Comparing nodal versus bony metastatic spread using tumour phylogenies

Abstract

The role of lymph node metastases in distant prostate cancer dissemination and lethality is ill defined. Patients with metastases restricted to lymph nodes have a better prognosis than those with distant metastatic spread, suggesting the possibility of distinct aetiologies. To explore this, we traced patterns of cancer dissemination using tumour phylogenies inferred from genome-wide copy-number profiling of 48 samples across 3 patients with lymph node metastatic disease and 3 patients with osseous metastatic disease. Our results show that metastatic cells in regional lymph nodes originate from evolutionary advanced extraprostatic tumour cells rather than less advanced central tumour cell populations. In contrast, osseous metastases do not exhibit such a constrained developmental lineage, arising from either intra or extraprostatic tumour cell populations, at early and late stages in the evolution of the primary. Collectively, this comparison suggests that lymph node metastases may not be an intermediate developmental step for distant osseous metastases, but rather represent a distinct metastatic lineage.

Figure 1

(A) Heat map of gene copy numbers across chromosomes. The copy number profiles were consistently fragmented across samples to allow comparative analyses. (B) Loess regression of the CNA profiles of the sampled clones adjusted for total ploidy. The value of 0 of the y axes represents no increase/decrease compared to a diploid genome.

Figure 2. Cancer evolutionary hierarchies for patients 167, 179, 421 (including lymph node metastasis) and 001, 299, 498 (including distant metastasis).

The roman numbers indicate the predicted ancestral cell populations. The decimal numbers at branches represent the confidence values of the clade prediction (when smaller then 1).

Figure 3. Loess regression of the increase/decrease CNA profiles, obtained from the difference in copy numbers across transition between predicted ancestral representative of the four histo-pathological categories (i.e., from diploid-to-central tumour, central-to-EPE, EPE-to-lymph node metastasis, and central/EPE-to-bony metastasis) for each patient (one per patient, per category).

The value of 0 of the y axes represents no variation compared to the parent histo-pathological category. The light-blue shades represent the genomic regions having a differential amount of copies across the four transitions (p-value <0.1, after method Benjamini, Hochberg, and Yekutieli multiple test correction44).

Figure 4. Multi-dimensional scaling plot of the centroids of the increase/decrease copy number profiles representative of the transitions between the four histo-pathological categories (i.e., from diploid-to-central tumour, central-to-EPE, EPE-to-lymph node metastasis, and central/EPE-to-bony metastasis).

Representative copy number profiles for distant metastases were taken from patients 001, 177, 299, 498; representative copy number profiles for the other histo-pathological categories were taken from patients 167, 179 and 421.