OS and PFS were estimated using the Kaplan-Meier method. Differences in prognostic factors were assessed by the log-rank test in univariate analysis. The Cox proportional hazards model evaluated the impact of patient characteristics on PFS. Data analyses were performed using Statistical Package for the Social Sciences version 25.0 and GraphPad Prism version 9.0. All tests were two-sided, with a significance level set at 0.05.
Between January 2022 and September 2023, 35 previously untreated patients with non-GCB DLBCL were enrolled. One patient was excluded due to CNS involvement, leaving 34 patients who received treatment (Fig. 1). Baseline characteristics are summarized in Table 1 and Table S7 in Additional file 2. The median age was 55 years (range, 22-71 years), and 58.8% (20/34) were male. Advanced-stage disease (stage III-IV) was present in 73.5% (25/34), 26.5% (9/34) had an ECOG-PS ≥ 2, and 47.1% (16/34) had an International Prognostic Index (IPI) score ≥ 3. Extranodal involvement was observed in 88.2% (30/34), and 64.7% (22/34) had double expressor lymphoma (DEL), defined as MYC expression ≥ 40% and BCL2 expression ≥ 50%. Molecular subtype data were available for 28 patients, determined using the LymphGen algorithm [29]. In 75% (21/28) of patients, genotyping was performed using tumor tissue DNA; in 25% (7/28) of patients, plasma ctDNA was used due to unavailable tumor samples. The MCD, BN2, ST2, N1, and other subtypes accounted for 39.3% (11/28), 14.3% (4/28), 3.6% (1/28), 3.6% (1/28), and 39.3% (11/28) of patients, respectively. Based on mutational and fluorescence in situ hybridization data, TP53 abnormalities were present in 34.5% (10/29) of patients.
A total of 88.2% (30/34) of patients completed all planned cycles of ZR2-CHOP. The median relative dose intensities (RDI = total actual dose/total planned dose × 100%) for doxorubicin, zanubrutinib and lenalidomide were 100%, 96% and 100%, respectively. The proportion of patients receiving ≥ 80% of the intended dose intensity was 97.1% (33/34) for doxorubicin, 85.3% (29/34) for zanubrutinib, and 91.2% (31/34) for lenalidomide. Patients aged ≥ 60 years received fewer cycles of ZR2-CHOP (median 5 vs. 6) and shorter durations of zanubrutinib and lenalidomide treatment (both median 5 vs. 6 cycles) compared to those < 60 years (Table 2). Seven patients (20.6%) underwent autologous stem cell transplantation (ASCT) for consolidation, and one patient (P16) received zanubrutinib maintenance due to persistent ctDNA positivity. All eight of these patients had achieved CR before initiating consolidation therapy. As they received additional treatment beyond protocol-defined therapy, they were included in the full analysis set (FAS) but excluded from the per-protocol set (PPS; Fig. 1).
At the data cut-off (February 9th, 2025), all patients were evaluable for response. The best ORR was 100% (34/34) and the best CR rate was 97.1% (33/34). After 3 cycles of treatment, 70.6% (24/34) of patients achieved CR/complete metabolic response (CMR), and 29.4% (10/34) achieved PR; after 6 cycles, the CR/CMR rate increased to 94.1% (32/34) (Fig. 2A). Six patients relapsed. One patient died from progressive disease, and another died from secondary acute myeloid leukemia (AML). In the FAS population, with a median follow-up of 28 months (range, 15-36 months), 79.4% (27/34) of patients maintained CR/CMR. The 2-year PFS and OS rates were 84.8% and 96.8% (Fig. 2B and C). In the PPS population, with a median follow-up of 27 months (range, 15-36 months), 80.8% (21/26) of patients sustained CR/CMR, with 2-year PFS and OS rates of 79.9% and 95.8%, respectively (Fig. 2D and E). Subgroup analysis of PFS showed that traditionally unfavorable factors -- such as DEL, the MCD subtype, bulky disease (≥ 7.5 cm), and multiple extranodal involvements -- did not significantly impact the efficacy of ZR2-CHOP in either the FAS or PPS populations. However, patients with an IPI ≥ 3 demonstrated a trend toward poorer PFS. No significant differences in PFS or OS were observed between patients aged ≥ 60 and those < 60 years (Fig. 3 and Additional file 2: Fig. S1-S3).
