Fig. 6.

Morphologic evolutionary trajectories in ccRCC. (a) Architectural evolution model, t-SNE plot based on the architectural composition of each tumor, dashed arrow indicating hypothesized evolution directions with overlaid (b) tumor size, (c) nucleolar grade. Each point on the plot represents a single sample, and the color is based on the dominant pattern present in that tumor. Tumors with a single pattern are indicated with a black border to represent possible truncal patterns. As t-SNE uses relative distances to generate embedding, the axes are essentially arbitrary and are hence not displayed; however, the same scale is used for both axes. The dashed arrow was drawn manually to represent the hypothetical evolutionary trajectory. (d) Morphologic patterns in thrombus evolution. Plot showing the percentage of dominant architectural patterns in primary tumors that seeded the tumor thrombus. Each arrow represents an individual ccRCC with a paired thrombus. The tail of the arrow represents the percentage of that architecture in the primary tumor, and the head of the arrow represents the percentage in the paired thrombus. (e) Comparison of the engraftment rates in mice based on architectural pattern. Plot showing the percentage of successful stable (lines that could be passaged at least twice in mice) engraftment from 174 ccRCCs implanted in mice and the architectural pattern observed in the flanking H&E stained sections. The odds ratio for engraftment of aggressive patterns as compared to indolent patterns was 3.02 (95% CI 1.31, 6.92; p = 0.0104).