Synthesis: The IMpassion / KEYNOTE evidence base in TNBC

Evidence grades (GRADE-adapted): A high — multiple well-conducted RCTs or systematic reviews converge. B moderate — single pivotal RCT or consistent observational evidence. C limited — single observational study, mechanistic, or expert consensus. D preclinical / hypothesis-generating.

The IO landscape in TNBC, in one paragraph

Pembrolizumab (anti-PD-1) is the only immune checkpoint inhibitor currently approved for any TNBC indication. Pembrolizumab + chemotherapy is the first-line metastatic standard for PD-L1 CPS ≥ 10 disease (KEYNOTE-355) and the neoadjuvant standard for high-risk stage II–III early-stage disease (KEYNOTE-522). Atezolizumab (anti-PD-L1) was briefly approved in metastatic disease based on IMpassion130 but was withdrawn in 2021 after the confirmatory IMpassion131 failed. Other ICIs (durvalumab, nivolumab) have had isolated positive readouts in specific settings but are not approved. The biological signal across the program is consistent — IO + chemo benefits a substantial subset of TNBC patients — but the regulatory outcomes have differed sharply based on trial-design choices, chemotherapy backbones, and biomarker assay decisions.

The pembrolizumab program (KEYNOTE)

KEYNOTE-086 (monotherapy phase II)

Adams and colleagues conducted a single-arm phase II trial of pembrolizumab monotherapy in metastatic TNBC across two cohorts: Cohort A enrolled previously treated patients regardless of PD-L1 status; Cohort B enrolled first-line PD-L1-positive patients^[1]B.

• Cohort A response rate: 5.3% (pre-treated, unselected) — modest activity
• Cohort B response rate: 21.4% (first-line PD-L1+) — substantial
• Cohort B median OS: 18.0 months

The Cohort B signal motivated the KEYNOTE-355 and KEYNOTE-119 randomized trials.

KEYNOTE-119 (monotherapy vs chemo, second-line+)

Winer and colleagues randomized 622 patients with previously treated metastatic TNBC to pembrolizumab monotherapy vs investigator's choice chemotherapy^[2]A. Pre-specified analyses examined PD-L1 CPS thresholds of 1, 10, and 20.

• Overall trial OS: 9.9 vs 10.8 months — not significant
• CPS ≥ 10 subgroup OS: 12.7 vs 11.6 months — numerical benefit, not significant
• CPS ≥ 20 subgroup OS: 14.9 vs 12.5 months — trend, hierarchy precluded testing

The trial did not meet its primary endpoint. Pembrolizumab monotherapy was not approved for any TNBC indication on the basis of KEYNOTE-119. The result reinforced the principle that IO + chemo combinations outperform IO monotherapy in TNBC.

KEYNOTE-355 (first-line metastatic combination)

Cortes and colleagues randomized 847 patients with previously untreated metastatic TNBC to pembrolizumab + chemo vs placebo + chemo^[3]A. Strong positive result in the CPS ≥ 10 subgroup (median PFS 9.7 vs 5.6 months; median OS 23.0 vs 16.1 months in the 2022 update). See the first-line metastatic synthesis for detail.

KEYNOTE-522 (neoadjuvant + adjuvant early-stage)

Schmid and colleagues randomized 1,174 patients with stage II–III TNBC to pembrolizumab + chemo → pembrolizumab adjuvant vs placebo + chemo → placebo adjuvant^[4]A. The pCR primary endpoint (64.8% vs 51.2%) and the EFS update (3-year EFS 84.5% vs 76.8%; HR 0.63) both confirmed substantial benefit. See the KEYNOTE-522 synthesis for detail.

The atezolizumab program (IMpassion)

IMpassion130 (first-line metastatic, nab-paclitaxel backbone)

Schmid and colleagues randomized 902 patients with previously untreated metastatic TNBC to atezolizumab + nab-paclitaxel vs placebo + nab-paclitaxel^[5]A. PD-L1 was measured with the SP142 assay using IC scoring; the ≥ 1% IC threshold defined the PD-L1+ subgroup (~40% of trial enrollment).

