Randomized study of induction with bendamustine-rituximab ± bortezomib and maintenance with rituximab ± lenalidomide for MCL Free

Mantle cell lymphoma (MCL) is an uncommon B-cell lymphoma with a variable but often aggressive course.¹ Initial therapy for MCL is commonly selected based on patient age and fitness as determining appropriateness of an intensive immunochemotherapy approach including high-dose therapy with autologous stem cell support or a less intensive immunochemotherapy regimen.² Because the median age at diagnosis is in the seventh decade of life, less intensive approaches apply to the majority of patients with MCL. Over the past decade, the combination of bendamustine plus rituximab (BR) has become the dominant therapy for patients with MCL not eligible or desiring intensive therapy, based on efficacy and tolerability.^3-5 Unfortunately, the response rate to therapy is high but not universal, and eventual recurrence is expected. Adding novel agents to BR represents 1 strategy to improve outcomes, with multiple such trials underway or having been completed. Among the more promising agents for a BR-based combination is bortezomib, approved in 2006 by the US Food and Drug Administration for previously treated MCL.⁶ Moreover, substituting bortezomib for vincristine in the R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) regimen (VR-CAP) resulted in an improvement in not only progression-free survival (PFS) but also overall survival (OS).⁷ In 2 phase 2 trials, the combination of BR plus bortezomib demonstrated promising efficacy without a clear impact on safety.^8,9 Given these promising data, we hypothesized that the addition of bortezomib to BR induction (BVR) would generate higher complete response rates and longer PFS compared with BR alone.

Because induction therapy is not curative, maintenance with anti-CD20–targeted therapies has been investigated to consolidate remission.^10-14 Maintenance rituximab (R) after R-CHOP induction prolonged PFS and OS compared with interferon in a randomized trial in older patients with MCL,¹³ and also prolonged PFS and OS in younger patients after autologous stem cell transplant (ASCT).¹⁴ Considering options to improve upon R maintenance outcomes, lenalidomide and R may potentiate each other’s activity and can be safely coadministered. The combination is active in relapsed^15,16 as well as in frontline MCL therapy.¹⁷ Given the known efficacy of R maintenance and the promising activity of the LR regimen, we hypothesized that the addition of lenalidomide to rituximab (LR) would improve PFS compared with R alone.

E1411 was an open-label, randomized phase 2 trial for adult patients with treatment-naïve, histologically confirmed MCL requiring therapy. Although initially restricted to patients aged >60 years, after the early closure of S1106,¹⁸ eligibility was amended to include all ages >18 years. At full study accrual, only 13% of patients were aged <60 years. Patients were required to have measurable disease (at least 1 lesion of >1.5 cm), although this could include splenomegaly as the sole evaluable site if therapy was indicated. In addition, Eastern Cooperative Oncology Group performance status score of 0 to 2, adequate organ function (absolute neutrophils of ≥1.5 × 10³/μL, platelet count of ≥100 × 10³/μL, total bilirubin and transaminase levels less than or equal to twice the upper limits of normal, and calculated creatinine clearance of ≥30 mL/min), no clinical evidence of central nervous system involvement, no baseline neuropathy of grade ≥2, and no contraindication to thrombosis prophylaxis required for lenalidomide administration. The study was approved by each relevant institutional review board, and all participants provided written informed consent.

Positron emission tomography (PET) scan was mandated for initial staging, and repeated for response evaluation after cycle 3 and again at end of induction. If the end-of-induction PET remained positive, PET was to be repeated after cycle 2 (8 weeks) of maintenance. Bone marrow aspirate and biopsy was part of initial staging evaluation, and was mandated at end of induction for any patient whose on study marrow was involved with lymphoma and who otherwise met criteria for complete remission (CR), to confirm CR assessment. Response criteria were as per International Working Group 2007 criteria.

Follow-up visits during maintenance were every 2 cycles. Imaging during maintenance for PET-negative patients could be either PET/computed tomography or computed tomography at investigator discretion, performed every 6 months. After completion of therapy, imaging was every 6 months for 3 more years, then annually for 5 years or until documentation of disease progression or death. Imaging after the 5 years after end of maintenance (∼7.5 years from study entry) was as clinically indicated.

