Genetic Landscape of Clear Cell Endometrial Cancer and the Era of Precision Medicine

Endometrial cancer is the 4th most common cancer and the 6th leading cause of cancer-related death in women in the United States. Histological types of endometrial cancer include endometrioid, serous, clear cell, and carcinosarcoma. Although endometrioid histology is the most common, the non-endometrioid types account for a disproportionate number of cancer deaths due to their aggressive clinical behavior. Understanding the molecular drivers of aggressive endometrial cancer is critical to the development of effective strategies to reduce morbidity and mortality from this disease.

Previously, The Cancer Genome Atlas (TCGA) Project performed an integrated genomic characterization of endometrioid and serous endometrial cancer.¹ This led to a molecular classification of endometrial cancer into 4 types defined by their nucleotide substitutions and patterns, microsatellite instability (MSI) status, and copy number alterations: POLE (ultramutated), MSI (hypermutated), copy number low (endometrioid), and copy number high (serous-like). Substantial biological differences distinguish the groups with respect to their genomic and proteomic aberrations as well as clinical outcomes. The poorest disease-free survival was observed in the copy number high (serous-like) group, which includes approximately 25% of high-grade endometrioid tumors. The serous and serous-like endometrioid tumors are characterized by frequent mutations in TP53, extensive copy number alterations, and few DNA methylation changes. In contrast, the majority of endometrioid tumors demonstrate few copy number alterations and rare mutations in TP53 but frequent mutations in PTEN and KRAS. Molecular classification strategies are presently undergoing broad implementation and evaluation in the clinical setting with the goal of improving accuracy in diagnosis and treatment of endometrial cancer.

Clear cell endometrial cancer (CCEC) is an uncommon but aggressive histologic type that accounts for 1%-6% of endometrial cancer. This tumor type was not included in the TCGA endometrial cancer study, and much less is known about the molecular drivers of this malignancy. The study by Le Gallo et al² in this issue of Cancer represents the largest effort to date to characterize the mutational landscape of CCEC. Whole exome sequencing was performed using paired tumor-normal DNA samples from 16 cases of CCEC to identify recurrently mutated genes. For 22 genes of interest, an additional 47 cases of CCEC were investigated to more precisely define mutational frequency. MSI status was evaluated at 5 sites of mononucleotide repeats. Somatic mutations were observed in genes previously identified in endometrioid and serous endometrial cancers, at intermediate frequencies (Table 1). The most commonly altered gene was TP53 (39.7%). This is consistent with a prior immunohistochemical analysis of p53 in CCEC, which showed altered p53 in 37.5% of pure CCEC.³ In a study by Bae et al of 16 pure cases of CCEC, PIK3CA mutations were seen in 18.8% of cases. ARID1A loss by immunohistochemistry was observed in 25% of cases, higher than the mutational frequency (15.9%) reported by Le Gallo et al. Another small study of 14 cases of CCEC reported mutations of telomerase reverse transcriptase promoter in 21% of cases.⁴

TABLE 1.

Common Mutations in CCEC Compared With Uterine Serous and Endometrioid Cancer

	CCEC	Uterine Serous	Endometrioid
TP53	39.7%	71.1%-88.4%	7.5%-14%
PIK3CA	23.8%	32.6%-35.6%	50%-52.3%
PIK3R1	15.9%	4.7%-6.7%	39.9%-45%
ARID1A	15.9%	6.7%-9.3%	32.5%-38.9%
PPP2R1A	15.9%	24.4%-25.6%	7.3%-7.5%
SPOP	14.3%	7%-8%	9.3%
TAF1	9.5%	4.7%-13.5%	17.1%
Microsatellite instability	11.3%	2.2%-11.8%	17.5%-40%

Adapted from Le Gallo et al,² Supplemental Figure 2, and references ¹ and ¹⁵.

A novel discovery by Le Gallo et al was the finding of recurrent mutations in TATA box binding protein-associated factor 1 (TAF1), occurring in 9.5% of CCEC cases. A major subunit of the basal transcription factor TFIID, TAF1 possesses DNA-binding activity, histone acetyltransferase activity, 2 kinase domains, and ubiquitin-activating/conjugating activity.^5,6 Germline mutations in this gene result in dystonia 3, torsion, X-linked, a dystonia-parkinsonism disorder.

We performed an exploratory analysis of TAF1 alterations in cancer using the cBio Cancer Genomics Portal.^7,8 Among diverse cancer types, the highest frequency of somatic TAF1 mutations has been observed in uterine cancers, including serous (13.5%), endometrioid (14.6%), and carcinosarcoma (3.6%-13.6%). The only other cancer type with >10% reported mutation frequency is small cell lung cancer (10.3%). Interestingly, TAF1 amplification is a common event in neuroendocrine prostate cancer (27%) and metastatic prostate adenocarcinoma (4.7%-8.2%).