• ITT PFS: 7.2 vs 5.5 months (HR 0.80; p=0.0025) — positive primary
• PD-L1+ subgroup PFS: 7.5 vs 5.0 months (HR 0.62; p<0.0001)
• PD-L1+ subgroup OS: 25.4 vs 17.9 months (HR 0.71) — numerical benefit, hierarchy precluded testing
• ITT OS: 21.0 vs 18.7 months (HR 0.87) — not significant

FDA granted accelerated approval for atezolizumab + nab-paclitaxel for SP142 IC ≥ 1% metastatic TNBC in March 2019.

IMpassion131 (first-line metastatic, paclitaxel backbone)

Miles and colleagues randomized 651 patients (similar to IMpassion130's population) to atezolizumab + paclitaxel vs placebo + paclitaxel^[6]A. The trial was designed as the confirmatory study for the accelerated approval, with the key difference being paclitaxel (not nab-paclitaxel) as the chemotherapy backbone.

• PD-L1+ subgroup PFS: 6.0 vs 5.7 months (HR 0.82; p=0.20) — not significant
• PD-L1+ subgroup OS: 22.1 vs 28.3 months (HR 1.12) — numerically worse with atezolizumab

Roche voluntarily withdrew the atezolizumab indication for metastatic TNBC in August 2021. The mechanism for the discordance with IMpassion130 has been hypothesized to involve the corticosteroid premedication required for solvent-based paclitaxel (which may blunt the immune response) or chance variation in a moderately powered trial.

IMpassion031 (neoadjuvant early-stage)

Mittendorf and colleagues randomized 333 patients with stage II–III TNBC to atezolizumab + nab-paclitaxel → atezolizumab + AC vs placebo + the same chemo^[7]A.

• pCR rate: 58% vs 41% (absolute difference 17 percentage points; p=0.0044)
• Long-term outcomes: numerical EFS benefit, not statistically tested due to small sample

The trial was positive on its primary pCR endpoint. Atezolizumab's withdrawal in metastatic TNBC effectively curtailed development in the early-stage setting; the IMpassion031 result has not been pursued as a regulatory basis.

Other notable IO trials

NeoTRIPaPDL1 (Italian, atezolizumab + chemo no anthracycline)

Gianni and colleagues randomized 280 patients with high-risk early-stage TNBC to atezolizumab + carboplatin + nab-paclitaxel (8 cycles, anthracycline-free) vs the same chemo without atezolizumab^[8]A. The trial did not meet its primary 5-year EFS endpoint, despite a numerically higher pCR (43.5% vs 40.8%). The chemotherapy backbone difference (anthracycline-free) is hypothesized as a contributing factor to the failure.

GeparNuevo (durvalumab + chemo, neoadjuvant)

Loibl and colleagues randomized 174 patients with stage II–III TNBC to durvalumab + standard chemo (only neoadjuvant; no adjuvant durvalumab) vs placebo + chemo^[9]A. Initial pCR primary endpoint was numerically positive (53% vs 44%) but not statistically significant. A 2022 long-term follow-up reported substantial 3-year iDFS benefit (HR 0.48) despite the missed initial primary, suggesting the neoadjuvant-only durvalumab signal was meaningful but the trial was underpowered.

Cross-trial comparison — what the evidence base shows

Lessons from the program

• IO + chemo > IO monotherapy in TNBC. KEYNOTE-086 vs KEYNOTE-119 vs KEYNOTE-355 establish this dose-response, replicated across atezolizumab. Single-agent IO in TNBC is unlikely to be developed further without clear biomarker selection.
• Chemotherapy backbone matters. IMpassion130 (nab-paclitaxel) succeeded; IMpassion131 (paclitaxel + steroid premedication) failed; NeoTRIPaPDL1 (anthracycline-free) failed. The anthracycline + steroid-effect hypotheses both suggest IO-chemo synergy depends on the specific backbone.
• PD-L1 assay heterogeneity is consequential. SP142/IC and 22C3/CPS are not interconvertible (see the PD-L1 assays synthesis). The atezolizumab/IMpassion program used SP142; the pembrolizumab/KEYNOTE program used 22C3. Cross-program comparison is complicated by this assay-level difference.
• Early-stage IO benefit may be larger than metastatic. KEYNOTE-522's hazard ratios for EFS (0.63) and the absolute pCR benefit (13.6 percentage points) suggest early-stage IO benefit may be larger than the metastatic-setting magnitude. The mechanism is debated but may involve the better-preserved immune system of the early-stage patient.
• Confirmatory trial requirements matter. Accelerated approval based on IMpassion130 was withdrawn after IMpassion131's failure — the regulatory framework worked as intended. KEYNOTE-522's pCR-then-EFS confirmation followed the textbook path.

