MICROSATELLITE INSTABILITY IN ENDOMETRIAL CANCER: NEW PURPOSE FOR AN OLD TEST

Précis:

Historically, microsatellite instability testing has been used to identify endometrial cancer patients with Lynch Syndrome. Now, it is also being used to identify those who may be immunotherapy candidates.

Introduction

The approval of pembrolizumab for solid tumors with high microsatellite instability (MSI-high) has greatly fueled interest in this molecular alteration. Although many tumor types have been shown to demonstrate MSI-high with significant frequency¹, endometrial cancer is one of the few tumor types in which MSI status is already assessed semi-routinely. Herein, we will discuss strategies and implications for MSI-high testing in endometrial cancer and how this is being leveraged in new clinical treatment paradigms.

Microsatellite Instability in Endometrial Cancer

Although Lynch Syndrome traditionally evokes concerns about colorectal cancer first, data suggest that a significant proportion of women with Lynch Syndrome will actually present with an endometrial cancer as their initial cancer diagnosis². Currently, somatic assessment of the endometrial tumor is a common means of evaluating patients for Lynch Syndrome. This testing is conducted in one of two ways: direct assessment for microsatellite instability (MSI), or indirect assessment reflected in loss of mismatch repair protein (MMR) expression (Table 1). Although these techniques are sometimes used interchangeably, they are functionally evaluating different factors. Microsatellite instability is a polymerase chain reaction (PCR) test which evaluates for insertions and deletions of microsatellite repeats in five to seven DNA regions³. MSI testing categorizes tumors as having high microsatellite instability (MSI-high), low microsatellite instability (MSI-low), or as being microsatellite stable (MSS) as outlined in Table 1³. Tumors considered to be MSI-low are of much lower prevalence and not as well understood, but in practice these tumors are usually considered to be similar to MSS tumors.

Table 1:

Summary of assessments for mismatch repair protein and microsatellite instability status.

Molecular aberration	Technique	Results	DNA regions evaluated
Mismatch repair protein	Immunohistochemistry	Evaluate for loss of expression of the protein	MSH2
			MSH6
			MLH1
			PMS2
Microsatellite instability	DNA for Polymerase Chain Reaction (PCR)	Evaluate for mutations in 0 (microsatellite stable, or MSS), 1 (low microsatellite instability, or MSI-low) or ≥ 2 (high microsatellite instability, or MSI-high) genes	BAT25
			BAT26
			D2S123
			D5S346
			D17S250
MLH1 methylation	DNA for PCR	Evaluate for methylation of the promoter of the MLH1 gene (methylated implies a somatic loss of MLH1)	MLH1

PCR: polymerase chain reaction

Alternatively, MMR protein expression is generally assessed by immunohistochemistry (IHC). This testing evaluates loss of expression of any of the four most common mismatch repair proteins: MLH1, MSH2, MSH6, and PMS2⁷. Loss of expression of any of those proteins on IHC suggests the possibility of a germline mutation in one of the genes associated with Lynch Syndrome. However, MLH1 loss can be commonly due to epigenetic modification⁸, specifically via methylation of the MLH1 promoter. For this reason, MLH1 deficiency is usually followed by an assessment of MLH1 promoter methylation prior to embarking on more extensive germline sequencing. If methylation is not present, the likelihood that MLH1 loss is due to a germline mutation is much higher. Unlike in colorectal cancer, evaluation for EPCAM loss and BRAF mutation have not been relevant to endometrial cancer patients. While EPCAM-associated Lynch Syndrome is frequently associated with colorectal cancer, it is far less common for these patients to develop endometrial cancers⁹. For this reason, EPCAM assessment is not recommended in endometrial cancer MMR evaluations. Similarly, although BRAF mutations are common in colorectal cancers, and specifically MMR proficient colorectal cancers, BRAF mutations are uncommon in endometrial cancers¹⁰. Thus, assessment for BRAF mutations is not useful as a means to help differentiate sporadic from Lynch-associated endometrial cancers.

