NEW YORK (GenomeWeb News) – An international team of researchers has used genomic profiling to identify expression patterns and copy number changes associated with aggressive endometrial cancers.
A team of Norwegian, American, and German researchers assessed genome-wide expression profiles, copy number changes, clinical data, and histopathological information for dozens of endometrial cancer samples from a Norwegian tissue bank. Using this approach, they uncovered a 29-gene signature that classified tumors in two clusters — one linked to poor outcomes and shorter disease-free recurrence times and another linked to better outcomes.
The team also identified pathways that are amplified in poor outcome endometrial cancers and found that over-expression of a protein called stathmin 1 was an independent prognostic marker for poor outcomes. The research is scheduled to appear online this week in the Proceedings of the National Academy of Sciences.
"Endometrial cancer is a great cancer to work on because, frankly, I think it's understudied," senior author Rameen Beroukhim, a researcher affiliated with Harvard Medical School, the Dana-Farber Cancer Institute, and the Broad Institute, told GenomeWeb Daily News. Now that information about the human genome is available, he added, there are new possibilities for understanding what gives rise to cancers.
Endometrial cancer, cancer of the cells that line the uterus, affects roughly two to three percent of women and is the most common pelvic gynecological cancer in the industrialized world. Although about three quarters of cases are caught when they're still contained within the uterus, up to a fifth of the tumors recur after surgery.
Beroukhim argued that the key to improving treatments for endometrial and other cancers will be figuring out the underlying problems in cancerous cells rather than simply trying to find compounds that kill more cancerous than normal cells.
For instance, endometrial cancers are usually classified as type I or type II tumors based on tumor stage and grade, endometrial histology, and prognosis. But the molecular and genetic characteristics of type I and type II tumors are still murky.
In an effort unravel the molecular underpinnings of endometrial cancer, the researchers used Agilent microarrays to assess gene expression in 57 endometrial carcinomas from a Norwegian population-based tissue bank.
In the past, Beroukhim noted, it has been difficult to do these types of studies given a limited availability of tissue, especially in the US, where tumors are often removed by a community surgeon and fixed in wax, decreasing the quality of the DNA and RNA in the sample.
Developing systems to collect tumor tissues and clinical data and properly annotate that information also takes a great deal of time, money, and organization, he explained. In the past five or ten years there have been efforts to improve tissue collection and an increase in fresh frozen tumor samples. But because follow-up times can last many years, tumor banks that were created up to a decade ago are just starting to yield results, Beroukhim said.
The researchers found two distinct gene expression profiles, dubbed cluster 1 and cluster 2, which showed two-fold or more differential gene expression at 138 genes: 64 of the genes were up-regulated in cluster 2, while 74 were down-regulated. They also pinpointed a 29-gene group that could accurately classify tumors into one of the two clusters.
Individuals with cluster 2 tumors had significantly worse outcomes and shorter recurrence-free survival times than those with cluster 1 tumors. Compared to the traditional classification scheme, cluster 2 tumors represented all of the type II tumors as well as about a third of the type I tumors.
"Segregation into cluster 2 predicted recurrence better than FIGO stage, histologic subtype, ore receptor status and slightly less well than grade but did not exhibit independent prognostic impact, probably because of the limited number of cases and events," the authors explained.
Next, the researchers used Affymetrix arrays to look at copy number changes in 84 tumors from the tissue bank. After evaluating 11 amplified and 13 deleted regions of the genome in endometrial cancer, the team found two amplifications — containing KRAS and the PI3K activator PIK3CA, among other genes — that were linked with poor survival and shorter recurrence-free survival times.
Whereas previous animal studies suggested mutations in PIK3CA could activate the PI3K pathway, no one had reported links between amplification or over-expression of the normal form of PIK3CA and these cancers, Beroukhim said. In contrast, the researchers found that poor outcome tumors tended to have amplifications in PIK3CA while PIK3CA mutations were not detected in the poor outcome group. That suggests more research is needed to distinguish between lab effects and actual human tumors, Beroukhim said.
But it was another member of the PI3K pathway — stathmin 1, or STMN1 — that turned out to be an independent prognostic marker for poor endometrial cancer outcomes. High STMN1 expression was linked to poor prognosis regardless of tumor stage, grade, and histology, patient age, and other prognostic indicators. Even in otherwise low-risk tumors, high stathmin levels were tied to poor outcomes.
Consequently, while the 29-gene signature can distinguish between aggressive and less-aggressive tumors and could have potential prognostic applications down the road, Beroukhim said he is most excited about STMN1, which may be clinically applicable relatively quickly.
Since STMN1 is an independent prognostic marker, he explained, it would be possible to stain for this protein quickly, easily, and affordably, making stathmin a more viable assay in the short term. "If you stain for stathmin, you get additional information that can be used to identify patients at most risk," Beroukhim said.
In the future, Beroukhim and his co-workers hope to do similar genomic profiling studies on a larger number of tumors, perhaps hundreds or even thousands. They also plan to delve into the biology behind some of the associations identified in the current study. "A lot more effort should go into understanding endometrial cancer on a number of levels," Beroukhim said.