Combination of Proactive Molecular Risk Classifier for Endometrial cancer (ProMisE) with sonographic and demographic characteristics in preoperative prediction of recurrence or progression of endometrial cancer

ABSTRACT Objective To evaluate the ability of demographic and sonographic variables and the Proactive Molecular Risk Classifier for Endometrial cancer (ProMisE) classification to predict preoperatively tumor recurrence or progression in women with endometrial cancer. Methods The study included 339 women with histologically confirmed endometrial cancer who underwent expert transvaginal ultrasound in a single center before surgery as part of the prospective International Endometrial Tumor Analysis 4 study or who were evaluated using the same protocol. The tumors were classified according to histotype, FIGO (International Federation of Gynecology and Obstetrics) grade and FIGO stage. In addition, molecular analysis was performed for classification into the four ProMisE subtypes: polymerase‐ϵ exonuclease domain mutations (POLE EDM), mismatch repair proteins deficiency (MMR‐D), protein 53 wild type (p53 wt) and protein 53 abnormal (p53 abn). Demographic and preoperative sonographic characteristics, tumor recurrence or progression and survival were compared between the ProMisE subgroups. Cox regression analysis was used to identify prognostic factors associated with recurrence or progression, using univariable models to study crude associations and multivariable models to study adjusted associations. Logistic regression and receiver‐operating‐characteristics (ROC)‐curve analysis were used to assess the predictive ability of the preoperative prognostic factors regarding recurrence or progression of cancer within 3 years after surgery, and to compare their predictive ability to that of the European Society for Medical Oncology (ESMO) preoperative (based on depth of myometrial invasion, histotype and grade) and postoperative (based on histotype, grade, surgical stage and lymphovascular space invasion) risk classifications. In a separate subanalysis, cases were stratified according to ProMisE p53 abn status (present vs absent) and sonographic tumor size (anteroposterior (AP) diameter < 2 cm vs ≥ 2 cm). Results Median follow‐up time from surgery was 58 months (interquartile range, 48–71 months; range, 0–102 months). Recurrence or progression of cancer occurred in 51/339 (15%) women, comprising 14% of those with MMR‐D, 8% of those with POLE EDM, 9% of those with p53 wt and 45% of those with p53 abn ProMisE subtype. On multivariable analysis, age, waist circumference, ProMisE subtype and tumor extension and AP diameter on ultrasound were associated with tumor recurrence or progression. A multivariable model comprising ProMisE subtype, age, waist circumference and sonographic tumor extension and size (area under the ROC curve (AUC), 0.89 (95% CI, 0.85–0.93)) had comparable ability to predict tumor recurrence/progression to that of a multivariable model comprising histotype, grade, age, waist circumference and sonographic tumor extension and size (AUC, 0.88 (95% CI, 0.83–0.92)), and better predictive ability than both the preoperative (AUC, 0.74 (95% CI, 0.67–0.82); P < 0.01) and postoperative (AUC, 0.79 (95% CI, 0.72–0.86); P < 0.01) ESMO risk classifications. Women with a combination of non‐p53 abn subtype and tumor size < 2 cm (164/339 (48%)) had a very low risk (1.8%) of tumor recurrence or progression. Conclusions The combination of demographic characteristics, sonographic findings and ProMisE subtype had better preoperative predictive ability for tumor recurrence or progression than did the ESMO classification, supporting their use in the preoperative risk stratification of women with endometrial cancer. The combination of p53 status with ultrasound tumor size has the potential to identify preoperatively a large group of women with a very low risk of recurrence or progression. © 2020 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology. ‐ Legal Statement: This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.


What are the novel findings of this work?
The combination of demographic variables (age, waist circumference), sonographic variables (tumor size and extension) and the Proactive Molecular Risk Classifier for Endometrial cancer (ProMisE) classification had better ability than the European Society for Medical Oncology (ESMO) risk classification to predict preoperatively tumor recurrence or progression in women who underwent surgery for endometrial cancer. Ultrasound tumor size < 2 cm combined with absence of protein 53 abnormal (p53 abn) ProMisE subtype can identify a large group (∼50%) of women at very low risk of recurrence or progression of disease.