MSI testing and MMR assessment can both be done by an oncologic pathology lab. Several studies have compared MSI testing and MMR assessment in endometrial cancer patients, and found reasonable concordance between the two methodologies. Discordance between MSI-high and MMR deficiency ranged from 2–8% in several studies from different institutions^11–16. Unfortunately, cross comparisons are difficult due to discrepancies in handling MSI-low tumors, which comprise approximately 3% of endometrial tumors^{14, 17}. However, although this discrepant proportion is relatively low, the fact that there is any discrepancy at all suggests that there may be clinical differences between the group identified by MSI assessment compared to that identified by immunohistochemistry assessment. In general, though, both techniques are currently considered acceptable, and it is considered unnecessary to perform both tests in initial testing for Lynch Syndrome as long as one is conclusive. From a resource perspective, IHC for MMR assessment is generally less expensive to perform than PCR for MSI analysis¹⁸, and provides insight into which gene is mutated if a germline mutation is suspected. Methodologically, however, both techniques are relatively easy to perform, and choice of technique is often driven by preferences within a given laboratory.

One of the more central debates in MMR and MSI testing for endometrial cancer—which is likely to become even more pronounced—regards the identification of patients in whom testing should be performed. Historically, patients who were at high risk for Lynch Syndrome were identified using clinical criteria followed by genetic counseling^3–5. However, more recently, patients have been evaluated for Lynch Syndrome using tumor (i.e., somatic) testing. While this was initially limited to those who had clinical indicators concerning for Lynch Syndrome (e.g., young age or family history), practices have expanded and include tumor testing for all uterine cancer patients in some centers (“universal testing”). If this testing is suspicious for Lynch Syndrome, patients are then referred to a genetic counselor and offered germline testing^{5, 6}. Using tumor testing as an intermediate step theoretically allows for more focused referrals to genetic counselors and streamlined germline sequencing.

The Bethesda Criteria and the Amsterdam II are frequently utilized clinical algorithms for identifying patients at high risk of Lynch Syndrome, as are guidelines released by the Society of Gynecologic Oncology and the American College of Obstetricians and Gynecologists^3–6. However, given the ease at which testing can be performed and the concern that at-risk patients were being missed by following these clinical guidelines, multiple institutions have initiated universal tumor testing; that is, testing all patients with a histologically confirmed diagnosis of endometrial cancer for MMR or MSI^{15, 30–32}. According to recent data, up to 80% of endometrial cancer patients with Lynch Syndrome would be missed using family history criteria alone, and 24% of Lynch Syndrome patients are diagnosed with endometrial cancer after age 60¹⁴. For these reasons, both the Society of Gynecologic Oncology and the National Comprehensive Cancer Network now recommend universal testing. Critics of universal testing cite the relatively low prevalence of Lynch Syndrome among an unselected endometrial cancer cohort, as well as the increased costs associated with testing all endometrial tumors^{15, 33}. Currently, there is no consensus, and thus the decision about whom should be tested is often clinician or institution dependent. Regardless of the observed variation in implementation of testing, however, tumor testing for MMR or MSI status has become a well-established mainstay in endometrial cancer, especially now that MSI status has therapeutic implications.

Clinically, many studies have sought to evaluate other characteristics of MSI-high or MMR deficient endometrial tumors. Histologically, some patterns have emerged. Tumors with MMR deficiency or MSI-high are more often – but not always – endometrioid histology, and some studies suggest that these tumors may be more frequently associated with poor prognostic factors such as advanced stage, deep myometrial invasion, and lymphovascular space invasion^19–21. From a clinical management standpoint, however, there are less clear associations, aside from the known relationship with Lynch Syndrome. Data regarding prognosis for patients whose tumors demonstrate MMR deficiency or MSI-high are mixed. Some studies show better survival outcomes^{19, 22}, some show worse survival outcomes^{23, 24}, and many show no association at all^25–28 (Table 2). Additionally, The Cancer Genome Atlas study of endometrial cancer demonstrated similar progression-free survival in the two subgroups comprising the majority of the endometrioid tumors: the MSI-high or “hypermutated” group and the microsatellite stable or “copy number low” group²⁹. Thus, until recently, the benefit of MMR or MSI testing was limited to identification of patients with Lynch Syndrome.

Table 2:

Summary of select studies evaluating the impact of MMR/MSI status on prognosis.

