Article Text

Prognostic role of transcription factor ARID1A in patients with endometrial cancer of no specific molecular profile (NSMP) subtype
Free
  1. Arina Onoprienko1,
  2. Gerda Hofstetter2,
  3. Leonhard Muellauer2,
  4. Tim Dorittke1,
  5. Stephan Polterauer1,
  6. Christoph Grimm1 and
  7. Thomas Bartl1,3
    1. 1 Department of Obstetrics and Gynecology, Division of General Gynecology and Gynecologic Oncology, Medical University of Vienna, Vienna, Austria
    2. 2 Department of Pathology, Medical University of Vienna, Vienna, Austria
    3. 3 Translational Gynecology Group, Department of Obstetrics and Gynecology, Medical University of Vienna, Vienna, Austria
    1. Correspondence to Dr Christoph Grimm, Department of Obstetrics and Gynecology, Division of General Gynecology and Gynecologic Oncology, Medical University of Vienna, Vienna 1090, Austria; christoph.grimm{at}meduniwien.ac.at

    Abstract

    Objective As more than 50% of newly diagnosed endometrial cancers remain classified as ‘no specific molecular subtype’ (NSMP) due to a lack of established biomarkers to further improve molecular subtyping, this study aims to evaluate the prognostic value of ARID1A in endometrial cancers of NSMP subtype.

    Methods Prospectively collected molecular profiling data of all consecutive patients with endometrial cancer who underwent primary surgery at our department between August 2017 and June 2022 and for whom both molecular profiling and clinical follow-up data were available were retrospectively evaluated. Tumor specimens were evaluated by combined mismatch repair protein immunohistochemistry and targeted next-generation hotspot sequencing. ARID1A mutational status, as defined by full-length gene sequencing, was matched with risk of recurrence, progression-free and disease-specific survival within the NSMP cohort.

    Results A total of 127 patients with endometrial cancer were included. Among 72 patients with tumors of NSMP subtype (56.7%), ARID1A mutations were identified in 24 cases (33.3%). ARID1A mutations were significantly associated with a higher risk of recurrence (37.5% vs 12.5%, OR 4.20, 95% CI 1.28 to 13.80, p=0.018) and impaired progression-free survival (HR 3.96, 95% CI 1.41 to 11.15, p=0.009), but not with disease-specific survival. The results for both risk of recurrence (OR 3.70, 95% CI 1.04 to 13.13, p=0.043) and progression-free survival (HR 3.19, 95% CI 1.10 to 9.25, p=0.033) were confirmed in multivariable analysis compared with advanced tumor stage International Federation of Gynecology and Obstetrics (2009) (FIGO ≥III) and impaired Eastern Clinical Oncology Group performance status (ECOG ≥1).

    Conclusion ARID1A appears to identify patients with endometrial cancer of NSMP subtypes with a higher risk of recurrence and could be used as a future prognostic biomarker. After clinical validation, ARID1A assessment could help to further sub-classify selected endometrial cancers and improve personalized treatment strategies.

    • Endometrium

    Data availability statement

    Data are available upon reasonable request. Raw data were generated at the Medical University of Vienna. Derived data supporting the findings of this study are available from the corresponding author on request.

    Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Although the clinical implementation of the molecular classification system marked a paradigm shift in the therapeutic landscape of endometrial cancer, no therapeutic recommendations for more than 50% of newly diagnosed endometrial cancers—still labeled as ‘no specific molecular profile’ (NSMP)—are available to date.

    WHAT THIS STUDY ADDS

    • This study provides the largest sample available to date to evaluate the association between ARID1A mutations and both the risk of recurrence and of progression-free survival in patients with endometrial cancer of NSMP subtype. ARID1A mutations appear to be independently associated with an increased risk of recurrence and impaired progression-free survival. Nevertheless, progression-free survival appears not to translate into reduced disease-specific survival, which may be hypothesized to be related to better response to recurrence therapy.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • As the prognosis of patients with NSMP subtype remains intermediate due its molecular background of this tumor subgroup, there is an urgent clinical need to identify new biomarkers to refine the molecular sub-classification. After clinical validation, ARID1A could be a promising new biomarker to sub-classify NSMP endometrial cancers and further personalize therapeutic strategies.

