Adding an immunomodulator to standard R-CHOP therapy failed to improve progression-free survival in high-risk diffuse large B-cell lymphoma (DLBCL), an international randomized trial showed.
Although median progression-free survival (PFS) had yet to be reached, the trial did not meet the prespecified 37.5% reduction in hazard ratio (HR) with R-CHOP plus lenalidomide (Revlimid, R2-CHOP). Instead, adding the immunomodulator led to a nonsignificant 15% reduction.
Subgroup analysis suggested a benefit with R2-CHOP for patients with the highest-risk disease, reported Grzegorz S. Nowakowski, MD, of the Mayo Clinic in Rochester, Minnesota, and colleagues in the Journal of Clinical Oncology.
The study failed to confirm positive results from a phase II randomized trial published Feb. 8, which showed a 34% reduction in the PFS hazard with R2-CHOP, including a 33% reduction in high-risk activated B-cell (ABC) DLBCL. Given the stronger evidence from the phase III trial, the collective results do not support a change in practice, said Nowakowski, who was the lead author of both reports.
“The randomized phase II study was designed as a signal-seeking study, and it showed a positive signal in terms of improvement in progression-free survival and a trend for improvement in overall survival [OS],” he told MedPage Today. “Why the studies had discordant results is the subject of much debate.”
Differences in the study designs might account for the discrepant results, Nowakowski continued. Patients in the phase II study received a higher dose of lenalidomide for a shorter treatment duration. Additionally, the phase II ECOG-ACRIN E1412 trial enrolled an unselected patient population, whereas the phase III ROBUST trial used biomarker analysis to limit enrollment to patients with the ABC subtype.
Phase III Data Wins the Day
An accompanying editorial enumerated the same factors as possible explanations for the conflicting results of E1412 and ROBUST, and added the additional lead time required for biomarker analysis for the phase III trial as another possible factor. The resulting 10-day difference in the start of therapy might have resulted in a more favorable-risk patient population for E1412.
“Fundamentally, the difference lies not in execution of the two studies but in their design,” wrote Andrew J. Davies, PhD, of the University of Southampton in England. “E1412 is less constrained by the rigors of phase III design.”
The design of E1412 “plays to our belief that the intervention will be beneficial and negate the risk of harm. This is not an unreasonable position to take in an exploratory study based on the preliminary observations of R2-CHOP. The conclusion from such a randomized phase II study is less definitive, and the observer may mistakenly take home the message that there is substantial evidence for a treatment effect in the experimental arm.”
“For now, the take-home conclusion for the community is that there remains a lack of strong evidence to support the use of lenalidomide in combination with R-CHOP in patients with ABC-DLBCL,” Davies concluded.
Both trials tested the hypothesis that the R2-CHOP regimen would improve PFS as compared with R-CHOP in newly diagnosed DLBCL. ROBUST involved investigators in 21 countries, who randomized 570 patients with ABC-DLBCL and planned treatment with assigned therapy for a maximum of six cycles. Most patients in both arms completed the six cycles.
The trial had the statistical power to detect an HR of 0.625 for R2-CHOP versus R-CHOP. The primary analysis yielded an HR of 0.85 (95% CI 0.63-1.14). The 2-year PFS was 67.2% for the R2-CHOP group and 64% for the R-CHOP arm.
ROBUST also failed to meet the key secondary endpoint of event-free survival (HR 1.04, 95% CI 0.80-1.34). Though OS data remained immature, the 2-year OS was 79% with R2-CHOP and 80% with R-CHOP. Both regimens achieved objective response rates of 91%, including complete responses in 69% of the R2-CHOP cohort and 65% with R-CHOP.
Exploratory analyses yielded trends in 2-year PFS favoring R2-CHOP in patients with international prognostic index scores ≥3 (59% vs 50%, P=0.09), nonbulky disease (73% vs 66%, P=0.05), and lower baseline creatinine clearance (69% vs 45%, P=0.03).
Grade ≥3 treatment-emergent adverse events (TEAEs) occurred in 78% and 71% of patients randomized to R2-CHOP and R-CHOP, respectively, and rates of serious TEAEs were 37% and 31%.
Conducted entirely in North America, the E1412 trial involved 349 patients with newly diagnosed DLBCL. The study population consisted of 94 patients with ABC subtype, 122 with germinal center subtype, 18 with unclassified subtype, and 46 with unknown subtype.
Beyond the 33%-34% reduction in the PFS hazard, R2-CHOP led to a 3-year PFS of 73% versus 61% for R-CHOP (P=0.03). The OS at 3 years was 83% with R2-CHOP and 75% with R-CHOP (P=0.05). Response rates did not differ significantly (97% vs 92%), including complete response (73% vs 68%).
The ECOG-ACRIN E1412 trial was supported by the National Cancer Institute. The ROBUST trial was supported by Celgene.
Nowakowski disclosed relevant relationships with Celgene, MorphoSys, Genentech, Selvita, Debiopharm Group, Kite/Gilead, and NanoString Technologies.
Davies disclosed relevant relationships with Gilead Sciences, Roche, Janssen, Celgene, Acerta Pharma, Kite, MorphoSys, Regeneron, Bayer, GlaxoSmithKline, Takeda, Pfizer, Karyopharm, and ADC Therapeutics.