Study	Technique	Patient Population	Outcome
McMeekin DS et al. J Clin Oncol 2016.	PCR for MSI status; PCR for MLH1 methylation; IHC for expression of MSH6, MSH2, and MLH1 for all; IHC for expression of PMS2 for a subset. Used combination of IHC and PCR to categorize	n = 1024; patients on GOG0210; endometrioid endometrial adenocarcinoma, all grades and all stages	No statistically significant difference in progression-free survival or endometrial cancer-specific survival
Black D et al. J Clin Oncol 2006.	PCR for MSI status (considered MSI if at least 2 of 5 markers demonstrated mutations)	n = 473; adenocarcinomas of endometrioid, serous, clear cell, or undifferentiated histology; all stages and grades	MSS patients had statistically significantly worse disease-free survival and disease-specific survival than MSI (both persisted on multivariable analyses)
Mackay HJ et al. Eur J Cancer 2010.	PCR for 2 markers (BAT25 and BAT26) to determine MSI status	n = 163; compiled patients enrolled on several NCI CTG studies; all stages, grades; histologies included both endometrioid and non-endometrioid (did not specify further)	No difference in survival for combined cohort stratified by MSI status; when limited to patients with stage I or II disease, MSI tumors had worse disease-free survival and overall survival (both persisted on multivariable analyses)
Cosgrove CM et al. Gynecol Oncol 2017.	PCR for MSI status; PCR for MLH1 methylation if MSI-high or MLH1/PMS2 loss on IHC; IHC for MSH2, MSH6, MLH1, and PMS2	n = 466; patients from the Ohio Colorectal Cancer Prevention Initiative study and other endometrial cancer patients treated at Ohio State University; all grades and stages; endometrioid, serous, clear cell, carcinosarcoma, undifferentiated/ dedifferentiated, and mixed histology	No difference when all histologies were included for recurrence-free survival or overall survival; when limited to patients with endometrioid histology, epigenetic MMR defect group had statistically significantly worse recurrence-free survival but not endometrial-cancer specific or overall survival
Ruiz I et al. Gynecol Oncol 2014.	IHC for MLH1, MSH2, MSH6, and PMS2	n = 212; endometrioid endometrial adenocarcinoma only; all grades and stages	No difference in progression-free survival or overall survival; no difference when limited to either early stage or advanced stage patients
Diaz-Padilla I et al. Crit Rev Oncol Hematol 2013.	Studies using either PCR for MSI status or IHC for mismatch repair protein expression	For disease-free survival, n = 4 studies were pooled; for overall survival, n = 6 studies were pooled; all grades and stages; endometrioid and non-endometrioid histology	No difference in disease-free survival for early-stage patients or overall survival for all patients
Arabi H et al. Gynecol Oncol 2009.	IHC for MLH1, MSH2, and MSH6; considered MSI when there was loss of at least 2 of 3 proteins	n = 91; only grade 3 tumors; endometrioid, serous, and clear cell histology; all stages	No difference in overall survival
Zighelboim I et al. J Clin Oncol 2007.	PCR for MSI, considered MSI+ if at least 2 of 5 markers were mutated and low-level MSI if 1 of 5 markers was mutated; PCR for MLH1 methylation	n = 446; endometrioid histology only; all stages and grades	No difference in disease-free survival or overall survival

MSI: microsatellite instability

PCR: polymerase chain reaction

IHC: immunohistochemistry

MMR: mismatch repair

Implications for Therapy

In a seminal paper by Le et al, pembrolizumab, a PD-1 inhibitor, was found to have therapeutic efficacy in colorectal cancer patients whose tumors were MMR deficient³⁴. A subsequent study by the same group evaluated 86 patients with MMR deficiency with 12 different tumor types, including 15 patients with endometrial cancer (second only to colorectal cancer)³⁵. Although survival estimates are not mature, the progression-free survival at two years for this study was estimated at 53%, which is significantly higher than what would be expected for this population³⁵. In conjunction with several other studies, these data resulted in a Food and Drug Administration (FDA) approval for the use of pembrolizumab in patients whose tumors demonstrate MMR deficiency or MSI-high in May 2017. This was particularly notable as it was the first approval to be biomarker-based but independent of primary tumor site. This was followed closely first by an approval for nivolumab in patients with metastatic colorectal cancer demonstrating MMR deficiency or MSI-high (July 2017), and then by ipilimumab in combination with nivolumab this same patient population (July 2018). However, no other tumor types currently have nivolumab approved for this indication. With these approvals, the consideration of MMR or MSI testing is no longer limited to its implications for Lynch Syndrome.

