Zanubrutinib monotherapy in CD79B-mutated relapsed or refractory diffuse large B-cell lymphoma: A multicenter Phase 2 study

Abstract

Diffuse large B-cell lymphoma (DLBCL) remains a clinically and molecularly heterogeneous disease,¹ with approximately one-third of patients developing relapsed or refractory (R/R) disease following frontline immunochemotherapy.^{2, 3} Among molecularly defined subgroups, CD79B mutations occur in 5%–25% of DLBCL, and tumors harboring CD79B mutations represent a biologically distinct entity characterized by chronic active B-cell receptor (BCR) signaling and constitutive activation of the NF-κB pathway, conferring inferior outcomes with conventional therapy.^3-7 CD79B mutations are enriched in the ABC subtype, consistent with their role in subtype-specific BCR signaling.^7-9

Bruton tyrosine kinase (BTK) is a central mediator of BCR signaling, providing a strong biological rationale for BTK inhibition in CD79B-mutated DLBCL.^{10, 11} However, prospective data evaluating BTK inhibitor monotherapy specifically in this genetically defined population remain limited. Here, we report the efficacy, safety, and exploratory biomarker analyses from a multicenter, Phase 2 study of zanubrutinib monotherapy in patients with CD79B-mutated R/R DLBCL. This was an open-label, single-arm, multicenter, Phase 2 study conducted in China (NCT05068440). Adult patients with histologically confirmed DLBCL harboring CD79B mutations detected using a validated amplification refractory mutation system polymerase chain reaction (PCR) assay and R/R disease after at least one prior systemic therapy were eligible. Zanubrutinib was administered orally at 160 mg (as two 80-mg capsules) twice daily in continuous 28-day cycles until progressive disease (PD), unacceptable toxicity, or study discontinuation. All patients who permanently discontinued zanubrutinib remained in long-term survival follow-up. Safety follow-up visits occurred ~30 days (±7 days) after the last dose of zanubrutinib and included planned safety assessments, laboratory assessments, and recording of alternative anticancer therapy given after the last dose of the study drug. Long-term follow-up continued until PD, start of alternative anticancer therapy, loss to follow-up, or study end, whichever occurred first. The primary endpoint was investigator-assessed overall response rate (ORR) according to the Lugano 2014 criteria.¹² Secondary endpoints included complete response (CR) rate, duration of response (DOR), progression-free survival (PFS), overall survival (OS), and safety. Detailed eligibility criteria and full methodological details, including molecular profiling, circulating tumor DNA (ctDNA) analysis, RNA sequencing, and statistical analysis, are provided in the Supplemental Methods.

Demographics, baseline clinical characteristics, and safety data were summarized descriptively. For response rates, two-sided Clopper–Pearson 95% confidence intervals (CIs) were generated. Time-to-event analyses, including DOR, PFS, and OS, used the Kaplan–Meier method to estimate the event-free curves and corresponding quartiles (including the median). Brookmeyer and Crowley method was used for generating the 95% CIs of all quartiles, and Greenwood's formula was used for generating the 95% CIs of the event-free rates at selected landmark times. Additionally, the expression levels and ctDNA abundances between groups were compared using the two-sided Mann−Whitney U test.

The study was conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonization Guidelines for Good Clinical Practice. Written informed consent was obtained from each patient before participation in any study procedure, and institutional review board approval was obtained at each study site.

Between August 2021 and March 2025, 65 patients with CD79B-mutated R/R DLBCL were enrolled and received at least one dose of zanubrutinib (Figure S1). Baseline clinical characteristics are summarized in Table 1 and presented in full in Tables S1–S3. The median age was 66 years (range, 42–92), and 52.3% of patients were male. Most patients had an Eastern Cooperative Oncology Group performance status of 0–1 (84.6%) and advanced-stage disease (Ann Arbor Stage III–IV, 81.5%). Forty-one (63.1%) patients had relapsed disease, while 24 (36.9%) were refractory to their last-line therapy. The median number of prior treatment lines was 1 (range, 1–6). The non–germinal center B-cell (non-GCB) subtype predominated (86.2%), consistent with the known enrichment of CD79B mutations in this molecular subgroup.

