CONFIRM
Nivolumab versus placebo in patients with relapsed malignant mesothelioma (CONFIRM): a multicentre, double-blind, randomised, phase 3 trial
Phase 3N = 332NCT03063450
- Population
- Adults with histologically confirmed pleural or peritoneal mesothelioma of any histological subtype who had received at least one course of platinum-based chemotherapy and then had radiological progression; ECOG performance status 0 or 1; 316 of 332 (95%) had pleural disease and 322 (97%) had previously received pemetrexed. The authors state that most patients were in the third-line setting.
- Design
- Multicentre, double-blind, placebo-controlled, parallel-group phase 3 trial at 24 UK hospitals. Randomised 2:1, stratified by epithelioid versus non-epithelioid histology, to nivolumab at a flat dose of 240 mg or placebo every 2 weeks intravenously until progression or a maximum of 12 months. Co-primary endpoints investigator-assessed PFS and OS, analysed by the treatment-policy estimand (described by the authors as equivalent to the intention-to-treat principle). PFS was added as a co-primary endpoint on 14 February 2019 (protocol amendment 6), with a hierarchical testing procedure and an alpha of 0.04 kept for OS. Recruitment 10 May 2017 to 30 March 2020. The 2021 paper reports a preliminary analysis (database lock 4 January 2021, 210 deaths, 72% of the 291 deaths required for the final analysis); the 2025 paper reports the final analysis with median follow-up 37.2 months.
- Primary endpoint
- PFS, OS
- Arms
- Nivolumab (experimental, n = 221): Nivolumab 240 mg flat dose every 2 weeks for up to 12 months
- Placebo (control, n = 111): Placebo every 2 weeks for up to 12 months
- People
- Fennell DA
OSHR 0.69 (95% CI 0.52–0.91)Δ median OS+3.3 monthsPFSHR 0.67 (95% CI 0.53–0.85)Δ median PFS+1.2 monthsOSHR 0.81 (95% CI 0.64–1.04)Δ median OS+2.72 months
Show the numbers
| Endpoint | Analysis | Nivolumab | Placebo | Effect |
|---|---|---|---|---|
| OSRandomised patients (treatment-policy estimand, described as equivalent to ITT) | interimmedian follow-up 11.6 monthsFennell 2021 | median 10.2 months (8.5–12.1) | median 6.9 months (5–8) | HR 0.69 (95% CI 0.52–0.91)P=0.0090Adjusted HR (Cox model adjusted for epithelioid histology, a stratification factor). Preliminary analysis at database lock on 4 January 2021: 134 deaths (61%) of 221 with nivolumab and 76 (68%) of 111 with placebo, 72% of the 291 deaths required for the final analysis. Median follow-up 11.6 months (IQR 7.2-16.8). One-year OS 43.4% (95% CI 36.3-50.4) with nivolumab versus 30.1% (21.0-39.6) with placebo. The final analysis (2025 report) is recorded as a separate result. |
| PFSRandomised patients (treatment-policy estimand, described as equivalent to ITT)Investigator-assessed progression-free survival | interimmedian follow-up 11.6 monthsFennell 2021 | median 3 months (2.8–4.1) | median 1.8 months (1.4–2.6) | HR 0.67 (95% CI 0.53–0.85)P=0.0012Adjusted HR (adjusted for epithelioid histology). Preliminary analysis; investigator assessment without central radiology review, which the authors list as a limitation. One-year PFS 14.2% (95% CI 9.9-19.3) with nivolumab versus 7.2% (3.1-13.8) with placebo. |
| OSRandomised patients (treatment-policy estimand, described as equivalent to ITT) | finalmedian follow-up 37.2 monthsFennell 2025 | median 9.49 months (7.69–11.01) | median 6.77 months (5.03–7.92) | HR 0.81 (95% CI 0.64–1.04)P=0.096Final analysis planned at 291 events: 198 of 221 deaths with nivolumab and 100 of 111 with placebo. Adjusted HR 0.81 (95% CI 0.64-1.04), p=0.096; unadjusted HR 0.81 (0.64 to 1.04), p=0.093. Median follow-up with censoring at death 37.2 months (IQR 30.3-44.4). The authors state that the PFS endpoint was met but the OS endpoint was not. A post hoc restricted mean survival time analysis at 12 months reported 1.1 months longer average survival with nivolumab (95% CI 0.2 to 2.0; p=0.015). The 2021 preliminary result is recorded as a separate result. |
| PFSRandomised patients (treatment-policy estimand, described as equivalent to ITT)Investigator-reported progression-free survival | finalmedian follow-up 37.2 monthsFennell 2025 | median 2.89 months (2.76–4.11) | median 1.64 months (1.38–2.56) | HR 0.65 (95% CI 0.51–0.82)Unadjusted HR; 216 of 221 events with nivolumab and 108 of 111 with placebo. The adjusted HR and p value are in the supplementary note, which was not read. |