These studies and their resultant FDA approvals do have some limitations, however. First, histologic data are not available for the endometrial cancer patients included in these studies. Given that the vast majority of endometrial carcinomas are endometrioid, these patients most likely had tumors with endometrioid histology. As such, we probably have even fewer data on the likelihood of response to checkpoint inhibitors in endometrial cancer patients with non-endometrioid histologies, such as clear cell, serous, and carcinosarcoma. Unfortunately, these non-endometrioid endometrial cancers also comprised a small proportion of patients included in studies evaluating the utility of universal testing for the identification of Lynch Syndrome. Thus, the benefit of universal testing for non-endometrioid endometrial cancers still remains unclear from a therapeutic standpoint, even in light of the recent FDA approvals. This also highlights a second broader issue, which is that endometrial cancer patients comprised a significant proportion of the non-colorectal cancer patients in these studies. Thus, the benefit of checkpoint inhibition in MMR deficiency in ovarian cancer, for example, is based largely upon data from patients with colorectal, endometrial, and other gastrointestinal cancers³⁵. As we have learned with other targeted therapies, such as BRAF inhibitors in colorectal cancers compared with melanoma³⁶, different tumor types with the same alteration may still respond differently to the same therapeutics.

Another important limitation to the Le et al study is the breakdown of germline versus somatic causes of MMR deficiency. Although we know that 45% of cases were germline or suspected germline and 37% were somatic, the distribution the patients with endometrial cancer is not described. Furthermore, while we know that there was no statistically significant difference in response between the patients with germline compared with somatic alterations, we again do not know this at the level of endometrial cancer patients specifically³⁵. A study by Pakish et al compared the immune microenvironment in endometrial tumors³⁷. They found differences in several factors including increased numbers of CD8+ cells in the stroma of Lynch Syndrome MSI-high tumors, and increased numbers of CD68+ macrophages in both the stroma and tumor regions of sporadic MSI-high endometrial tumors. These data suggest that it is plausible that there may be differences in responses to immunotherapy in patients with sporadic and germline MSI-high tumors. Unfortunately, these subgroups of endometrial cancer patients have not yet been separated out in clinical studies of checkpoint inhibitor efficacy. Further research will be needed to determine whether the origin of MSI-high will have therapeutic implications for endometrial cancer patients.

Last, these studies were based on a relatively small absolute number of patients, and an even smaller number of endometrial cancer patients. The study by Le et al was, in fact, a Phase II study, and there was no standard therapy arm to which the intervention patients were compared. Preliminary data from another Phase II trial of 15 patients with MSI-high or POLE mutated endometrial cancer treated with avelumab demonstrated 3 of 8 evaluable patients had a partial response, and another 4 patients had stable disease³⁸. The plan is to proceed with the second stage of that Phase II study. Currently, there is a confirmatory Phase III study that is ongoing for the use of pembrolizumab in MSI-high colorectal cancer patients. However, to our knowledge there are no ongoing Phase III trials evaluating the utility of pembrolizumab as a single-agent therapy in MSI-high endometrial cancer patients or other MSI-high tumor types. There is a Phase III trial of the combination of pembrolizumab and lenvatinib in endometrial cancer patients that is ongoing, and several other Phase III trials of immunotherapy in combination with chemotherapy are planned. Additionally, multiple other Phase II trials of other checkpoint inhibitors both as single agents and in combinations for endometrial cancer patients are ongoing (Table 3). The addition of these studies to the scientific literature will be critical, as they will help to better delineate which MSI-high endometrial cancer patients might benefit most from checkpoint inhibition.

Table 3:

Select clinical trials on ClinicalTrials.gov involving checkpoint inhibitors evaluating a significant proportion of endometrial cancer patients.

