Endometrial Cancer Detection by DNA Methylation Analysis in Cervical Papanicolaou Brush Samples

Abstract

Background: Endometrial cancer (EC) is the leading gynecological cancer worldwide, yet current EC screening approaches are not satisfying. The purpose of this retrospective study was to evaluate the feasibility and capability of DNA methylation analysis in cervical Papanicolaou (Pap) brush samples for EC detection. Methods: We used quantitative methylation-sensitive PCR (qMS-PCR) to determine the methylation status of candidate genes in EC tissue samples, as well as cervical Pap brushes. The ability of RASSF1A and HIST1H4F to serve as diagnostic markers for EC was then examined in cervical Pap brush samples from women with endometrial lesions of varying degrees of severity. Results: Methylated RASSF1A and HIST1H4F were found in EC tissues. Further, methylation of the two genes was also observed in cervical Pap smear samples from EC patients. Methylation levels of RASSF1A and HIST1H4F increased as endometrial lesions progressed, and cervical Pap brush samples from women affected by EC exhibited significantly higher levels of methylated RASSF1A and HIST1H4F compared to noncancerous controls (P < .001). Receiver operating characteristic (ROC) curves and area under the curve (AUC) analyses revealed RASSF1A and HIST1H4F methylation with a combined AUC of 0.938 and 0.951 for EC/pre-EC detection in cervical Pap brush samples, respectively. Conclusion: These findings demonstrate that DNA methylation analysis in cervical Pap brush samples may be helpful for EC detection, broadening the scope of the commonly used cytological screening. Our proof-of-concept study provides new insights into the field of clinical EC diagnosis.

Introduction

Endometrial cancer (EC) is the most prevalent female genital tract malignancy in developed countries. ¹ In 2017, EC's incidence and mortality were 61 380 and 10 920 (new cases) in the United States, respectively. ² In China, EC affected 81 964 women and led to 16 607 deaths in 2020. ³ For localized/early stage EC patients, the 5-year overall survival is almost 95% for stage I disease, whereas it decreases to 20% in stage IV. ⁴ Hence, effective early detection and diagnosis of EC is essential for EC clinical management. Current methods for early EC detection/screening include endometrial biopsy, transvaginal ultrasound, pelvic magnetic resonance imaging (MRI) scan, hysteroscopy, and endometrial cytology. Among them, an endometrial biopsy for histopathological assessment is the gold standard to confirm EC diagnosis. However, due to inadequate sampling, it still has a 10% false negative rate. Sometimes, patients should endure invasive sampling repeatedly, which is unpleasant and stressful. Although hysteroscopic biopsy can improve the positive rates, invasive workups may potentially lead to the spread of cancer cells. Transvaginal ultrasound by measuring endometrium thickness is commonly used for EC screening in postmenopausal women without clinical symptoms. Unfortunately, poor specificity limits its solo use in a routine clinical manner, because serous EC can occur in patients with atrophic endometrium. ⁵ For pelvic MRI scan, it is costly and normally used to assess neoplasm invasiveness region/range. Endometrial cytology is a commonly used method in medical care with the aim of EC diagnostic screening among high-risk populations.

Endometrial cytology consists of a transvaginal or transuterine collection of endometrial exfoliative cells. At present, the former is commonly used in clinical care to obtain endometrial “sloughed off” cells just from the cervix uteri to check the presence or absence of cancer cells. As a result, the small number of epithelial cells shedding from endometrium results in low positivity rates and sensitivity. Alternatively, the collection of endometrial cells from the uterine cavity has been used to screen postmenopausal EC patients and follow-up or monitor asymptomatic women with a high risk of developing EC in some countries. Nevertheless, cytological evaluation of smears derived from endometrial cells is quite tricky. The lack of general consent for diagnosing EC limits its comprehensive utilization in clinical medicine. ⁶ In fact, current methods of EC screening are inaccurate and unsatisfactory, even in symptomatic patients. Therefore, the development of an effective, noninvasive, or minimally invasive method for EC detection or triage of high-risk patients is a high priority.

