NEW YORK (GenomeWeb) – Through a molecular analysis, researchers from the Cancer Genome Atlas Research Network have found that three-quarters of primary prostate cancers fall into one of seven subtypes.
A team led by Memorial Sloan-Kettering Cancer Center's Chris Sander examined more than 300 primary prostate cancers using seven genomic platforms. As Sander and his colleagues reported today in Cell, most tumors fell into certain subtypes based on the gene fusions or mutations they harbored. They also noted substantial epigenetic heterogeneity and variation in androgen receptor activity among tumors. Some of these alterations could represent potential treatment targets, they added.
"[O]ur integrative assessment of 333 primary prostate cancers has confirmed previously defined molecular subtypes across multiple genomic platforms and identified novel alterations and subtype diversity," Sander and his colleagues wrote in their paper. "It provides a resource for continued investigation into the molecular and biological heterogeneity of the most common cancer in American men."
Using a combination of whole-exome sequencing, arrays, and mRNA sequencing, the researchers examined and characterized the somatic mutations, copy-number alterations, DNA methylation status, and transcriptomes of 333 primary prostate cancer samples. For a subset of those samples, they also performed microRNA sequencing and reverse-phase protein array analysis, and, for a set of tumor-normal pairs, they turned to low-pass and high-pass whole-genome sequencing.
Overall, the researchers reported unsupervised clustering of results from each of these platforms, and integrative clustering using iCluster could assign 74 percent of the tumors to one of seven molecular classes based on their oncogenic drivers: ERG, ETV1, ETV4, or FLI1 fusions, or SPOP, FOXA1, or IDH1 mutations.
Slightly more than half of the tumors contained an ETS family gene fusion, a fusion involving ERG, ETV1, ETV4, or FLI1. TMPRSS2 was the most common fusion partner, though fusions involving SLC45A3 and NDRG1 also occurred, the researchers noted.
Still, about a quarter of tumors the researchers analyzed appear to have been driven by as-yet-unknown molecular alterations.
A portion of these tumors, they said, seemed to be enriched for copy-number alterations or DNA hypermethylation. The most frequent abnormality involved recurrent gains of chromosomes 7 and 8q and heterozygous losses of chromosomes 8p, 13q, 16q, and 18. Recurrent focal amplifications spanned oncogenic genes like MYC and CCND1, while losses included the region harboring the PTEN locus.
While the researchers weren't able to perform an outcome analysis, they noted that the subset of tumors with the greatest burden of somatic copy-number alterations had higher Gleason scores and PSA levels.
Clustering analysis also revealed four prostate cancer groups that were epigenetically distinct, which the researchers then overlaid onto the mutational data they'd gathered.
From this, they noted that some two-thirds of ERG-fusion tumors belonged to a cluster of tumors with only moderately raised DNA methylation levels. The other third, though, belonged to a distinct hypermethylated cluster. That cluster exhibited nearly twice the number of hypermethylated sites as the first and was nearly exclusively associated with ERG-fusion tumors.
This, Sander and his colleagues added, supports further ETS fusion-based subtyping of prostate cancer and also suggests there's greater molecular and biological diversity among ERG fusion tumors than thought.
Meanwhile, SPOP and FOXA1 mutant tumors showed homogenous epigenetic profiles, and IDH1 mutant tumors had elevated levels of DNA hypermethylation.
By combining their epigenetic and mRNA data, the researchers unearthed 164 genes that were silenced in subsets of their cohort. For instance, they noted that SHF, FAXDC2, GSTP1, ZNF154, and KLF8 were silenced in nearly all tumors, though STAT6 was silenced mostly in ETS fusion-positive tumors and HEXA in SPOP mutant tumors.
Sander and his colleagues also reported that androgen receptor activity was variable in primary prostate tumors — primary prostate cancer is, they noted, androgen dependent. ETS fusion groups had variable AR transcriptional activity, they reported, while tumors with SPOP or FOXA1 mutations had the highest AR transcriptional activity.
These changes in primary prostate tumors also present potential clinical targets. Some 19 percent of tumors, the researchers said, had inactivated DNA repair genes that could be targeted by therapeutics. For instance, BRCA2 inactivation affected about 3 percent of tumors, while 1 percent of tumors had loss-of-function mutations or homozygous deletions of CDK12 and 3 percent of tumors had mutations affecting RAD51C.
Other potential clinically actionable lesions are those affecting PI3K and Ras signaling, the researchers added. PTEN, they noted, was deleted or mutated in 17 percent of their samples.
About a quarter of tumors also harbored mutations affecting genes involved in the MAPK pathway.
"As DNA sequencing of tumor samples becomes more widely adopted earlier in the clinical care of cancer patients, such alterations may emerge as candidates for inclusion in clinical trials after front-line therapy," Sander and his colleagues wrote.