    Introduction

    The introduction of the endometrial cancer molecular classification into polymerase-epsilon mutated (POLEmut), mismatch-repair deficient (MMRd), copy-number high with mutations driven by TP53 (p53abn) and no specific molecular subtype (NSMP) induced a paradigm shift in early disease risk stratification and led to its incorporation in the International Federation of Gynecology and Obstetrics (FIGO) staging in 2023.1 2 The molecular classification aims to improve the reproducibility and interobserver variability of the traditional classification system of the World Health Organization based on histologic subtype, grade, tumor stage, myometrial invasion, and lymph vascular space invasion.3 4 However, according to the molecular classification, more than 50% of all primary endometrial cancers still fall into the category of NSMP, representing tumors of heterogenous molecular background and intermediate prognosis. For these tumors, no validated specific molecular biomarker allowing for further subclassification and/or personalized therapeutic targeting has entered clinical routine to date.5

    The AT-rich interacting domain-containing protein 1A gene (ARID1A) is an emerging molecular biomarker which is mutated in approximately 6% of all solid cancers and was previously associated with more pro-immunogenic tumor microenvironments and response to checkpoint inhibitor therapies in various solid tumors.6–10 ARID1A acts as a transcription factor and is the most frequently mutated subunit of chromatin remodeling complex SNF/SWI responsible for binding DNA.11 As previous research estimates that 30–40% of endometrial cancers appear to harbor ARID1A alterations, ARID1A could modify endometrial cancer phenotypes and clinical behaviors.12 13 Therefore, ARID1A could represent a promising biomarker to further sub-classify NSMP subtypes. This study aims to assess the prevalence of ARID1A mutations in NSMP endometrial cancers and to evaluate potential prognostic implications.

    Materials and Methods

    The study was designed as a retrospective analysis of a prospectively maintained single-center database of all consecutive patients with endometrial cancer who underwent primary surgical resection ± adjuvant treatment at the Medical University of Vienna between August 2017 and June 2022. Data of all patients with stages IA–IVB according to FIGO 2009, and for whom oncologic follow-up data at our institution was available, were analyzed. The assessments of mismatch repair protein status and molecular profiling were performed as part of the ongoing clinical routine by the Department of Pathology of the Medical University of Vienna. As shown in Figure 1, patients undergoing fertility-sparing therapy concepts, those who received neoadjuvant therapy, patients with malignant mixed mesodermal tumors, patients who underwent primary surgical therapy in another hospital, those with other active secondary malignancies within 5 years from diagnosis, and patients who did not undergo at least a 6-month oncologic follow-up at our department were excluded from the final analysis. Only patients with both a complete immunohistochemical mismatch repair protein panel and molecular profiling including ARID1A sequencing were considered for further assessment.

    Figure 1

    Flowchart showing the constitution of the patient cohort evaluated at final analysis (n=127). Of the 241 patients with endometrial cancer who underwent primary treatment initially evaluated, 114 were excluded for not meeting the primary inclusion criteria: missing or not complete molecular profiling (n=52), missing follow-up (n=28), rare histotype of tumor (n=13), primary surgery in another cancer center (n=9), secondary malignancy (n=6), hormone neoadjuvant treatment for fertility-sparing reason (n=5), and neoadjuvant chemotherapy (n=1).

    Surgical Approach

    Patients with early-stage endometrial cancer underwent routine laparoscopic hysterectomy, bilateral salpingo-oophorectomy, and pelvic sentinel lymphadenectomy using indocyanine green. In cases of suspected advanced disease, radical cytoreductive surgery via median laparotomy was performed. Adjuvant treatment and/or recurrence treatment consisting of systemic and/or radiotherapy, if applicable, were performed as indicated by our institution’s multidisciplinary oncology tumor board in accordance with the ESMO-ESGO-ESTRO (European Society for Medical Oncology, European Society of Gynecological Oncology, European Society for Radiotherapy and Oncology) guidelines.14 15 All patients were included in our department’s routine oncologic follow-up program, which recommends regular clinical examinations and imaging studies. Recommended follow-up visits are performed four times annually for 3 years, followed by two visits until year 5 and annual visits until year 10. Only patients with at least a 6-month follow-up were included in the statistical analysis.

