13 Assessing Intratumor Heterogeneity in Programmed Death-Ligand 1 (PD-L1) Protein Expression in Early-Stage Breast Cancer

Background

Programmed death–ligand 1 (PD-L1) expression is required for benefit from immune checkpoint inhibitors in metastatic triple-negative breast cancer (TNBC), although in the neoadjuvant setting, patients benefited regardless of PD-L1 expression. We aimed to examine intratumor spatial heterogeneity of PD-L1 protein expression in multiple regions of primary breast cancers.

Materials and Methods

Biopsies were collected from 3 separate tumor regions of surgically resected primary tumors without prior neoadjuvant systemic therapy. E1L3N antibody was used to assess PD-L1 status. The combined positivity score (CPS) method was used for scoring (PD-L1 positive defined as CPS >10). Fleiss Multirater Kappa assessed heterogeneity of PD-L1 expression among the different samples of the same tumor. Optimal agreement is suggested by a kappa agreement coefficient of 0.6 or more. Association between PD-L1 status and tumor-infiltrating lymphocyte (TIL) counts was assessed using Mann-Whitney test.

Results

Of the 33 included patients, 9 (27%) had TNBC, 5 (15%) had HER2-positive disease, and 19 (58%) had hormone receptor–positive, HER2-negative disease. Four of 33 patients (12%) had 1 biopsy sample showing PD-L1 positive status with additional biopsies of different regions of the tumor showing negative results. Fleiss’ kappa analysis on patients with all 3 biopsies per tumor available and able to be stained for PD-L1 (29 of 33 [88%]) showed an overall agreement of –0.36 (95% CI, –0.246 to 0.174; P = .739) when using CPS as categorical variable (positive vs negative) and 0.410 (95% CI, 0.286-0.535; P <.01) when using CPS as continuous variable. PD-L1 status was associated with TIL counts, with a median TIL count of 5% for PD-L1 positive and 0% for PD-L1 negative (P = .2949).

Conclusions

Our study showed that in 12% of the cases, PD-L1 was found to be positive in one of the tumor areas and all remaining biopsy samples of the same tumor were found to be negative, suggesting intratumor spatial heterogeneity regarding PD-L1 expression in early-stage breast cancer. However, due to the overall low incidence of PD-L1 positivity, it seems unlikely that the benefit of immunotherapy in early-stage breast cancer would be simply explained by sampling error. Further studies are required to investigate effective predictive biomarkers of response to immunotherapy, as well as to standardize PD-L1 assessment in breast cancer.

Author Affiliations:

Adriana Matutino Kahn,¹ Reza Golestani,² Malini Harigopal,² Lajos Pusztai¹

¹Section of Medical Oncology, Yale School of Medicine, New Haven, CT

²Department of Pathology, Yale School of Medicine, New Haven, CT