Drugs (immunotherapy-containing arms)	Endometrial Cancer Population	ClinicalTrials.gov Identifier	Status (as of 1/27/2019)
Durvalumab, Vigil (bi-shRNAfurin and GMCSF Augmented Autologous Tumor Cell Immunotherapy)	Locally advanced or metastatic endometrial cancer, treatment-naïve or resistant/refractory to checkpoint inhibitors	NCT02725489	Enrolling by Invitation
Durvalumab; Durvalumab, Tremelimumab	Persistent or recurrent endometrial carcinoma, epithelial histology	NCT03015129	Recruiting
Pembrolizumab	Persistent or recurrent endometrial carcinoma, epithelial histology, that is hypermutated (MMR gene defect) or ultramutated (POLE mutation) on next-generation sequencing and/or comprehensive genomic profiling	NCT02899793	Recruiting
Avelumab	Persistent or recurrent endometrial cancer that are either 1) POLE mutated or MMR loss or 2) microsatellite stable on IHC	NCT02912572	Recruiting
Nivolumab; Nivolumab, Cabozantinib	Advanced, recurrent, or metastatic endometrial cancer, epithelial histology, with MSI/MMR results available	NCT03367741	Recruiting
Pembrolizumab, Carboplatin, Taxol	Advanced or recurrent endometrial carcinoma, 0–1 prior platinum regimens (at least 6 months prior) and 0–1 prior non-platinum regimens	NCT02549209	Recruiting
Durvalumab, Radiation Therapy; Durvalumab, Tremelimumab, Radiation Therapy	Advanced or recurrent endometrial cancer, must have progressed on platinum-based chemotherapy	NCT03277482	Recruiting
Nivolumab, Ipilimumab	Advanced or metastatic endometrial cancer, grade 3 endometrioid, serous, clear cell, or mixed high grade; must have loss of expression of at least one MMR protein on IHC	NCT02982486	Not yet recruiting
Nivolumab	Metastatic or recurrent uterine cancer, both epithelial histologies and sarcomas, must be MSI-high, MMR-deficient, or hypermutated	NCT03241745	Recruiting
Pembrolizumab, Immune Modulatory Cocktail (Vitamin D, Lansoprazole Teva, Cyclophosphamide, Aspirin), Radiation Therapy, Curcumin	Persistent or recurrent uterine cancer, epithelial histologies or sarcomas, received at least one line of chemotherapy	NCT03192059	Recruiting
Spartalizumab, MCS110 (Anti-M-CSF Monoclonal Antibody)	Advanced endometrial carcinoma	NCT02807844	Recruiting
Pembrolizumab, Carboplatin, Paclitaxel	Newly diagnosed Stage III-IV high grade endometrial cancer requiring adjuvant therapy	NCT02630823	Active, not recruiting
Atezolizumab, Carboplatin, Paclitaxel	Newly diagnosed Stage III-IV endometrial cancer with residual disease after surgery, inoperable newly diagnosed Stage III-IV endometrial cancer, or recurrent endometrial cancer who have not received prior first-line systemic anti-cancer therapy	NCT03603184	Recruiting
Pembrolizumab, Lenvatinib	Advanced/recurrent endometrial cancer, epithelial histology (not carcinosarcoma), who have not received more than 1 prior chemotherapy in the recurrent setting (this does not include neoadjuvant or adjuvant therapy)	NCT03517449	Recruiting
Pembrolizumab; then Pembrolizumab, Carboplatin, Paclitaxel	Newly diagnosed Stage I-II clear cell or serous endometrial cancer, or Stage III clear cell, serous, or grade 3 endometrioid endometrial cancer	NCT03694834	Not yet recruiting
Pembrolizumab, Doxorubicin	Advanced/recurrent endometrial cancer, any epithelial histology, who have had only 1 prior line of platinum-based chemotherapy	NCT03276013	Recruiting
Pembrolizumab	Primary treatment of endometrial cancer, epithelial histology who have not yet had surgery or any treatment	NCT02728830	Recruiting
Atezolizumab, Bevacizumab	Recurrent endometrial cancer, epithelial histology, who have received at least 1 platinum-based chemotherapy regimen but not more than 2 prior chemotherapy regimens	NCT03526432	Recruiting
Atezolizumab, Bevacizumab, Rucaparib	Persistent/progressive endometrial cancer, epithelial histology, 1–2 prior chemotherapy regimens	NCT03694262	Not yet recruiting
Atezolizumab, carboplatin, cyclophosphamide	Stage IV endometrial cancer treated with no more than 1 line of chemotherapy in the advanced setting	NCT02914470	Active, not recruiting
Nivolumab, Rucaparib; Nivolumab	Metastatic endometrial cancer, unlimited number of prior regimens	NCT03572478	Recruiting
Avelumab, Carboplatin, Taxol, with or without Avelumab maintenance	Recurrent or advanced endometrial cancer, epithelial histology (except carcinosarcoma), may have received adjuvant chemotherapy but must have completed treatment at least 6 months prior	NCT03503786	Not yet recruiting

