Reviews/Targeted Therapy Breast Cancer/Pembrolizumab in PD-L1-positive metastatic triple-negative breast cancer
Ici BC · BC-ICI-02 · v2
Borderline: within 4 points of the Established/Likely boundary, so a re-grade could land either side.
One large, well-run randomised trial tested adding pembrolizumab to standard chemotherapy as the first treatment for metastatic triple-negative breast cancer, and its final analysis found that patients whose tumours had high PD-L1 expression (CPS of 10 or more) lived longer on the combination than on chemotherapy alone, 23.0 versus 16.1 months [12]. The benefit was specific to that high-expression group: when the same trial looked at tumours with any PD-L1 expression or at everyone enrolled, the survival difference was not statistically meaningful [12]. Progression-free survival in the same high-expression group pointed the same way [1][14], and a smaller analysis of Asian patients from the same trial trended the same direction [15]. The weakness the panel focused on is that every piece of supporting evidence comes from this one trial, with no independent study confirming it in the corpus. The panel therefore treated the finding as well-established within this single trial but not yet confirmed by a second one.
The final overall-survival analysis of KEYNOTE-355 [12] reports a prespecified, significance-met survival benefit in exactly the claimed first-line CPS 10 or more population: median 23.0 versus 16.1 months, hazard ratio 0.73 (0.55 to 0.95), P=0.0185. The progression-free survival readouts of the same trial [1] (HR 0.65 in CPS 10 or more) and [14] support the direction in the identical population. A subgroup analysis of Asian patients from the same trial [15] shows a directionally consistent overall-survival hazard ratio of 0.54 (0.28 to 1.04). By contrast, pembrolizumab monotherapy in a later-line setting did not significantly improve survival even at CPS 10 or more [6] (HR 0.78, 0.57 to 1.06), underlining that the benefit here is observed in the combination first-line context the claim specifies. The contested point: The evidence divides on independence, not direction: every supporting entry [12][1][14][15] is the same trial, so the corpus holds no independent confirmation. The benefit is also confined to the CPS 10 or more subgroup, since in the same trial [12] the CPS 1 or more (HR 0.86, P=0.1125) and intention-to-treat (HR 0.89) analyses were not significant.
All 17 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Targeted Therapy Breast Cancer, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
Single-trial dependence: every supporting entry ([12] OS, [1] and [14] PFS, [15] subgroup) is the same study, KEYNOTE-355. The corpus contains no independent confirmatory trial of pembrolizumab+chemo versus chemo in first-line CPS>=10 mTNBC.
CPS>=10 is an enrichment subgroup: in [12] OS was not significant in CPS>=1 (HR 0.86, P=0.1125) or ITT (HR 0.89, not tested), and the Asian subgroup OS [15] HR 0.54 crosses 1 (0.28-1.04). The benefit is real only within the specific subgroup the claim names.
Where it remains uncertain. All supporting evidence comes from one trial [12][1][14][15], the benefit disappears outside the CPS 10 or more subgroup [12], and the significance margin is modest (P=0.0185).
Aleatory
The overall-survival result in [12] met its prespecified bar at P=0.0185 with the confidence interval reaching 0.95, leaving little room above chance variation.
Epistemic
Every supporting entry [12][1][14][15] is KEYNOTE-355; the corpus contains no second trial of pembrolizumab plus chemotherapy versus chemotherapy in this population.
Inferential
The benefit holds only within CPS 10 or more; in the same trial [12] CPS 1 or more (HR 0.86) and intention-to-treat (HR 0.89) were not significant, so extrapolation beyond the cutoff is not supported.
The load-bearing papers from a corpus of 17, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Establishes the trial, population and the concordant CPS>=10 PFS benefit underpinning the OS result
“Among patients with CPS of 10 or more, median progression-free survival was 9·7 months with pembrolizumab-chemotherapy and 5·6 months with placebo-chemotherapy (hazard ratio [HR] for progression or death, 0·65, 95% CI 0·49-0·86; one-sided p=0·0012 [primary objective met])”
Panel auditInternal validity: high · scope of inference: First-line locally recurrent inoperable or metastatic TNBC; PFS (not OS) in the PD-L1 CPS of 10 or more subgroup.Key limitations: OS immature at this analysis · interim PFS readout only
Mature OS benefit meeting prespecified significance in exactly the claimed CPS>=10 first-line population
“In the CPS-10 subgroup, the median overall survival was 23.0 months in the pembrolizumab-chemotherapy group and 16.1 months in the placebo-chemotherapy group (hazard ratio for death, 0.73; 95% confidence interval [CI], 0.55 to 0.95; two-sided P = 0.0185 [criterion for significance met])”
Panel auditInternal validity: high · scope of inference: First-line locally recurrent inoperable or metastatic TNBC with PD-L1 CPS of 10 or more; overall survival.Key limitations: CPS-1 and ITT OS not significant · single pivotal trial
Confirms the CPS>=10 PFS signal in the same trial at interim
“Pembro combined with several chemo partners showed a statistically significant and clinically meaningful improvement in PFS vs chemo alone in pts with previously untreated locally recurrent inoperable or metastatic TNBC whose tumors expressed PD-L1 (CPS ≥10).”
Panel auditInternal validity: high · scope of inference: First-line locally recurrent inoperable or metastatic TNBC with CPS at least 10; PFS (not OS).Key limitations: OS not yet reported · conference abstract interim data
The OS result (CPS of 10 or more HR 0.78, not significant) is for monotherapy in a previously treated population, a different regimen and line than the claim's first-line combination, so it does not establish direction for the claim.
Panel auditInternal validity: high · scope of inference: Second- or third-line metastatic TNBC, pembrolizumab monotherapy versus chemotherapy.Key limitations: monotherapy not combination · later-line not first-line · open-label
4 of 17 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
A single high-quality phase 3 RCT meeting its prespecified OS endpoint on a hard outcome in the precise claimed population supports a strong but not multiply-confirmed verdict.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
Independent confirmatory trial. The verdict relies on one trial; an independent randomised comparison would convert a single-trial result into a replicated one. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A second phase 3 RCT of pembrolizumab plus chemotherapy versus chemotherapy in first-line CPS 10 or more metastatic TNBC, with overall survival as the endpoint.
Open question: Does an independent trial confirm the first-line overall-survival benefit of pembrolizumab plus chemotherapy in CPS 10 or more metastatic TNBC? Why it is still open: All supporting evidence in the corpus is from the single KEYNOTE-355 trial [12][1][14][15].
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 23 Oct 2026 · protocol v0.19