Patients were randomized to 1 of 4 arms (Figure 1A) consisting of an induction (step 1) phase (BR or BVR) and a maintenance (step 2) phase (R or LR). Randomization was stratified according to MCL International Prognostic Index (MIPI) risk status (low, intermediate, or high) and age (<60 or ≥60 years) using permuted blocks and dynamic balancing on main ECOG-ACRIN Cancer Research Group institutions plus affiliates. Partway through enrollment (after ∼245 patients had enrolled), it was recognized that some patients had been stratified according to a miscalculated MIPI score based on wrong white blood cell units, resulting in misclassification of MIPI risk status. The corrected MIPI¹⁹ is reported here, and sensitivity analysis for the primary end point resulted in similar study conclusions.

Arm A (BR→R) included induction with BR and maintenance with R. Arm B (BVR→R) included BVR followed by maintenance with R. Arm C (BR→LR) included induction with BR followed by maintenance with addition of lenalidomide to R (LR). Arm D (BVR→LR) included induction with BVR followed by LR maintenance (Figure 1A). Bendamustine was administered at 90 mg/m² IV on days 1 and 2 along with R 375 mg/m² administered IV on day 1 of each 28-day cycle for 6 cycles. Bortezomib administered IV or subcutaneously was initially scheduled as 1.3 mg/m² on days 1, 4, 8, and 11, but this was amended (in December 2013 at 18% of final study accrual) to 1.6 mg/m² on days 1 and 8 of each induction cycle to reduce the number of required patient visits. Most patients received the amended dose schedule, 128 of 165 had subcutaneous treatment only, and a separate (sensitivity) analysis for differences in bortezomib administration was not performed. Bortezomib for the study was supplied by Millennium (now Takeda). After restaging after induction, patients responding or with stable disease reregistered to protocol step 2 and then received R maintenance once every 8 weeks for 12 doses and, for arms C and D, lenalidomide 15 mg orally on days 1 to 21 of each 28-day cycle for 24 cycles.^13,14 Note that even the younger patients in E1411 did not receive ASCT consolidation, which, despite its inclusion in many treatment recommendations for younger patients, is not frequently applied in the United States.³ Lenalidomide for the study was supplied by Celgene (now Bristol Myers Squibb). All participants randomized to receive lenalidomide were required to enroll in the RevAssist program and have thrombosis prophylaxis with aspirin, or a full anticoagulation regimen if indicated. Supportive care, including growth factors and antimicrobials, were at investigator discretion.

The coprimary objectives of the study were to determine whether PFS was improved by the addition of bortezomib to BR induction (BVR vs BR) and by the addition of lenalidomide to R maintenance (LR vs R). We hypothesized that patients treated with BR followed by R-based maintenance would have a 2-year PFS rate of 70% (estimated from R-CHOP→R study data and preliminary data showing noninferiority of BR compared with R-CHOP).^13,20 A planned enrollment of 372 patients accrued over 45 months, 360 of whom were expected to be eligible and receive therapy, provides 93.8% statistical power to detect a 37.4% (or PFS hazard ratio [HR] of 0.626 for BVR vs BR) reduction in the PFS risk/hazard between BVR (arms B + D) and BR (arms A + C), corresponding to a 2-year PFS rate of 80% with BVR using a stratified log-rank test (based on MIPI risk, age, and step 2 treatment [LR vs R]) with a 1-sided type 1 error of 10%. Full statistical information for this PFS comparison required 149 PFS events (disease progressions or deaths) expected to occur within 23 additional months of follow-up after the end of accrual. The main secondary endpoints were PET-documented CR rate and objective response rate.

For the primary end point for the maintenance (step 2) question, the analysis pooled the 2 R maintenance arms (A + B) and the 2 LR maintenance arms (C + D) stratified by MIPI risk, age, and induction (step 1) regimen (BVR vs BR). We projected that 290 patients with a best induction treatment response of stable disease or better and also meeting other eligibility criteria would proceed to maintenance therapy. With 290 patients, there would be 89.4% power to detect a similar 37.4% reduction in PFS hazard (ie, PFS HR of 0.626 for LR vs R), with a 1-sided type 1 error of 10% using a stratified log-rank test. Full statistical information required 120 PFS events, with PFS time defined from the date of step-2 registration.