    Molecular Profiling

    According to our institution’s internal standards, combined immunohistochemistry of mismatch repair proteins MLH1, PMS2, MSH6 and MSH2, followed by molecular profiling by targeted next generation sequencing, was implemented into clinical routine assessment for all patients with newly diagnosed endometrial cancer starting from August 2017.16

    Immunohistochemistry for mismatch repair proteins was performed on 2 µm thin tissue tumor sections with a Benchmark Ultra autostainer (Ventana Medical Systems, Tucson, Arizona, USA).16 For DNA sequencing, DNA was extracted from macro-dissected formalin-fixed, paraffin-embedded tissue treated with uracil-DNA glycosylase.17 DNA was quantified using a Qubit 2.0. DNA sequencing was obtained by the 50-gene mutation hotspot Ion AmpliSeq Cancer Hotspot Panel v2, which was expanded to the 161-gene panel of Oncomine Comprehensive Assay v3 (both Thermo Fisher Scientific, Waltham, Massachusetts, USA) by mid-2018 as previously described.18 ARID1A sequencing was only sporadically performed until mid-2018 but, starting from then, ARID1A sequencing became part of routine assemessment. All patients were categorized into molecular profiles as previously described.1 19 If multiple classifier endometrial cancer was identified, the tumor was classified as MMRd or as POLE-mut, respectively, while TP53 mutations were considered as a subclonal event occurring during tumor progression.20 21

    Statistical Analysis

    Statistical analysis was performed using Statistical Package for Social Sciences (SPSS) (IBM SPSS Statistics for Windows, v26.0. IBM Corp, Armonk, New York, USA) for Windows. Patient data were analyzed by descriptive statistics. Categorical variables were described using percentages and medians with IQR. For all statistical tests, p values <0.05 were considered statistically significant. To identify associations between ARID1A mutations and clinicopathological variables, Student’s t-test, χ2 test, or one-way analysis of variance were performed where appropriate. To test for associations between ARID1A mutation status and risk of recurrence, univariate and multivariable binary logistic regressions were fitted. Survival analyses were performed for the endpoints of progression-free survival and disease-specific survival with a follow-up data cut-off of September 30, 2023. Regression models were fitted applying FIGO staging and ECOG performance status as clinically relevant covariates. To achieve maximal statistical power, all eligible patients who received primary treatment between the timepoints of first available routine tumor profiling in August 2017 and June 2022 to allow for a minimal follow-up of at least 1 year from the date of initial data analysis were included. Progression-free survival and disease-specific survival were defined as the time between primary surgery and cancer recurrence or endometrial cancer-related death. Survival analysis was performed fitting Cox regressions and patient survival was visualized by Kaplan–Meier curves. Survivors were censored at the last timepoint of available follow-up data.

    Results

    In total, 127 patients were assessed for final statistical analysis of which 10 were classified as POLE-mut (7.9%), 33 as MMRd (26.0%), 12 as p53abn (9.4%), and 72 as NSMP (56.7%). ARID1A mutations were observed in 60.0% of those classified as POLE-mut (6/10), 30.3% of MMRd (10/33), 8.3% of p53abn (1/12) and 33.3% of NSMPs (24/72).

    Clinicopathologic baseline characteristics of the NSMP cohort are shown in Table 1. No associations between ARID1A status and pre-therapeutic parameters could be identified. Four patients with FIGO IVb disease (three cases of localized peritoneal carcinosis with low tumor load, one positive inguinal lymph node) all underwent successful primary cytoreductive surgery without residual disease; three were ARID1A wildtype (ARID1A-wt), one was ARID1A mutation (ARID1A-mut). Median follow-up within the subgroup of ARID1A-mut NSMPs was 22.5 (IQR 13.0–37.8) months.

