COMMIT
COMMIT: A Randomized Study of mFOLFOX6/Bevacizumab/Atezolizumab or Atezolizumab Alone as First-Line Treatment of Deficient DNA Mismatch Repair Metastatic Colorectal Cancer
Phase 3N = 102NCT02997228
- Population
- Adults with measurable metastatic colorectal cancer, no prior chemotherapy for metastatic disease, locally determined dMMR or MSI-H status and ECOG performance status 0 to 2. Enrolled November 2017 to March 2025: 102 patients, including 20 in the closed mFOLFOX6 plus bevacizumab arm; 82 were randomised to atezolizumab alone (41) or the combination (41). Median age 63.3 years; 23.2% had a BRAF V600E mutation.
- Design
- Open-label phase 3 trial (NRG-GI004/SWOG-S1610), run in North America, in metastatic colorectal cancer with dMMR or MSI-H status by local testing (no central review). Patients were first randomised 1:1:1 to mFOLFOX6 plus bevacizumab, atezolizumab alone or all three. The mFOLFOX6 plus bevacizumab arm closed after 20 patients (June 2020 amendment, after KEYNOTE-177), leaving a 1:1 comparison of atezolizumab alone with mFOLFOX6, bevacizumab and atezolizumab. Stratified by BRAF V600E status, liver-only versus extrahepatic disease and prior adjuvant chemotherapy. No crossover. Primary endpoint PFS (randomisation to progression, symptomatic deterioration or death) in the ITT population; tumour assessments by investigators using RECIST v1.1, without central radiology review. The target sample size for the two arms was cut from 211 to 100 (April 2023); accrual was suspended on 31 March 2025 after 82 were randomised, and the study was closed permanently after a July 2025 interim analysis. From April 2023, one cycle of FOLFOX or CAPOX with or without bevacizumab before enrolment was allowed (11 patients, 13.4% of the 82, enrolled afterwards). The paper assigns no control or experimental roles. The roles recorded here follow the comparison it reports (the hazard ratio is for the combination against atezolizumab alone, and the forest-plot legend reads values below 1 as favouring the combination), so atezolizumab alone is the reference arm; the registry instead lists mFOLFOX6 plus bevacizumab as the active comparator and both atezolizumab-containing arms as experimental.
- Primary endpoint
- PFS
- Arms
- mFOLFOX6 plus bevacizumab (arm closed early) (control, n = 20): mFOLFOX6 (oxaliplatin 85 mg/m2, leucovorin 400 mg/m2, fluorouracil 400 mg/m2 bolus then 2400 mg/m2 over 46 hours) plus bevacizumab 5 mg/kg
- Atezolizumab alone (control, n = 41): Atezolizumab 840 mg IV every 2 weeks, stopped after cycle 48
- mFOLFOX6, bevacizumab and atezolizumab (experimental, n = 41): mFOLFOX6 plus bevacizumab as in the first arm (oxaliplatin stopped after cycle 10) plus atezolizumab 840 mg IV every 2 weeks, stopped after cycle 48
- People
- Rocha Lima CMSP, NRG Oncology / SWOG
PFSHR 0.439 (95% CI 0.23–0.84)PFSHR 0.42 (95% CI 0.22–0.80)Δ median PFS+19.2 monthsΔ 12-month PFS+31.6 pointsΔ ORR+40.1 pointsOSHR 1.04 (95% CI 0.47–2.28)
Show the numbers
| Endpoint | Analysis | mFOLFOX6, bevacizumab and atezolizumab | mFOLFOX6 plus bevacizumab (arm closed early) / Atezolizumab alone | Effect |
|---|---|---|---|---|
| PFSITT, patients randomised to the two compared arms (41 per arm)Progression-free survival, mFOLFOX6/bevacizumab/atezolizumab versus atezolizumab alone (primary endpoint): randomisation to progression, symptomatic deterioration or death | interimRocha 2026 | — | — | HR 0.439 (95% CI 0.23–0.84)P=0.0103Interim analysis reviewed by the data monitoring committee in July 2025: data through 31 March 2025, 45 events (65% information), crossing the prespecified O'Brien-Fleming efficacy boundary (one-sided P = 0.01523 in the full text; the abstract gives the critical value as 0.0152). The text compares the P value with that one-sided boundary but does not state whether the P value itself is one-sided or two-sided. The abstract places this hazard ratio next to a median follow-up of 46 months, which the full text attributes to the later analysis (cutoff 6 November 2025). |
