Reviews/Targeted Therapy Breast Cancer/Pembrolizumab neoadjuvant therapy in triple-negative breast cancer
Ici BC · BC-ICI-01 · 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 pre-surgery chemotherapy in women with early triple-negative breast cancer, and it reported more patients clearing their tumour before surgery, fewer cancer recurrences, and longer survival [12][1][3]. The benefit also reached patients who did not fully clear their tumour [4], and smaller real-world clinics reported similarly high tumour-clearance rates [19][21]. The trial enrolled patients without selecting them by the PD-L1 marker, which is why the panel accepted a benefit that does not depend on PD-L1 status, and one small regional subgroup pointed the same way [16]. The main caution is that almost all of the randomised evidence comes from this single trial, with no independent pembrolizumab trial in the corpus to confirm it. The PD-L1-independent part was only directly measured in a small group whose numbers could not rule out no difference, and a sister trial of a different drug in the same setting did not show a clear survival gain [20]. The panel therefore treated the core finding as solid while holding back from the highest tier.
A single phase 3 randomised trial raised pathological complete response from 51.2% to 64.8% [12] and improved event-free survival (hazard ratio 0.63) [1], with a later overall-survival benefit of 86.6% versus 81.7% at 60 months [3]. A preplanned residual-cancer-burden analysis showed pembrolizumab 'improved EFS among most patients who do not have a pCR' [4]. Enrolment was not selected by PD-L1 and a Japan subgroup reported 'findings were similar regardless of PD-L1 expression' [16]. Real-world retrospective cohorts reproduced high complete-response rates of 56.0% [19], 64.7% [11] and 77.2% [21], though these are uncontrolled or non-matched. The contested point: The division is over how strongly the 'regardless of PD-L1 status' claim is supported: the primary papers [1][3][12] report intention-to-treat results rather than PD-L1 subgroups, and the only direct subgroup test [16] used 76 participants with intervals that cross no difference (pCR difference 95% CI -17.6% to 27.1%; EFS HR 0.54, 95% CI 0.20-1.50). A second contested point is reliance on one trial, since all randomised pembrolizumab evidence in the corpus is KEYNOTE-522, while a different-drug trial [20] in the same setting missed significance (HR 0.80, P=0.083).
All 21 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: essentially all randomized evidence for both pCR and EFS ([1],[3],[4],[12],[14],[16]) is KEYNOTE-522; no independent confirmatory pembrolizumab RCT exists in the corpus, so the causal claim rests on one trial.
The 'regardless of PD-L1 status' rider is directly tested only in the N=76 Japan subgroup [16], where both the pCR difference (95% CI -17.6% to 27.1%) and EFS HR 0.54 (95% CI 0.20-1.50) cross the null; the primary papers [1],[3],[12] report ITT, not PD-L1 subgroups.
Where it remains uncertain. Nearly all randomised evidence is from the single KEYNOTE-522 trial, and the PD-L1-agnostic claim is directly tested only in a 76-patient subgroup whose intervals cross no difference [16].
Aleatory
The only direct PD-L1-stratified test [16] has 76 participants, so random variation leaves wide intervals (pCR difference 95% CI -17.6% to 27.1%; EFS HR 95% CI 0.20-1.50) that cannot distinguish benefit from no difference.
Epistemic
All randomised pembrolizumab evidence in the corpus is KEYNOTE-522 [1][3][4][12][14][16]; no independent confirmatory RCT exists here to show the effect replicates.
Inferential
Inferring a checkpoint-inhibitor class benefit is undercut because the atezolizumab RCT [20] in the same early-TNBC setting did not meet EFS significance (HR 0.80, P=0.083).
The load-bearing papers from a corpus of 21, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Pivotal randomized evidence for the EFS component (HR 0.63).
“The estimated event-free survival at 36 months was 84.5% (95% confidence interval [CI], 81.7 to 86.9) in the pembrolizumab-chemotherapy group, as compared with 76.8% (95% CI, 72.2 to 80.7) in the placebo-chemotherapy group (hazard ratio for event or death, 0.63; 95% CI, 0.48 to 0.82; P<0.001)”
Panel auditInternal validity: high · scope of inference: Previously untreated stage II or III triple-negative breast cancer; EFS endpoint only (pCR reported in an earlier analysis).Key limitations: abstract does not report pCR result here · no PD-L1 subgroup reported in this abstract
Mature survival benefit reinforcing durability of the pCR/EFS wins.
“The estimated overall survival at 60 months was 86.6% (95% confidence interval [CI], 84.0 to 88.8) in the pembrolizumab-chemotherapy group, as compared with 81.7% (95% CI, 77.5 to 85.2) in the placebo-chemotherapy group (P = 0.002)”
Panel auditInternal validity: high · scope of inference: Previously untreated stage II or III TNBC; overall survival as a secondary endpoint, with pCR/EFS cited from prior analyses.Key limitations: this abstract's new result is OS not pCR/EFS · no PD-L1 subgroup reported in this abstract
Shows EFS benefit is not confined to pCR responders.
“These findings demonstrate that pembrolizumab not only increased pCR rates, but also improved EFS among most patients who do not have a pCR.”
Panel auditInternal validity: medium · scope of inference: Stage T1c/N1-2 or T2-4/N0-2 TNBC; EFS stratified by residual cancer burden.Key limitations: exploratory endpoint · wide CIs in RCB-0/1/3 · no PD-L1 subgroup reported
The abstract reports a single-arm pCR rate (48.4%) and a TIL-pCR association, with no pembrolizumab-versus-chemotherapy contrast to judge the claim.
Panel auditInternal validity: low · scope of inference: Locally advanced TNBC treated with the K522 regimen; predictors of pCR, chiefly TIL.Key limitations: no comparator arm · small sample · single institution
4 of 21 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.
Strong, internally consistent RCT evidence for the pCR and EFS core, held just below the Established band by single-trial dependence and a PD-L1-agnostic rider that is only thinly tested directly.
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 pembrolizumab RCT. All randomised evidence is KEYNOTE-522 [1][3][12], so the core causal finding rests on one trial. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A separate randomised trial of neoadjuvant pembrolizumab plus chemotherapy versus chemotherapy alone in early TNBC with pCR and EFS endpoints.
Open question: Does the pembrolizumab pCR and EFS benefit replicate in a trial independent of KEYNOTE-522? Why it is still open: Every randomised result in the corpus [1][3][4][12][14][16] comes from KEYNOTE-522; no independent confirmatory RCT is present.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 23 Oct 2026 · protocol v0.19