WHAT IS ALREADY KNOWN ON THIS TOPIC

The use of immunotherapy for women with advanced or recurrent endometrial cancer has changed rapidly over recent years, but little is known about the use of immune checkpoint inhibitors in a real-world setting.

WHAT THIS STUDY ADDS

We present results from an observational longitudinal retrospective study using a real-world database among women with advanced or recurrent endometrial cancer who received systemic treatment between 2014–2021. We found that 14.7% of patients received any immune checkpoint inhibitor, and that receipt and line of immune checkpoint inhibitor varied by genomic and clinical characteristics and included repeated immunotherapy usage.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Given the variety of lines of immune checkpoint inhibitor use and the large subset of women who received repeat immune checkpoint inhibitors, future research is needed to determine optimal sequencing of immune checkpoint inhibitors.

Introduction

Endometrial cancer remains the most common gynecologic malignancy in the United States.1 For women with metastatic or recurrent disease, limited efficacious treatments have historically resulted in low survival rates.2 However, a better understanding of the genomic composition of endometrial cancer has led to the recent incorporation of immune-based therapy as an integral treatment strategy.3

Immune checkpoint inhibitors blocking the programmed cell death 1 (PD-1) receptor or ligand (PD-L1) have shown efficacy in subsets of endometrial cancer patients.2 The Food and Drug Administration (FDA) approval of pembrolizumab for progressive, unresectable or metastatic microsatellite instability-high (MSI-H)/mismatch repair deficient (dMMR) tumors in 20174 represented the first immune checkpoint inhibitor applicable for certain endometrial cancers. In advanced or recurrent endometrial cancer after frontline treatment, the anti-PD-1 immune checkpoint inhibitors pembrolizumab, dostarlimab, and nivolumab, as well as the anti-PD-L1 immune checkpoint inhibitor avelumab, are now listed in consensus guidelines and/or are US FDA-approved for MSI-H/dMMR tumors based on improved oncologic outcomes in clinical trials.5–8 For patients with recurrent mismatch repair proficient (pMMR)/microsatellite stable (MSS) endometrial cancer, pembrolizumab plus lenvatinib received FDA approval in 2019 due to improved overall survival and progression-free survival vs chemotherapy.9 10 Adding immune checkpoint inhibitors to chemotherapy has also shown efficacy in the upfront setting11–13 in a landscape that continues to rapidly change.

Despite the successes of immune checkpoint inhibitors in endometrial cancer, data regarding their real-world use are limited. As immune checkpoint inhibitor use becomes more widespread in endometrial cancer, questions surrounding patient selection and sequencing increasingly arise, and understanding post-immune checkpoint inhibitor progression treatment patterns is important to inform future treatment.

The goal of the present study was to use a real-world database to describe treatment patterns of immune checkpoint inhibitor use for women with endometrial cancer during a period with rapidly evolving treatment paradigms. We specifically examined factors associated with receiving any immune checkpoint inhibitor, treatment sequence of immune checkpoint inhibitor use, and post-immune checkpoint inhibitor treatments.

Methods

We conducted a retrospective, cross-sectional, observational study using the nationwide Flatiron Health advanced-endometrial database, a longitudinal electronic health record (EHR)-derived database from approximately 280 cancer clinics (~800 sites) across the United States, comprising patient-level structured and unstructured data curated via technology-enabled abstraction.14 15 Manual clinical documentation abstraction available in the EHR (eg, clinic notes, radiology reports, pathology reports) is generally performed by trained abstractors (eg, oncology nurses or tumor registrars), following iterated-on protocols and procedures, overseen by medical oncologists.16 Data are de-identified and subject to obligations to prevent re-identification and protect patient confidentiality.

