Results from The Cancer Genome Atlas' analysis of several hundred endometrial cancers have classified the disease into four molecularly defined groups with differing prognoses. According to the study, some endometrial tumors with similar histologic features actually differ in their molecular profile and might benefit from different treatments.
While the new classification does align in some ways with earlier histological categories, some patients that would have been classified in a lower-risk histological category according to current subtypes would now be grouped in the category with the worst prognosis according to the molecular results from TCGA's study published today in Nature. For patients that fall in this molecular subtype, this finding could mean that they should receive alternative treatments.
"One of the problems is that, especially for higher-grade tumors, it is difficult for pathologists to classify them. There can be disagreement among pathologists and those have important ramifications on what treatments we recommend," the study's lead author, Douglas Levine from Memorial Sloan-Kettering Cancer Center, told PGx Reporter this week. "There are cases that are just difficult and this is where molecular information can help to add another layer of clarity."
The latest study also separated a small group of subjects into a new category that looks to have extremely good prognosis. If the results hold over time, it could mean these patients, if identified based on their molecular signature, would require less or even no treatment compared to the other subtypes.
Current methods of histologic assessment classify an endometrial carcinoma as one of two types — either endometrioid tumors, with relatively better prognoses, or serous tumors, with relatively worse prognoses. Endometrioid cancers are most often treated with radiation, while serous tumors are treated with chemotherapy.
TCGA's new genomic data breaks these cancers instead into four groups: a new subgroup categorized by mutations in the gene POLE and an overall-high mutation rate; a group with high mutation rate and high microsatellite instability, but without POLE mutations; a group with high microsatellite instability but low rates of copy number alterations; and a group with high copy number alterations.
Overall, the study profiled 373 tumor samples using whole-exome sequencing, microarrays, RNA-seq, and other methods.
The POLE group, with 17 tumors, or less than 10 percent of the cohort, was associated with the best prognosis in the study, while those in the copy number-high group had the worst outcomes. The other two subgroups fell in the middle.
According to Levine, the fact that the fourth subgroup had much worse prognosis was not surprising. "The unique thing, he said, "is that we put some of the endometrioid cases with the serous cases based on the molecular features, suggesting that we can identify a subset of endometrioid cases that may have a poorer outcome."
Currently, known-serous cases frequently get treated with chemotherapy, but those classified as endometrioid, "depending what stage they are, they could get anything from nothing to radiation to chemotherapy and radiation," he said. "The suggestion from the molecular data is that maybe they should just get chemo like the serous group."
However, the molecular subgroups identified in this study must first hold up in terms of their association to cancer prognoses in follow-up investigations. "Once we do that we can then design trials that stratify based on these types."
He said researchers behind the study are planning to validate the findings in an upcoming prospective trial of chemotherapy and alternative treatments in endometrial cancer patients by measuring a variety of molecular targets to classify patients in the trial into the same four subtypes. They will evaluate whether the four subtypes hold true, that is, whether patients with the copy number-high subtype really do have worse outcomes than those in the other groups.
If so, it would suggest that some patients with these copy number-high or "serous-like" endometriod tumors might benefit from more aggressive treatment, as do those with clearly serous tumors.
The study authors suggested the results could provide a map for future clinical trials of targeted therapy. "Each tumor subtype might warrant dedicated clinical trials because of the marked genomic differences between them that are indicative of different drivers of cancer … and developing therapies for each subtype independent of the other may improve outcomes, as has been shown in breast cancer," said study co-leader Elaine Mardis of Washington University School of Medicine in a statement.
The new four-tier categorization of endometrial carcinoma is a highlight of the TCGA results, but the researchers also found other data that could potentially impact how physicians treat these tumors in the future.
Cancers in the copy number-high group — serous and serous-like endometrioid tumors — shared a number of molecular similarities with both serous ovarian tumors and basal-like, or triple-negative, breast cancers. For example, the cancers share a high frequency of TP53 mutations — between 84 and 96 percent — and a low frequency — 1 to 2 percent — of PTEN mutations.
"For ovarian serous tumors, the standard treatment is combination chemotherapy, which works very well," Levine said. "We use that same regimen for uterine serous tumors now. But in basal-like breast cancer, treatments are often different and some investigators want to test these in other groups. But it’s a question whether it will work as well in all these tumor types, because they do have all these similarities, but they also have lots of differences."
Additionally, Levine said, the study results shore up previous observations that the PI3K – AKT pathway is highly active in endometrial cancers. They also showed that the pathway is activated to a different extent in serous and in non-serous cases — significantly more so in non-serous tumors.
"Lots of drug companies [working on PI3 kinase inhibitors] are interested in studying endometrial cancer because they know this pathway has a lot of activation in this tumor type," he said. From the results, he said, "the pathway seems in fact to be so active in the non-serous cases that it might be hard to suppress it with only one of the targeted drugs available." That, he said, may be important information for these companies to use going forward.
In the group's upcoming validation study, Levine said he and his colleagues are prospectively collecting tumor samples in a three-arm trial involving just over 300 patients and testing various treatment methods and doing a "cadre of molecular tests" to reproduce these subtypes identified by the recent TCGA study. "We are screening for P53 mutations, POLE, PIK3CA, MSI — and we should be able to reproduce the four subsets within the context of this trial," he said.
"Then we can ask the question, do these endometrioid cases that have this molecular feature of being serous-like actually [do] worse— does that affect [their] response to treatment?"
If Levine and his colleagues are able to reproduce the findings, "the next trial will be to see if we give chemo to these patients who would normally get radiation or no treatment, does that actually make a difference? That would be step two of the validation process."