Bone marrow aspirate and blood samples at baseline, and blood after cycle 6 of induction were drawn into separate tubes and sent overnight directly to Mayo Laboratories for flow cytometry and for next-generation sequencing (NGS), initially to Sequenta, later to Adaptive Biotechnologies (Seattle, WA). Blood was also sent after cycle 3 for NGS only. Of patients with available measurable residual disease (MRD) data (ie, samples submitted, and a trackable clone identified from diagnostic tissue), a malignant clone was identified by NGS in the baseline blood or bone marrow samples in 98.2%% of cases, consistent with the known high frequency of MCL involvement in the bone marrow.

Flow cytometric analysis of baseline blood and bone marrow aspirates and end-of-induction blood for clonal MCL cells was performed as previously described for chronic lymphocytic leukemia (antibody panel against CD19, CD20, CD23, CD5, CD45, CD38, and κ and λ light chains).²¹ 

NGS assessment of residual disease was as previously described.²² Briefly, MCL clonotype specific identifying (ID) sequences were determined in diagnostic samples and then measured in follow-up MRD samples using the clonoSEQ Assay (Adaptive Biotechnologies). Once suitable MCL ID sequences were identified, these ID sequences were compared with successive MRD sample(s) for tracking.

Immunohistochemical staining for KI-67 and TP53 was performed for the subset of patients who had signed informed consent for their remaining tissue specimens to be analyzed and for whom adequate tissue remained. This staining and interpretation were carried out centrally by expert hematopathologists.

The clinical trial protocol was institutional review board approved per institutional guidelines at all participating centers.

Between August 2012 and September 2016, 373 patients were enrolled at 101 institutions (United States and Canada; supplemental Table 1, available on the Blood website). The last induction cycle was administered February 2017, and last maintenance therapy February 2019, before the COVID-19 pandemic. Patient characteristics (Table 1) included predominantly non-Hispanic White men, median age of 67 years (range, 42-90), of whom 87% were aged ≥60 years. There were no imbalances among the treatment arm with respect to patient characteristics. MIPI scores distributed as 25% low risk, 41% intermediate risk, and 34% high risk.

Patient disposition is described by the CONSORT diagram (Figure 1B). Reasons for ineligibility included non-MCL pathology in 5, and missing/out-of-range laboratory values in 4. The median number of cycles of induction completed in each arm was 6. The number of patients completing 6 cycles of BR and BVR was 155 of 179 (87%) and 149 of 179 (83%), respectively.

The best overall response rate (supplemental Table 2) was 88% with either induction regimen, with complete response rates of 59% vs 65% in BR vs BVR, respectively. The 2-year PFS was 74.8% with BR, and 79.7% with BVR (Figure 2), and 5-year PFS was 56.6% for BR and 56.2% for BVR. With a median (25th percentile, 75th percentile) survival follow-up of 7.5 years (6.5, 8.5), the estimated median PFS was 5.5 years (95% CI, 4.5-6.6) and 6.4 years (95% CI, 4.8-7.7) in the BR and BVR arms, respectively. There were no differences in the distribution of stratification variables by treatment arms. The stratified HR (using corrected MIPI score and other randomization stratification factor) was 0.90 (90% CI, 0.70-1.16); 1-sided stratified log-rank P = .251, indicating no significant benefit with addition of bortezomib.

BR with or without bortezomib, led to MRD levels in peripheral blood of <1 in 10⁴ in >90% of patients, and achieved this by the end of cycle 3 induction. Whether this was assessed by NGS or multicolor flow cytometry did not affect these observations. The few patients who did not improve to this level at midinduction or end induction had shorter PFS (not shown). A more detailed analysis of MRD by NGS at higher sensitivity levels is underway.

Patient flow by each of the 4 arms is in the CONSORT diagram (Figure 1); however, outcomes for maintenance R with or without lenalidomide were analyzed per protocol independently from induction by pooling the induction arms. The efficacy populations are 136 and 140 in the R and LR arms, respectively. The 2-year estimate of PFS was 77.8% (95% CI, 69.7-84.0) in the R arm and 85.8% (95% CI, 78.6-90.7) in the LR arm (Figure 2), with a 1-sided P value from the stratified log-rank test (using corrected MIPI and other stratification factors) of .178, indicating no statistically significant benefit for the addition of lenalidomide. Again, there were no differences when analyzed in patients aged ≥60 years (not shown), nor within MIPI cohorts (Figure 3A). Patients achieving CR during induction appeared to have longer PFS; further discussed in supplemental Materials and supplemental Figure 1.