The Flatiron Health electronic database inclusion criteria selected patients ≥18 years old, diagnosed with primary advanced stage III–IV endometrial cancer (defined by International Classification of Diseases (ICD)-10: C54.1 or ICD-9: 182.0 codes) at initial diagnosis who developed a locoregional or distant recurrence between January 1, 2011 and May 31, 2021, and who presented for at least two clinic visits. To be included in this analysis, patients were further required to have received at least one line of systemic therapy recorded between January 1, 2014 and November 1, 2020. The treatment-based inclusion criteria end date (November 1, 2020) allowed for possible follow-up time post- immune checkpoint inhibitor treatment to the end of the study observational window (May 31, 2021). Women with uterine sarcoma, another primary cancer diagnosed during the treatment period, or those without relevant documents for review in the database were excluded.

Systemic treatment was identified by administration of an antineoplastic recorded in the EHR and concurrent manually abstracted oral therapies. Treatment lines of therapy were categorized as first line (1 L), second line (2 L), and third line or later (≥3 L). Exposure to immune checkpoint inhibitor exposure was defined as receipt of pembrolizumab, dostarlimab, nivolumab, ipilimumab, avelumab, durvalumab, atezolizumab, and/or cemiplimab. Treatment index date was taken from the first exposure to these agents if patients had received more than one line of immunotherapy. Regimens were divided into five main categories: cytotoxic chemotherapy (±bevacizumab), single-agent or combination immune checkpoint inhibitor, targeted/hormonal therapy without concomitant chemotherapy, or other therapies (including clinical trials). A rules-based line of therapy was defined as the first eligible drug episode plus other eligible drugs given within 28 days. A subsequent line of therapy was based on a change in drug regimen or a gap in drug episodes of more than 120 days. For patients receiving immune checkpoint inhibitors after a prior immune checkpoint inhibitor, a 70-day rule for true line switch was implemented to account for additions of lenvatinib to pembrolizumab.

Demographic and clinical characteristics were described. Descriptive comparisons were performed using a Chi-Squared test or Welch’s 2-sample t-tests as appropriate between patients with vs without immune checkpoint inhibitor exposure; between patients based on treatment line of immune checkpoint inhibitors; and between patients based on post-initial immune checkpoint inhibitor treatment. In the case of small cell counts (<5 patients), a minimum of two cells per variable were masked to preserve patient privacy. Time to next treatment, a proxy for progression-free-survival, was indexed from the time of initial immune checkpoint inhibitor to first subsequent therapy or death; patients who remained on treatment were censored. Prespecified survival analyses were performed using Kaplan–Meier methods for immune checkpoint inhibitor line of therapy. Sankey plots were used to depict patient-level temporal treatment utilization. Statistical significance was set with 2-sided p<0.05. All analyses were conducted using R statistical software.

Results

Data from 4504 women with primary advanced, metastatic, or recurrent endometrial cancer were available for analysis, of which 2224 received ≥1 line of systemic treatment and met our eligibility criteria. Of this population, 326 women (14.7%) received their first immune checkpoint inhibitor therapy during our study period, with no immune checkpoint inhibitor treatment recorded before 2016. The majority were White (66.6%), had an Eastern Cooperative Oncology Group (ECOG) performance score ≤ 1 (75.5%), and were treated at a community site (77.0%). First treatment with checkpoint inhibitors increased between 2016 to November 2020 from 12 patients to 148 patients treated, respectively. Of a total 3517 treatment starts between 2016 and November 2020, 9.3% were first immune checkpoint inhibitor therapies, increasing year-on-year (2016: 2.0%; 2017: 3.5%; 2018: 7.5%; 2019: 11.0%; January to October 2020: 21.0%).

Women who received immune checkpoint inhibitor treatment were most likely to be originally diagnosed with advanced stage disease (59%) reflecting metastatic (vs recurrent) disease. There were more women treated with immune checkpoint inhibitors with MSS/pMMR disease (46.6%) than MSI-H/dMMR disease (32.8%). Of note, 20.6% of patients treated with immune checkpoint inhibitors had an unknown MMR status. Compared with women who never received immune checkpoint inhibitors, women who ever received immune checkpoint inhibitors differed significantly in ECOG status (p=0.001), disease stage (p=0.002), MMR/MSI status (p<0.001), and prior radiation receipt (p<0.001) and modality (p=0.003), but with no differences based on race, region, payer category, practice type, or histology (Table 1).