Aberrant DNA methylation is an early event during tumorigenesis and is sufficiently stable for molecular testing. In addition to chromosomal instability by global hypomethylation, DNA hypermethylation in the promoter region of tumor-related genes usually causes transcriptional silencing, which subsequently affects a series of downstream signaling events and finally results in carcinogenesis. ⁷ Accumulating evidence shows that methylation-based early cancer detection and patient stratification are becoming a hotspot in the field of cancer diagnostics. For example, detecting plasma methylated Septin 9 has been applied to screening of early colorectal cancer. ⁸ To overcome the shortcoming of cervical cytology, methylation analysis of FAM19A4 and mir124 has been used to predict cervical cancer, ⁹ indicating its promising prospect for cancer diagnostic screening. Besides, methylation analysis of target genes in urine provides a completely noninvasive method for detecting EC. ¹⁰ Given the invasive procedures of endometrial scratching, someone tried the minimally invasive cervical scrapings-based EC diagnosis because tumor cells can be shed into the lower genital tract. For example, Huang et al reported the identification of EC by detecting methylated BHLHE22, CDO1, and CELF4 in cervical scrapings with high sensitivity (83.7%-96.0%) and specificity (78.7%-96.0%). ¹¹ Similar studies furthermore demonstrated that cervical scrapings, instead of endometrium and blood, are a good source of DNA for molecular testing, and DNA methylated genes as biomarkers for EC detection from cervical scrapings is feasible and appealing.^11–13 However, relevant research is relatively uncommon.

In previous studies and our work,^14–20 hypermethylation of RASSF1A, HIST1H4F, Septin 9, and PAX1 has been found in tissue samples from patients with EC and other malignancies including colorectal cancer, and cervical cancer. However, the feasibility of DNA methylation testing in cervical Pap brush samples and its value for EC detection remains largely undetermined. In the present study, we checked the methylation status of a panel of methylation biomarkers in EC tissues and cervical Pap brush samples from subjects with various endometrial lesions along with normal conditions. Our data show that using qMS-PCR, RASSF1A, and HIST1H4F methylation analysis in cervical Pap brush samples may readily distinguish EC from non-EC control. The detection of female genital tract malignancies and risk-stratified cancer screening by combining cervical Pap brushes with DNA methylation markers is an attractive concept.

Materials and Methods

Sample Collection

Thirty-two tissue samples with histopathologically proven EC were used to screen hypermethylated genes in EC, which include fresh endometrial tissues from 19 EC patients with total hysterectomy at the department of gynecology of the study hospital as well as formalin-fixed paraffin-embedded (FFPE) specimens from 13 EC women. Type I EC comprises endometrioid adenocarcinomas and type II EC consists of serous-type, clear-cell, and poorly differentiated carcinomas. Cervical Pap brush samples were collected from 94 subjects with EC (n = 19), atypical hyperplasia (AH) of the endometrium (n = 21), normal endometrium (NE) with cervical intraepithelial neoplasia (CIN) 1 to 2 (NE + CIN 1-2, n = 20), and NE (n = 34) based on histopathology. All cervical samples were initially collected for cervical screening according to the guidelines ²¹ and the manufacturer's instructions from June 2021 to March 2022. The reporting of this retrospective study conforms to MDAR (Materials Design Analysis Reporting) guidelines. ²² This work was approved by the Institutional Ethics Board of the study hospital (Nos. 202001142 and 202301304). Written (universal/general) informed consent was obtained from the study participants and all subjects agreed to the probable use of their biological samples for medical research. Patients were selected consecutively in this study and patient details have been de-identified. An overview of patient characteristics is given in Table 1.

Table 1.

Clinical Characteristics of Patients With Cervical Brushes.

	EC (n = 19)		AH (n = 21)	Normal endometrium with CIN 1–2 (n = 20)	Normal endometrium (n = 34)
	Type I, n = 7	Type II, n = 12
Age, mean ± SD	42.50 ± 4.21	59.03 ± 6.04	48.00 ± 5.03	49.70 ± 6.98	48.50 ± 7.01
BMI, mean ± SD	22.42 ± 1.90	23.42 ± 2.60	25.31 ± 2.44	24.47 ± 2.55	24.93 ± 2.99
Hypertension, n (%)	2 (28.6%)	5 (41.7%)	9 (42.9%)	9 (45%)	17 (50%)
Diabetes, n (%)	2 (28.6%)	4 (33.3%)	5 (23.8%)	3 (15%)	6 (17.6%)
FIGO stage, n
I	5	8
II	1	3
III	1	1
FIGO 2009 Grade, n
G1	3	10
G2	3	1
G3	1	1
CA125 (0-35 U/mL)	41.99 ± 21.16	38.99 ± 16.08	19.53 ± 11.52	ND	ND

Abbreviations: ND: not detected; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics; EC, endometrial cancer; AH, atypical hyperplasia; CIN, cervical intraepithelial neoplasia.