What are the clinical implications of this work?
In women with endometrial cancer, preoperative ultrasound assessment has an independent prognostic role beyond that of the preoperative ESMO classification, as sonographic tumor size combined with p53 status can identify a large group of women with an excellent prognosis, in whom sentinel-node biopsy or adjuvant treatment may not be considered necessary. Our findings support the use of the ProMisE molecular classification in the preoperative risk stratification of women with endometrial cancer.

INTRODUCTION
In patients with endometrial cancer, transvaginal ultrasound can be used in conjunction with tumor histotype and grade from endometrial biopsy to predict preoperatively the risk of lymph-node metastasis according to the European Society for Medical Oncology (ESMO), European Society for Radiotherapy & Oncology (ESTRO) and European Society of Gynaecological Oncology (ESGO) classification (referred to as 'ESMO classification' in this paper, for brevity) 1 . The preoperative ESMO risk classification (based on depth of myometrial invasion on imaging and tumor histotype and grade from preoperative biopsy) guides decision-making for lymphadenectomy based on the risk for lymph-node metastasis, whereas the postoperative ESMO risk classification (based on surgical stage, grade, histotype and lymphovascular space invasion (LVSI) from the surgical specimen) guides the use of adjuvant therapy based on the risk of recurrence.
Ultrasound is an established modality in the preoperative risk assessment for lymph-node metastasis 1 , while the ability of ultrasound to predict before surgery recurrence or progression has not been studied. Moreover, the value of biometric variables in the prediction of adverse prognosis in women with endometrial cancer needs to be further explored.
Tumor histotype and grade are important in both the pre-and postoperative ESMO classification, but have limited reproducibility, particularly in high-grade tumors [2][3][4][5] . Moreover, agreement on tumor grade between endometrial biopsy and the hysterectomy specimen is only moderate 6 . These limitations hinder a reproducible categorization of endometrial cancer and limit the value of tumor histotype and grade as risk predictors.
The Cancer Genome Atlas (TCGA) Research Network developed a genomic classification of endometrial cancer into four prognostic subgroups (polymerase-ε (POLE) ultramutated, microsatellite instability hypermutated, copy number low and copy number high 7 ), which, however, requires costly and complex methodologies, making the TCGA classification not yet fit for clinical use. The Proactive Molecular Risk Classifier for Endometrial cancer (ProMisE) was developed and validated as a clinically applicable surrogate molecular classifier [8][9][10][11] , rendering four corresponding prognostic subgroups: POLE exonuclease domain mutations (POLE EDM), mismatch repair proteins deficiency (MMR-D), protein 53 wild type (p53 wt) and protein 53 abnormal (p53 abn). A molecular classification system is more robust and objective than tumor histotype and grade, as it is based on the presence or absence of a protein or mutation. It allows classification of all endometrial cancers in the preoperative setting, when several prognostic factors, such as surgical tumor stage and LVSI, are not available. Moreover, in contrast to histotype and grade, ProMisE classification on diagnostic endometrial biopsy has been shown to be highly concordant with subsequent classification on the hysterectomy specimen 11,12 .
The objective of this study was to evaluate the ability of demographic and sonographic variables and the ProMisE classification to predict preoperatively tumor recurrence or progression in women with endometrial cancer.

459
Endometrial Tumor Analysis (IETA) 4 study 13 or were evaluated using the same protocol. Inclusion lasted from 1 January 2011 to 31 December 2015. End of follow-up was on 31 August 2019. The inclusion criterion was histologically confirmed endometrial cancer based on the preoperative biopsy and/or the hysterectomy specimen. Only women with epithelial malignant tumors (endometrial carcinoma: endometrioid, mucinous, serous, clear cell, mixed cell and undifferentiated carcinomas) or mixed epithelial and mesenchymal malignant tumors (carcinosarcomas) were included. Exclusion criteria were: hysterectomy not performed or carried out at another hospital or performed more than 120 days after the ultrasound examination; final diagnosis other than endometrial cancer; incomplete ultrasound data; duplicate entries; error in the identification key; and insufficient material for the construction of a tissue microarray and isolation of genomic DNA from formalin-fixed paraffin-embedded tumor tissue.