Table 1

Demographic and clinical differences in patients who had never received vs ever received immune checkpoint inhibition

Of the 107 patients with MSI-H/dMMR tumors who received immune checkpoint inhibitors, 90 (84%) received initial pembrolizumab monotherapy while four (4%) received combination therapy with pembrolizumab and lenvatinib. Of the 152 patients with MSS/pMMR tumors who received immune checkpoint inhibitors, 88 (58%) patients received the pembrolizumab/lenvatinib combination while 37 (24%) received pembrolizumab monotherapy. Immune checkpoint inhibitors were given as 1 L, 2 L, and ≥3 L of therapy in 24.5%, 41.7%, and 46.1% of evaluable patients, respectively. Median time to next treatment on immune checkpoint inhibitors was 12.46, 6.94, and 4.41 months when given in the 1 L, 2 L, and 3 L settings, respectively (p=0.008; Figure 1). Significant differences existed between patients who received immune checkpoint inhibitors in the 1 L, 2 L, or 3 L setting by region (p=0.004), stage (p<0.001), and prior radiation receipt (p=0.014) and modality (p=0.009) (Table 2).

Figure 1

Unadjusted Kaplan–Meier curves of time to next treatment by line of immune checkpoint inhibitor receipt. 1 L, first line; 2 L, second line; 3L+, third and subsequent lines; ICI, immune checkpoint inhibitor.

Table 2

Demographic and clinical differences in patients by immune checkpoint inhibitor line of treatment

In visualizing the sequence of treatment regimens (Figure 2), most patients received chemotherapy as their 1 L treatment (66.3%) compared with immune checkpoint inhibitors (24.5%). After frontline chemotherapy treatment, more patients received immune checkpoint inhibitors (41.7%) than a second chemotherapy regimen (36.8%), targeted therapy (16.5%), or other treatment in the 2 L setting. Immune checkpoint inhibitor use was more common than chemotherapy as 2 L and 3 L treatments among evaluable patients. After initial immune checkpoint inhibitor treatment, a subset of patients received additional immune checkpoint inhibitors.

Figure 2

Sankey diagram of patient-level systemic therapy regimen by line of treatment.

Of the 326 patients who received immune checkpoint inhibitors, with a recorded median follow-up of 8.1 months from starting immune checkpoint inhibitor therapy (interquartile range (IQR) 3.6 to 13.8 months), 212 patients (65%) had no recorded subsequent therapy and 99 (46.7%) died <90 days post-immune checkpoint inhibitor treatment. Of the 114 women (35.0%) who received subsequent therapy after immune checkpoint inhibitor treatment, 56 (49.1%) remained alive at the end of the observation period and 24 (21.0%) died <90 days post-immune checkpoint inhibitor treatment. For the 114 women who received post-immune checkpoint inhibitor treatment, the next treatment they received was chemotherapy for 50 women (43.9%), additional immune checkpoint inhibitor treatment for 34 women (29.8%), and other treatment for 30 women (26.3%). There were no significant differences in which post-immune checkpoint inhibitor treatment patients received by evaluated characteristics, although analyses were limited by low numbers (Table 3).

Table 3

Demographic and clinical differences in patients by post-initial immune checkpoint inhibitor treatment

DiscussionSummary of Main Results

We found that in an observational retrospective study using real-world data of women with advanced or recurrent endometrial cancer who were treated at the beginning of this rapidly evolving period of immunotherapy paradigms, immune checkpoint inhibitor treatment increased year-on-year from a first observed treatment in 2016. Receipt of immune checkpoint inhibitor treatment and the line of immune checkpoint inhibitor treatment varied by genomic and clinical characteristics. While most women received appropriate regimens per FDA approval, many received treatment outside of consensus guidelines, including repeat immune checkpoint inhibitor treatment.

