DEveloping Tests for Endometrial Cancer deTection (DETECT): protocol for a diagnostic accuracy study of urine and vaginal samples for the detection of endometrial cancer by cytology in women with postmenopausal bleeding

Introduction Postmenopausal bleeding (PMB), the red flag symptom for endometrial cancer, triggers urgent investigation by transvaginal ultrasound scan, hysteroscopy and/or endometrial biopsy. These investigations are costly, invasive and often painful or distressing for women. In a pilot study, we found that voided urine and non-invasive vaginal samples from women with endometrial cancer contain malignant cells that can be identified by cytology. The aim of the DEveloping Tests for Endometrial Cancer deTection (DETECT) Study is to determine the diagnostic test accuracy of urine and vaginal cytology for endometrial cancer detection in women with PMB. Methods and analysis This is a multicentre diagnostic accuracy study of women referred to secondary care with PMB. Eligible women will be asked to provide a self-collected voided urine sample and a vaginal sample collected with a Delphi screener before routine clinical procedures. Pairs of specialist cytologists, blinded to participant cancer status, will assess and classify samples independently, with differences settled by consensus review or involving a third cytologist. Results will be compared with clinical outcomes from standard diagnostic tests. A sample size of 2000 women will have 80% power to establish a sensitivity of vaginal samples for endometrial cancer detection by cytology of ≥85%±7%, assuming 5% endometrial cancer prevalence. The primary objective is to determine the diagnostic accuracy of urogenital samples for endometrial cancer detection by cytology. Secondary objectives include the acceptability of urine and vaginal sampling to women. Ethics and dissemination This study has been approved by the North West–Greater Manchester West Research Ethics Committee (16/NW/0660) and the Health Research Authority. Results will be disseminated through publication in peer-reviewed scientific journals, presentation at conferences and via charity websites. Trial registration number ISRCTN58863784.

1. Did the authors include patients with recurrent episodes of PMB during the study? If included, was each presentation with PMB considered as a new case? 2. Arranging investigations too soon after a negative initial assessment may result in duplication of some results. Have the authors considered the time interval to recommend further investigation, following a negative evaluation of women with PMB? 3. The authors plan to consider atypical endometrial hyperplasia as a positive result on sensitivity analysis. Can the authors explain the rationale for this? As the index test/tests aim to detect endometrial cancer, it may be more appropriate to consider atypical hyperplasia as a negative result.