The study was approved by the local ethics committee (LU 2016/362). Permission for biobanking of tissue samples was granted by the regional biobank review board (2018-00479). All women gave written consent for use of their biobank-stored tissue for research purposes.
All women underwent preoperative ultrasound examination by the same ultrasound expert (E.E.). Tumor size (anteroposterior (AP) tumor diameter), extension and morphology were assessed on ultrasound according to the IETA examination technique and terminology 14 . Tumor AP diameter was measured in the sagittal plane. Color/power Doppler examinations were carried out at a pulse repetition frequency of 0.3-0.9 kHz. All women underwent hysterectomy and bilateral salpingo-oophorectomy with or without lymphadenectomy. Pathologists classified tumor stage according to the FIGO (International Federation of Gynecology and Obstetrics) 2009 staging system 15 and tumor grade according to the FIGO grading system 16 . Detailed medical and reproductive history, using a standardized questionnaire, and biometric data (height, weight and waist circumference, measured by the physician/nurse on the day of the ultrasound examination) were included in the study protocol. All demographic and sonographic data were entered into internet-based data capture software (Clinical Data Miner (https://cdm.esat.kuleuven.be)) 17 on the day of the ultrasound examination. Results regarding histological outcome and tumor stage were entered into the database after surgery. Data on recurrence or progression of cancer and survival were obtained through review of the patient's digital medical records.
Molecular analysis for classification into the four ProMisE subtypes (MMR-D, POLE EDM, p53 wt and p53 abn) was performed retrospectively using biobank-stored tissue from the hysterectomy specimen. Two pathologists (J.W.C., D.N.), blinded to the patient characteristics and outcomes, independently reviewed all immunohistochemistry stains, and any interpretative discrepancies were resolved by consensus, using a multiheaded microscope. As p53 immunohistochemistry staining was performed as part of routine clinical practice in all cases with endometrial cancer, it was obtained from full tumor sections. Immunohistochemistry results for p53 were categorized into three groups: Group 0, completely negative staining; Group 1, 1-80% of the tumor nuclei showed heterogeneous staining; and Group 2, > 80% of tumor nuclei showed strong positive staining. Group 1 was considered as p53 wt and Groups 0 and 2 as p53 abn. Mismatch repair (MMR) status was analyzed by immunohistochemistry for the presence of the microsatellite-stability proteins MSH2, MSH6, PMS2 and MLH1 on tissue microarrays from formalin-fixed paraffin-embedded tumor tissue. MMR proteins were considered absent (MMR-D) if one or more of the four microsatellite-stability proteins were missing, or intact if all four microsatellite-stability proteins were present. POLE EDM status was analyzed from genomic DNA, which was isolated from two 1-mm core punches of formalin-fixed paraffin-embedded tissue. Mutations of the POLE gene (NM.006231) exons 9 to 14 were analyzed by Sanger sequencing. The following POLE mutations were considered pathogenic: P286R, V411L, S297F, A456P and S459F. These are the five most common pathogenic variants described for which there are strong data linking them to the ultramutated phenotypes 8,[18][19][20] . ProMisE classification of the tumors was performed according to the pragmatic model of Talhouk et al. 10 . According to the ProMisE decision tree, tumors were assigned to a specific molecular subtype in the following order: first, tumors with MMR proteins deficiency were classified as 'MMR-D'; of the remaining cases, tumors with EDM in the POLE gene were classified as 'POLE EDM'; of the remaining cases, tumors with p53 wild type were classified as 'p53 wt' and the rest of the tumors with p53 null or missense mutations were classified as 'p53 abn'.
Endometrial biopsies, through simple biopsy, dilatation and curettage or hysteroscopic resection, were performed before study inclusion and analyzed in various pathology departments in Stockholm, whereas the hysterectomy specimens were analyzed in the same department at the Karolinska University Hospital where the surgery took place. For practical reasons (i.e. access to tissue blocks), the ProMisE molecular analysis was performed on tumor tissue from the hysterectomy specimen, but the classification was used as a proxy for the preoperative biopsy specimen throughout the study, as agreement in ProMisE classification between the diagnostic biopsy and hysterectomy is high 11,12,21 .
