Summary of clinical trials testing immune checkpoint inhibitors in early breast cancer (EBC), for which preliminary or final results have been presented and/or published

TrialDesignSettingPopulation (n)Drugs (n per arm)pCR rates % (95% CI)OR (95% CI), PBiomarker analysisRef.
GeparNuevoII RandomizedNeoadjuvantTN (174)
Any PD-L1
Durvalumab (2w)a → Durvalumab + NabP → Durvalumab + EC (88)53.4 (42.5–61.4) Window cohort: 61.01.45 (0.80–2.63), P = 0.224sTILs and PD-L1 (trend) associated with pCR, not predictive for Durvalumab benefit[14]
Durvalumab (2w)a → placebo + NabP → placebo + EC (86)44.2 (33.5–55.3) Window cohort: 41.4Window cohort: 2.22 (1.06–4.64), P = 0.035iTILs (increase between baseline and end of the window phase) predictive for Durvalumab benefit (OR 9.36, 95% CI 1.26–69.65, P = 0.029)
KEYNOTE-173IbNeoadjuvantTN (60) Any PD-L1Pembrolizumab + 6 CT regimens (cohort A-F)bOverall: 60 (range: 49–71)NAsTILs and PD-L1 associated with pCR[16]
KEYNOTE-522IIINeoadjuvantTN (602) Any PD-L1Pembrolizumab + NabP-carboplatin → A/E-C + pembrolizumab (401)64.8 (59.9–69.5)Estimated treatment difference: 13.6% (5.4-21.8, P > 0.001)PD-L1 associated with pCR, not predictive for Pembrolizumab benefit[17]
Placebo + NabP-carboplatin → A/E-C + placebo (201)51.2 (44.1–58.3)
NeoTRIPaPDL1IIINeoadjuvantTN (280) Any PD-L1Atezolizumab + NabP-carboplatin (138)43.5 (35.1–52.2)1.11 (0.69–1.79), P = 0.66cPD-L1 associated with pCR, not predictive for Atezolizumab benefit[19]
NabP-carboplatin (142)40.8 (32.7–49.4)
IMpassion031IIINeoadjuvantTN (333) Any PD-L1Placebo + NabP → placebo + A (168)41.1Delta pCR 16.5 (5.9– 27.1), P = 0.0044dPD-L1 associated with pCR, not predictive for Atezolizumab benefit[18]
Atezolizumab + NabP → atezolizumab + A (165)57.6
I-SPY 2II Adaptive-randomizedNeoadjuvantHER2- (205) Any PD-L1Pembrolizumab + paclitaxel → AC (69)Total: 44 (33–55) e
TN: 60 (44–75) e
HR+: 30 (17–43) e
Probability superior to control: Total: > 99.9%
TN: > 99.9%
HR+: 99.6%
NA[15]
Paclitaxel → AC (181)Total: 17 (11–23) e
TN: 22 (13–30) e
HR+: 13 (7–19) e
GIADAIINeoadjuvantLuminal-B (43) Any PD-L1EC → Nivolumab + triptorelin + exemestane16.3 (7.4–34.9) fTILs and specific immune infiltrate characteristics at baseline associated with pCR; anthracycline-induced increase in cytotoxic T-cells and decrease in T-regulatory T cells[41]

window phase stopped after 117 patients recruited (ethical concerns);

CT regimens; cohort A: NabP → doxorubicin-cyclophosphamide (AC); cohort B: NabP-Carboplatin AUC6 → AC; cohort C: NabP-Carboplatin AUC5 → AC; cohort D: NabP-Carboplatin AUC2 → AC; cohort E: Paclitaxel-Carboplatin AUC5 → AC; cohort F: Paclitaxel-Carboplatin AUC2 → AC;

pCR was a secondary endpoint;

one-sided significance boundary P = 0.0184;

estimated rates of pCR;

2-step statistical design: second step (8 pCR/43 patients) not met;

w: weeks; NA: not available; pCR: pathologic complete response; OR: odds ratio; EC: epirubicin-cyclophosphamide; sTILs: stromal TILs; iTILs: intra-tumoral TILs; AC: doxorubicin-cyclophosphamide