When analyzed as 4 separate treatment arms, (Figure 2C), the 2-year estimated PFS was 77.7% (95% CI, 65.7-85.9) for BR followed by R, 78.0% (95% CI, 65.7-86.4) for BVR followed by R, 80.1% (95% CI, 68.2-88.0) for BR followed by LR, and 91.2% (95% CI, 81.4-95.9) for BVR followed by LR. The median PFS was 5.5 years (95% CI, 4.8-6.0), 6.9 years (95% CI, 4.0 to not reached), 7.3 years (95% CI, 3.9 to not reached), and 6.9 years (95% CI, 5.5-8.0), respectively.

Overall, for the entire study population with median follow-up of 7.5 years, the median PFS was 6.9 years. Analysis of the 87% of patients who fit the original study design limited to patients aged ≥60 years, median PFS was 5.5 years and 5-year PFS was 54.7%. For the 64% of E1411 patients in the commonly used age cutoff of ≥65 years, median PFS was 5.2 years and 5-year PFS was 51.5%.

Adequate tissue was evaluable for Ki-67, TP53, and SOX11 biomarkers in 138 patients. The risk factors of Ki-67 ≥ 30% observed in 18% of analyzable patients, and TP53 immunohistochemistry ≥50% observed in 11% of patients (Figure 3B) each correlated with worse outcomes for both PFS and OS. As expected in a cohort of patients with MCL requiring therapy, SOX11 positivity was 92%. The low number of patients analyzed for Ki-67 and TP53 precludes by arm comparison.

Hematologic adverse events, and others of interest occurring at a frequency of >5% on any arm, of grade ≥3 are listed in Table 2. Compared with BR, patients treated with BVR more commonly experienced grade 3/4 neutropenia but not febrile neutropenia, and grade 3 sensorimotor neuropathy. Three grade 5 adverse events occurred during induction: tumor lysis with BR and 1 event each of liver failure and cardiac arrest in BVR.

Compared with R, patients receiving LR had an increased incidence of grade ≥3 toxicities of neutropenia, (10 vs 5 events), rash (4 vs none), and fatigue (8 vs 1). Three deaths considered treatment related included myocardial infarction on LR, invasive squamous cell carcinoma on LR, and myelodysplasia on R. Patients in the LR arms did generally receive the intended treatment. Analysis (supplemental Table 3) shows that 66% of the cohort received 23 to 24 cycles, and 78% a minimum of 12 cycles. Of patients completing 23 to 24 cycles, 82% of these cycles were without dose reduction or delay.

Discontinuation of therapy before completion of the planned 24 maintenance cycles was more commonly because of adverse events in (15 vs 7) and less commonly because of progressive disease (11 vs 22) in the LR arms compared with the R alone arms.

Second malignancies by arm were: BR-R (18), BVR-R (21), BR-LR (27), and BVR-LR (34); these data represent 66 unique patients because 20 patients had >1 second malignancy. Of the 51 second cancers that were noninvasive, nonmelanoma skin cancers, 17 occurred in the R maintenance arms vs 34 in the LR arms (17 with each induction). Focusing on the invasive cancers and melanoma, there were 8 cases in the BR-R arm, 14 in BVR-R, 10 in BR-LR, and 17 in BVR-LR. These include 3 cases of MDS/AML, 1 in each arm except none in BVR-LR, as well as 3 cases coded as aggressive B-cell lymphoma, all in the BVR-R arm, which may be transformation of their MCL. Although numbers are small, there appears to be an increase in localized nonmelanoma skin cancers associated with lenalidomide inclusion in maintenance.