Accessed: 1/27/2019

MMR: mismatch repair

MSI: microsatellite instability

IHC: immunohistochemistry

In general, though, these studies have begun to demonstrate that there are a subset of patients who will respond to checkpoint inhibition and identification of these patients is therapeutically important. At a population level, endometrial cancer patients with MSI-high and/or MMR deficient tumors appear to have better responses to checkpoint inhibitor therapy³⁵. However, these studies also demonstrate that some MSI-high/MMR deficient endometrial cancer patients do not respond well to checkpoint inhibitors. Although the objective response rate for endometrial cancer patients in the Le et al study reached an impressive 53%³⁵, 20% of the endometrial cancer patients with MSI-high/MMR deficiency still had progressive disease after treatment with pembrolizumab. Thus, although these biomarkers are narrowing the group of patients who are most likely to respond, nascent biomarker imprecision still challenges efficacy expectations on an individual level.

One of the ways to theoretically improve patient identification is to better characterize the patients whose tumors respond to checkpoint inhibition. To accomplish this, we need a better understanding of why some of these MSI-high or MMR deficient tumors respond while others remain resistant. From a mechanistic standpoint, MSI-high/MMR deficiency is thought to be a surrogate for the actual predictive biomarker: high tumor mutational burden and resultant high neoantigen load. For this reason, biomarkers such as tumor mutational burden and mutation associated neoantigens have been explored and have shown promise^{35, 39–41}.

Immune infiltrates such as T-cells, macrophages, and other tumor-infiltrating lymphocytes (TILs) conceptually make sense in terms of predicting which tumors are more likely to respond to checkpoint inhibition. Unfortunately, secondary analyses of responses using these biomarkers have been mixed^{42, 43}. Although these biomarkers do appear to be upregulated in MSI-high/MMR deficient tumors^{37, 44, 45}, it remains to be seen whether these biomarkers prospectively predict response to checkpoint inhibitors in clinical trials. Furthermore, more information is needed as to whether they are as easy and cost-efficient to use as MSI/MMR status in a CLIA-certified setting.

Importantly, the mixed nature of current data also implies that there are endometrial cancer patients without MSI-high/MMR deficiency who might respond well to this class of drugs. PD-L1 status is currently being evaluated as a biomarker for checkpoint inhibitors in many tumor types, including endometrial cancer⁴⁶, although its utility as a predictive biomarker is still under debate⁴⁷. There has been some success in specific tumor types, however, and based on data from KEYNOTE 158⁴⁸, pembrolizumab was recently approved for patients with recurrent or metastatic cervix cancer who progressed after chemotherapy and whose tumors express PD-L1 with a combined positive score (CPS) ≥1. Similarly, although less clinical research has been completed, preclinical data suggest that endometrial carcinomas with POLE mutations have an immune microenvironment that should respond to checkpoint inhibition^{49, 50}. A number of clinical trials of immunotherapy agents are currently planned or ongoing in endometrial cancer patients, which will hopefully help address some of these important considerations (Table 3).

Future Directions

Clinically, the optimal timing and allocation of MSI/MMR testing is still being determined. As many pathology laboratories only perform MMR/MSI assessments for a targeted population, there are many endometrial cancer patients who are not currently being evaluated for MMR or MSI status. With the approval of pembrolizumab, performing tumor testing in endometrial cancer patients who have recurrence or progression of disease makes practical sense. Effective therapies, such as chemotherapy and other biologics, are limited, and the only FDA-approved therapy for these patients is megestrol acetate. However, limiting the use of pembrolizumab to the recurrent or progressive setting may miss an opportunity to prevent these poor outcomes in the first place. The role of checkpoint inhibitors as first line treatment in patients with MSI-high or MMR deficient endometrial cancer has not yet been evaluated. Practically speaking, implementation of immunotherapy as front-line therapy would most likely occur in patients with advanced disease at diagnosis, as these are the patients for whom outcomes are the poorest. One could argue, however, that there may be a role in patients with early stage disease whose tumors are at high risk for recurrence. Furthermore, in an ideal setting, future clinical studies would not only evaluate for MSI/MMR status, but would stratify treatment arms by tumor status so as to truly address the question of whether or not these biomarkers are predictive.

