Background: The efficacy of cancer chemotherapy is generally assessed by the ability to directly kill or inhibit the growth of cancer cells. However, there is evidence suggesting that anticancer immune responses also contribute to the efficacy of conventional chemotherapy. Recent studies have shown that chemotherapeutic agents (anthracyclines in particular) can elicit innate immune responses via the release of proteins from dying tumor cells that interact with Toll-like receptors (TLR) on macrophages (mΦ) and dendritic cells (DC). The subsequent induction of cytotoxic T cell responses leads to the elimination of residual cancer cells not killed by the chemotherapy. To test this idea of chemotherapy-induced immunogenic cell death in breast cancer, we examined gene expression data for women with breast cancer treated neoadjuvantly, starting with adriamycin, on the I-SPY TRIAL (2003-2006), to see if genes that predict this type of immune response would also predict complete pathologic response (pCR) Methods: A panel of 47 genes were selected, including those expressed by the cancer cells as well as infiltrating immune cells: TLR/mΦ/DC activating proteins, TLR signaling genes, chemokines & chemokine receptors, cytokines & cytokine receptors, and mO, DC, & T cells markers. This gene set was then evaluated using expression array data from the I-SPY TRIAL, available on both Affymetrix (n=) and Agilent (n= 153) platforms. Initial analyses were performed using the UCSC Cancer Genomics Browser, a suite of web-based tools to integrate, visualize and analyze cancer genomics/expression data and clinical data. Differentially expressed genes were compared between patients who achieved a pathological complete response (pCR) following neoadjuvant chemotherapy vs. those who did not. A second independent data set available on the Cancer Genomics browser served as a validation set. These data, from MD Anderson, included 133 patients treated with neoadjuvant therapy (TFAC) with available Affymetrix array data.
Results: Of the 47 selected genes, 1 was not present on the Affymetrix array and 7 were absent from the Agilent array. Twelve of the 46 genes (26%) on the Affymetrix array and 20 of the 40 genes (50%) on the Agilent array were significantly differentially expressed comparing responders vs. non-responders. Overall, 25 of the 47 genes (53%) were differentially expressed on one or both arrays. Gene set enrichment and Paradigm pathway inference analyses confirmed that these genes are significantly enriched among responders. In the validation set, 20 of 46 genes (43%) were significantly differentially expressed.
Conclusion: We identified a panel of genes associated with immunogenic cell death and demonstrated their differential expression in patients who responded to neoadjuvant chemotherapy vs. non-responders. These results suggest a role for the host immune system in the response to chemotherapy. If the success of currently used chemotherapies depends on synergistic interactions with the immune system, then a better understanding of the mechanisms involved in chemotherapy-induced immunogenic cell death will provide a foundation for the design of novel chemo-immunotherapeutic strategies for breast cancer treatment.