Whole-exome sequencing of advanced and lethal prostate cancer tumors has led to the identification of a number of genes that are regularly altered, researchers report. The team, at the University of Washington and the Fred Hutchinson Cancer Research Center, in addition found that a subset of prostate cancer tumors exhibit hypermutated genomes, which may be linked with drug resistance.
Reporting in PNAS, the University of Washington’s Jay Shendure, M.D., Fred Hutchinson’s Peter S. Nelson, M.D., and colleagues, say comparison of castration-resistant and castration-sensitive cells derived from the same prostate cancer has highlighted mutations in the Wnt pathway as potentially contributing to the development of castration resistance. Their results are described in a paper titled “Exome sequencing identifies a spectrum of mutation frequencies in advanced and lethal prostate cancers.”
Drs. Shendure, Nelson et al. carried out whole-exome sequencing to investigate the mutational landscape of 23 prostate cancers from 16 different lethal metastatic tumors including three high-grade primary carcinomas. They separately compared the exomes of three tumors representing castration-resistant variants of original cancers. All tissues were propagated in immunocompromised mice as tumor xenografts to model the heterogeneity in tumor growth, response to treatment, and lethality.
Because corresponding normal tissue wasn’t available for a number of the tumor samples, the researchers only sequenced tumor tissue but discarded genes that mapped to the mouse genome, as well as all variants that had been identified in the pilot dataset of the 1,000 Genomes Project and those present in any of some 2,000 additional exomes sequenced at the University of Washington.
The team then narrowed down the resulting set of novel nonsynonymous single nucleotide variants (nov-nsSNVs) to identify the most likely protein-altering point mutations across different tumors. This reduced the 14,705 novel variants identified in the initial sequencing phase down to 20 genes in which nov-nsSNVs were found in two or more exomes and 10 genes with nov-nsSNVs in three or more exomes.
To segregate candidate genes further, they annotated positions in terms of conservation, using the Genomic Evolutionary Rate Profiling (GERP) score. This technique predicted which variants at highly conserved positions would be functionally significant and resulted in the identification of a subset of best candidates including several previously identified mutations in advanced prostate cancer (including the top candidate TP53) and others ( including DLK2 and SDF2) that have been linked with tumorigenesis but not with prostate cancer per se.
Nov-nsSNVs in TP53 were found in 5 of the 16 independent tumors used to evaluate recurrence, and these variants were predicted to cause premature termination of the protein. Three tumors harbored nov-nsSNVs within the gene encoding DLK2, a protein that shares similarity with the delta transcription factor and has recently been shown to be involved in notch1 signaling during development, the authors note. Three tumor genomes encoded variants in the calcium-binding protein stromal-derived factor (SDF4). Non-nsSNVs found in SDF4 and DLK2 included variations in conserved regions.
Other genes that harbored nov-nsSNVs in at least two or more advanced prostate cancers included CDH15, LAMC1, and GPC6. These were recently identified in a whole-genome sequencing study of localized primary prostate cancers.
A comparison of the exomes of castration-sensitive tumors with their castration-resistant derivatives identified 12–50 genes with nonsynonymous mutations that were only present in the castration-resistant xenografts, even though each of these genes had also been identified in at least one of the 16 independent tumors.
Of note, the castration-resistant tumors displayed a significant enrichment for genes involved in in Wnt signaling: of 86 mutations found in CRPCs but not the castration-sensitive cancer from which they were derived, each tumor had at least one mutation in a member of the Wnt pathway. These included FZD6, GSK3B, and WNT6.
The genomes of three prostate cancers displayed a nearly 10-fold higher number of nov-nsSNVs than the other advanced or lethal prostate cancer tumors. These hypermutated tumors were consequently excluded from the initial analysis and filtering and evaluated separately. Interestingly, the researchers report, none of the three tumors demonstrated distinguishing features: All were derived from Caucasian patients, one represented a primary neoplasm, one a lymph node metastasis, and one a liver metastasis.
One of the tumors was evaluated further to confirm that the hypermutator phenotype arose before passaging in mice. Further characterization of this tumor indicated that the pattern of somatic mutations was heavily dominated by transition mutations, with G→A and C→T transitions accounting for greater than 70% of mutations observed.
The mutation frequencies in these hypermutated tumors far exceed those found in primary prostate cancers and in most cancers such as breast, pancreas, and brain for which comprehensive exome or genome sequencing has been carried out, the authors point out.
However, they add, “cancers in the colon with mismatch repair gene defects and those that arise in the lung and skin, where environmental genotoxins like tobacco or UV sun exposure are implicated in disease etiology, have numbers of mutations that approach those present in these hypermutated prostate cancers.” Even so, the patterns of mutations observed in the hypermutated prostate tumors didn’t match those that occur in cancers associated with tobacco exposure.
One potential explanation for the large number of mutations in the three prostate cancer samples is the acquisition of a mutator phenotype, in which alterations in DNA polymerase or DNA repair genes result in an accelerated rate of mutations, they suggest. Supporting this notion was the finding that one of the hypermutated tumors possessed three candidate mutations in MSH6, a gene known to promote mismatch repair and microsatellite stability, and which has previously been implicated in prostate cancers with increased overall mutation rate.
However, the other two hypermutated tumors didn’t appear to harbor non-nsSNVs within DNA mismatch repair genes. “Thus, a plausible explanation for the elevated mutation frequencies in these cancers remains to be established,” the authors admit.
“Collectively, our results indicate that point mutations arising in coding regions of advanced prostate cancers are common, but with notable exceptions, very few genes are mutated in a substantial fraction of tumors,” they conclude. “Our results also suggest that increasingly deep catalogs of human germline variation may challenge the necessity of sequencing matched tumor normal pairs.”