    Table 1

    Descriptive statistics of patients with NSMP endometrial cancer undergoing primary surgery

    Assessing potential associations between ARID1A mutational status and recurrence within the NSMP cohort (n=72), ARID1A mutations were univariately associated with an increased risk of recurrence (ARID1A-mut: 37.5% vs ARID1A-wt: 12.5%; OR 4.20 (95% CI 1.28 to 13.80), p=0.018). The association between ARID1A mutational status and recurrence remained significant in multivariable analysis (ARID1A-mut: OR 3.70 (95% CI 1.04 to 13.13), p=0.043) compared with advanced FIGO stage (FIGO ≥III: OR 2.25 (95% CI 1.15 to 4.41), p=0.018) and impaired Eastern Clinical Oncology Group (ECOG) performance status (ECOG ≥1: OR 2.19 (95% CI 0.29 to 16.58), p=0.451) (see Online supplemental table S1A). In line with this, ARID1A mutations were associated with significantly impaired progression-free survival in NSMP endometrial cancers (univariate: HR 3.96 (95% CI 1.41 to 11.15), p=0.009). The results were confirmed in multivariable analyses (ARID1A-mut: HR 3.19 (95% CI 1.10 to 9.25), p=0.033) compared with advanced FIGO stage (FIGO ≥III: HR 1.84 (95% CI 1.06 to 3.18), p=0.029) and impaired performance status (ECOG ≥1: HR 1.07 (95% CI 0.23 to 5.06), p=0.932); see Online supplemental table S1B). A Kaplan–Meier curve depicting progression-free survival according to ARID1A mutational status is shown in Figure 2A.

    Supplemental material

    Figure 2

    (A, B) Kaplan–Meier curves showing progression-free survival (PFS) and disease-specific survival (DSS) from the timepoint of primary surgery stratified by ARID1A mutational status. ARID1A-mutated tumors were associated with an impaired PFS. Despite a log-rank test of p=0.049, the association of ARID1A mutations and improved DSS was not considered statistically significant as a related Cox regression analysis did not yield significant results but very wide confidence intervals.

    To allow for contextualization of observed effects, both models were analogously translated to the overall cohort of endometrial cancers independent of molecular subtype (n=127) which did not yield significant associations between ARID1A mutational status in binary regression (ARID1A-mut: OR 0.93 (95% CI 0.38 to 2.27), p=0.870) nor in Cox regression (ARID1A-mut: HR 1.03 (95% CI 0.47 to 2.26), p=0.942). Multivariate models compared with advanced FIGO stage and impaired performance status showed equally non-significant results.

    In a subgroup analysis of all FIGO I and II endometrial cancers of NSMP subtype (n=56), ARID1A remained predictive of disease recurrence in both univariate analysis (ARID1A-mut: 31.3% vs ARID1A-wt: 5.0%; OR 8.6 (95% CI 1.47 to 8.64), p=0.017) and multivariable analysis (ARID1A-mut: OR 9.49, 95% CI 1.53 to 59.05), p=0.008) compared with impaired performance status (ECOG ≥1: OR 2.42 (95% CI 0.18 to 32.41), p=0.505). In line with this, ARID1A mutations were associated with significantly impaired progression-free survival (univariate: HR 7.83 (95% CI 1.51 to 40.66), p=0.014). The results could be confirmed in multivariable analyses (ARID1A-mut: HR 8.43 (95% CI 1.60 to 44.48), p=0.012) compared with impaired performance status (ECOG ≥1: HR 2.85 (95% CI 0.31 to 26.58, p=0.357).

