Intra-Tumor Genetic Heterogeneity in Wilms Tumor: Clonal Evolution and Clinical Implications

Abstract

The evolution of pediatric solid tumors is poorly understood. There is conflicting evidence of intra-tumor genetic homogeneity vs. heterogeneity (ITGH) in a small number of studies in pediatric solid tumors. A number of copy number aberrations (CNA) are proposed as prognostic biomarkers to stratify patients, for example 1q+ in Wilms tumor (WT); current clinical trials use only one sample per tumor to profile this genetic biomarker. We multisampled 20 WT cases and assessed genome-wide allele-specific CNA and loss of heterozygosity, and inferred tumor evolution, using Illumina CytoSNP12v2.1 arrays, a custom analysis pipeline, and the MEDICC algorithm. We found remarkable diversity of ITGH and evolutionary trajectories in WT. 1q+ is heterogeneous in the majority of tumors with this change, with variable evolutionary timing. We estimate that at least three samples per tumor are needed to detect >95% of cases with 1q+. In contrast, somatic 11p15 LOH is uniformly an early event in WT development. We find evidence of two separate tumor origins in unilateral disease with divergent histology, and in bilateral WT. We also show subclonal changes related to differential response to chemotherapy. Rational trial design to include biomarkers in risk stratification requires tumor multisampling and reliable delineation of ITGH and tumor evolution.

Keywords: Copy number aberrations; Intra-tumor genetic heterogeneity; Molecular biomarkers; Pediatric solid tumors; Tumor evolution; Tumor multisampling; Wilms tumor.

Fig. 1

Scope of intra-tumor copy number heterogeneity in 20 Wilms tumor cases. Case numbers, indicated on the y-axis, are split into three groups i) homogeneous, ii) bilateral and iii) heterogeneous. Chromosomes 1 to 22 and X are ordered on the x-axis from left to right, separated by vertical dashed lines. Cases are separated by solid black horizontal lines and samples from each case by dashed horizontal marks at each chromosome boundary. In the bilateral group, samples from contralateral kidneys are split by horizontal markers either side of the plot. Copy number states are displayed as white (expected copy number state), red (gain state), blue (loss state) and grey (expected copy number state with loss of heterozygosity).

Fig. 2

Branched evolution of a multi-sampled Wilms tumor from Case 19. (A) Phylogenetic tree displaying the evolutionary relationship between the normal kidney sample (NK, green node) and tumor samples (R1–6, red). Vertical edges are weighted by the number of copy number alterations that were acquired as the tumor evolved and the events themselves are labelled next to the appropriate edge. Horizontal edges are not weighted and are used to separate graphically the nodes. (B) Copy number profiles of the normal kidney and tumor samples. Plots display the Log R ratio on the y-axis and chromosomes 1–22 and X on the x-axis. Data are shown for every tenth SNP probe. Data points corresponding to normal diploid copy number states are coloured grey, those representing copy number gains are in green and losses in dark red. (C) An annotated photograph of the opened nephrectomy specimen, showing the locations of the sampled tumor and normal kidney regions.

Fig. 3

Chromosome 1q gain (1q +) is variably heterogeneous in eight cases. (A) Each sample is shown as a circle, filled in green if there is 1q + and in white if there is not. Bilateral cases are split into contralateral samples with a horizontal black line. (B) Chromosome 1 copy number profiles for samples from Case 20. Log R ratios (y-axis) are shown for all chromosome 1 SNP probes, ordered by genomic position on the x-axis. Data points are coloured green if they correspond to copy number gains, and grey if not. Samples R1 and R4 display whole chromosomal arm gains.

Fig. 4

Phylogenetic analysis reveals two genetically distinct Wilms tumors within a single mass in Case 13. The phylogenetic tree shows the evolutionary relationship between the normal kidney sample (NK, green node) and tumor samples (R1–5, red). The edges were drawn using the same rules as in Fig. 2. The tree shows that sample R5 is genetically distinct from R1–4, since their only common ancestral state corresponds to the normal kidney sample. The histology of the middle nodule, represented by R3, is triphasic, with blastemal, epithelial and stromal elements, whereas the superior nodule, represented by R5, is markedly different and composed exclusively of more mature epithelial elements (original magnification, × 100).

Fig. 5

Wilms tumor phylogeny mirrors the differential treatment responses shown by different parts of the same tumor in Case 8. (A) Pre- and post-chemotherapy T2-weighted magnetic resonance (MR) images show the two masses indicated by red (small nodule) and orange (large nodule) dashed lines. (B) Estimated tumor volumes pre- and post-treatment shows that the small nodule (red) shrank by 78% and the large nodule shrank by 42% (orange). (C) The small nodule (red) showed a greater post-chemotherapy gain in the mean assessed diffusion coefficient (ADC) than the large nodule (orange), indicating a better response to chemotherapy. (D) A phylogenetic tree shows the genetic relationships between samples from the large nodule (R3–4, orange) and the small nodule (R1, red). The normal kidney sample is represented by a green node (NK). The small nodule is related to the rest of the tumor, but has evolved additional changes, including focal gain of MYCN.