In this trial BR-based induction followed by 2 years of R-based maintenance was efficacious in treatment-naïve patients with MCL largely aged >60 years, yielding an objective response rate of 88%, an MRD-negative (at 10⁻⁴ sensitivity) rate at end of induction of 91%, and a median PFS of 6.9 years, with median follow-up 7.5 years. In terms of the first primary objective of this trial, addition of bortezomib was not associated with superior PFS compared with BR alone, nor with improved rate of MRD negativity or CR, whereas it was associated with an increased risk of neutropenia and sensory neuropathy. Bortezomib conferred benefit when replacing vincristine in R-CHOP⁷ as initial MCL therapy and when added to R-high dose cytarabine/dexamethasone (R-HAD)²³ in the relapsed/refractory setting. Lack of demonstrable benefit in our trial may reflect the higher baseline efficacy of the BR regimen. In fact, neither of the 2 BR-bortezomib phase 1 trials^8,9 achieved their postulated efficacy end point. Alternatively, although weekly bortezomib dosing appears effective and less toxic in the myeloma setting, it is possible that there is a dose-response effect in MCL²⁴ and the dose may not have been adequate in this trial. Finally, there may be additional, as yet unknown, biologic factors, such as drug or microenvironmental interactions, that play a role. Additive toxicity with bortezomib was not significant.

Applying molecular and flow cytometric methods to assess depth of response and to monitor disease recurrence and growth is becoming increasingly useful in B-cell malignancies. MRD in MCL has been mostly investigated as a prognostic marker in the setting of intense therapies, although samples and methods have varied. MRD is currently being studied as a potential way to make treatment decisions (EA4151). Our study compared multicolor flow cytometry and NGS at what is now considered a relatively insensitive cutoff of 10⁻⁴, and found the methods comparable. These methods are each feasible in a cooperative group setting in MCL. The high rate of MRD negativity after BR was not expected at the time of study design, indicating the efficacy of BR induction, yet precluding analysis of added efficacy of maintenance approaches. Not surprisingly, patients with deeper responses have a longer time to progression. Molecular data at a more sensitive cutoff from additional sequential samples acquired per protocol during and after maintenance will be forthcoming (manuscript in preparation). Such data should complement kinetic data after ASCT from the FIL MCL0208 trial.²⁵ 

The second question addressed by the E1411 trial tested the addition of lenalidomide to 24 months of R maintenance. LR maintenance did not improve the primary end point of PFS compared with single agent R, whereas toxicities were generally as expected. The role of R maintenance itself after BR remains a subject of debate. R maintenance after R-CHOP in older patients with higher MIPI risk profile than our study or cytarabine-based induction/ASCT consolidation resulted in an improvement in both PFS and OS. Observational data for the E1405 frontline MCL trial suggested that, after an anthracycline-based regimen also containing bortezomib, maintenance R yielded similar outcomes to ASCT.¹¹ The only randomized trial to evaluate R maintenance after BR in MCL reportedly has not shown evidence of benefit,²⁶ again pointing out potential issues when comparing maintenance results after different induction regimens. Nonetheless, the encouraging PFS results in E1411 are in line with other data, suggesting that there may be benefit of R after BR induction in MCL. In the original Study Group of Indolent Lymphomas (StiL) trial,⁴ the median PFS for patients with MCL treated with BR was ∼36 months, results similar to the subsequent BRIGHT trial.⁵ These results fall roughly 2 years short of the median PFS seen in E1411, as well as the control arm of the SHINE trial (see below). Additionally, a real world analysis from a large data set indicates maintenance R after BR improves both time to next treatment and OS.³ This remains an important question because a number of ongoing studies in MCL are based on BR and yet maintenance anti-CD20 antibody use varies. For example, the BGB-3111-306 trial comparing zanubrutinib-R with BR does not include maintenance R after BR. Also, the rituximab-bendamustine-cytarabine (R-BAC) regimen²⁷ reports encouraging long-term results by adding cytarabine to build on BR, without maintenance.

If R maintenance is of benefit, there remains the question of why there was no benefit from the addition of lenalidomide, an effective agent against MCL, particularly in combination with R at doses of 20 to 25 mg daily.^15-17 Preliminary analysis suggests that the prescribed 15 mg daily dose was largely delivered, although additional analyses will be performed, so it is possible that this is an ineffective dose in MCL. Another hypothesis is that prior therapy may alter the immune environment required for lenalidomide antilymphoma activity. In particular, T-cell depletion by bendamustine might compromise lenalidomide efficacy,²⁸ although our trial lacks data to address this.