| PFSITT, patients randomised to the two compared arms (41 per arm)Progression-free survival, mFOLFOX6/bevacizumab/atezolizumab versus atezolizumab alone (primary endpoint), updated analysis | primarymedian follow-up 46 monthsRocha 2026 | median 24.5 months; 66.7% at 12 mo; 53.7% at 2 y | median 5.3 months; 35.1% at 12 mo; 31.6% at 2 y | HR 0.42 (95% CI 0.22–0.80)P=0.0068Updated analysis, cutoff 6 November 2025, median follow-up 46 months (95% CI 35 to 52); stratified Cox model. The paper reports these results under its 'Primary Efficacy Endpoint' heading, after the interim analysis (recorded as a separate result). Median PFS 95% CIs as published: 10.1 to not estimable (combination) and 2.2 to 18.2 (atezolizumab alone); a median with an open upper bound cannot be shown as an interval here. The text does not say whether this P value is one-sided or two-sided (the design used a one-sided type 1 error of 0.05). An inverse-probability-weighted sensitivity analysis was described as consistent. |
| ORRResponse-evaluable patients: 36 with the combination and 37 with atezolizumab aloneObjective response rate by investigator-assessed RECIST v1.1 (secondary endpoint) | primaryRocha 2026 | 86.1% | 46% | Secondary endpoint; the data cutoff for this value is not stated. Complete response 36.1% vs 18.9%, partial response 50.0% vs 27.0%, progressive disease as best response 2.8% vs 32.4%. The 12-month disease control rate was 64.7% vs 32.4% (P = 0.009). No test of the ORR difference is reported. |
| OSITT, patients randomised to the two compared arms (41 per arm)Overall survival (secondary endpoint; the trial was not powered for survival) | primaryRocha 2026 | 67% at 2 y | 67% at 2 y | HR 1.04 (95% CI 0.47–2.28)P=0.90Not powered for survival; the report does not state the data cutoff for this value. Crossover was not allowed and post-progression therapy was not restricted. Subsequent anti-cancer therapy was given to 50% of patients on the combination and 70% on atezolizumab alone; later chemotherapy 8% vs 35%, later immunotherapy 42% vs 35% (combination vs atezolizumab alone). |
| Grade 3 or worse adverse eventsAll 41 patients per arm as tabulated in Table 2 (the report does not define the safety population)Adverse events of grade 3 or higher (CTCAE v4.0), regardless of attribution | primaryRocha 2026 | 82.9% | 43.9% | Grade 3 or worse events in 34 of 41 patients on the combination and 18 of 41 on atezolizumab alone; the data cutoff is not stated. Grade 3 or worse neutrophil count decrease 26.8% vs 0.0%, hypertension 19.5% vs 2.4% and sepsis 9.7% vs 0.0% (combination vs atezolizumab alone). Five grade 5 events: four with the combination and one with atezolizumab alone (disease progression). Median treatment duration 11 months with the combination and 5 months with atezolizumab alone. |
| Endpoint | Analysis | mFOLFOX6, bevacizumab and atezolizumab | mFOLFOX6 plus bevacizumab (arm closed early) / Atezolizumab alone | Effect |
|---|---|---|---|---|
