Disseminated tumors in metastatic prostate cancer (mPC) share a common monoclonal origin. However, the selective pressure exerted by therapies promotes diversity as resistance mechanisms emerge. This heterogeneity adds to the already diverse spectrum of oncogenic aberrations present in localized prostate cancer (PC), which have been well characterized by several approaches, including single-cell sequencing, proteomics, and spatial transcriptomics. Still, characterization of molecular diversity at the intraindividual and intratumoral level in mPCs is limited, and further characterization could inform recurrence risk assessment and treatment selection.
Researchers in the Nelson lab in the Human Biology Division partnered with the cancer diagnostic tool company NanoString Technologies, Inc. to characterize the intra- and inter- tumor variation in a cohort mPCs patients using digital spatial gene expression profiling (DSP), a technique that can attribute RNA and protein expression to discrete features in a tissue of interest. Their study is now published in the journal Nature Communications.
The researchers, led by postdoctoral fellow Dr. Lauren Brady and NanoString scientist Dr. Michelle Kriner, first build tissue microarrays from two distinct dissemination sites in 27 patients with mPC. They selected three spatially distinct regions from each tumor (tumor cores) and then, guided by histological analyses to identify the largest percentage of tumor cells, identified a region of interest (ROI) in each tumor core. This approach limited the inclusion of benign cells, fat, or stroma. Nonetheless, each ROI was categorized based on its cellular composition (i.e., tumor cells per sampled area). The DSP panel included mRNA or protein probes for over 2000 unique genes encoding molecular targets for specific therapeutics, tumor phenotype markers for androgen receptor (AR) and neuroendocrine (NE) differentiation activity, immune cells markers, and immunomodulatory messenger molecules.
Metastatic PCs are classified into six phenotypic classes based on the expression of AR-regulated genes and NE-associated genes, which promote survival and proliferation of PC cells and are involved in therapeutic resistance mechanisms, respectively. All of the tumors evaluated by DSP profiling had matched bulk tumor whole-transcriptome RNA-seq data, which the authors then compared with the DSP findings. DSP assignment of mPC tumors to AR/NE-defined classes matched the RNAseq assignment, as well as the expression of individual genes involved in PC pathobiology or genes that serve as therapeutic targets. However, detecting the splice variant AR-V7, which promotes resistance to treatment, was variable between DSP, bulk tumor RNAseq, and immunohistochemistry, with DSP showing greater heterogeneity.
DSP classification in disseminated tumors within the same patient was primarily uniform, but some patients had more than one mPC type. In patients with mixed tumor classes, the investigators observed divergence in individual gene expression for transcripts involved in NE differentiation, suggesting that distinct therapy resistance mechanisms might arise within individual patients, which may require treatment combinations. Among patients, DSP revealed that gene expression of individual therapeutic targets varied significantly, suggesting that patients might benefit from precise target expression evaluation.
Dr. Brady shared future directions with DSP as a tool for intra- and inter-tumor heterogeneity: “Our next steps are to further harness the capability of digital spatial profiling by examining the interactions between tumor cells and the surrounding microenvironment. Building on our existing dataset, that primarily focused on intra-tumoral regions, we are currently investigating expression profiles within peri-tumor regions and regions distal to the tumor cells – with a specific focus on immune composition across the different phenotypes of advanced prostate cancer.”