Breast cancer stem cells (CSCs) contribute to therapeutic resistance. I-SPY is a multi-center neoadjuvant trial for women with locally advanced breast cancers (3cm or greater) who received neoadjuvant doxorubicin and cyclophosphamide then paclitaxel. We hypothesized that CSC markers will be increased by neoadjuvant chemotherapy in patients who did not achieve a pathologic complete response, and that women with elevated CSC markers will have worse survival in this multicenter study.

Methods: 215 women completed surgery, of those 156 had residual tumor, of whom 56 had sufficient tumor sampled pre- and post-neoadjuvant chemotherapy to enable measurements of CSCs by automated quantitative analysis (AQUA) of immunofluourescent staining for aldehyde dehydrogenase (ALDH1), CD44, and ALDH1/CD44. A threshold of background staining was determined for each marker from a tissue microarray representing multiple cell lines and patient specimens and was used to define samples positive for CSCs. Effect of treatment on proportion of samples positive for CSCs was assessed using the McNemar test. Association between CSC markers and recurrence free survival (censored at 5 years) (RFS) were evaluated using Kaplan-Meier analysis of cohorts dichotomized at optimal marker thresholds as determined from ROC analysis for RFS prediction, as well as univariate Cox analysis of continuous CSC marker scores.

Results: Of the 56 patients studied median tumor size was 6cm, 30% HR-neg, and 75% HER2 neg (by central and/or local IHC and/or FISH). In the HER2-neg subset, despite reducing tumor size in approximately 76% of cases, chemotherapy significantly increased the proportion of samples positive for CSCs: ALDH1 (38% to 74%), CD44 (67% to 88%), and ALDH1 in CD44 compartment (43% to 76%) (McNemar test p = 0.0007, 0.03, and 0.001, respectively). Within this HER2-neg subset, tumors with high pre-treatment CSC expression (i.e. CSC marker scores greater than optimal thresholds for RFS prediction – ALDH1 score ≥470.5, CD44 score ≥8662, ALDH1 in CD44 compartment score ≥742.5) appeared to have worse RFS with a hazard ratio of 3.7, 2.8, and 2.1 respectively (log rank p = 0.03, 0.07, 0.17). Interestingly, Cox regression analysis revealed that pretreatment ALDH1 score, as well as the ALDH1 in CD44 compartment score appeared significantly associated with RFS in the HR positive HER2 neg subset (n = 33) (hazard ratio: 2.2 (1.3–3.8), p = 0. 003, and hazard ratio: 1.8 (1.1–2.8), p = 0.01 respectively). Surprisingly, post-treatment CSC markers expression did not show significant RFS associations within the HER2 neg or the HR positive HER2 neg subsets.

Conclusion: These results from this multicenter study validate the hypothesis of treatment-resistant cancer stem cells, with an increase in CSC markers post chemotherapy, and suggest worse prognosis in women whose tumors have high expression of these stem cell markers, especially for HER2 negative patients. Surprisingly, only pre- (and not post-) treatment CSC marker expression appears to associate with RFS. Studies to evaluate inhibitors of stem cell self-renewal and conventional therapy should be considered for HER2-neg breast cancers.