Results in the Context of Published Literature

Few prior studies have examined real-world use of immunotherapy in endometrial cancer.17–21 The most contemporaneous study describing real-world use in patients with advanced or recurrent endometrial cancer who received systemic treatment between 2013 and 2021 similarly described that immune checkpoint inhibitor treatment was more common in 2 L (36%) compared with 1 L (16%) treatment and was more common in MSI-H/dMMR groups.18 These rates are higher than those in older database studies describing the use of immune checkpoint inhibition in endometrial cancer up to 2019, which found that only 2.7% of women received immune checkpoint inhibition as 2 L treatment,19 20 2.3% as 3 L treatment,20 and 8.9% as 4 L treatment,20 but increased to 9.8% in the 2 L of treatment after 2017, when pembrolizumab received FDA approval.4 20 In a smaller real-world study describing treatment patterns of women with MSS/pMMR endometrial cancer after progression on systemic therapy between 2016–2019, immune checkpoint inhibitor treatment was only used in the 3 L treatment setting, with 36.8% of women receiving pembrolizumab and lenvatinib in this setting but only representing 8% of their study population who initiated at least 2 L treatment.21

The current work includes more patients and examines a longer longitudinal period of immune checkpoint inhibitor use in endometrial cancer compared with older studies. We were also able to examine repeat immune checkpoint inhibitor use, finding a lack of clinical or genomic factors guiding these decisions, although we were limited by small numbers. Although increased compared with previous studies, immune checkpoint inhibitor treatment was only used in 14.7% of women overall in our cohort. The majority of patients in this study were treated with immune checkpoint inhibitors post-2017, with over half of them (n=235, 72%) treated in 2019 and 2020; these treatment patterns align with the tumor-agnostic approval of pembrolizumab in unresectable or metastatic microsatellite instability-high or mismatch repair deficient patients who have progressed on prior therapy in May 2017 and pembrolizumab’s accelerated approval with lenvatinib for advanced endometrial carcinoma in September 2019.4 6 With the recent regulatory approval of immune checkpoint inhibitors in the frontline setting and increasing physician experience with immune checkpoint inhibitors, we anticipate immune checkpoint inhibitor use to continue increasing. This hypothesis is supported by a recent survey of providers in Italy demonstrating increased administration of immune checkpoint inhibitors among its members between 2021 and 2022.22 23 We believe our study expands on prior literature by demonstrating increased use of immune checkpoint inhibitors over time, showing a shift in immunotherapy use in earlier lines of treatments, and pointing toward the next frontier of immune checkpoint inhibitor use, including frontline and repeat usage.

Strengths and Weaknesses

Strengths of our study include the use of a nationwide, de-identified, real-world database, which provides results on a large, multicenter, geographically diverse dataset in a non-trial setting. This allows our findings to be generalizable to the greater population, particularly to patients in community practices that may not be represented by clinical-trial populations. Flatiron Health’s use of technology-enabled abstraction enables more complete data collection through both structured and unstructured sources than may be obtained from traditional chart review and allows for comparison across multiple electronic health records. Because the database is longitudinal, we were able to examine endometrial cancer treatments over time and provide visualization of the care that women are receiving, including repeat immune checkpoint inhibitors.

Limitations of this study include those inherent to a real-world database including missing and unknown data. Notably, 48% of women in our study had missing MMR/MSI testing data, including up to 20.6% of women who received immune checkpoint inhibitors. This low rate of MMR/MSI testing is likely partially due to limitations of the EHR and data abstraction, but could also point to a gap with universal and standard testing. The use of time to next treatment as a progression-free-survival-proxy estimate is limited by being unable to differentiate between reasons for treatment discontinuation such as tumor progression, adverse events, or other patient-specific factors. Many of these patients were censored over time, which limits interpretation of later lines of systemic treatment and post-immune checkpoint inhibitor regimens. We also had a short median duration of follow-up, which limited estimates of overall survival. Notably, our study period does not include approved use of immune checkpoint inhibitors in the frontline setting and cannot be extrapolated to this context.