The preoperative ESMO risk classification was based on findings on endometrial biopsy and transvaginal ultrasound (depth of myometrial invasion, histotype and grade) and comprised the following risk categories: low risk, Grade 1 . The advanced and metastatic risk groups were combined owing to the low number of women in these two groups.
Recurrence was defined as recurring tumor in a woman who had been tumor-free either directly after surgery or at the end of primary treatment. The date of recurrence was defined as the date of biopsy-confirmed recurrence in all cases but three, in which confirmative biopsy was not performed initially or not performed at all. In these three cases, recurrence date was defined as the date of recurrence according to computed tomography or clinical examination. Progression was defined as tumor progression in a woman who had remaining tumor at the end of the primary treatment. The date of progression was defined as the date of progression diagnosed by computed tomography. Overall survival was defined as time from surgery until death due to any cause, loss to follow-up or end of follow-up, whichever occurred first. Disease-free survival time was defined as the time from surgery to detection of recurrence, loss to follow-up, death due to any cause or end of follow-up, whichever occurred first.

Statistical analysis
Statistical analysis was performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, USA), STATA/IC version 12.1 (Stata Corp., College Station, TX, USA) and R version 3.6.1 (https://www.r-project.org/). Fisher's exact test was used for comparison of categorical variables, and the Mann-Whitney U-test and the Kruskal-Wallis test were used for comparison of continuous variables between two groups and between more than two groups, respectively. Categorizations for age 13 , body mass index (BMI) 22 , waist circumference 23 and ultrasound tumor size 24 on univariable and multivariable analyses were based on previous publications. Demographic and sonographic characteristics, recurrence or progression of cancer and survival were compared between the ProMisE subgroups, with focus on the p53 abn subtype, which is known to be associated with adverse outcome [8][9][10]12,20 . Women were also stratified according to p53 status (present vs absent) and ultrasound tumor size (AP diameter < 2 cm vs ≥ 2 cm), which are clinically easily obtained prognostic factors known to be associated
with adverse outcome 24,25 , and compared regarding risk of recurrence or progression. Survival analysis was performed using Kaplan-Meier curves, with pairwise comparison of ProMisE subtypes using the log-rank test. Cox regression analysis was used to identify preoperative variables associated with recurrence or progression of endometrial cancer. Univariable models were used to study crude associations and multivariable models to study adjusted associations. All variables found to be significant in the univariable models were analyzed in multivariable models. The variables histotype/grade and ProMisE subtype were correlated strongly on multivariable analysis. Therefore, they were analyzed in separate, otherwise identical, multivariable models: (1) the 'histotype-and-grade model', which included tumor histotype and grade, age, waist circumference, ultrasound tumor extension and ultrasound tumor size; and (2) the 'ProMisE model', which included ProMisE subtype, age, waist circumference, ultrasound tumor extension and ultrasound tumor size. The ProMisE model was also adjusted for the ESMO postoperative classification, to determine if the variables had an independent association with tumor recurrence or progression, beyond the ESMO classification. All women were followed up until recurrence or progression (event), or censured owing to death, loss to follow-up or end of follow-up, whichever occurred first.
To evaluate whether preoperative prognostic factors associated with progression or recurrence on multivariable Cox regression analysis also had a predictive value, the histotype-and-grade model, the ProMisE model and the preoperative and postoperative ESMO risk classifications were analyzed using logistic regression with a fixed time (recurrence or progression within 3 years (yes/no), as all women underwent surgery at least 42 months before the study end). Their ability to predict recurrence or regression was assessed using the areas under receiver-operating-characteristics (ROC) curves (AUC). The statistical significance of a difference in AUC was determined using pairwise comparison through the DeLong test. These analyses were performed to assess the ability of the preoperative prognostic factors evaluated to predict recurrence or progression of endometrial cancer, and to compare their predictive ability with that of the ESMO pre-and postoperative risk classifications. All tests were two-sided, and P < 0.05 was considered to indicate statistical significance.