At a median follow-up of 13.9 (range, 0.5–36.4) months, zanubrutinib demonstrated meaningful antitumor activity. The investigator-assessed ORR was 46.2% (95% CI, 33.7–59.0), including a CR rate of 29.2% (Table S4), which was higher than the prespecified historical control of 30% based on a binomial exact test with a one-sided Type I error rate of 0.025 (P = 0.0044). Best percent change from baseline is presented in Figure S2 for patients with evaluable post-baseline measurements. Among responders (n = 30), the median time to response was 2.8 (range, 0.9–5.5) months and the median DOR was 22.7 months (95% CI, 2.8–not estimable [NE]; Figure 1A); 54.7% (95% CI, 35.0%–70.7%) of responders (partial response [PR] or CR) maintained a response for ≥12 months. The median PFS was 4.3 months (95% CI, 2.7–5.5; Figure 1B). The median OS was 18.1 months (95% CI, 11.4–NE; Figure 1C). At data cutoff, 32 deaths were reported, including 24 attributed to the underlying disease. Responses were predominantly observed in patients with non-GCB DLBCL, whereas no objective responses were recorded in the small GCB subgroup (Table S5), highlighting the biological dependence of CD79B-mutated non-GCB tumors on BCR signaling.

Zanubrutinib monotherapy was generally well tolerated. Treatment-emergent adverse events (TEAEs) of any grade occurred in 90.8% of patients, with Grade ≥3 TEAEs reported in 43.1% (Table S6). The most common Grade ≥3 TEAEs included pneumonia (9.2%) and hematologic toxicities such as neutropenia (7.7%) and thrombocytopenia (6.2%; Table S7). Five (7.7%) patients experienced TEAEs leading to death. These fatal events included acute coronary syndrome, sudden cardiac death, upper gastrointestinal hemorrhage, pneumonia, and completed suicide. Of these, acute coronary syndrome and pneumonia were reported as treatment-related AEs and occurred in patients with pre-existing comorbidities. Of note, the patient who deceased from acute coronary syndrome was 81 years old with a significant history of cardiovascular disease. Importantly, no cases of atrial fibrillation or flutter, hypertension, tumor lysis syndrome, opportunistic infections, or second primary malignancies were observed. Overall, the safety profile was consistent with the established tolerability of zanubrutinib in other B-cell malignancies.^{11, 13, 14}