REVIEWER
Clarke, Megan National Cancer Institute REVIEW RETURNED 16-Apr-2021

GENERAL COMMENTS
Endometrial cancer is the most common gynecologic malignancy diagnosed in the U.K., and like in many other countries, incidence rates of endometrial cancer in the U.K. have been on the rise. The majority of women diagnosed with endometrial cancer present with postmenopausal bleeding (PMB); however, PMB has many benign etiologies, and only 5-10% of women with PMB will be diagnosed with endometrial cancer. Current diagnostic approaches, including transvaginal ultrasound and endometrial biopsy, are costly and invasive. Moreover, 10-30% of endometrial biopsies yield insufficient or inadequate samples for diagnosis. The identification of minimally invasive, accurate testing approaches to rule out endometrial cancer among women with PMB would be of great clinical benefit. The protocol for the DEveloping Tests for Endometrial Cancer detection (DETECT) study aims to evaluate the diagnostic accuracy of urine and vaginal samples for endometrial cancer detection by cytology in women with PMB. The study will recruit approximately 2,000 women over a three-year period, with follow-up extending an additional year. I am highly supportive of this research but have some questions that if addressed would help clarify the protocol as written.
Major points: 1. The greatest challenge I have with the protocol is understanding the clinical implications of the potential findings and how the investigators envision incorporating such a test into clinical practice. I realize that recommendations would be based on the performance of these tests, but I think it would still be good to consider some scenarios up front. a. For example, as described in the intro, this test is being proposed as a triage test for PMB that would safely reassure women testing negative. In the primary analysis, atypical hyperplasia is considered negativedoes that mean women with PMB testing negative but with atypical hyperplasia would not undergo additional testing and be safely reassured? Would you recommend surveillance with repeat testing at a follow-up interval? 2. Hormone therapy use can lead to bleeding or spotting in postmenopausal women, particularly within the first 6 months of use. Do the investigators plan to account for PMB that might be secondary to HRT use, and therefore not necessarily associated with underlying malignancy? 3. The inclusion/exclusion criteria do not mention anything about previous endometrial biopsy or other prior diagnostic workup procedures. Do you anticipate having patients enrolled who may have been previously evaluated for PMB? Along those lines, will the investigators distinguish between women presenting with initial PMB versus recurrent given that recurrent PMB is associated with higher risk? 4. Will the investigators obtain data on stage at diagnosis? It would be interesting to know if both early and late stage tumors shed malignant cells into the lower genital tract at similar rates/amounts. Also, this will have important implications for determining whether this testing strategy could have clinical benefit in terms of early detection. 5. How will clinical follow-up of negative women be carried out? Will this be active or passive? For how long?
Minor points: 1. Can the authors provide more detail regarding the rationale for the second urine collection? 2. Why aren't the investigators using the endometrial intraepithelial neoplasia diagnostic schema which is preferred to the WHO94? 3. Can the authors clarify the purpose of the multivariate model and what these associations would mean in terms of clinical practice? It seems like stratified analyses of predictive values by salient risk factors would be the preferred approach. 4. What is meant by "random endometrial biopsy" (page 12)? 5. It will be very important to calculate the complement of the negative predictive value corresponding to the risk of cancer in those testing negative. 6. The classification of endometrial cancers diagnosed within 3 months of discharge as false negative seems like a short time window. Do the authors think cancers diagnosed within 6-12 months e.g. would have not been prevalent at the time of evaluation?

VERSION 1 -AUTHOR RESPONSE
Reviewer: 1 Dr. Juan Alcázar, Clinica Universidad de Navarra Comments to the Author: Interesting and well-designed study that addresses a relevant issue commonly found in clinical practice Thank you for your kind words.
I have only minor comments 1. Why did you choose 5% prevalence? Endometrial cancer in PBM is a little bit higher We chose 5% based on an audit of postmenopausal bleeding at our centre.

