Obinutuzumab vs rituximab for transplant-eligible patients with mantle cell lymphoma Free

Mantle cell lymphoma (MCL) is an aggressive B-cell malignancy for which therapeutic innovation has led to significant improvements in outcomes over the past 20 years. Rituximab (R)–containing high-dose cytarabine chemotherapy regimens followed by autologous stem-cell transplantation (ASCT) and rituximab maintenance (RM) are the standard of care¹ for transplant-eligible patients. We recently reported that 78% of patients who underwent transplantation are alive and remain free of disease at 7 years. However, patients with MCL experiencing a progressive disease or early relapse within 2 years from treatment initiation (progression of disease within 24 months [POD24])² or relapsing after maintenance³ both represent unmet medical needs. Bruton tyrosine kinase inhibitor targeted therapies and chimeric antigen receptor T-cell approaches have demonstrated high efficacy in relapse MCL. The addition of ibrutinib to standard frontline chemotherapy has recently shown promising results.⁴ The emergence of new targeted therapies, immunotherapy, and CD20 antibody (Ab) maintenance has thus deeply modified the outcomes in patients with MCL.

Obinutuzumab (O) is a glycoengineered, type 2, anti-CD20 monoclonal Ab designed to improve Ab-dependent cell–mediated cytotoxicity compared with R. O is approved for follicular lymphoma (FL).⁵ Conversely, O failed to demonstrate its superiority in diffuse large B-cell lymphoma.⁶ In vitro data suggest that O provides better antimantle cell leukemia activity than R.⁷ The LyMa-101 phase 2 trial investigated the activity of O plus DHAP (O-DHAP) followed by ASCT and O maintenance (OM).⁸ However, no head-to-head trial comparing O vs R in MCL has been performed so far and is unlikely to happen, given the rarity of the disease and the current therapeutic landscape.

Thus, the question of the best CD20 Ab in MCL remains open, although it has been demonstrated that RM prolongs overall survival (OS) in both young and elderly patients.^1,9 We performed an indirect comparison of O and R in patients with first-line MCL.

All patients enrolled in the LyMa and LyMa-101 trials (NCT00921414 and NCT02896582, respectively) were eligible for the present work. Inclusion and exclusion criteria were similar in both trials and have been previously reported.^1,8 These 2 trials enrolled transplant-eligible patients aged 18 to 65 years with newly diagnosed untreated MCL. Trial designs are presented in supplemental Figure 1, available on the Blood website. Briefly, patients received 4 courses of anti-CD20 plus DHAP regimen followed by ASCT. In the phase 3 LyMa study (N = 299), 240 patients were randomized between RM or observation (120 patients in each arm).¹ In the phase 2 LyMa-101 study (N = 86), O was used, treatment design was similar, and all patients received maintenance. The timing of computed tomography evaluation was identical in both studies, during and after the maintenance phase.

The statistical analysis plan was divided into 2 parts. We first updated the LyMa-101 trial and then compared O and R using propensity score matching (PSM) analysis. The results of this last analysis were challenged by a stabilized inverse probability of treatment weighting (sIPTW) analysis.

We first updated the outcome of patients enrolled in the LYMM-101 trial. Time-to-event survival curves were estimated using the Kaplan-Meier method. Progression-free survival (PFS) was defined as the time from inclusion into the study to the first observation of documented disease progression or death due to any cause. If a participant had not progressed or died, PFS was censored at the time of the last visit with adequate assessment. OS was measured from the date of inclusion to the date of death from any cause. Living patients were censored on their last contact date.

Measurable residual disease (MRD; quantitative polymerase chain reaction for clonal immunoglobulin gene, as previously reported⁸) was monitored during the maintenance period at months 6, 12, 18, 24, 30, and 36. After the end of maintenance, an “on demand” maintenance period started, with O on-demand administration in case of MRD positivity. During this on-demand period, the MRD was assessed at months 3, 6, 9, 12, 15, 18, 21, 24, 27, and 30 after the end of OM.

The aim of PSM was to balance the covariates between the LyMa and LyMa-101 protocols to account for all possible measured confounding variables.¹⁰ Clinical characteristics at inclusion (sex, Ann Arbor stage, Mantle Cell Lymphoma International Prognostic Index (MIPI) score, B symptoms, presence of a blastoid variant, and/or presence of bulky disease) were used for PSM to compare overall response rate (ORR)¹¹ and bone marrow (BM) MRD negativity rate among responders at end of induction (EOI), between LyMa and LyMa-101 patients. To compare the efficacy of RM and OM, a second matching process, relying on the same characteristics plus EOI BM MRD negativity within patients initiating maintenance, was performed. Notably, in LyMa-101, 2 patients were not eligible for matching (1 due to consent withdrawal and 1 due to missing data) and in LyMa, 3 patients were not eligible: 1 due to consent withdrawal and 2 due to missing data.