| ORRPatients with a response assessment (195 nivolumab, 102 placebo)Objective response rate by modified RECIST, recomputed centrally by the trials unit, in patients with a response assessment | finalFennell 2025 | 10.3% | 0% | 20 of 195 patients with nivolumab and 0 of 102 with placebo. The authors state that the recomputed rate is 1% lower than the investigator-reported rate (11%) in the preliminary publication. |
| Endpoint | Analysis | Nivolumab | Placebo | Effect |
|---|---|---|---|---|
| OSRandomised patients (treatment-policy estimand, described as equivalent to ITT) | interimmedian follow-up 11.6 monthsFennell 2021 | median 10.2 months (8.5–12.1) | median 6.9 months (5–8) | HR 0.69 (95% CI 0.52–0.91)P=0.0090Adjusted HR (Cox model adjusted for epithelioid histology, a stratification factor). Preliminary analysis at database lock on 4 January 2021: 134 deaths (61%) of 221 with nivolumab and 76 (68%) of 111 with placebo, 72% of the 291 deaths required for the final analysis. Median follow-up 11.6 months (IQR 7.2-16.8). One-year OS 43.4% (95% CI 36.3-50.4) with nivolumab versus 30.1% (21.0-39.6) with placebo. The final analysis (2025 report) is recorded as a separate result. |
| PFSRandomised patients (treatment-policy estimand, described as equivalent to ITT)Investigator-assessed progression-free survival | interimmedian follow-up 11.6 monthsFennell 2021 | median 3 months (2.8–4.1) | median 1.8 months (1.4–2.6) | HR 0.67 (95% CI 0.53–0.85)P=0.0012Adjusted HR (adjusted for epithelioid histology). Preliminary analysis; investigator assessment without central radiology review, which the authors list as a limitation. One-year PFS 14.2% (95% CI 9.9-19.3) with nivolumab versus 7.2% (3.1-13.8) with placebo. |
| OSRandomised patients (treatment-policy estimand, described as equivalent to ITT) | finalmedian follow-up 37.2 monthsFennell 2025 | median 9.49 months (7.69–11.01) | median 6.77 months (5.03–7.92) | HR 0.81 (95% CI 0.64–1.04)P=0.096Final analysis planned at 291 events: 198 of 221 deaths with nivolumab and 100 of 111 with placebo. Adjusted HR 0.81 (95% CI 0.64-1.04), p=0.096; unadjusted HR 0.81 (0.64 to 1.04), p=0.093. Median follow-up with censoring at death 37.2 months (IQR 30.3-44.4). The authors state that the PFS endpoint was met but the OS endpoint was not. A post hoc restricted mean survival time analysis at 12 months reported 1.1 months longer average survival with nivolumab (95% CI 0.2 to 2.0; p=0.015). The 2021 preliminary result is recorded as a separate result. |
| PFSRandomised patients (treatment-policy estimand, described as equivalent to ITT)Investigator-reported progression-free survival | finalmedian follow-up 37.2 monthsFennell 2025 | median 2.89 months (2.76–4.11) | median 1.64 months (1.38–2.56) | HR 0.65 (95% CI 0.51–0.82)Unadjusted HR; 216 of 221 events with nivolumab and 108 of 111 with placebo. The adjusted HR and p value are in the supplementary note, which was not read. |
| ORRPatients with a response assessment (195 nivolumab, 102 placebo)Objective response rate by modified RECIST, recomputed centrally by the trials unit, in patients with a response assessment | finalFennell 2025 | 10.3% | 0% | 20 of 195 patients with nivolumab and 0 of 102 with placebo. The authors state that the recomputed rate is 1% lower than the investigator-reported rate (11%) in the preliminary publication. |
Data table
| Endpoint | HR | 95% CI |
|---|---|---|
| OS | 0.69 | 0.52–0.91 |
| PFS | 0.67 | 0.53–0.85 |
| OS | 0.81 | 0.64–1.04 |
| PFS | 0.65 | 0.51–0.82 |
- Nivolumab (experimental)
- Placebo (control)
Median-only curves assume a constant hazard (exponential) and fit plateaus, as seen with some immunotherapies, poorly.
Data table (modelled values)
| Months | Nivolumab | Placebo |
|---|---|---|
| 0 | 100.0% | 100.0% |
| 6 | 66.5% | 54.7% |
| 13 | 41.3% | 27.1% |
| 19 | 27.5% | 14.8% |
| 26 | 17.1% | 7.3% |
- Nivolumab (experimental)
- Placebo (control)
Median-only curves assume a constant hazard (exponential) and fit plateaus, as seen with some immunotherapies, poorly.
Data table (modelled values)
| Months | Nivolumab | Placebo |
|---|---|---|
| 0 | 100.0% | 100.0% |
| 2 | 63.0% | 46.3% |
| 4 | 39.7% | 21.4% |
| 6 | 25.0% | 9.9% |
| 8 | 15.7% | 4.6% |
Appears in
Sources
Fennell DA, Ewings S, Ottensmeier C, Califano R, Hanna GG, Hill K, et al. Nivolumab versus placebo in patients with relapsed malignant mesothelioma (CONFIRM): a multicentre, double-blind, randomised, phase 3 trial. Lancet Oncol. 2021;22(11):1530-1540.
Fennell DA, Hill K, Zhang M, Poile C, Ewings S, Baitei EY, et al. Constitutive inflammation and epithelial-mesenchymal transition dictate sensitivity to nivolumab in CONFIRM: a placebo-controlled, randomised phase III trial. Nat Commun. 2025;16(1):6688.