Do you consider exclude patients with previous history of cervical or ovarian cancer?
The aim is to have a representative sample of women referred with unexplained postmenopausal bleeding. We will exclude women with a history of hysterectomy. Most women with a history of ovarian or cervical cancer will have had a hysterectomy. Those who have not are unlikely to be referred to a diagnostic service with 'unexplained postmenopausal bleeding', but rather undergo expedited follow up by their oncologist in the event of symptoms.
3. The 5th exclusion criterion is quite "open". Please be more specific This is a requirement of our ethics committee and it is a 'catch-all' to ensure patient safety remains paramount throughout conduct of the study.
4. Which actions are planned to overcome problems with patients recruitment due to COVID-19 pandemic?
The In the pilot study, we did not have many unsatisfactory results (<5%) and anticipate similarly small proportions in this larger study.
6. Vaginal sampling is taken by the patient herself? If so, sampling quality could affect data obtained?
In this study, vaginal sampling is carried out by a research practitioner immediately prior to standard diagnostics. This ensures sampling is carried out to a strict protocol and ensures reliability of the results. If the DETECT study shows clinical utility of this tool, the potential to expand its use to homebased vaginal self-sampling is an exciting prospect worthy of further study.
Reviewer: 2 Dr. Nikolaos Burbos, Norfolk and Norwich University Hospital Comments to the Author: The authors present a well-written study protocol evaluating new diagnostic strategies for women with postmenopausal vaginal bleeding (PMB). The protocol is based on the results of a pilot study conducted by the same group (1). The study addresses an important clinical question and has the potential to improve the diagnostic pathways for women with PMB. The study population is clearly defined. The methods for the index and reference tests are described well. Patient and public involvement has also taken place. Statistical analysis is appropriate.
Thank you for these generous comments.
The authors may wish to clarify the following points: 1. Did the authors include patients with recurrent episodes of PMB during the study? If included, was each presentation with PMB considered as a new case?
We will include a representative sample of women referred for investigation of postmenopausal bleeding in this study. This may include women with recurrent episodes of postmenopausal bleeding. However, each participant will only be recruited once to this study.
2. Arranging investigations too soon after a negative initial assessment may result in duplication of some results. Have the authors considered the time interval to recommend further investigation, following a negative evaluation of women with PMB?
We will only recruit women with recurrent symptoms once to the study. In practice, it is unusual for women to be re-referred as a new case of unexplained PMB <3 months after the last episode.
3. The authors plan to consider atypical endometrial hyperplasia as a positive result on sensitivity analysis. Can the authors explain the rationale for this? As the index test/tests aim to detect endometrial cancer, it may be more appropriate to consider atypical hyperplasia as a negative result. Endometrial cancer is the most common gynecologic malignancy diagnosed in the U.K., and like in many other countries, incidence rates of endometrial cancer in the U.K. have been on the rise. The majority of women diagnosed with endometrial cancer present with postmenopausal bleeding (PMB); however, PMB has many benign etiologies, and only 5-10% of women with PMB will be diagnosed with endometrial cancer. Current diagnostic approaches, including transvaginal ultrasound and endometrial biopsy, are costly and invasive. Moreover, 10-30% of endometrial biopsies yield insufficient or inadequate samples for diagnosis. The identification of minimally invasive, accurate testing approaches to rule out endometrial cancer among women with PMB would be of great clinical benefit. The protocol for the DEveloping Tests for Endometrial Cancer detection (DETECT) study aims to evaluate the diagnostic accuracy of urine and vaginal samples for endometrial cancer detection by cytology in women with PMB. The study will recruit approximately 2,000 women over a three-year period, with follow-up extending an additional year. I am highly supportive of this research but have some questions that if addressed would help clarify the protocol as written.
Thank you for your support.
Major points: 1. The greatest challenge I have with the protocol is understanding the clinical implications of the potential findings and how the investigators envision incorporating such a test into clinical practice. I realize that recommendations would be based on the performance of these tests, but I think it would still be good to consider some scenarios up front.
a. For example, as described in the intro, this test is being proposed as a triage test for PMB that would safely reassure women testing negative. In the primary analysis, atypical hyperplasia is considered negativedoes that mean women with PMB testing negative but with atypical hyperplasia would not undergo additional testing and be safely reassured? Would you recommend surveillance with repeat testing at a follow-up interval? 2. Hormone therapy use can lead to bleeding or spotting in postmenopausal women, particularly within the first 6 months of use. Do the investigators plan to account for PMB that might be secondary to HRT use, and therefore not necessarily associated with underlying malignancy?
Thank you for raising this point. We think it's really important that the test can distinguish unscheduled bleeding on HRT from bleeding secondary to sinister underlying pathology. A high proportion of our referrals for urgent investigation of postmenopausal bleeding are HRT-related, and it is therefore important to include a representative sample of all comers, including those taking HRT, to ensure the tool is fit for purpose in its intended population.
3. The inclusion/exclusion criteria do not mention anything about previous endometrial biopsy or other prior diagnostic workup procedures. Do you anticipate having patients enrolled who may have been previously evaluated for PMB? Along those lines, will the investigators distinguish between women presenting with initial PMB versus recurrent given that recurrent PMB is associated with higher risk?
We will recruit all comers with postmenopausal bleeding in order to establish the utility of the tool in a standard referral population. This may include women who have previously been investigated for PMB, however, they will only be recruited once to this study. We will record if the participant has recurrent PMB, but we will treat the current episode of PMB as the index episode against which we will judge the accuracy of the tool. 4. What is meant by "random endometrial biopsy" (page 12)?
We mean a sample taken blindly from the uterine cavity using a pipelle endometrial sampler, as opposed to one taken under direct vision at hysteroscopy.
5. It will be very important to calculate the complement of the negative predictive value corresponding to the risk of cancer in those testing negative.
Yes, we agree. This will be a very important metric to consider, when assessing the value of urogenital cytology for endometrial cancer detection.
6. The classification of endometrial cancers diagnosed within 3 months of discharge as false negative seems like a short time window. Do the authors think cancers diagnosed within 6-12 months e.g. would have not been prevalent at the time of evaluation?