To compare the efficacy of O vs R for PFS and OS from treatment initiation, half of the 59 nonrandomized patients in LyMa (n = 29) were randomly allocated to the RM arm to create a pseudopopulation with outcome data from inclusion (LyMa-intention to treat [ITT]; n = 149 patients of whom 29 nonrandomized and 120 randomized) in an RM ITT fashion. PSM based on the initial characteristics was then performed between the LyMa-ITT population (n = 149) and all the LyMa-101 (N= 86) patients.

Another statistical approach using sIPTW analysis was also performed to confirm or refute the PSM results.¹⁰ In IPTW, weights are assigned to patients based on the inverse probability of receiving one treatment or another, as estimated by the propensity score. For each treatment, the sIPTW was calculated as the inverse of the propensity score associated with the treatment, multiplied by the marginal probability of receiving the treatment.

After matching, the balance between the populations and pseudopopulations was checked using standardized mean differences.

Time-to-event end points in the different groups were compared using log-rank tests and Cox proportional hazard regression. Patients who were lost to follow-up (eg, all the patients for whom an outcome was not updated for >1 year at the time of the final analysis) who did not have a PFS event had their data censored at the time of their last visit. The incidence of progression or lymphoma-related death within 24 months from treatment initiation (POD24 events) was compared between the O and R groups using the PSM-matched population of LyMa-ITT and LyMa-101.

These 2 prospective trials were performed according to the principles of the Declaration of Helsinki, and the protocols were approved by ethics committees.

The 5-year PFS and OS estimates were 83.4% (95% confidence interval, 73.5-89.9) and 86.9% (95% confidence interval, 77.6-92.5), respectively (Figure 1B). Twelve patients died. The causes of death were MCL in 5 cases (42%), COVID-19 in 3 cases (25%, during OM), myocardial infarction in 1 case, related to second-line treatment in 1 case, and unknown in 2 cases, without disease progression. Seventy-four patients presented with an adverse event (AE) of at least grade 3 or higher (total 809 AEs). A total of 52.3% of AEs occurred during induction, 31% during ASCT, and 16.7% during OM (Table 1; supplemental Table 2A). Notably, as an exploratory analysis, treatment with oxaliplatin within the induction regimen had no impact on PFS as compared with treatment with caroplatine/cisplatine, with a 5-year PFS of 79% (95% CI, 63.5-88.5) vs 87.9% (95% CI, 73.3-94.8) for patients treated with oxaliplatin vs caroplatine/cisplatine, respectively (P = .38; hazard ratio [HR], = 0.631; 95% CI, 0.224-1.780; supplemental Figure 2).

Of 86 included patients, 40 (46.5%) had premature treatment discontinuation (mean time after inclusion: 16.2 months [median, 10.4]), 57.5% during the OM phase (n = 23). Most were due to AE (n = 25, 62.5%) and 5 were due to progressive disease (12.5%; supplemental Table 2).

During the OM phase, all evaluable patients reached MRD negativity at months 6, 12, 24, 30, and 36, and only 1 patient had a positive MRD at M18 (1/48 evaluable patients, 2.4%). During the OM on-demand phase (ie, after the end of the 3 years of OM planned in the protocol), positive MRD results (>10-4) were detected in 5 patients. MRD positive time points were at months 15, 18, 21, 24, and 30 (no positive sample detected at months 3, 6, 9, and 12). OM was then restarted (ie, on demand) for 4 of these patients, resulting in a negativation of MRD results in 3 of them. The patient with persistent MRD remained in a clinical complete response (CR) and under OM.