The promising results of the overall treatment strategy studied in E1411, BR followed by R maintenance, indicate that this remains a strong backbone upon which to build, although trials become more challenging, given high clinical and molecular response rates and durable longer PFS. In light of the activity of Bruton tyrosine kinase inhibitors in MCL, their activity in combination with chemotherapy is being investigated. The SHINE trial evaluating the addition of ibrutinib to BR in older patients with MCL started after this trial began and was recently reported.²⁹ It demonstrated outcomes for the BR with R maintenance control arm similar to, perhaps slightly shorter, than those of our BR-R cohort, with median PFS of 52.9 months, in a slightly older population all aged >65 years, but with similar MIPI risk profiles. Addition of continuous ibrutinib improved the median PFS to 80.6 months, although with some added toxicity. The question remains as to whether concurrent or sequential ibrutinib use would be a better strategy. A similar study using acalabrutinib rather than ibrutinib is underway (ACE-LY-308). In patients with MCL aged <65 years, the TRIANGLE trial demonstrated that ibrutinib improved outcomes when added to high-dose cytarabine-based induction and ASCT.³⁰ Similarly, the EA4181 trial is evaluating addition of acalabrutinib to a backbone of BR plus high-dose cytarabine. These studies have the potential to yield important results, but differences in study design, as well as variable populations in terms of age distribution and MIPI risk profiles, may complicate interstudy comparisons and the treatment decisions to follow. Promising data from Bruton tyrosine kinase inhibitor–based “chemotherapy-free” approaches have been reported^31-35 and randomized trials are underway. The BGB-311-306 trial is comparing zanubrutinib-R with BR, whereas the ENRICH study is comparing ibrutinib plus R with R-chemotherapy plus 2 years of maintenance R. At present, while awaiting results of these and other studies, bendamustine-R followed by maintenance R still appears to be an appropriate initial therapy and benchmark upon which to build or as a comparator for newer regimens, which are needed particularly for high-risk patients with MCL who are not candidates for more intensive approaches.

This study was coordinated by the ECOG-ACRIN Cancer Research Group (Peter J. O'Dwyer and Mitchell D. Schnall, joint group chairs) and supported by the National Institutes of Health (NIH), National Cancer Institute (NCI) (award numbers U10CA180820, U10CA180794, UG1CA189828, U10CA180821, U10CA180888, UG1CA189863, UG1CA189957, UG1CA189971, UG1CA232760, UG1CA233234, UG1CA233277, UG1CA233320, UG1CA233328, UG1CA233331, UG1CA233339, U10CA180863, UG1CA233253, and P01CA214274), Canadian Cancer Society (grant 704970), and the Leukemia Lymphoma Society (LLS) Mantle Cell Lymphoma Research Initiative. P.M. is supported by the NIH, NCI (grant P01 CA214274) and the LLS Mantle Cell Lymphoma Research Initiative.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Contribution: M.R.S. designed research, performed research, analyzed data, and wrote the initial draft of the manuscript; O.A.J. designed research, analyzed data, and provided major input into writing/revision of the manuscript; P.M. and B.G.T. designed research, performed research, analyzed data, and provided major input into writing/revision of the manuscript; S.S.P., T.W., S.D., D.S., L.K.S., N.A.-S., C.C., N.L.B., P.F.C., T.A.B., K.A.B., M.D.R., D.J.I., R.E.L., and L.I.W. performed research; D.T.Y., E.D.H., and C.H. performed research and analyzed data; R.F.L. designed research; J.W.F. and J.P.L. designed and performed research; and B.S.K. designed research, performed research, and provided major input into writing/revision of the manuscript.

Conflict-of-interest disclosure: P.M. consulted for AstraZeneca, Bayer, BeiGene, Bristol Myers Squibb (BMS), Epizyme, Gilead, Janssen, Karyopharm, Merck, MorphoSys, Regeneron, and Takeda. B.G.T. holds a patent with/receives royalties from Mustang Bio; received research funding from Mustang Bio, BMS/Celgene; and serves as a consultant for Mustang Bio and Proteios Technologies. E.D.H. received research funding from Eli Lilly; serves as consulted for Novartis; and holds stock options in Abcon therapeutics. S.D. holds equity in Data Driven Bioscience. The remaining authors declare no competing financial interests.

Correspondence: Mitchell R. Smith, Follicular Lymphoma Foundation, 2516 Q St NW, Suite Q104, Washington, DC 20007; email: lymphomadocdoc@gmail.com.