Tumor Mutational Burden Guides Therapy in a Treatment Refractory POLE‐Mutant Uterine Carcinosarcoma

Gynecologic carcinosarcomas are uncommon malignancies that have an aggressive biology and lack a standard therapeutic approach. Molecular analyses have shown recurrent alterations in chromatin remodeling genes, but clinical support for therapeutic significance is lacking. This article is the first report of a case of a patient with refractory uterine carcinosarcoma that responded to immune checkpoint inhibitor therapy.

Abstract

Gynecologic carcinosarcomas, previously known as malignant mixed Müllerian tumors, are uncommon malignancies that demonstrate an aggressive biology and lack a standard therapeutic approach. Molecular analyses have revealed recurrent alterations in chromatin remodeling genes, but clinical support for therapeutic significance is lacking. We prospectively identified a patient with refractory uterine carcinosarcoma whose tumor was subject to molecular profiling at diagnosis and again at radiographic progression. Initial molecular testing did not assess tumor mutational burden, DNA polymerase ɛ (POLE), or microsatellite status. After the failure of several lines of chemotherapy, comprehensive genomic profiling of a repeat biopsy identified two missense mutations of the exonuclease domain of POLE (P286R and T323A). Tumor mutational burden was elevated (169 mutations per DNA megabase), consistent with an ultramutator phenotype. As seen in previously reported POLE‐endometrioid cases, our patient harbored alterations in PIK3CA, ARID1A, and PTEN and was microsatellite stable, with appreciable tumor‐infiltrating lymphocytes. She achieved an ongoing durable response with pembrolizumab. This is the first report of programmed cell death protein 1 response in uterine carcinosarcoma.

Key Points.

• Uterine carcinosarcoma is an uncommon and aggressive histologic variant of endometrial carcinoma with a poor prognosis.

• Inactivating DNA polymerase ɛ (POLE) mutations have been associated with high tumor mutational burden (TMB) and response to immune checkpoint inhibition.

• To the authors' knowledge, this is the first report of response to immune checkpoint inhibitor therapy in a patient with uterine carcinosarcoma.

• This case further supports expanding genomic profiling to include assessment of tumor mutational burden across tumor types, given the potential for immune checkpoint inhibitor therapy in TMB‐high tumors.

Patient Story

Inactivating DNA polymerase ɛ (POLE) mutations have been associated with high tumor mutational burden (TMB) uterine endometrioid adenocarcinomas. When restricted to uterine carcinosarcoma, an elevated TMB has been observed at variable rates [1], [2]. Our case provides orthogonal support for the responsiveness of tumors with elevated TMB to immune checkpoint inhibitors, regardless of histology. The importance and evolution of comprehensive genomic profiling is highlighted by the clinical impact of microsatellite instability (MSI), POLE, and TMB in subsequent testing for our patient.

A 55‐year‐old woman was diagnosed with a uterine carcinosarcoma after presenting with postmenopausal bleeding in July 2014 (Fig. 1A–C). She underwent curative attempt total abdominal hysterectomy, bilateral salpingo‐oophorectomy, and pelvic lymphadenectomy. Surgical pathology revealed an International Federation of Gynecology and Obstetrics stage IIIA (pT3aN0M0) uterine carcinosarcoma with lymphovascular invasion (Fig. 1D, E). Surgical margins were negative. However, an 8‐week postoperative positron emission tomography and computed tomography in September 2014 revealed persistent and recurrent nodal metastases. Re‐resection was attempted but not possible. Broad molecular hot‐spot testing from the patient's original specimen was not considered to reveal actionable therapeutic alterations (Table 1).

Figure 1.

Results of biopsies. Panels (A–C) show the endometrial biopsy performed for postmenopausal bleeding in July 2014. The tumor shows a biphasic pattern of epithelial (carcinomatous) and spindled (sarcomatous) differentiation. (D, E): The hysterectomy specimen from September 2014 shows a large, friable, fungating tumor with >50% myometrial invasion. The tumor consists predominantly of crowded, fused, and cribriform glands, as well as solid nests making up >50% of the tumor and large areas of comedonecrosis, consistent with an endometrioid‐type, International Federation of Gynecology and Obstetrics grade 3, endometrial adenocarcinoma. This matches the carcinomatous component seen on the previous biopsy. (F): The recurrence biopsy from April 2016 shows predominantly carcinomatous differentiation with some areas of spindled pattern consistent with the original diagnosis of uterine carcinosarcoma. Tumor nests are accompanied by a moderate peritumoral lymphocytic response; however, tumor infiltrating lymphocytes are, by and large, non‐brisk.

Table 1. Comparison of molecular testing results conducted on original surgical specimen and repeat testing from progression biopsy in a case of endometrial carcinosarcoma.

Additional variants of unknown significance were identified but not reported because of space constraints.

Abbreviations: Ab, antibody; IHC, immunohistochemistry; Mb, DNA megabase; MI, Molecular Intelligence; MSI, microsatellite instability; MSS, microsatellite stable; NGS, next‐generation sequencing; PD‐1, programmed cell death protein 1; PDL‐1, programmed cell death ligand 1; TMB, tumor mutational burden; TIL, tumor infiltrating lymphocytes.

