| Summary: | <p>Immune signatures are important tools to investigate the prognostic significance and therapeutic possibilities of different immune populations in the context of the tumour microenvironment (TME). Accompanied by the wealth of cancer transcriptomic data, many immune signatures have been developed using different samples, technologies, and methods. But there has been limited effort to
comprehensively assess these tools in cancer, especially in the context of macrophage signatures.</p>
<p>In this study, I first exanimated the performance of six bioinformatic immune dissection tools (Chapter 3). I conducted in-depth comparisons on two of the most comprehensive and methodologically distinctive tools (CIBERSORT and xCell) in 33 different cancer types across the Cancer Genome Atlas (TCGA). The effort identified high agreements on five immune subsets and an unexpected positive relationship between the theoretically polarized M1 and M2 signatures. Subsequently, I assessed the impact of tumour hypoxia on infiltrating immune cells (Chapter 4), since hypoxia is a hallmark of the immunosuppressive microenvironment. I uncovered a reciprocal relationship between tumour hypoxia and the stromal contents within the breast cancer TME.</p>
<p>In the work concerned with tumour associated macrophage (TAM) (Chapter 5), I systematically compared 16 different ex vivo and in vivo derived macrophage signatures. This comparison indicated these macrophage signatures have high divergence and the ex vivo derived signatures were not representative of TAMs in breast cancer. Following this, I derived a new independent TAM signature from single-cell data, which featured lipid metabolism and lysosomal activities. I further refined and improved this signature using the least absolute shrinkage and selection operator (LASSO). Using the refined TAM signature, I demonstrated the lipid-associated macrophages were immunosuppressive, enriched in aggressive HER2-enriched and Basal-like subtypes, and associated with poor prognoses across breast cancer subtypes. Therefore, this lipid-associated TAM could be a promising target in breast cancer.</p>
|
|---|