Clonal evolution of chemotherapy-resistant urothelial carcinoma

Abstract

Chemotherapy-resistant urothelial carcinoma has no uniformly curative therapy. Understanding how selective pressure from chemotherapy directs the evolution of urothelial carcinoma and shapes its clonal architecture is a central biological question with clinical implications. To address this question, we performed whole-exome sequencing and clonality analysis of 72 urothelial carcinoma samples, including 16 matched sets of primary and advanced tumors prospectively collected before and after chemotherapy. Our analysis provided several insights: (i) chemotherapy-treated urothelial carcinoma is characterized by intra-patient mutational heterogeneity, and the majority of mutations are not shared; (ii) both branching evolution and metastatic spread are very early events in the natural history of urothelial carcinoma; (iii) chemotherapy-treated urothelial carcinoma is enriched with clonal mutations involving L1 cell adhesion molecule (L1CAM) and integrin signaling pathways; and (iv) APOBEC-induced mutagenesis is clonally enriched in chemotherapy-treated urothelial carcinoma and continues to shape the evolution of urothelial carcinoma throughout its lifetime.

Figure 1. Clinical characteristics of study cohort

(a) Bar graph illustrating number of tumor samples sequenced from each study subject. Treatment status of each sample is color-coded within each bar (see inset). Gender, smoking status development of metastases and history of intravesical Bacillus Calmette-Guerin (BCG) for pre-existing non-muscle invasive bladder cancer are represented on the bottom (b) Schematic illustrating anatomical sites of primary and metastatic tumor samples.

Figure 2. Clonal mutational heterogeneity in chemotherapy-treated UC

(a) Percentage of shared and unique SNVs in matched chemotherapy-naive and chemotherapy-treated UC tumors. (b) Discordance in the mutational status in selected driver genes between chemotherapy-naive and chemotherapy-treated UC tumors. Each column represents a paired set of pre- and post-chemotherapy tumors obtained from an individual subject. Individuals with more than one paired set of tumors are highlighted with matching color designations. Primary-primary tumors are represented on the left (vertical hatching) and primary-metastatic tumors are represented on the right (slanted hatching). Note: ERCC2 missense mutations were identified in two tumor samples, WCM259_3 (p.Q755H) and WCM141_1 (p.T484M), matched pre- and post-chemotherapy samples were not available from either patient.

Figure 3. Early branching evolution in UC

Phylogenetic trees (top), shared and private clonally-adjusted mutations (bottom) from 6 patients with three or more tumor samples per patient. All cases showed early metastatic spread occurring as branching evolution occurred early in the natural history of the disease. All primary tumors temporally preceded the development of metastasis but were assigned “branch” values based on genomic distance suggesting early branching evolution and metastatic spread.

Figure 4. Reconstructing the spatio-temporal evolution of UC over time and through different treatments

Analysis of 12 tumor samples collected during disease progression and at the time of death of patient WCM117 (a) (top) Circles represent sites of sequenced tumors. (bottom) Timeline and clinical course (vertical lines) in the natural history of the disease. (b) CLONET-adjusted variant allele frequencies of selected mutations. Each dot represents the fraction of cells harboring the corresponding mutation. (c) Fluorescence in-situ hybridization for CDKN2A (red) and reference (green) probes in primary cystectomy tumor #4 (right) and right supraclavicular lymph node collected at time of autopsy (left) (scale bar 5 μm). Middle panel: allele-specific copy number. Axes correspond to different alleles. Dots represent CDKN2A allele-specific copy number. (d) CLONET-adjusted shared and private mutations. Fractions of tumor cells harboring each mutation represented by shades of green (scale upper left corner of panel). Gray: clonality information was not available. (e) Reconstruction of evolutionary tree. WT: taken last ancestor with wild-type genome that acquires a series of mutations during oncogenesis. The length of the branches represents the distance between two tumors based on the number of shared mutations. Samples are color-coded consistently across panels: Yellow: primary pre-chemotherapy TURBT. Orange: 4 different areas of post-gemcitabine cisplatin bladder tumors and pelvic lymph node metastatic lesion removed through cystectomy and lymph node dissection. Green: post-docetaxel-ramucirumab metastatic lesions removed during autopsy.

Figure 5. Hierarchical clusters of 44 UC tumor samples by copy-number alterations

(a) Copy number gains are represented in red and copy number losses are represented in blue. Each column represents one tumor sample. Clinical annotations are represented on top. Matched samples from the same patient are represented in the same color in the “matched samples” annotation track. Selected genes harboring frequent copy number alterations are listed on the left. (b) Dots represent pairs of samples in Fig. 5a from the same patient (left) and from different patients (right). Left boxplot: median 0.2, IQR (0.13, 0.28), whiskers (0.05, 0.49). Right boxplot: median 0.53, IQR (0.42, 0.66), whiskers (0.07, 0.93).

Figure 6. Clonal enrichment of mutations in chemotherapy-treated UC

(a) Density plot representing clonality of non-silent single nucleotide variants (SNVs) on the X-axis and the density distribution on the Y-axis. Pre-chemotherapy tumor samples are represented in blue color and post-chemotherapy tumor samples in green. Box plots (right) represent the percentage of clonal SNVs in pre-chemotherapy tumor and post-chemotherapy tumors demonstrating a significant increase in clonality in post-chemotherapy UC tumors. Left boxplot: median 0.5, IQR (0.41, 0.71), whiskers (0.24, 0.88). Right boxplot: median 0.83, IQR (0.61, 0.89), whiskers (0.22, 0.94). (b) Results of Gene set Enrichment Analysis (GSEA) of mutations in chemotherapy-treated UC. Each node represents one REACTOME pathway. Node size represents number of genes in each pathway. Enriched pathways post-chemotherapy samples are represented by nodes in red color.

Figure 7. Mutagenesis in advanced UC is shaped by chemotherapy and APOBECs

(a) Composite bar graphs representing the distribution of all possible nucleotide substitutions in non-silent SNVs in sequenced pre-chemotherapy (left) and post-chemotherapy (right) UC tumors. (b) Nucleotide motif contexts for each category of single nucleotide substitutions. Blue: pre-chemotherapy tumors, green: post-chemotherapy tumors. (c) Four mutational signatures identified in UC. (d) Significant enrichment of APOBEC signatures in post-chemotherapy tumors. (e) Significant increases in the clonality of APOBEC-induced mutations in post-chemotherapy UC. Number of SNVs, exact p-values, and total number of samples are reported in Supplementary Note.