Eighty-four of the 86 and 296 of the 299 patients included in the LyMa-101 and LyMa protocols were eligible for matching, respectively (supplemental Figure 1). Matching based on propensity score resulted in a total population of 252 patients eligible for response comparison (PS set), of whom 168 were treated with R-DHAP and 84 with O-DHAP (Table 2). After matching, the absolute standardized mean difference values were <0.1 for all matching covariates (supplemental Table 3). The ORR were 90.5% and 91.7% for PSM patients treated with R-DHAP and O-DHAP, respectively (Table 3). The incidence of primary refractory disease (ie, stable or progressive disease) was higher in the R group (5.6%) than in the O group (1.3%). Notably, in LyMa-101, 10 patients were classified in partial response (PR) and not CR due to the absence of a BM biopsy at the EOI (all were in CR according to positron emission tomography). EOI MRD negativity assessed in BM was superior with O (83.1% vs 63.4%; χ², P = .007) and blood (95.3% vs 72.9%; χ², P < .001) (Table 3). Similar findings were observed using the sIPTW approach (supplemental Table 4A-C). In the sensitivity analysis, we performed PSM using the same clinical variables and ki67 (data available for 214 patients in LyMa and 68 patients in LyMa-101). The matching resulted in 2 groups of 68 patients with consistent results with regard to ORR and MRD negativity at EOI (supplemental Table 5A-D).

Seventy-eight of 120 patients who initiated RM and 59 of 69 who initiated OM were assessed for MRD before maintenance. PSM resulted in 2 sets of 43 patients (Table 4; supplemental Table 6). No difference was observed after ASCT PFS (P = .5) or OS (P = .9) between the R and O groups (Figure 2A-B). The sIPTW approach led to similar results (supplemental Table 7; supplemental Figure 3).

The matching between LyMa-ITT and LyMa-101 populations (see “Methods”; supplemental Figure 1C) resulted in 2 groups of 82 patients (namely O and R-ITT groups; Table 5). Patients treated in the O group presented a prolonged PFS (P = .029) and OS (P = .039). The estimated 5-year PFS and OS for O-ITT vs R-ITT were 82.8% vs 66.6% (HR, 1.99; 95% CI, 1.05-3.76) and 86.4% vs 71.4% (HR, 2.08; 95% CI 1.01-4.16; Figure 2C-D), respectively. The sIPTW approach yielded similar results, although OS was not statistically significant (supplemental Table 8; supplemental Figure 4A-B). Finally, the incidence of POD24 events (progression or lymphoma-related death) was 19.5% for R-treated patients and 7.2% in the O arm.

Per protocol, maintenance durations were identical (29 months for RM and 29.4 months for OM). The incidence of grade 3 to 4 neutropenia during maintenance was also comparable (44.2% in RM vs 37.6% in OM), as was the rate of grade 3 to 5 infections (12.5% vs 15.9% in RM vs OM, respectively). The rate of premature maintenance discontinuation was identical with 30.8% in R and 33% in O, who stopped maintenance before 3 years (supplemental Table 9). The reasons for discontinuation were AE in 15 of 37 (40.5%) vs 14 of 23 (61%) and progression in 10 of 37 (27%) and 3 of 23 (13%) in the RM and OM groups, respectively. The rate of grade 3 to 4 infusion-related reaction during induction tended to be higher in LyMa-101 than in LyMa (4.7% vs 0.7%). The median treatment intervals between the R and G-DHAP cycles were identical (21 days), such as the relative dose intensity of R and O during induction (mean 97.4% (standard deviation 5.4), median 98.6% and mean 92.1% (standard deviation 16.33), and median 100% for R and O, respectively). The incidences of POD24 (progression or lymphoma-related death during the first 2 years of treatment) were 19.5% for R-treated patients vs 7.2% in those treated with O.

Overall, lymphoma was the leading cause of death in both the LyMa and LyMa-101 protocols (42% in O and 53% in R). The rate of infectious deaths with O (n = 3, 25% of all deaths) tended to be higher than that with R (8% of all deaths), but importantly, all infectious deaths in LyMa-101 (O-treated patients) were COVID-19 related, whereas the LyMa trial was conducted before the pandemic (supplemental Table 10). Finally, as an exploratory analysis, responding patients relapsing after OM had a similar OS-2 compared with those who relapsed after RM (supplemental Figure 5).

The final LyMa-101 phase 2 trial analysis confirmed that O provided long-term disease control and a high MRD-negativity rate. Indeed, we showed the superiority of O vs R in terms of MRD negativity at EOI, reduction in the incidence of early relapse (POD24), and longer PFS and OS. The predominant benefit of O at EOI in terms of BM MRD negativity and the lack of significant difference after ASCT suggest that O provides a better quality of response than R, which is mainly due to induction.

