A phase Ib trial of pembrolizumab plus paclitaxel or flat-dose capecitabine in 1st/2nd line metastatic triple-negative breast cancer

Abstract

Chemoimmunotherapy with anti-programmed cell death 1/ligand 1 and cytotoxic chemotherapy is a promising therapeutic modality for women with triple-negative breast cancer, but questions remain regarding optimal chemotherapy backbone and biomarkers for patient selection. We report final outcomes from a phase Ib trial evaluating pembrolizumab (200 mg IV every 3 weeks) with either weekly paclitaxel (80 mg/m² weekly) or flat-dose capecitabine (2000 mg orally twice daily for 7 days of every 14-day cycle) in the 1st/2nd line setting. The primary endpoint is safety (receipt of 2 cycles without grade III/IV toxicities requiring discontinuation or ≥21-day delays). The secondary endpoint is efficacy (week 12 objective response). Exploratory aims are to characterize immunologic effects of treatment over time, and to evaluate novel biomarkers. The trial demonstrates that both regimens meet the pre-specified safety endpoint (paclitaxel: 87%; capecitabine: 100%). Objective response rate is 29% for pembrolizumab/paclitaxel (n = 4/13, 95% CI: 10-61%) and 43% for pembrolizumab/capecitabine (n = 6/14, 95% CI: 18-71%). Partial responses are observed in two subjects with chemo-refractory metaplastic carcinoma (both in capecitabine arm). Both regimens are associated with significant peripheral leukocyte contraction over time. Response is associated with clinical PD-L1 score, non-receipt of prior chemotherapy, and the H&E stromal tumor-infiltrating lymphocyte score, but also by a novel 27 gene IO score and spatial biomarkers (lymphocyte spatial skewness). In conclusion, pembrolizumab with paclitaxel or capecitabine is safe and clinically active. Both regimens are lymphodepleting, highlighting the competing immunostimulatory versus lymphotoxic effects of cytotoxic chemotherapy. Further exploration of the IO score and spatial TIL biomarkers is warranted. The clinical trial registration is NCT02734290.

Fig. 1

CONSORT diagram illustrating enrollment of subjects.

Fig. 2. Tumor response.

a Waterfall plot of percent change in tumor size at 12 weeks by RECIST 1.1 criteria in capecitabine and paclitaxel arms, respectively. Two patients are not depicted in capecitabine arm and 3 in paclitaxel arm due to progression of disease prior to week 12. Some subjects experienced RECIST1.1 progression due to progression of non-target or new lesions, despite radiographic shrinkage of target lesions. b Spider plot of percent change in tumor size, by RECIST 1.1 criteria in capecitabine and paclitaxel arms, respectively. RECIST 1.1 Response Evaluation Criteria in Solid Tumors guidelines, version 1.1, CR complete response, PR partial response, SD stable disease, PD progression of disease, NE not evaluable.

Fig. 3. Kaplan–Meier survival curves.

Kaplan–Meier estimates of progression-free survival (a) and overall survival (b). Dotted lines represent the median survival. Statistical comparisons of the survival curves are not performed because the study is not randomized or controlled for covariates. Cape Capecitabine, PFS progression-free survival, OS overall survival.

Fig. 4. Associations of response with clinical parameters.

Intervals represent the 80% confidence interval of the point estimate of response rate. PD-L1: programmed death ligand 1; CPS combined positive score, TIL tumor-infiltrating lymphocytes, ALC absolute lymphocyte count, RECIST response evaluation criteria in solid tumors, *(neo)-adjuvant systemic therapy counted as one line if received <6 months from study enrollment.

Fig. 5. Outcomes in metaplastic TNBC patients.

a Summary of clinical findings of metaplastic patients; b radiographic response of patient 1. Findings overall demonstrate a partial response, but with an initial non-target left lung nodule gradually increasing in size; c radiographic response of patient 2, showing a mixed, but overall partial response. Cape capecitabine, CPS combined positive score, H&E hematoxylin & eosin, sTILs stromal tumor-infiltrating lymphocytes, PR partial response, PD progression of disease, R right, L left, AC/T doxorubicin, cyclophosphamide, and paclitaxel, Cb carboplatin, RCB residual cancer burden score.

Fig. 6. 27-gene IO score as a predictive biomarker of response.

a Distribution of 27-gene IO score according to week 12 RECIST response; b Correlation of pre-treatment versus on-treatment IO-score among matched biopsies. CR complete response, PR partial response, SD stable disease, PD progression of diseases.

Fig. 7. Associations of response with stromal T cell density and/or skewness.

The X axis reports cellular density and the Y axis reports skewness scores, with high scores indicating hotspot regions within the tumor. Using 350 cells/mm² density cutoff and skew = 2 cutoff, responses and PFS are enriched among tumors with high T cell density and/or T cell skew (outer & upper quadrants). a CD3 + CD8 + T cells; b CD3 + CD8- T cells. PFS progression-free survival.