Discordance between PAM50 intrinsic subtyping and immunohistochemistry in South African women with breast cancer

Purpose Breast cancer is a heterogeneous disease with different gene expression profiles, treatment options and outcomes. In South Africa, tumors are classified using immunohistochemistry. In high-income countries multiparameter genomic assays are being utilized with implications for tumor classification and treatment. Methods In a cohort of 378 breast cancer patients from the SABCHO study, we investigated the concordance between tumor samples classified by IHC and the PAM50 gene assay. Results IHC classified patients as ER-positive (77.5%), PR-positive (70.6%), and HER2-positive (32.3%). These results, together with Ki67, were used as surrogates for intrinsic subtyping, and showed 6.9% IHC-A-clinical, 72.7% IHC-B-clinical, 5.3% IHC-HER2-clinical and 15.1% triple negative cancer (TNC). Typing using the PAM50 gave 19.3% luminal-A, 32.5% luminal-B, 23.5% HER2-enriched and 24.6% basal-like. The basal-like and TNC had the highest concordance, while the luminal-A and IHC-A group had the lowest concordance. By altering the cutoff for Ki67, and realigning the HER2/ER/PR-positive patients to IHC-HER2, we improved concordance with the intrinsic subtypes. Conclusion We suggest that the Ki67 be changed to a cutoff of 20–25% in our population to better reflect the luminal subtype classifications. This change would inform treatment options for breast cancer patients in settings where genomic assays are unaffordable.