Beginning in late 2014, the patient was treated with several lines of systemic therapy, initially carboplatin and paclitaxel, followed by gemcitabine and oxaliplatin and later single‐agent topotecan. Ultimately, in 2016, she developed increasing abdominal pain and significant left lower extremity lymphedema with radiographic disease progression (Fig. 2). Given the widespread and symptomatic progression despite several lines of conventional cytotoxic chemotherapy, a new biopsy was submitted for comprehensive genomic profiling (CGP; Fig. 1E; FoundationONE, Foundation Medicine, Cambridge, MA, http://www.foundationone.com), revealing an elevated tumor mutational burden as well as inactivating missense mutations in the exonuclease domain of POLE.

Figure 2.

Response to immune checkpoint inhibition in a case of endometrial carcinosarcoma with an elevated tumor mutational burden and POLE mutation. Axial computed tomography images demonstrate progression in a periportal nodal conglomerate and necrotic pelvic sidewall followed by response after initiation of pembrolizumab, now lasting over 12 months.

Molecular Tumor Board

Uterine carcinosarcomas, formerly known as malignant mixed Müllerian tumors, compose a minority of endometrial cancers and morphologically appear biphasic, containing foci of carcinoma and sarcoma, which gives rise to the carcinosarcoma label [1], [3], [4]. The clinical course for most uterine carcinosarcomas is worse compared with the more common endometrioid adenocarcinomas or even with those with pure serous histology [5], [6], [7]. Surgery remains the mainstay of therapy for localized disease, and advanced disease is managed with systemic chemotherapy, largely carboplatin and paclitaxel [8], [9], [10].

Large collaborative molecular classification efforts such as the Cancer Genome Atlas (TCGA) have improved biologic understanding and identified molecular subtypes across multiple tumor types, including endometrial cancers [11]. Although the endometrial TCGA identified two groups (POLE mutant and hypermutated microsatellite instable [MSI‐H]) with increased numbers of nonsynonymous mutations, the analysis was largely restricted to early‐stage, surgically resected cases with endometrioid or serous histology (360 of 373 cases) [11]. Smaller series have investigated the molecular landscape of uterine carcinosarcomas, and cases with increased TMB have been reported, although clinical response data are lacking [1], [12].

Current National Comprehensive Cancer Network guidelines (version 1.2017) endorse MSI testing for all endometrial cancers, but determination of tumor mutational burden or POLE status is not standard. However, across other tumor types, elevated MSI‐H and elevated TMB have been associated with responsiveness to immune checkpoint inhibitor therapy based on the presumed increase in tumor‐specific neoantigens in these patients [13], [14], [15]. Within this context, tumor tissue from the index patient was submitted for CGP to identify additional therapeutic options.

Genotyping Results and the Interpretation of the Molecular Results

Formalin‐fixed, paraffin‐embedded material from the April 2016 biopsy was sent for CGP using another commercial platform, as previously described [16], [17]. The CGP results revealed a microsatellite‐stable, ultramutated tumor with a TMB of 169 mutations per DNA megabase (Table 1). Two missense mutations of the exonuclease domain of POLE (P286R and T323A) were also identified. Notably, despite the biphasic histologic pattern of the carcinosarcoma, our patient shared clinicopathologic features with POLE‐mutant endometrioid endometrial adenocarcinomas, including mutations in PIK3CA, ARID1A, and PTEN; MSI status; increased TILs; and relative insensitivity to platinum agents, suggesting shared biologic mechanisms despite divergent appearance (Fig. 1, Table 2) [18], [19].

Table 2. Literature review of reported POLE‐mutant, high‐tumor mutational burden, solid tumors with response to immune checkpoint inhibitor therapy.

^aDenotes incident case reported herein.

^bTumor mutational burden (TMB) from next‐generation sequencing, whereas other reported cases determined TMB from whole exome sequencing.

Abbreviations: IV, intravenously; N/A, not applicable; PR, partial response; q2wk, once every 2 weeks; q3wk, once every 3 weeks; Ref, reference.

Potential Strategies to Target the Pathway and Implications for Clinical Practice

POLE‐mutant microsatellite stable (MSS) and MSI‐H endometrial adenocarcinomas have been linked with increased immunogenic mutations [20], [21]. Both POLE‐mutant and MSI‐H endometrial cancers demonstrate enhanced antitumor immune activity with infiltration of T lymphocytes [20], [21], [22], [23], [24]. The histologic appearance of the endometrial tumor is not a reliable predictor of the underlying molecular defect, especially for the POLE‐mutant subgroup, which has been previously reported to show low‐grade or high‐grade endometrioid, serous, or mixed histology. Taken together, these preclinical data provide a scientific rationale for the use of programmed cell death protein 1 (PD‐1) inhibitors among endometrial tumors with a relevant molecular profile, even if they demonstrate uncommon histologic phenotypes. This hypothesis is further supported by several case reports demonstrating significant clinical responses to PD‐1 inhibitors in patients with advanced, recurrent tumors with an ultramutated molecular phenotype (Table 2) [25], [26], [27]. Previously, Castelucci et al. reported two cases of POLE‐mutant colorectal carcinoma with an ultramutated phenotype; one patient experienced partial and durable response to single‐agent pembrolizumab [22]. Our case differs in that the histopathologic classification is uterine carcinosarcoma and expands the description of POLE across tumor types.