Implications for Practice and Future Research

As expected, we found a significant association between MMR/MSI status and receipt of immune checkpoint inhibitors. Interestingly, MMR/MSI status was not associated with treatment line of immune checkpoint inhibitors or post-immune checkpoint inhibitor treatment regimen, suggesting that MMR/MSI status mattered in deciding to give women immune checkpoint inhibitors but was not used to determine sequencing of immune checkpoint inhibitor treatment. When immune checkpoint inhibitors were used, MMR status was used to guide treatment decision-making but did not always ensure guideline-concordant treatment, with 84% of women with MSI-H/dMMR disease receiving pembrolizumab and 58% of women with MSS/pMMR disease receiving pembrolizumab and lenvatinib. Overall, many regimens were used, including immune checkpoint inhibitors without consensus or FDA approval during our study period. These findings may be due to contemporaneous studies investigating immune checkpoint inhibitor treatment, a perceived lack of differentiation within this class of agents among prescribers, and the dearth of clear best-treatment options. Importantly, the time to next treatment was significantly longer when immune checkpoint inhibitor treatment was used in earlier settings and is in-line with recent Phase III clinical trials demonstrating significantly improved progression-free-survival with frontline combination chemotherapy and immunotherapy.11–13

Not surprisingly, stage of disease and prior radiation treatment were associated with ever receiving immune checkpoint inhibitor treatment as well as with the line of treatment in which immune checkpoint inhibitor was first used. Interestingly, women were more likely to receive immune checkpoint inhibitors if they had better ECOG scores, despite favorable health-related quality of life results in patients treated with immune checkpoint inhibitors.24 25 Given prior findings demonstrating racial and socioeconomic disparities in endometrial cancer treatment in the United States,26–28 we were encouraged that our study did not find race, payer, and treatment setting to be associated with receiving immune checkpoint inhibitors. As immune checkpoint inhibitor use becomes increasingly common, it will be important to ensure that access to immunotherapy is equitable.

The most interesting results of our study may be the use of immune checkpoint inhibitors after immune checkpoint inhibitor treatment, which has also been reported in some case series.29 The reason for administration of immune checkpoint inhibitor treatment in the frontline systemic setting was not recorded. Nevertheless, our data showing that nearly 30% of women received additional post-immune checkpoint inhibitor immunotherapy provides novel and important real-world information as there are no currently approved indications for re-treating with immunotherapy in women with endometrial cancer. There was no significant association between use of post-immune checkpoint inhibitor chemotherapy, immunotherapy, or other treatment based on clinical or demographic variables in our study. If immune checkpoint inhibitor treatment is used increasingly as a repeated treatment, it will be crucial to identify patients who will derive the most benefit from these regimens. Unfortunately, the low numbers in our post-immune checkpoint inhibitor cohort precluded a time to next treatment or survival analysis. More research is needed to determine what the best treatment after immune checkpoint inhibition therapy is for our patients.

Conclusions

We found that in a real-world nationwide database of women with advanced or recurrent endometrial cancer during a time of rapidly evolving immunotherapy paradigms, 14.7% of women received at least one immune checkpoint inhibitor between January 2014 and November 2020 and use increased over time. Patients received immune checkpoint inhibitors across various lines of treatment, including after an initial immune checkpoint inhibitor. Further work remains to determine optimal sequencing of immune checkpoint inhibitors and which patients may benefit from immune checkpoint inhibitor re-challenge.

Data availability statement

Data are available upon reasonable request. In accordance with the journal’s guidelines, we will provide our data for independent analysis by a selected team by the Editorial Team for the purposes of additional data analysis or for the reproducibility of this study in other centers if such is requested.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

Not applicable.

Acknowledgments

Third-party editorial support was provided by Brian Law, PhD, of Nucleus Global, an Inizio Company, and funded by F. Hoffmann-La Roche Ltd.