AUA 2017: Leveraging the Power of Ultra-Deep Sequencing (When Deep Sequencing Fails)
His group (along with YZ Wang, PhD) has leveraged patient-derived xenograft (PDX) murine models to help functionalize heterogeneity. He and his collaborators took five separate biopsies from a single patient and established five different PDXs with different growth rates and metastatic capability. Interestingly, the largest primary never metastasized in the murine model, calling into question that purely anatomic criteria can be used to define the "index lesion". Instead, he hypothesized that the molecular characteristics of these tumors would explain their differential biology and aggressiveness.
Using these PDXs, his group performed whole genome sequencing (60X), transcriptome sequencing, and mass spec to identify drivers of metastasis. DNA mutational and copy number analysis did not explain the metastatic phenotype. Neither did gene expression. He was extremely frustrated by this and generated an alternative hypothesis that the stromal compartment drives metastasis and was missed by the human cell-centric bioinformatics approaches described above. His group found that mouse stroma replaces human stroma in PDX models and contributes 4-8% of the cellular volume. Species-specific sequences allowed differential identification of stromal signal on RNA sequencing.
By focusing on stromal gene expression, they found a correlation with metastatic phenotype that had been missed with tumor-specific approaches. They then took this stromal expression profile, and looked at these genes in human primary prostate cancer samples. Indeed, they found that the stromal expression profile seemed to be associated with metastatic outcome. Focusing on gleason 7 patients in conjunction with GenomeDx, they were able to predict metastatic biology in humans. They then asked whether combining tumor based signatures and stromal signatures could improve the prognostic ability, and sure enough it did. The stromal signature included cell-cell and cell-matrix interactions (LGALS1, LGALS3, SPP1, ADAM12 and COL14A).
He concluded by stating that the stromal gene signature seemed to be independent of grade and PSA and that it may be reflective of a field effect of the prostate as a whole (but this has not yet been proven).
These are provocative experiments that remind us of the importance of the stromal component. However, as the mouse models are genetically identical, I would posit that the tumor is driving these stromal phenotypes, and studies to identify these paracrine factors released by the tumors are ongoing.
Presented by: Colin Collins, PhD University of British Columbia
Contributed by: Jed Ferguson, MD/PhD and Ashish Kamat, MD. MD Anderson Cancer Center, Department of Urology.
at the 2017 AUA Annual Meeting - May 12 - 16, 2017 – Boston, Massachusetts, USA