New Diagnostic Algorithms Have Prognostic Implications in Endometrial Cancer

Endometrial cancer tumor types are traditionally categorized through a combination of histologic molecular subtyping and clinical staging; however, the incorporation of new diagnostic algorithms may have clinical implications for treatment.

Endometrial cancer tumor types are traditionally categorized through a combination of histologic molecular subtyping and clinical staging; however, the incorporation of new diagnostic algorithms may have clinical implications for treatment, said Lauren Ritterhouse, MD, PhD, in a virtual presentation during the 4th Annual International Congress on Oncology and Pathology™.1

The most common way to classify endometrial cancer is through histological subtype, according to Ritterhouse, who is associate director of the Center for Integrated Diagnostics and an assistant professor of pathology at Massachusetts General Hospital, as well as an assistant professor at Harvard Medical School.

The most common histologic subtype is endometrioid-type endometrial cancer, which is graded by the percent of gland-forming architecture versus solid growth in the tumor. As such, low-grade tumors, grade 1 or 2, predominately have glandular growth patterns, while higher-grade tumors, grade 3 or higher, would have greater than 50% of solid-growth pattern.

Serous carcinoma of the endometrium is another subtype that is also characterized by glandular architecture and capillary growth, according to Ritterhouse, which is characterized by high cytologic and nuclear atypia. Another subtype, clear cell carcinoma of the endometrium, which is categorized by cleared cytoplasm and hobnailing of the cells into the lumen. Several other histologic subtypes exist in this space, including mucinous, undifferentiated/dedifferentiated, mixed-type, and carcinosarcoma; however, they are less common.

Clinical staging for endometrial cancer has traditionally been based on histology as well as FIGO staging, said Ritterhouse. “FIGO staging is how much the tumor has involved both the uterus and myometrial thickness, as well as either adjacent organs or distant organs,” said Ritterhouse. “These two [factors] are predominantly what is currently used to predict prognosis and therapy for [patients with] endometrial cancer.”

In the Cancer Genome Atlas Study in 2013, 4 molecular subtypes for endometrial cancer were proposed: POLE ultramutated, microsatellite instability (MSI) hypermutated, copy-number low, and copy-number high.2 These subtypes were based on driver mutations, molecular mutational signatures and profiling, and copy number alterations in tumor mutational burden (TMB), explained Ritterhouse.

The first type, POLE-mutated tumors, have a very high TMB and are characterized by a C > A transversion mutational signature. The microsatellite unstable tumors are known to have defects in mismatch repair and also have very high numbers of mutations, according to Ritterhouse. The copy-number high group did not have many mutations, although it had a high number of copy changes; this group was also characterized as having TP53 gene mutations. Notably, the study included many patients with serous carcinomas of the endometrium and it is known that these tumors typically harbor TP53 mutations. The final subtype was the copy-number low group, which is sometimes referred to as having no specific molecular profile due to its lack of mutations and dearth of copy number changes.

“While it is very interesting biologically that we can form these molecular subtypes of endometrial cancer, what made this study most intriguing was they were stratified by survival,” said Ritterhouse. Of the 4 subtypes mentioned, POLE-mutated tumors had an 100% progression-free survival in the particular cohort analyzed. Not surprisingly, the group that had many serous tumors, copy number alterations, and TP53 mutations had the worst survival of all 4 subtypes, according to Ritterhouse. The other 2 groups—MSI and copy-number low—were found to have intermediate outcomes.

Another observation made among these 4 molecular subtypes was that the copy-number high group, while it had many patients with serous tumors, it also had a notable number of patients with endometrioid cancers. POLE-mutated and MSI tumors were enriched for having higher-grade endometrioid tumors, while the copy-number low group was enriched for having lower-grade endometrioid tumors, explained Ritterhouse.

The POLE-mutated and the MSI unstable tumors had high TMB. The POLE-mutated tumors had hundreds of mutations per megabase, so very high mutational burden. “The reason I wanted to highlight this is because of the recent FDA approval that has to do with having high TMB,” said Ritterhouse. The MSI unstable tumors also had a high TMB, although the average was not as high as the POLE-mutated tumors, which averaged over 200 mutations per megabase.

The immunotherapy implications for these 2 mutational subtypes that have high TMB are associated with tumor agnostic FDA approvals. “For example, there's pembrolizumab (Keytruda), which has been approved for patients with unresectable or metastatic tumors who have progressed following prior therapy,” Ritterhouse explained. “The POLE-mutated tumors would fall under the recent FDA approval for tumors with TMB high or greater than 10 mutations per megabase.” Most of these tumors have hundreds of mutations per megabase, noted Ritterhouse, and as such, they would meet the criteria for immune checkpoint inhibition.