DNA Extraction and Bisulfite Conversion

Genomic DNA was extracted from endometrial tissues and cervical Pap brush samples using the TIANamp Genomic DNA Extraction Kit (Tiangen, Beijing) according to the recommendations of the manufacturer. Following isolation, DNA was eluted in 30 μL elution buffer and stored at −20 °C until analysis.

Isolated DNA was modified with bisulfite using the DNA Methylation Kit (Guangdong Bright-Innovation BioMed Co., Ltd, Shunde) following the manufacturer's protocol.

Quantitative Methylation-Sensitive PCR (qMS-PCR)

Bisulfite-converted DNA was subjected to qMS-PCR testing for methylated RASSF1A, HIST1H4F, Septin 9, and PAX1. In brief, the PCR cycling conditions shared for four genes were as follows: 1 cycle at 50 °C for 5 min, followed by 1 cycle at 94 °C for 20 min, and 50 cycles at 93 °C for 30 s, 56 °C for 60 s, and 65 °C for 30 s. The housekeeping gene beta-actin (ACTB) was used as an internal reference for quantification and quality assessment. For the qMS-PCR testing, ≥ 1 out of 2 or 2 out of 3 replicates, where appropriate, were considered positive if a Ct value was < 40 as well as if the signal intensity of targets was greater than that of ACTB. Differences in methylation levels of the candidate genes among each group (ie, EC, AH, NE + CIN 1-2, and NE) were determined by methylation scores using the formula: 2^{[Ct (ACTB)−Ct (target)]}× 100, as described previously. ²³ All analyses were conducted on an ABI 7500 Real-Time PCR system (Applied Biosystems) and tests for each sample were conducted in duplicate or triplicate. The relevant primer and probe sequences for qMS-PCR are summarized in Table 2. All the target regions of methylation are promoters.

Table 2.

Primers and Probes Used for qMS-PCR.

Primer/probe	Sequence (5′–3′)
RASSF1A-F	GCGTTGAAGTCGGGGTTC
RASSF1A-R	CCCGTACTTCGCTAACTTTAAACG
RASSF1A-P	FAM-AAACGCGAACCGAACGAAACC-BHQ1
Septin 9-F	TTTAGTTAGCGCGTAGGGTTC
Septin 9-R	AACTAATAAACAACGAATCGCG
Septin 9-P	FAM-ACGCCCCCGACGAAACC-BHQ1
HIST1H4F-F	AGGAATATTTTAAGAATATCGC
HIST1H4F-R	CCATCGCAAAATACTACG
HIST1H4F-P	FAM- GGTTAAGCGACGGATGG-BHQ1
PAX1-F	GGCGGTAGGTTTTGGAGC
PAX1-R	CGCAATCCGAAAAAACTTAAACG
PAX1-P	FAM-TACGCGGCGGCGGCGG-BHQ1
ACTB-F	ATAATAAAAAGGAGGTTGGAT
ACTB-R	CTCCCRCAAAACAACCAC
ACTB-P	VIC-CCACCTTACCCTAAACACTACAAC-BHQ1

Abbreviations: ACTB, beta-actin; qMS-PCR, quantitative methylation-sensitive polymerase chain reaction.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism 5.0. To compare the levels of methylation markers among all groups, t-tests, analysis of variance (ANOVA), and post hoc tests for multiple comparisons were used, as appropriate. All significant differences were assessed using P < .05 (two-sided).

Results

DNA Methylation Detection in Endometrial Cancer Tissues

We used qMS-PCR to investigate the methylation status of a panel of molecular markers including RASSF1A, HIST1H4F, Septin 9, and PAX1 in 19 resected endometrial tissues from EC-affected women. To enlarge the sample size, we also collected an additional 13 FFPEs from EC patients within 3 years of hysterectomy in this study. Similar trends were seen in both sample types. Our qMS-PCR tests showed that methylated RASSF1A and HIST1H4F were detectable in EC. Conversely, Septin 9 and PAX1 are barely amplified in all EC specimens. Further, we compared the methylation levels of the four genes in EC and found that RASSF1A and HIST1H4F had considerably higher methylation levels compared to Septin 9 and PAX1 (Figure 1, P < .001).

Figure 1.

