5%) patients received prior PD-1 inhibitors (Table 1). Five (62.5%) patients had refractory diseases, and three (37.5%) experienced relapses. Patient 1 had PCNSL. Four (50%) patients were germinal center B-cell (GCB) subtype and four (50%) had non-GCB per the Hans algorithm. MYC/BCL2/BCL6 triple expression was detected in four patients (50%), and double expression in two (25%) patients. Five (62.5%) patients had stage IV disease using Ann Arbor staging. Five patients (62.5%) were assigned to the low-intermediate risk group and three (37.5%) were included in high-intermediate risk group per IPI score, respectively.

All the eight patients had AEs of grade 3 or higher (Table 2). Three patients (37.5%) experienced grade 1 CRS that resolved spontaneously, with a median onset of 4 (range, 2–9) days and a median duration of 3 (range, 1–8) days. Patient 2 developed grade 3 ICANS, manifested as confusion, barylalia, tremor, and agitation, which occurred after CRS and responded to corticosteroids. ICANS occurred on day 9 after infusion, lasted for 9 days, and was thus judged as a DLT. Consequently, another four patients were treated at the dose of 2 × 106 TCR + T cells/kg. Apart from patient 2, DLTs were not observed in the patient cohort (Additional file 1: Fig. S2). Tocilizumab was not administered. Patient 8 had pulmonary infection on day 15, which lasted for 4 days after antibiotic treatment. Other infectious complications were not observed within 1 month, which was possible due to administrating antiviral and antifungal prophylaxis in these patients. Patient 4 had lymphoma involvement in the intestinal tract, and he suffered acute intestine perforation, and received an emergency surgery on day 16 after infusion. The surgical pathology reconfirmed lymphoma infiltration. All the acute AEs were reversible with supportive treatment.

The increasing folds of inflammatory cytokines from baseline to peak were modest, except for the elevated serum interleukin (IL)-6 greater than tenfold of the baseline value was observed in patient 2, 4 and 8 (Fig. 2A–I). The elevation of serum IL-6 levels generally coincided in serum C-reaction protein levels and was concurrent with the onset of CRS and ICANS in patient 2, intestine perforation in patient 4, and pulmonary infection in patient 8 (Fig. 2J). Therefore, in addition to ET019003 treatment, there might be alternative etiologies for the elevated inflammatory markers in these three patients.

Changes in serum inflammatory markers within 1 month after ET019003 infusion. A Fold changes of inflammatory cytokines from baseline to peak (n = 10). Patient 5 received the repeated infusions in the outpatient department and data were not available. B–I Changes in the serum interleukin (IL)-6, C-reactive protein (CRP), IL-2, IL-4, IL-10, interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α) and ferritin in individuals. J Changes in serum IL-6, CRP, and ET019003 counts and copies in peripheral blood (PB) of patient 2, 4 and 8

Hematologic toxicities were the most common AEs, including neutropenia, leukopenia, and thrombocytopenia of grade 3 or 4 in seven (87.5%), six (75%), and two (25%) patients after ET019003 infusion, respectively (Table 2; Additional file 1: Fig. S3a–f). Severe anemia was not observed. The preconditioning regimens exhibited significant adverse effects on leucocytes, lymphocytes, monocytes, and hemoglobin levels, but not on platelets and neutrophils (Additional file 1: Fig. S3g). The median time from infusion to recovery of ≤ grade 2 neutropenia and leukopenia was 13 (range, 4–26) days and 13 (range, 4–26) days, respectively. Delayed recovery from severe thrombocytopenia was observed in patient 2 for over 2 months.

B-cell aplasia, defined as CD19+ B-cells representing < 3% of lymphocytes in peripheral blood (PB) [16], was observed in all patients (100%) at baseline (Additional file 1: Fig. S4a). The preconditioning chemotherapy exhibited significant inhibition on T and NK cells in the PB, and ET019003 cells showed effects on T cells (Additional file 1: Fig. S4b–c). CD4+ T and CD8+ T cells decreased significantly after the preconditioning chemotherapy and expanded remarkably on day 14 after ET019003 infusion with an invert CD4/CD8 ratio (Additional file 1: Fig. S4d–f). Three patients (37.5%) had preexisting hypogammaglobulinemia, defined as serum IgG < 800 mg/dL, IgM < 50 mg/dL, and IgA < 100 mg/dL [17]. The reduction of serum IgG, IgA, and IgM after ET019003 infusion was observed in seven (87.5%), eight (100%) and six (75%) patients, respectively. The recovery of serum IgG, IgA, and IgM to their normal levels during follow-up was observed in three (42.8%), two (25%), and four (66.7%) patients, respectively (Additional file 1: Fig. S4g–i).

Other long-term AEs included viral encephalitis at month 18 and MOG + encephalomyelitis at month 30 of patient 1, and both were treatable. The diagnosis and treatment were detailed in the supplemental results (Additional file 1). ET019003 cells were undetectable at that time in the patient; thus, we supposed that the two delayed AEs were not directly ET019003 cell-mediated.