RESULTS
Eligible for inclusion were 409 women from the Stockholm center at Karolinska University Hospital who were part of the prospective IETA 4 cohort, and an additional two women who were examined at the Stockholm center according to the same protocol (n = 411). Of these, 72 women were excluded (underwent surgery in another hospital (n = 6), insufficient or no remaining tumor sample for analysis (n = 38), incorrect personal security number (n = 2), incomplete ProMisE analysis (n = 8), ProMisE analysis not performed (n = 17), duplicate case (n = 1)), leaving a study cohort of 339 women. Their demographic, sonographic and histopathological characteristics are presented in Table 1.
Lymphadenectomy was performed in 91 (27%) women, of whom 21 (23%) had lymph-node metastasis. Median follow-up time from surgery was 58 months (interquartile range, 48-71 months; range, 0-102 months). Two women died within 1 month after surgery: one after 30 days owing to postoperative complications and the other of unknown cause in her home after 3 days. Three women were lost to follow-up after 29, 30 and 40 months because they moved to another county. Overall, 51 (15%) women had tumor recurrence or progression. This was detected because of symptoms in half of the women and at routine follow-up in the other half. The vast majority (38/43 (88%)) of recurrences occurred within 3 years and all progressions (100% (8/8)) within 2 years after surgery. Women with tumor recurrence or progression, compared with those without, were significantly older and had a larger waist circumference, their tumors were more often non-endometrioid, of ProMisE subtype p53 abn and of higher stage, and on ultrasound their tumors were larger, with a higher color score and more advanced tumor extension ( Table 1).
The clinical, sonographic and histopathological characteristics of the women according to their ProMisE subtype are presented in Table 2. Compared with the other subtypes, p53 abn was associated with older age, larger tumors on preoperative ultrasound, non-endometrioid cancer, higher cancer stage, more advanced postoperative ESMO risk group, higher likelihood of death from disease and lower 5-year disease-free survival and overall survival. A Kaplan-Meier plot on tumor recurrence or progression according to ProMisE subtype is presented in Figure 1  shows that women with p53 abn had a higher probability of recurrence or progression than did the other three subgroups. All evaluated preoperative variables apart from BMI were associated with tumor recurrence or progression on univariable analysis (Table 3). Among the ProMisE subtypes, only p53 abn was associated with recurrence or progression. On multivariable analysis (which included all preoperative variables significantly associated with tumor recurrence or progression on univariable analysis) only age, waist circumference, ProMisE subtype, tumor extension on ultrasound and tumor AP diameter on

DISCUSSION
We have verified that the ProMisE subtype p53 abn is an adverse prognostic factor, as it was associated with larger tumors on preoperative ultrasound, non-endometrioid cancer, higher stage, increased risk of recurrence or progression and reduced survival, compared with the other ProMisE subtypes. The combination of non-p53 abn status and sonographic tumor AP diameter < 2 cm showed the potential to identify a large group of women (48%) at very low risk (1.8%) of recurrence or progression of disease. The ProMisE model, including ProMisE subtype and demographic (age, waist circumference) and sonographic (tumor size and extension) prognostic factors, had better predictive ability for the recurrence or progression of disease than did the current pre-and postoperative ESMO risk classifications, supporting its use in preoperative risk stratification.
Strengths of this study include the prospective study cohort, which represents a general population and not a high-risk sample, recruited during a recent and limited period of time, and the comprehensive prospective cohort database containing detailed clinical information, in which all data were locked after being saved. Nearly half of the study population had a follow-up of 5 years or longer and all had high-quality ultrasound data.