Scatter plot of RASSF1A, HIST1H4F, Septin 9, and PAX1 methylation status in EC tissues. Methylation levels were determined by qMS-PCR. Kruskal-Wallis statistics and Dunn's multiple comparison tests were conducted.

Abbreviations: EC, endometrial cancer; qMS-PCR, quantitative methylation-sensitive polymerase chain reaction.

Identification of RASSF1A and HIST1H4F Methylation in Cervical Pap Brush Samples

Since highly methylated RASSF1A and HIST1H4F were identified in EC tissues, we wonder if the molecular signatures can be detected in cervical Pap smear samples. That would greatly facilitate the identification of EC because of its minimal invasiveness and convenience. To this end, we collected cervical Pap brush samples from 19 patients with EC and 34 healthy individuals to analyze the methylation status of RASSF1A and HIST1H4F using qMS-PCR. As expected, RASSF1A and HIST1H4F methylation were detectable in all EC cervical Pap brush samples. Moreover, RASSF1A methylation was found in 7 of 34 healthy women's cervical Pap brush samples, whereas only one healthy woman had positive testing for HIST1H4F in cervical Pap brush samples, indicating superior performance of HIST1H4F compared to RASSF1A.

Methylation of RASSF1A and HIST1H4F in Cervical Pap Brush Samples From Subjects With Varying Degrees of Endometrial Lesions

Subsequently, we assessed the viability of cervical Pap brush sample-based methylation analysis of RASSF1A and HIST1H4F in detecting EC. We used qMS-PCR to measure the methylation levels of RASSF1A and HIST1H4F in cervical Pap brush samples from 94 female participants, including 19 patients with EC, 21 women with AH, 20 subjects with NE but with CIN 1 to 2, and 34 healthy individuals. Overall, methylation levels of RASSF1A and HIST1H4F increased as endometrial lesions progressed (Figure 2). In particular, cervical Pap brush materials from women with EC had the highest level of methylated RASSF1A and HIST1H4F among the groups (P < .001, Kruskal-Wallis test), which could be potential markers to monitor EC.

Figure 2.

Comparison of RASSF1A and HIST1H4F methylation levels in cervical Papanicolaou (Pap) brushes from participants with various endometrial disorders as well as normal conditions.

RASSF1A and HIST1H4F Methylation Hold the Potential to be Diagnostic Markers for EC

To evaluate the clinical performance of RASSF1A and HIST1H4F methylation for detecting EC, we constructed the receiver operating characteristic (ROC) curves and calculated the area under the curve (AUC). Figure 3 exhibits RASSF1A and HIST1H4F methylation with a combined AUC of 0.938 and 0.951 for EC/pre-EC detection in cervical Pap brush samples, respectively. Our results demonstrated that analysis of methylated RASSF1A and HIST1H4F in cervical Pap brush materials provides a minimally invasive and promising tool for distinguishing EC from noncancerous changes of endometrium.

Figure 3.

AUC-ROC of RASSF1A and HIST1H4F methylation in cervical Pap brush samples for the detection of EC.

Abbreviations: EC, endometrial cancer; Pap, Papanicolaou; AUC-ROC, area under the receiver operating characteristic curve.

Discussion

Liquid-based cytology Pap smears and cervical Pap brushes have been widely used for cervical cancer screening in routine clinical practice, which has revolutionized the management of cervical cancer patients and dramatically reduced the incidence and mortality of cervical cancer in screened populations. In sharp contrast, endometrial and ovarian cancers continue to pose considerable challenges for risk assessment. Access to the cells of origin for these malignant tumors can be achieved with invasive procedures such as endometrial biopsy or a laparoscopy, limiting its potential for population-based screening. ²⁴ Could regular endocervical sampling be used to identify other tumors arising in the internal female genital tract? Compared with the DNA derived from normal cells brushed from the endocervical canal, the relatively small amount of DNA derived from neoplastic cells in the cervical Pap brushes might be a challenge to reliably detect a rare population of genetic or epigenetic abnormalities among a high background of DNA. To address this issue, nucleic acid amplification techniques are helpful.