| PFSITT, patients randomised to the two compared arms (41 per arm)Progression-free survival, mFOLFOX6/bevacizumab/atezolizumab versus atezolizumab alone (primary endpoint): randomisation to progression, symptomatic deterioration or death | interimRocha 2026 | — | — | HR 0.439 (95% CI 0.23–0.84)P=0.0103Interim analysis reviewed by the data monitoring committee in July 2025: data through 31 March 2025, 45 events (65% information), crossing the prespecified O'Brien-Fleming efficacy boundary (one-sided P = 0.01523 in the full text; the abstract gives the critical value as 0.0152). The text compares the P value with that one-sided boundary but does not state whether the P value itself is one-sided or two-sided. The abstract places this hazard ratio next to a median follow-up of 46 months, which the full text attributes to the later analysis (cutoff 6 November 2025). |
| PFSITT, patients randomised to the two compared arms (41 per arm)Progression-free survival, mFOLFOX6/bevacizumab/atezolizumab versus atezolizumab alone (primary endpoint), updated analysis | primarymedian follow-up 46 monthsRocha 2026 | median 24.5 months; 66.7% at 12 mo; 53.7% at 2 y | median 5.3 months; 35.1% at 12 mo; 31.6% at 2 y | HR 0.42 (95% CI 0.22–0.80)P=0.0068Updated analysis, cutoff 6 November 2025, median follow-up 46 months (95% CI 35 to 52); stratified Cox model. The paper reports these results under its 'Primary Efficacy Endpoint' heading, after the interim analysis (recorded as a separate result). Median PFS 95% CIs as published: 10.1 to not estimable (combination) and 2.2 to 18.2 (atezolizumab alone); a median with an open upper bound cannot be shown as an interval here. The text does not say whether this P value is one-sided or two-sided (the design used a one-sided type 1 error of 0.05). An inverse-probability-weighted sensitivity analysis was described as consistent. |
| ORRResponse-evaluable patients: 36 with the combination and 37 with atezolizumab aloneObjective response rate by investigator-assessed RECIST v1.1 (secondary endpoint) | primaryRocha 2026 | 86.1% | 46% | Secondary endpoint; the data cutoff for this value is not stated. Complete response 36.1% vs 18.9%, partial response 50.0% vs 27.0%, progressive disease as best response 2.8% vs 32.4%. The 12-month disease control rate was 64.7% vs 32.4% (P = 0.009). No test of the ORR difference is reported. |
| OSITT, patients randomised to the two compared arms (41 per arm)Overall survival (secondary endpoint; the trial was not powered for survival) | primaryRocha 2026 | 67% at 2 y | 67% at 2 y | HR 1.04 (95% CI 0.47–2.28)P=0.90Not powered for survival; the report does not state the data cutoff for this value. Crossover was not allowed and post-progression therapy was not restricted. Subsequent anti-cancer therapy was given to 50% of patients on the combination and 70% on atezolizumab alone; later chemotherapy 8% vs 35%, later immunotherapy 42% vs 35% (combination vs atezolizumab alone). |
| Grade 3 or worse adverse eventsAll 41 patients per arm as tabulated in Table 2 (the report does not define the safety population)Adverse events of grade 3 or higher (CTCAE v4.0), regardless of attribution | primaryRocha 2026 | 82.9% | 43.9% | Grade 3 or worse events in 34 of 41 patients on the combination and 18 of 41 on atezolizumab alone; the data cutoff is not stated. Grade 3 or worse neutrophil count decrease 26.8% vs 0.0%, hypertension 19.5% vs 2.4% and sepsis 9.7% vs 0.0% (combination vs atezolizumab alone). Five grade 5 events: four with the combination and one with atezolizumab alone (disease progression). Median treatment duration 11 months with the combination and 5 months with atezolizumab alone. |
Data table
| Endpoint | HR | 95% CI |
|---|---|---|
| PFS | 0.439 | 0.23–0.84 |
| PFS | 0.42 | 0.22–0.8 |
| OS | 1.04 | 0.47–2.28 |
- mFOLFOX6, bevacizumab and atezolizumab (experimental)
- Atezolizumab alone (control)
Data table (modelled values)
| Months | mFOLFOX6, bevacizumab and atezolizumab | Atezolizumab alone |
|---|---|---|
| 0 | 100.0% | 100.0% |
| 6 | 82.1% | 60.7% |
| 13 | 64.9% | 34.3% |
| 19 | 58.1% | 32.0% |
| 26 | 53.7% | 31.6% |