In the phase 1 trial, seven (87.5%) patients attained clinical responses, six (75%) achieved CR, and five (62.5%) had ongoing CR (Fig. 3A). The Kaplan–Meier estimated OS at 12–36 months were 75.0% (95% CI, 31.5–93.1) (Fig. 3B). The Kaplan–Meier estimated PFS at 12–36 months were 62.5% (95% CI, 22.9–86.1) (Fig. 3C), with DOR at 12–36 months of 71.4% (95% CI, 25.8–92.0) (Fig. 3D).

Swimmer’s plot and long-term outcomes of the treated patients. A Swimmer’s plot of the eight treated patients. B–D Kaplan–Meier estimates of the overall survival (OS), progression-free survival (PFS), and duration of response (DOR)

Patient 1 with PCNSL was refractory to previous 8 lines of therapies. She got a continuing CR for over 3 years after ET019003 infusion (Fig. 4A). Numerous ET019003 cells were detectable in both the PB and CSF after infusion (Fig. 4B), indicating that ET019003 cells could sufficiently traffic from the periphery to the CNS. Patient 2 had extensive lesions and attained a quick partial response (PR) at month 1 (Fig. 4C), but the diseases progressed at month 2 (Additional file 1: Fig. S5a). A second tissue biopsy demonstrated DLBCL with expression of CD19, BCL2, C-Myc, P53, and Ki67 (LI: 90%). The patient received a second infusion with poor expansion, and the diseases progressed on day 14; consequently, the patient withdrew from the study for other salvage therapy. Patient 3 had two major lesions in the right eyeball and the pelvic cavity, and she attained a PR on day 14 and an ongoing CR for over 2 years (Fig. 4D). Patient 4 achieved a CR at month 2 and kept CR for over 2 years (Additional file 1: Fig. S5b). ET019003 cells exhibited rapid clearance and durable control of a bulky tumor (8.1 × 6.6 × 7.0 cm, SUVmax 9.4–12.7) in patient 5 (Additional file 1: Fig. S5c). Patient 6 with lymphoma mainly in the abdominal cavity did not respond to ET019003 treatment and withdrew from the study at month 2 (Additional file 1: Fig. S5d). Patient 7 had extensive lesions mainly in the lung and the abdominal cavity and attained a CR at month 5 (Fig. 4E). However, new lesions appeared at month 9, and a second infusion failed to work (Additional file 1: Fig. S5e). Patient 8 had two lymph node lesions in the left heart diaphragm angle and retroperitoneal space, and got a CR on day 24, and kept durable CR at month 24 (Additional file 1: Fig. S5f).

Clinical responses of ET019003 cells. A Changes in cranial magnetic resonance imaging (MRI) scans of patient 1. B ET019003 copies per microgram (µg) of genomic DNA in peripheral blood (PB) and cerebrospinal fluid (CSF) and body temperature changes in patient 1 within 1 month after infusion. C Changes in positron emission tomography-computed tomography (PET-CT) scans of patient 2. D Changes in ocular enhanced MRI and abdominal-enhanced CT of patient 3. E Changes in PET-CT scans of patient 7

Three patients received a second infusion. Patient 2 and 7 for salvage therapy after disease progression, but response was not observed. Patient 5 kept CR per the Lugano criteria at month 6, but PET-CT scans showed minimal residual lesions in the hepatogastric space-pancreatic head. Despite low levels of ET019003 cells in his PB, we decided to give the patient a repeated infusion in the outpatient department without the preconditioning chemotherapy. He experienced severe neutropenia, leukopenia, and thrombocytopenia, which were self-limiting. Apart from hematologic toxicities observed in patient 2 and 5, other AEs were not observed (Additional file 1: Fig. S6).

ET019003 cells showed peak expansion 9–21 days post infusion (Fig. 5A). ET019003 cells peaked at a median of 318 (range, 32–4,308,109) cells per milliliter of PB as measured by flow cytometry and 76,897 (range, 21,278–273,032) copies per microgram (µg) of genomic DNA as measured by qPCR, respectively. The median area under the curve from 0 to 28 days after infusion (AUC0–28d) was 585,493.5 copies per µg × days (Fig. 5B). ET019003 cells continued to be detectable in the PB in 50% of the patients at 12 months. ET019003 expansion was poor in the second infusions.

In vivo kinetics of ET019003 cells. A ET019003 expansion and persistence were measured as copies per microgram (µg) of the genomic DNA by qPCR in the eight treated patients within 1 year. The detectable threshold was 100 copies per µg of the genomic DNA. The black arrow indicates the second infusion. B The violin plot of peak ET019003 cells per milliliter of PB (cells/mL PB) as measured by flow cytometry, peak copies per µg of genomic DNA (copies/µg DNA) as measured by qPCR and area under the curve from 0 to 28 days after infusion (AUC0–28d)