    Assessing potential associations between ARID1A mutational status and survival, ARID1A mutations were not associated with disease-specific survival in univariate Cox regression (cancer-related death: 12.5% (ARID1A-wt) vs 0% (ARID1A-mut)) within the NSMP cohort (univariate: HR 0.02 (95% CI 0.00 to 19.93), p=0.272)). A Kaplan–Meier curve showing disease-specific survival according to ARID1A mutational status is shown in Figure 2B. As the log-rank test showed a significant association (p=0.049) but the Cox regression failed to confirm statistical significance, calculating very wide confidence intervals, the result was not considered statistically significant. In line with this, ARID1A mutations were not associated with disease-specific survival in multivariable analysis (HR 0.00 (95% CI 0.00 to not reached), p=0.988) compared with advanced FIGO stage (FIGO ≥III: HR 9.05 (95% CI 1.63 to 50.26), p=0.012) and impaired performance status (ECOG ≥1: HR 0.00 (95% CI 0.00 to not reached), p=0.967).

    Of 15 patients with NSMP endometrial cancers who experienced disease recurrence, 13 received recurrence treatment and underwent subsequent oncologic follow-up at our department. Of eight eligible patients with ARID1A-mut tumors, six (75.0%) responded to treatment and all patients experienced disease control (100.0%) following second-line therapy. Of five eligible patients with ARID1A-wt tumors, one responded to recurrence treatment (20.0%) and three experienced disease control (60.0%) following second-line treatment. Two patients (25.0%) with ARID1A-mut tumors remain in complete remission following second-line treatment at 24 months follow-up compared with none (0%) with ARID1A-wt tumors. One patient with an ARID1A-mut tumor who progressed following second-line chemotherapy after 8 months received pembrolizumab 200 mg fixed-dose every 3 weeks as third-line therapy and remained in complete remission at the 8-month follow-up.

    Details of disease recurrence, treatment of recurrence, and treatment response are shown in Online supplemental table S2. An exploratory survival comparison of all patients who received recurrence therapy and underwent subsequent oncologic follow-up at our department yielded a significant progression-free survival 2 difference in favor of ARID1A-mut tumors (log-rank p=0.038; see Kaplan–Meier curve in Online supplemental figure S1).

    Supplemental material

    Supplemental material

    Discussion

    Summary of Main Results

    ARID1A mutations appear to be associated with an increased risk of disease recurrence and significantly impaired progression-free survival, but not in endometrial cancers independent of the molecular classification. The higher risk of recurrence in patients within the NSMP subgroup and ARID1A mutations did not translate into impaired disease-specific survival.

    Results in the Context of Published Literature

    In line with previous studies, one-third of all NSMP endometrial cancers in the present cohort harbored ARID1A mutations.12 13 The association between ARID1A mutations and impaired progression-free survival is also in line with previously reported studies. Yokoyama et al observed that decreased expression of ARID1A was associated with resistance to platinum-based chemotherapy and impaired progression-free survival in patients with advanced clear-cell ovarian cancer.22 Cho et al showed that loss of BAF250a expression, the protein encoded by ARID1A, was a predictor of shorter disease-specific survival and overall survival in patients with cervical cancer.23 A meta-analysis by Liu et al indicated that missing expression of ARID1A is associated with impaired progression-free survival in patients with ovarian clear cell carcinoma.24

    To our knowledge, four studies have previously assessed the prognostic value of ARID1A expression on survival in endometrial cancer to date. Three studies retrospectively analyzed a total of 177 endometrial cancers with loss of ARID1A expression in stored paraffin-embedded tissue and reported no significant association between ARID1A expression and survival.25–27 A pooled meta-analysis of these three studies confirmed the negative results.24 As these studies were published between 2013 and 2014, no subgroup analysis according to molecular subtypes is available. Both the percentage of observed ARID1A aberrations and the missing association with survival among endometrial cancers unselected for molecular subytpes is in line with the present study. Only one study has assessed the prognostic significance of ARID1A mutations in NSMP endometrial cancers to date, reporting ARID1A mutations to be associated with an increased Ki67 proliferative index and increased risk of recurrence. This study features a valuable comparison of both immunohistochemical ARID1A expression levels and ARID1A sequencing, demonstrating concordant results. However, the number of ARID1A-mut tumors within the NSMP cohort was small (n=19/33) and survival analysis data are limited.28