For MSI and POLE-mutated tumor types that have many mutations, a subset of those mutations will be expressed and translated into novel peptides that are then processed and loaded onto the MHC complex and recognized as a neoantigen by the immune system, explained Ritterhouse. The concept is that generating a large amount of neoantigens will help to stimulate T cells to kill tumor cells. By "releasing the breaks of the immune system," a synergistic response is achieved; this is why patients with POLE-mutated tumors are shown to do well, even without immunotherapy, explained Ritterhouse.

In another study, investigators did similar molecular subtyping to what had been done with the TCGA dataset, but they only focused on endometrioid histology.3 In this research, there were quite a few tumors that were high stage, but they also had low-stage tumors that had recurred, explained Ritterhouse. Patients with POLE-mutated tumors were found to have better outcomes compared with the other subtypes; however, some of those patients still experienced recurrence or died of their disease. With regard to the TP53-mutated tumors, even with endometroid histology, were still found to have worse outcomes. “That difference in outcome stayed the same, even when we exclude all the tumors with serous histology,” said Ritterhouse.

Additionally, investigators noticed that once tumor recurrence had occurred, the TCGA molecular subtypes might have less prognostic significance observed, at least with regard to the POLE-mutated, MSI, and no specific molecular profile groups. Regardless, patients with tumors that harbor TP53 mutations are still shown to have a worse outcome. “If there is a recurrence, this molecular subtyping might not have as much prognostic utility as it might in the up-front setting,” noted Ritterhouse.

Molecular subtyping can be incorporated into clinical practice through the implementation of diagnostic algorithms, said Ritterhouse. One such algorithm, which was published in the International Journal of Gynecologic Pathology,4 focuses on initially classifying patients through POLE hotspot mutation analysis. This can be approached in different ways; this can be done through finger sequencing of the 2 hotspots, as part of a next-generation sequencing targeted panel, or by utilizing SNaPshot, which is a cheaper, faster method to confirm a POLE hotspot mutation.

If patients test positive, they fall within the POLE-mutated molecular subtype, while negative testing would indicate the need to look at the DNA mismatch repair IHC. If deficient mismatch repair is observed, the tumor would be categorized as MSI-high. However, if the IHC is intact, this would indicate the addition of another IHC, which is P53, a commonly available and easy to biomarker to perform, according to Ritterhouse.

From there, if aberrant staining is observed, then the tumor would be categorized as copy-number high or TP53-abnormal molecular subgroup. Finally, if the tumor shows normal or wild-type staining, that puts it into the copy-number low group. “With the exception of the POLE hotspot mutation analysis, most of this would be pretty easy to incorporate into regular clinical practice,” Ritterhouse said.

Similar diagnostic algorithms have been tested in trials such as the PORTEC-4a study5, which was used to determine a molecular-integrated risk profile for patients with stage I high-intermediate risk endometrial carcinoma. Much like the aforementioned algorithm, patients with a POLE mutation were found to have favorable outcomes. Another molecular subtype found to fall into this favorable category were patients who did not have a mismatch repair deficiency; these patients were put into the category of CTNNB1 wild-type. Those who were found to have mismatch repair deficiency were categorized as having a CTNNB1 mutation, which indicated an intermediate prognosis. However, patients who were noted as having substantial LVSI, a TP53 mutation, or a >10% L1CAM expression, were ranked as having the most unfavorable prognosis.

Based on these findings, there is reason to believe that molecular subtyping combined with diagnostic algorithms hold not only not only interesting implications with regard to immunotherapy, but a better way to categorize and predict patient outcomes. Although more data are needed and implementing certain testing aspects of diagnostic algorithms may prove challenging in a clinical setting, there is potential to change practice, Ritterhouse concluded.


  1. Ritterhouse, L. New Diagnostic Algorithms and Implications for Therapy: Endometrial Cancer and Ongoing Trials. Presented at: 4th Annual International Congress on Oncology and Pathology™; June 20, 2020; Virtual.
  2. Kandoth C, Schultz N, Cherniack AD, et al. Cancer Genome Atlas Research Network Investigators. Nature. 2013;497(7447):67-73. doi:10.1038/nature12113
  3. Devereaux K. Molecular classification of metastatic and recurrent endometrial endometrioid carcinoma. USCAP 2020 Abstract.
  4. Cho KR, Cooper K, Croce S, et al. International Society of Gynecological Pathologists (ISGyP) Endometrial Cancer Project: Guidelines From the Special Techniques and Ancillary Studies Group. The International Journal of Gynecological Pathology. January 2019. Volume 38 Issue p S114-S122 doi: 10.1097/PGP.0000000000000496
  5. PORTEC-4a: Molecular Profile-based Versus Standard Adjuvant Radiotherapy in Endometrial Cancer (PORTEC-4a). Updated Feb. 11, 2020. Accessed June 22, 2020.