Better biomarkers are clearly needed to more precisely and consistently identify patients who will respond to checkpoint inhibitor therapy. To start, historically all studies evaluating MMR assessment compared with MSI assessment in endometrial cancer used Lynch Syndrome as the outcome of interest. It is possible that while there is reasonable concordance for identifying Lynch Syndrome patients, one of the techniques may be found to be superior when the outcome of interest is response to checkpoint inhibitors. Additionally, newer techniques such as TCRseq and RNA expression signatures may offer important insights into the mechanism of action, and consequently into ways to better identify this class of patients^{51, 52}. Novel assessments of tumor mutational burden and/or tumor-specific neoantigen load may also provide a more direct way to evaluate patients with high likelihood of responding to immunotherapies^{53, 54}.

From a broader perspective, the use of MSI/MMR status for therapeutic strategies in endometrial cancer patients has been limited primarily to checkpoint inhibitors. Now that there is a better appreciation for the therapeutically favorable immune microenvironment in endometrial tumors with MSI-high/MMR deficiency, novel strategies for targeting this patient population should be investigated. Combinations of immunotherapy agents have been found to be effective in other tumor types as highlighted in a recent metaanalysis⁵⁵. Although toxicities were higher, better survival outcomes were demonstrated. While these studies have not yet been completed in endometrial cancer, conceptually this tumor type is an excellent candidate for this combination approach.

Similarly, studies are now starting to address the question of whether a targeted therapy can prime the immune environment and lead to improved efficacy with an immunotherapy agent in other tumor types^{56, 57}. Several of these combinations with immunotherapy could be rationally extended to endometrial cancer, such as therapies targeting the MAPK pathway or antiangiogenic agents. Indeed, encouraging preliminary response and safety data of pembrolizumab in combination with lenvatinib, a multi-targeted tyrosine kinase inhibitor of VEGFR1–3, PDGFR-a, FGFR1–4, RET and KIT, was recently reported in women with confirmed metastatic endometrial cancer⁵⁸. In this open label phase II study, 11 of 23 (48%) women receiving the combination had a confirmed objective response; adverse events were hypertension, fatigue, arthralgia, diarrhea and nausea, but were manageable with dose interruption and/or modification.

Perhaps most exciting, however, is the prospect of novel immunotherapeutics in endometrial cancer patients. A recent Phase I study of six melanoma patients generated individual vaccines against up to 20 specific neoantigens found in each individual patient⁵⁹. In this study, the two patients who did not have a complete response after treatment with the vaccine did have sustained complete responses with subsequent pembrolizumab treatment⁵⁹. Although these melanoma patients were not specifically MSI-high/MMR deficient, the high neoantigen load seen with melanoma mirrors the high neoantigen load seen in endometrial tumors with MSI-high/MMR deficiency.

Last, because endometrial cancer patients represent a relatively small fraction of the patients who die from their cancers each year⁶⁰, they represent a relatively small proportion of the patients who are enrolled on the early phase trials evaluating many of the immunotherapy agents. However, the promising results seen in the subset of patients enrolled on those trials who have endometrial cancer should underscore the exciting potential of immunotherapy agents in these patients. Although EML4-ALK fusions occur in only 4–5% of the patients diagnosed with non-small cell lung cancer each year, the response to the ALK inhibitor crizotinib was so ground-breaking that universal testing is performed in all advanced non-small cell lung cancer patients⁶¹. We have used this finding to help triage patients into a variety of personalized oncology approaches, and the improvement in lung cancer survival has reflected this progress⁶². In this same way, universal immune microenvironment evaluation in all endometrial cancer patients could be a practical approach to therapy if a survival benefit is demonstrated. For this reason, more research evaluating the impact of molecular and immune characteristics of endometrial cancer patients on therapy response is critical to the ongoing progress of endometrial cancer clinical care.

Conclusion

Microsatellite instability and mismatch repair protein assessments have already been extensively evaluated in endometrial cancer patients. Given the great potential for immunotherapy in endometrial cancer patients whose tumors demonstrate MSI-high/MMR deficiency, expanding these assessments for the purpose of therapeutic decision making makes clinical sense. Future studies must continue to evaluate the utility of MSI-high/MMR deficiency as predictive biomarkers, and continue to identify novel biomarkers and novel immune therapeutics that could result in significant and sustained clinical responses in this important patient population.