    Interestingly, the increased risk of cancer recurrence and impaired progression-free survival associated with ARID1A mutations appears not to translate into impaired disease-specific survival in the present study. ARID1A mutations were previously associated with more pro-immunogenic tumor phenotypes in several solid tumors, which may provide a molecular rationale for an improved response to treatment. ARID1A mutations were previously hypothesized to promote tumor mutability, increase the expression of programmed death-ligand (PD-L1), and modulate the tumor immune microenvironment.7 Increased PD-L1 and tumor-infiltrating lymphocyte levels were previously reported for ARID1A-deficient clear cell and mucinous ovarian cancer.10 29 Okamura et al reported a significant association of ARID1A alterations and higher response rates to immune checkpoint inhibitor therapy response in nine different solid tumors including uterine and ovarian endometrial carcinoma.9

    In the present study the number of patients undergoing second-line treatment for disease recurrence is very limited. An exploratory data analysis hints in the direction of a longer progression-free survival 2 in favor of ARID1A-mut tumors, probably as a consequence of better treatment response linked to a different tumor phenotype; this observation may provide an explanation for the missing association between ARID1A and disease-specific survival. However, these considerations with regard to disease-specific survival and response to recurrence therapy are to be seen as purely exploratory and hypothesis-generating and do not allow clarification of this interesting finding.

    Strengths and Weaknesses

    The current study provides the largest dataset to date, featuring prospectively collected sequencing data and follow-up data obtained by a standardized timely internal follow-up program. The results underline the promising prognostic value of ARID1A for these specific molecular tumor subtypes.

    The present study, however, also has limitations. First, as is typical of retrospective analyses, the lack of random patient assignment, patient selection, and potentially incomplete data acquisition reduce its clinical applicability. Second, even though the present study features the largest reported cohort to date, the analyzed data were collected prospectively and cases were selected with the utmost methodical care to alleviate potential bias, the present results should be interpreted with caution due to the limited number of observed ARID1A-mut NSMP tumors. Third, because of the limited follow-up in a cohort with a comparatively good prognosis, the present data do not allow us to elucidate the association between ARID1A mutations and disease-specific survival or whether ARID1A mutations could be linked to an improved response to recurrence therapy. The findings are therefore to be considered as purely exploratory and hypothesis-generating.

    Implications for Practice and Future Research

    Despite emerging evidence on a more pro-immunogenic phenotype and a distinct clinical behavior associated with ARID1A mutations in various solid tumors, the data for endometrial cancer remain scarce. The relatively high prevalence of ARID1A mutations in endometrial cancer could allow us to define a clinically relevant biomarker and satisfy the high unmet clinical need to further subclassify the NSMP subtype. Of note, as ARID1A sequencing remains cost-intensive, immunohistochemistry may offer an alternative to identify the ARID1A mutational status in endometrial cancers. A large body of evidence shows that ARID1A immunohistochemistry reliably predicts ARID1A mutations.30 31

    Conclusion

    ARID1A mutations appear to identify a subgroup of endometrial cancers of the NSMP subtype which appears to be associated with an increased risk of recurrence and impaired progression-free survival. After clinical validation, ARID1A could be a promising new biomarker to further sub-classify NSMP endometrial cancers and potentially even further personalize therapeutic strategies in affected patients.

    Data availability statement

    Data are available upon reasonable request. Raw data were generated at the Medical University of Vienna. Derived data supporting the findings of this study are available from the corresponding author on request.

    Ethics statements

    Patient consent for publication

    Ethics approval

    The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Medical University of Vienna (IRB 2014/2019 and 1966/2020). Participants gave informed consent to participate in the study before taking part.

    References

    Supplementary materials

    Footnotes

    • Contributors Conceptualization: AO, TB and CG. Data curation: AO, TD and GH. Formal analysis: AO, TB, SP and LM. Investigation: TD, GH. Methodology: AO and TB. Project administration: CG. Resources: SP, CG, GH and LM. Software: AO and TB. Supervision: CG. Validation: CG, SP and GH. Visualization: AO and TB. Writing—original draft preparation: AO and TB. Writing—review and editing: GH and CG. Guarantor for overall content: AO.

    • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.