Open questions and active investigation

• The atezolizumab story. Is atezolizumab biologically inferior to pembrolizumab in TNBC, or did IMpassion131 fail for non-drug-related reasons (chemotherapy backbone, cohort drift, chance)? Definitive resolution requires head-to-head data unlikely to be funded.
• Beyond pembrolizumab + chemo for the biomarker-negative subgroup. The ~50% of metastatic TNBC patients who are CPS < 10 and germline-BRCA-wildtype have no IO benefit established. TIGIT inhibitors, LAG-3 inhibitors, and novel ICI combinations are being tested.
• Duration of pembrolizumab in long-term responders. The KEYNOTE-355 protocol allowed indefinite treatment until progression. Whether shorter durations are non-inferior in responders is a quality-of-life question with cost implications.
• Concurrent ICI + radiation, surgery, and other modalities. The immunomodulatory effects of cancer-directed therapy on IO outcomes are being studied; whether the order of modalities can be optimized is an active question.
• De-escalation of chemo in KEYNOTE-522 IO + chemo regimen. Several de-escalation trials are testing whether the chemo can be reduced in IO-receiving patients, particularly in high-TIL subsets.

For the first-line metastatic decision tree, see the first-line metastatic synthesis. For the early-stage IO + chemo standard, see the KEYNOTE-522 synthesis. For PD-L1 assay heterogeneity, see the PD-L1 assays synthesis.

References

Each citation links to the original publication via DOI. The same records are searchable in the evidence library by title or DOI.

1. Adams S, Schmid P, Rugo HS, et al. Pembrolizumab monotherapy for previously treated metastatic triple-negative breast cancer: cohort A of the phase II KEYNOTE-086 study. Ann Oncol. 2019;30(3):397–404. doi:10.1093/annonc/mdy517. ↩
2. Winer EP, Lipatov O, Im SA, et al. Pembrolizumab versus investigator-choice chemotherapy for metastatic triple-negative breast cancer (KEYNOTE-119). Lancet Oncol. 2021;22(4):499–511. doi:10.1016/S1470-2045(20)30754-3. ↩
3. Cortes J, Cescon DW, Rugo HS, et al. Pembrolizumab plus chemotherapy versus placebo plus chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer (KEYNOTE-355). Lancet. 2020;396(10265):1817–1828. doi:10.1016/S0140-6736(20)32531-9. ↩
4. Schmid P, Cortes J, Pusztai L, et al. Pembrolizumab for Early Triple-Negative Breast Cancer (KEYNOTE-522). N Engl J Med. 2020;382(9):810–821. doi:10.1056/NEJMoa1910549. ↩
5. Schmid P, Adams S, Rugo HS, et al. Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer (IMpassion130). N Engl J Med. 2018;379(22):2108–2121. doi:10.1056/NEJMoa1809615. ↩
6. Miles D, Gligorov J, André F, et al. Primary results from IMpassion131 (atezolizumab + paclitaxel). Ann Oncol. 2021;32(8):994–1004. doi:10.1016/j.annonc.2021.05.801. ↩
7. Mittendorf EA, Zhang H, Barrios CH, et al. Neoadjuvant atezolizumab in combination with sequential nab-paclitaxel and anthracycline-based chemotherapy versus placebo and chemotherapy in patients with early-stage triple-negative breast cancer (IMpassion031). Lancet. 2020;396(10257):1090–1100. doi:10.1016/S0140-6736(20)31953-X. ↩
8. Gianni L, Huang CS, Egle D, et al. Pathologic complete response and outcomes with neoadjuvant atezolizumab in triple-negative breast cancer (NeoTRIPaPDL1). Ann Oncol. 2022;33(5):534–543. doi:10.1016/j.annonc.2022.02.004. ↩
9. Loibl S, Untch M, Burchardi N, et al. A randomised phase II study investigating durvalumab in addition to an anthracycline taxane-based neoadjuvant therapy in early triple-negative breast cancer (GeparNuevo). Ann Oncol. 2019;30(8):1279–1288. doi:10.1093/annonc/mdz158. ↩

Last reviewed: 2026-06-04. Researcher-layer synthesis page. Evidence grades follow the GRADE-adapted rubric defined at the top of this page.