Exploratory biomarker analyses identified molecular and dynamic features associated with clinical benefit (Figure S3). Quantitative levels of ctDNA were measured in haploid genome equivalents (hGE) per milliliter of plasma, calculated from the mean variant allele frequency and cfDNA concentration, assuming 3.3 pg DNA per hGE.¹⁵ Patients with low baseline ctDNA burden (<275.898 hGE, median value as the cutoff) experienced significantly prolonged PFS versus those with high ctDNA levels (n = 32 per group; two-sided P = 0.0012; Figure 1D), as well as longer OS (two-sided P = 0.000 16; Figure S4). Furthermore, ctDNA samples at the time of first response (~C3) were collected from 25 responding patients (17 CRs and 8 PRs), including 21 patients with undetectable ctDNA and 4 patients with detectable ctDNA. Among the patients with undetectable ctDNA, 10 patients achieved CR at first response and maintained CR, while 6 of 11 patients with PR at first response subsequently converted to CR, resulting in an overall CR rate of 76.2% (16/21patients). In contrast, in patients with detectable ctDNA, no patients achieved CR at first response, and only 1 patient converted from PR to CR during the treatment, corresponding to a CR rate of 25.0% (1/4 patients; Figure S5A,B). The undetectable ctDNA rate at the first response assessment, approximately 2.8 months after starting zanubrutinib monotherapy, was more prevalent in patients with a CR (CR: 16/17 [94.1%] patients; PR: 5/8 [62.5%] patients) (Figure S5C). This suggests that undetected ctDNA within the first 3 months of treatment could potentially serve as an early biomarker for predicting a CR with zanubrutinib monotherapy. Analysis of the baseline mutational landscape demonstrated an overall sensitivity of 66.4% and a median sensitivity of 91.9% for mutations with a variant allele frequency > 0.2 in detecting tumor-confirmed mutations in ctDNA, indicating the reliability in detecting mutations with fresh isolated ctDNA samples (Figure S6). The frequency of mutations in MYD88 was 42.2% in baseline ctDNA samples (Figure S7). Of note, patients with DLBCL harboring co-existing mutated CD79B and MYD88^L265P (n = 23) demonstrated higher ORR and CR rate versus those without MYD88^L265P (n = 41): ORR, 60.9% versus 36.6% (two-sided P = 0.0726); CR rate, 43.5% versus 19.5% (two-sided P = 0.049; Figure 1E). Formalin-fixed, paraffin-embedded (FFPE) samples were also used for mutational analysis (Figure S8). Consistent with previous studies, the most frequently mutated genes included: PIM1 (68.2% of patients), MYD88 (59.1%), BCL2 (34.1%), MLL2 (34.1%), BTG1 (31.8%), and CDKN2A (29.5%; Figure S8A).¹⁶ Similar to the analysis of mutations in baseline ctDNA samples, a trend for numerically higher response rates was observed in patients with co-existing mutated CD79B and MYD88^L265P in baseline FFPE samples (Figure S8B). Patients with DLBCL harboring mutated TP53, detected in baseline ctDNA or FFPE samples, exhibited a lower CR rate and shorter PFS versus those with wild-type TP53 (Figure S9). TP53 alterations frequently co-occurred with mutations in PIM1 (10/16 patients), PRDM1 (7/16), PCLO (6/16), BTG1 (5/16), BCL6 (6/16), and MYC (5/16). Among these mutations, a statistically significant co-occurrence was observed exclusively between PRDM1 and TP53 (Figure S10A). PFS analysis of the mutations showed that, beyond TP53, MYC, and BCL6 mutations may be associated with shorter PFS (Figure S10B). However, PRDM1 mutations were not associated with PFS (Figure S10B). These data demonstrated that the TP53 mutations may be associated with shorter PFS independently of co-existing PRDM1 mutations. Consistently, molecular classification revealed that poor clinical outcomes were observed in the TP53^Mut subtype (Figure S11). These data suggest that ctDNA could serve as a real-time non-invasive biomarker sample for baseline mutation profiling and mutation evolution post-treatment. Treatment response with MYD88^L265P at baseline and TP53 mutation status at baseline and undetectable ctDNA at first response are shown (Figures S5, S8, and S9). Additionally, prolonged PFS occurred in patients with high expression of genes related to G1/S phase cell-cycle transition (Figures 1F and S12), and DNA synthesis and repair (Figures 1G and S12), which are the downstream genes of the BCR-dependent NF-kB signaling pathway. Gene signature details are summarized in Table S8.

In this prospective biomarker-selected cohort, zanubrutinib demonstrated clinically meaningful activity with manageable safety in CD79B-mutated R/R DLBCL. Responses were enriched in patients with concurrent CD79B/MYD88^L265P mutations, consistent with prior studies showing preferential BTK inhibitor activity in BCR-dependent disease. These findings provide prospective validation of BTK inhibition in this molecular subgroup rather than exceeding prior efficacy signals. The favorable safety profile, with no emergent atrial fibrillation or hypertension, further supports improved tolerability versus first-generation BTK inhibitors.

These results should be interpreted in the context of an evolving treatment landscape. Bispecific antibody or CAR-T therapies have achieved higher response rates and survival outcomes, setting a higher efficacy benchmark. However, cross-trial comparisons are limited. In this setting, zanubrutinib may remain clinically relevant for patients unsuitable for bispecific antibody or CAR-T therapies, including those with advanced age, comorbidities, or limited access. Ongoing studies are evaluating zanubrutinib in combination regimens and its role as a backbone or bridging strategy.

Several limitations should be mentioned, spanning clinical, statistical, and biomarker-related aspects. From a clinical perspective, interpretation of the primary endpoint should consider the prespecified historical control rate of 30%, derived from limited pre-2020 data in largely unselected and BTK inhibitor–naive populations. This may represent a relatively favorable comparator in the context of a molecularly defined population, particularly in the context of an evolving treatment landscape. In addition, investigator-assessed responses without independent central review may introduce bias in this single-arm study, and subsequent therapies received by 50.8% of patients—including antibody–drug conjugate, CAR-T, and bispecific antibodies—may have influenced OS. From a statistical perspective, subgroup analyses, including the DLBCL cell-of-origin subgroups, MYD88^L265P status, TP53 mutation status, and ctDNA levels, were limited by small sample sizes and should be considered exploratory. Multiple testing across biomarker analyses further increases the risk of Type I error. Future studies with large sample sizes are needed to verify the results. From a biomarker perspective, the use of archival FFPE samples represents a limitation, as >60% were not collected at baseline and may not reflect the mutational landscape at enrollment. Although baseline ctDNA analyses were performed to partially mitigate this limitation, some uncertainty remains. Overall, given the population heterogeneity and limited subgroup sizes, these findings should be interpreted with caution and considered hypothesis-generating.