As our understanding of the therapeutic relevance of specific genomic features expands, so does the need for more extended molecular assessments. The differing information included in the initial molecular testing (September 2014) and repeat testing (April 2016) likely reflects methodologic differences in testing at the given time points (20‐month interval; Table 1). Although the extent of molecular testing undertaken for this patient in 2014 would be considered beyond standard, it was nonetheless inadequate to identify the elevated tumor mutational burden subsequently identified in 2016. In a 2014 series of 22 gynecologic carcinosarcomas (17 of 22 uterine), two cases harbored oncogenic loss‐of‐function POLE alterations (POLE P286R and V411L); albeit at the time of publication this was not considered clinically actionable, reflecting how our understanding and annotation of genomic information evolves over time [1].

In this case history, the histologic analyses and overlap in genomic alterations identified strongly suggest that both samples shared a clonal origin and that an elevated TMB would have been found in the surgical specimen if this had been assessed on the original test (Fig. 1, Table 1). In fact, genomic profiling may aid in confirming clonality in similar clinical scenarios [28]. Although standard MSI testing captures the nearly 28% of endometrial cancers that are MSI‐H, patients with high TMB and MSS tumors because of POLE‐mediated mechanisms (7%) and others go undetected [11], [21], [25], and thus potentially relevant treatment options remain unexplored. A recent analysis suggests that genomic profiling aids in clinical decision‐making in gynecologic cancers [29].

The observation that histologic evaluation is often subject to interobserver variability and may not be a reliable predictor of biologic behavior is a driving consideration in adding molecular characteristics to the classification system for endometrial tumors. Clinical examples of broad genomic tumor analyses, which identify unexpected oncogenic alterations within an anatomic or histologic subtype that subsequently respond to molecularly paired therapy, are accumulating despite variations of methodology and concerns about temporal and geographic sample heterogeneity [30], [31], [32], [33].

This raises an important question: should all tumors with highly elevated TMB, regardless of histology, be considered for immune‐mediated therapies? We recognize that prospective clinical studies are needed to answer this question, but mounting evidence warrants pan‐cancer study. In fact, the nationwide American Society of Clinical Oncology TAPUR trial (NCT02693535) was recently amended to match pembrolizumab to any tumor with a pathogenic POLE mutation, and an elevated TMB immunotherapy arm is planned as well.

Patient Update

Based on this high TMB, the index patient was treated with pembrolizumab after informed consent discussion. She achieved rapid symptomatic improvement with decreasing lymphedema and an excellent radiographic partial response, with 65% decrease by immune‐related response criteria and 40% by RECIST version 1.1 criteria (Fig. 2) [34], [35]. She had continued pembrolizumab for over 12 months at the time of manuscript submission.

Glossary of Genomic Terms and Nomenclature

Comprehensive genomic profiling: Assay that examines multiple classes of genomic alterations across a broad panel of cancer‐related genes

Missense mutations: A point mutation in which a single nucleotide alteration results in a codon for a different amino acid

Microsatellite instability: Pattern of genetic hypermutability that results from dysfunctional DNA mismatch repair

Nonsynonymous mutation: A nucleotide mutation that results in altered amino acid sequence

Tumor mutational burden: Total number of genetic mutations per coding area of a tumor genome

Ultramutator phenotype: Genetic profile associated with high tumor mutational burden

Author Contributions

Conception/design: Munveer S. Bhangoo, Peter Boasberg, Samuel J. Klempner

Provision of study material or patients: Pareen Mehta, Julia A. Elvin, Siraj M. Ali, Winnie Wu, Samuel J. Klempner

Collection and/or assembly of data: Munveer S. Bhangoo, Peter Boasberg, Julia A. Elvin, Siraj M. Ali, Samuel J. Klempner

Data analysis and interpretation: Munveer S. Bhangoo, Peter Boasberg, Pareen Mehta, Julia A. Elvin, Siraj M. Ali, Winnie Wu, Samuel J. Klempner

Manuscript writing: Munveer S. Bhangoo, Peter Boasberg, Pareen Mehta, Julia A. Elvin, Siraj M. Ali, Winnie Wu, Samuel J. Klempner

Final approval of manuscript: Munveer S. Bhangoo, Peter Boasberg, Pareen Mehta, Julia A. Elvin, Siraj M. Ali, Winnie Wu, Samuel J. Klempner

Disclosures

Julia A. Elvin: Foundation Medicine (E, OI); Siraj M. Ali: Foundation Medicine (E, OI), Incysus (SAB); Samuel J. Klempner: Foundation Medicine (H), Lilly Oncology (C/A), Boston Biomedical, Leap Therapeutics (RF). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board