CD20 antibodies have various antilymphoma mechanisms of action, leading to varying efficacies according to lymphoma type. Although O has demonstrated its superiority compared with R in FL,⁵ no difference was reported in diffuse large B-cell lymphoma.⁶ In MCL, the question of the best CD20 Ab has never been addressed and probably never will be. However, this is a question of great value because CD20 maintenance has been the only frontline treatment that has enhanced OS for both young and elderly patients so far. In the present update of the LyMa-101 trial, the 5-year PFS and OS were 83.4% and 86.9%, respectively, which compares favorably with studies conducted in the same transplant-eligible populations with MCL.^4,12,13 We observed a lower incidence of POD24 events for patients treated with O than for those treated with R, which is in line with the results reported in FL,¹⁴ and this seems to translate into early relapse reduction and may prolong OS. In this study, the use of oxaliplatin in LyMa-101 did not translate into longer disease control. This might be due to the use of O during induction, which could have erased this gain.¹⁵ The safety profiles of R and O during treatment appear to be similar, as reported in FL.⁵ The O-based strategy led to a few more discontinuations due to AE (including infusion reaction during induction) than with R; this did not convert into an excess in toxic deaths in LyMa-101 compared with LyMa. Infection-related deaths in the LyMa-101 study, in which all patients received O-DHAP and OM, were COVID-19 related, suggesting that the benefit/risk balance for CD20 maintenance should probably be carefully addressed for patients at a high risk of severe infections. Overall, our study supports the use of O-based therapy rather than R-based in MCL, but it has some limitations. Indeed, it is a comparison between 2 trials that were not designed to be compared, and some parameters, such as Ki67 and p53 status, could not be included due to missing data. The fact that data were prospectively collected, follow-up performed in a clinical prospective, and the use of PSM should limit the bias between the 2 groups. Finally, the designs of the 2 trials are not perfectly comparable given that MRD was a primary end point in LyMa-101, whereas it was exploratory in LyMa with on-demand maintenance after 3 years. However, despite the discrepancies between the 2 studies regarding maintenance, our results support the benefit of O vs R during the induction phase, but not during maintenance.

In the 2 trials that we compared in the present work, ASCT was used to consolidate responses at EOI. Recently, the emergence of new therapies for MCL has started to challenge the benefits of ASCT. The initial results of the TRIANGLE trial⁴ conducted by the European Mantle Cell Lymphoma Network, suggest that adding a Bruton tyrosine kinase inhibitor to chemotherapy could avoid the need for ASCT. Moreover, combinations of targeted therapies challenge the use of standard chemotherapy including for treatment-naïve patients, such as those with chronic lymphocytic leukemia. Indeed, the ibrutinib, venetoclax (V) plus O¹⁶ or the acalabrutib VR,¹⁷ zanubrutinib VO combinations have shown very promising response rates with MRD-negativity rates >90% when used frontline, including for high-risk patients presenting with p53 abnormalities.¹⁸ These triplet chemotherapy-free treatments are currently compared with standard chemotherapy in phase 2 or 3 randomized trials in the elderly populations with MCL (ISRCTN11038174 and NCT04002297). It is interesting to note that all these studies included anti-CD20 antibodies with or without maintenance, which underlines that anti-CD20 Abs remain pivotal for the treatment of patients with MCL; thus, the question of the best anti-CD20 remains of great value, including in the chemotherapy-free era. This study, in addition to in vivo and in vitro models, supports O as the most effective CD20 Ab in MCL.⁷ 

In conclusion, O might be considered as the anti-CD20 of choice during induction and maintenance in MCL, as O enhances the response rate at the molecular level and prolongs PFS and OS without jeopardizing safety. This will require further investigation.

This study was funded by Roche France.

Roche France had no role in the writing of the manuscript or the decision to submit it for publication.

Contribution: C.S., S.L.G., O.H., and M.C. designed the study; C.S., M.C., C.T., L.O., B.B., K.B., G.D., B.T., V.R., R.H., F.M., V.C., V.D., V.S., R.G., M.C., M.-H.D.-L., O.H., E.M., and S.L.G. performed the data collection; C.S., S.L.G., S.G., C.J., M.-H.D.-L, M.C., and E.M. analyzed and interpreted the data; C.S., S.L.G., O.H., E.M., and M.C. wrote the manuscript; all authors confirm that they had full access to all the data in the study and accept responsibility to submit for publication; all the authors performed the literature search and reviewed the manuscript; and all the authors edited and agreed to submission.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Steven Le Gouill, Institut Curie, 35 rue Dailly, 92180 Saint Cloud, France; email: steven.legouill@curie.fr.