A shortcoming of the study is that the ProMisE molecular analysis was performed on the hysterectomy specimen and not on the preoperative endometrial biopsy sample. However, it has been shown that ProMisE classification is highly concordant between preoperative biopsy and subsequent hysterectomy specimen 11,12 , and it has been concluded that the results of molecular markers, such as p53, on hysterectomy specimen can be safely translated towards the preoperative endometrial biopsy 21 . Another limitation is that the prognostic factors were identified from the same dataset that was used to compare their predictive ability with that of the ESMO classification, which may have favored the ProMisE and histotype-and-grade models. The prognostic factors for tumor recurrence or progression identified in our study could serve as a basis for future studies aiming to create preoperative risk-prediction models for recurrence or progression of endometrial cancer, based on a larger cohort, with ProMisE analysis performed on preoperative biopsy specimens and variable selection based on a-priori knowledge and with external validation.
Previous studies have found that tumor size on hysterectomy specimen 24-26 , ultrasound 13 and MRI 27 are predictive of lymph-node metastasis [24][25][26][27] , survival 24,25 and high-risk disease 13 . In accordance with these data, we found that tumors with an AP diameter of ≥ 2 cm on preoperative ultrasound were associated with more advanced tumor extension on ultrasound, increased likelihood of lymph-node metastasis and worse survival outcomes, supporting the potential for sonographic tumor size to predict adverse outcome before surgery. In addition, our findings show that tumor size on preoperative ultrasound is an independent predictor of recurrence or progression of disease, even among Increased BMI, consistent with being overweight or obese, is associated with an increased risk of the development of endometrial cancer 28 , and obesity at diagnosis has been associated with worse survival, though there is insufficient evidence to establish an increased risk of recurrence 29 . Changes in insulin resistance, systemic inflammation and alterations in the levels of hormones and growth factors have all been implicated in promoting the development and progression of cancer in those who are overweight or obese 30 . In spite of these data, we found no association between BMI and tumor recurrence or progression (Table 3). However, a waist circumference of ≥ 88 cm was found to be an independent predictor of recurrence or progression of cancer, with the risk being more than double that of women with a waist circumference of < 88 cm (Tables 4 and 5). This discrepancy might be explained by the fact that BMI constitutes a poor proxy for adiposity, as it does not  describe the distribution of adipose tissue or distinguish adipose tissue from muscle mass 30 . In postmenopausal women with endometrial cancer, subcutaneous fat contributes more to estradiol levels than does visceral fat, indicating that subcutaneous fat might be relevant for the carcinogenesis of endometrial cancer 31 , and women with increasing visceral fat percentage have a significantly reduced disease-specific survival, independent of BMI 32 . This indicates that the location of body fat is prognostic, and it can be stated that our divergent findings on BMI and waist circumference indicate that abdominal adiposity is a worse prognostic factor than is general adiposity. To the best of out knowledge, this is the first time that waist circumference has been reported as an independent predictor of recurrence or progression of endometrial cancer. The ProMisE classification was validated in a population-based cohort of 452 women 11 , which was similar to our cohort. The ProMisE subtype distribution was comparable to that in our cohort, and the study also identified p53 abn as an adverse prognostic factor. The ProMisE classification has several advantages, as it shows a higher concordance between preoperative endometrial biopsy and hysterectomy compared with tumor histotype and grade 11,12 , is able to differentiate high-grade tumors with excellent prognosis (POLE EDM) from tumors with poor prognosis (p53 abn) [8][9][10] , can evaluate tumors in the gray zone between endometrioid and serous histotype and can identify women with MMR-D, who may have Lynch syndrome and who should be referred for genetic counseling and testing [8][9][10] .
In the era of sentinel-node biopsy, the clinical importance of preoperative ultrasound has been questioned, at least in situations in which sentinel-node biopsy is offered to all women. This study indicates, however, that ultrasound variables have an independent prognostic role beyond the ESMO risk classification and that the combination of ultrasound and p53 status, often obtained on routine histopathology assessment, can identify a large group of women (48%) at very low risk of tumor recurrence or progression (1.8%) in whom neither sentinel-node biopsy nor adjuvant treatment may be considered necessary. Moreover, the combination of demographic characteristics, sonographic findings and ProMisE subtype had a better preoperative predictive ability for the recurrence or progression of disease than did the ESMO risk classification, supporting its use in the preoperative risk stratification of women with endometrial cancer.