In conclusion, zanubrutinib represents a biologically rational, chemotherapy-free option for selected patients with CD79B-mutated R/R DLBCL, particularly those unsuitable for intensive therapies, and may serve as a foundation for future combination strategies.

Li Wang: Writing—review and editing; conceptualization; investigation; writing—original draft; methodology. Keshu Zhou: Writing—review and editing; conceptualization; investigation; methodology. Fei Li: Writing—review and editing; conceptualization; investigation; methodology. Qingyuan Zhang: Writing—review and editing; conceptualization; investigation; methodology. Hui Zhou: Writing—review and editing; conceptualization; investigation; methodology. Ou Bai: Writing—review and editing; conceptualization; investigation; methodology. Liping Su: Writing—review and editing; conceptualization; methodology; investigation. Chunhong Hu: Writing—review and editing; conceptualization; methodology; investigation. Zhi-Ming Li: Writing—review and editing; conceptualization; methodology; investigation. Kaiyang Ding: Writing—review and editing; conceptualization; methodology; investigation. Qunyi Guo: Writing—review and editing; conceptualization; methodology; investigation. Xiaoling Li: Writing—review and editing; conceptualization; methodology; investigation. Xiaoxi Zhou: Writing—review and editing; conceptualization; methodology; investigation. Wenjuan Yu: Writing—review and editing; conceptualization; methodology; investigation. Shuhua Yi: Writing—review and editing; conceptualization; methodology; investigation. Zhixia Wei: Writing—review and editing; conceptualization; methodology; investigation. Wenbin Qian: Writing—review and editing; conceptualization; methodology; investigation. Feiheng Chen: Writing—review and editing; conceptualization; methodology; investigation. Guohui Cui: Writing—review and editing; conceptualization; methodology; investigation. Zhiyu Liang: Writing—review and editing; conceptualization; methodology; validation; formal analysis. Qingchao Zeng: Writing—review and editing; conceptualization; writing—original draft; validation; methodology; formal analysis. Jiaoyan Lyu: Writing—review and editing; conceptualization; writing—original draft; validation; methodology; visualization; formal analysis. Yang Liu: Writing—review and editing; conceptualization; methodology; validation; visualization; formal analysis. Pan Zhang: Writing—review and editing; validation; formal analysis; visualization. Zhirong Shen: Writing—review and editing; conceptualization; formal analysis. Zaixing Shi: Writing—review and editing; conceptualization; methodology; formal analysis. Jing Rong: Writing—review and editing; conceptualization; methodology; formal analysis. Wei-Li Zhao: Writing—review and editing; conceptualization; investigation; writing—original draft; methodology; supervision.

Z. Liang, Q. Zeng, J. Lyu, Y. Liu, P. Zhang, Z. Shi, Z. Shen, and J. Rong are employees of and hold stock in BeOne Medicines Ltd. All other authors report no relevant conflicts of interest.

The study was conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonization Guidelines for Good Clinical Practice. Written informed consent was obtained from each patient before participation in any study procedure, and institutional review board approval was obtained at each study site.

This study was sponsored by BeOne Medicines, Ltd.

The data for this study are deposited in the Genome Sequence Archive of the BIG Data Center at the Beijing Institute of Genomics, Chinese Academy of Sciences, under the accession number HRA012405 (https://bigd.big.ac.cn/gsa-human).

On request, and subject to certain criteria, conditions, and exceptions, BeOne Medicines, Ltd. will provide access to individual, deidentified participant data from applicable BeOne Medicines-sponsored studies. BeOne Medicines shares data only when permitted by applicable data privacy and security laws and regulations, when feasible to do so without compromising the privacy of study participants, and when other considerations allow. Data requests may be submitted to [email protected].