Plasma ctDNA samples at baseline were available for 26 patients; among them, 96.2% (25/26) had at least one detectable variant. Of these, 84% (21/25) achieved ctDNA negativity during treatment. At the mid-term evaluation, 73.9% (17/23) of patients were ctDNA-negative, increasing to 81.0% (17/21) at EOT. All patients who achieved ctDNA negativity also achieved CR/CMR. Notably, five patients with PR and ctDNA negativity at mid-term analysis converted to CR/CMR by EOT. No patients who achieved ctDNA negativity at mid-term reverted to ctDNA-positive status at EOT. Two patients who were ctDNA-positive at mid-treatment achieved negativity by EOT. Four patients remained ctDNA-positive throughout treatment but achieved CR/CMR by PET-CT (Fig. 4 and Table S8 in Additional file 2). Among these, one patient relapsed at 16 months, while the remaining three maintained CR at 20, 21, and 28 months of follow-up. Two patients who had previously achieved ctDNA negativity relapsed. One developed CNS relapse of primary testicular DLBCL with MCD subtype and TP53 mutation at 18 months post-treatment. The other, harboring a TP53 deletion, experienced lymph node relapse at 28 months. Overall, the 2-year PFS rate among patients who achieved ctDNA negativity was 95.0%.
All patients experienced treatment-related AEs. The most common AEs of any grade were anemia (88.2%, 30/34), fatigue (79.4%, 27/34), thrombocytopenia (73.5%, 25/34), hypokalemia (73.5%, 25/34), and neutropenia (70.6%, 24/34). AE incidence rates were generally consistent across different age groups and lenalidomide dose cohorts (Table 3). Grade 3-4 AEs occurred in 67.6% (23/34) of patients, most frequently neutropenia (50.0%, 17/34), thrombocytopenia (32.4%, 11/34), febrile neutropenia (23.5%, 8/34), anemia (17.6%, 6/34), and pneumonia (17.6%, 6/34). Notably, one patient developed Grade 2 atrial fibrillation, and another experienced a Grade 3 gastrointestinal hemorrhage. One patient died of secondary AML at the 33-month follow-up.
AEs led to temporary dose holds of R-CHOP, zanubrutinib, or lenalidomide in 17.6% (6/34), 64.7% (22/34), or 41.2% (14/34) of patients, respectively (Table 2). Dose reductions were required in 17.6% (6/34) of patients for R-CHOP, 11.8% (4/34) for zanubrutinib, and 14.7% (5/34) for lenalidomide. Treatment discontinuation due to AEs occurred in 2.9% (1/34), 5.9% (2/34), and 11.8% (4/34) of patients for R-CHOP, zanubrutinib, and lenalidomide, respectively. All four patients who discontinued any component of therapy were aged ≥ 60 years. The AEs leading to discontinuation included thrombocytopenia and neutropenia (n = 2), pneumonia (n = 1), and severe rash (n = 1).
Six patients experienced relapse. Among them, two were classified as MCD subtype, one as BN2, and one as N1. Three patients had TP53 aberrations. Recurrently observed mutations in this cohort included those involved in sustained BCR signaling (CD79A, CD79B and PTPN6 mutation), constitutive NF-κB pathway activation (MYD88 and TNFAIP3 mutation), interruption of key epigenetic regulator (EP300, CREBBP and KMT2D mutation), impairment of TP53 pathway (TP53, BTG2 mutation) and loss of anti-tumor immunity (B2M, CD58, CD83 and CIITA mutation, PDCD1LG2 fusion, Table 4). Despite these findings, no specific baseline plasma ctDNA mutations were associated with subsequent disease progression (Additional file 2: Fig. S4). Two patients (P15 and P22) experienced CNS relapse at 18 and 19 months of follow-up, respectively. Both had previously received intrathecal prophylaxis during frontline therapy. Following relapse, they underwent salvage treatment regimens that included zanubrutinib and subsequently achieved CR.