Defining the role of TAF1 alterations in CCEC and other cancers will require further functional studies. However, the literature supports a role for TAF1 as a regulator of apoptosis and a modulator of sex hormone signaling, both of which may be relevant to CCEC development and progression. In esophageal cancer, TAF1 mutations are acquired or enriched during neoadjuvant chemotherapy, suggesting that these mutations promote cancer cell survival under stress.⁹ Experimental data shows that depletion of TAF1 was associated with substantial attenuation of apoptosis induced by oxidative as well as genotoxic stress.¹⁰ It has been shown that TAF1 phosphorylates p53, leading to dissociation of p53 from the p21 promoter in response to DNA damage.¹¹ The effect of estrogens and anti-estrogens on estrogen receptor transcriptional activity requires interaction with TAFs.^12,13 TAF1 also interacts with androgen receptor (AR) and functions as a coactivator of AR that binds and enhances AR transcriptional activity.¹⁴ The role of TAF1 in sex hormone signaling may underlie the recurrent TAF1 alterations observed in cancers derived from hormonally responsive tissue lineages (ie, endometrial, prostate). It merits further investigation whether the interaction of TAF1 with hormone receptors can be exploited for precision therapy in these cancer types.

In the study by Le Gallo et al, MSI was identified in 11.3% of CCEC cases, similar to the mutational frequency seen in serous endometrial cancer (11.8%)¹⁵ and lower than that observed in endometrioid cancer (40%). Previous studies of smaller cohorts of CCEC reported somewhat higher MSI frequency in 25% of patients.³ In the TCGA integrated genomic analysis of endometrioid tumors, MSI was most often due to MLH1 promoter methylation resulting in decreased MLH1 messenger RNA levels.¹ In other cases, MSI is a consequence of somatic or germline mutations in genes encoding mismatch repair (MMR) proteins, including MLH1, MSH2, MSH6, PMS2, and EpCAM. The genomic consequence of defective mismatch repair is a “hypermutator” phenotype, characterized by 18 × 10⁻⁶ mutations per Mb DNA compared with 2-3 × 10⁻⁶ mutations per Mb in microsatellite stable endometrioid or serous tumors.

MSI is a clinically important finding for several reasons. MMR protein deficiency unrelated to MLH1 methylation should prompt genetic testing for Lynch syndrome, an autosomal dominant hereditary syndrome resulting from a germline MMR mutation. In approximately half of women with Lynch syndrome, endometrial cancer is the sentinel cancer and may occur at an earlier age than colorectal cancer. Patients identified as having Lynch syndrome can benefit from screening and prevention strategies to decrease second cancers. Moreover, cascade genetic testing can be performed to identify affected relatives who can dramatically reduce their risk of cancer-related morbidity and mortality with screening and prevention strategies, including risk-reducing surgery.

Due to the hypermutator phenotype, MSI endometrial tumors present an increased number of neoantigens proportional to the mutational load.¹⁶ The immune response reflected in the increased tumor-infiltrating lymphocytes appears to be dampened by high expression of immune checkpoint proteins PD-1 and PD-L1 in the tumor microenvironment, suggesting a potential therapeutic benefit of immune checkpoint inhibitors. Clinical activity of immune checkpoint inhibitors in MSI metastatic colorectal cancer led to a breakthrough designation by the FDA. The overall objective response rate in a variety of metastatic MSI cancers was 25%-71%.¹⁷ Thus far, few endometrial cancer patients have been treated with immune checkpoint inhibitors. However, promising activity has been observed in individual patients harboring hypermutated or ultramutated endometrial cancer, with regression of chemo-refractory metastatic disease following single-agent treatment with anti–PD-1 antibody therapy.^18,19 MSI and expression of immune checkpoint proteins in CCEC suggest a potential role of immunotherapy in the management of this disease.

The study by Le Gallo et al provides new knowledge and insight into the molecular drivers of CCEC, yet a number of important unanswered questions remain. For example, what is the genetic heterogeneity among CCEC and if subjected to genomic classification, would distinct subtypes emerge? How similar or different is the genomic landscape of CCEC to other endometrial types and to other clear cell adenocarcinomas (eg, ovarian clear cell carcinoma, renal clear cell carcinoma). To answer these questions will require additional molecular profiling of more patients. Another feature of CCEC is that it is frequently admixed with other histological types of endometrial cancer. Of the 16 cases subjected by Le Gallo et al to whole exome sequencing, 12 were pure CCEC, whereas 4 were focal CCEC admixed with other endometrial cancer types. It is unclear whether pure CCEC differs in its development and genomic features compared with CCEC of mixed tumors. Interestingly, a recent multi-institutional study of 41 cases of mixed endometrioid and clear cell endometrial carcinoma found that 66% had MMR protein deficiency by immunohistochemical analysis, a considerably higher rate than that observed in pure CCEC.²⁰ Identical staining patterns in the endometrioid and clear cell component suggest that both components arise from a common clonal origin. Future studies analyzing multiple tumor samples from the same patients, and including different histological components and tumor locations (eg, primary and metastatic sites) would help elucidate the clonal evolution and progression of CCEC. In addition, comparison of genomic features of clear cell and non–clear cell components from the same tumor may lead to the identification of unique molecular drivers of the clear cell phenotype. Undoubtedly, the increased molecular understanding of CCEC will guide precision medicine toward more effective treatment and improved survival in patients with this aggressive type of cancer.