Previously, a pilot study demonstrated that DNA mutational analyses of cervical scrapings could determine occult EC as well as ovarian cancer. The detection rates for EC and ovarian cancer were 100% (24/24) and 41% (9/22), respectively. ²⁵ With the aid of technical advances, molecular alterations present in a tiny fraction of DNA templates can be reliably detected. When EC occurs, cancer cells and cell fragments can easily detach from the uterus and drain into the endocervical canal due to the anatomical proximity, which may be the most likely mechanism for the above-mentioned analyses. Based on the foundation of the Pap test, we hypothesized that routinely collected DNA from cervical brushes could be used to detect epigenetic changes present in rare cancer cells that accumulate in the cervix once shed from EC. The cytological assessment of neoplastic cells derived from EC in Pap brushes is infrequent. Typically, cervical cytology samples contain only a limited quantity of exfoliative cancer cells and cellular debris. Microscopic examination may not definitively distinguish between benign and malignant cells, as well as EC and cervical cancer. Given these factors, we opted to extract DNA directly from conventional Pap specimens and conduct methylation analyses of molecular markers.

In the present study, we identified RASSF1A and HIST1H4F methylation from four candidates in EC tissues as well as cervical brushes. Importantly, methylated RASSF1A and HIST1H4F had high discriminatory power for EC detection with an AUC of 0.938 and 0.951, respectively, indicating similar performance in this regard. Similarly, previous reports documented 3-gene panel (BHLHE22-CDO1-CELF4/BHLHE22-CDO1-TBX5) or single gene-based (PCDHGB7) DNA methylation for EC detection using cervical scrapings, the ROC for these markers ranged from 0.80 to 0.94, or 0.86, respectively.^11–13 Also, Wang et al ²⁶ reported a relevant study on methylation marker-based EC screening in endometrial cytological samples using Li Brush. Compared with endometrial sampling, cervical Pap brush sampling is a more acceptable screening tool for the general population. Our results demonstrate the potential application of the readily accessible specimens, cervical scrapes, in EC diagnostics, which provides a viable and minimally invasive method for sensitive EC detection or triage of high-risk patients across different clinical features and histologies to supplement current hysteroscopy diagnosis.

Interestingly, a recent study reported that a DNA methylation signature in Müllerian Duct-derived cervical cells has a high sensitivity and specificity for determining women with EC and ovarian cancer, with an AUC of 0.81 and 0.76, respectively. These findings depend on the fact that EC and ovarian cancer arise from Müllerian Duct epithelial cells. Aberrant epigenetic alterations in cervical cells may serve as surrogate markers for ovarian cancer and EC predisposition. ²⁷ Further research into whether DNA methylation markers in a single cervical screening sample can identify four types of women's cancers (breast, ovarian, endometrial, and cervical cancer), dubbed “one test for four cancers,” is promising and warranted.

Compared to the well-known RASSF1A, HIST1H4F has been less reported. Histones are essential components of chromatin and are conserved in eukaryotic cells, which are involved in the cell cycle and replication. Although histone modifications have been widely studied in the epigenetic field, epigenetic changes in histone genes have been poorly investigated. HIST1H4F encodes a member of the histone H4 family. Dong et al first described that HIST1H4F was universally hypermethylated in all 17 tumor types from TCGA datasets and validated in nine different types of human cancer, including EC. ¹⁵ These findings indicate that HIST1H4F may serve as a pan-cancer biomarker, which may improve screening for early cancer diagnosis. However, the functional role of HIST1H4F in EC remains unclear and requires further investigation.

Our study has several limitations. First, the relatively small sample size limits its statistical power, and larger cohorts are needed to validate the findings. Moreover, epigenetic studies have indicated that Type I and Type II EC exhibit distinct methylation patterns.^28,29 Promoter methylation is frequently observed in Type I EC, but is less common in Type II. In contrast, both Type I and II EC could be detected using a BHLHE22/CDO1-based methylation panel, suggesting that they may share a common epigenetic profile. ¹³ Regarding the classification of EC and methylation frequency of target genes, it is hard to draw definite conclusions between Type I (n = 7) and Type II (n = 12) EC in the present study. Second, we did not standardize the timing of cervical Pap brush sample collection, and further standardization of sample collection and optimization of DNA methylation testing is warranted. In addition, the most problematic clinical situation for cervical Pap smear is cervical stenosis in elderly patients and Type 3 transformation zone. Of note, although cervical Pap brush-based EC detection has some advantages over conventional methods, it is possible to miss diagnosis or misdiagnosis of EC because of its far-site sampling. The definite diagnosis of EC relies on histological examination of a given endometrial tissue sample. ³⁰ Although improvements need to be made before its wide application in clinical practice, it represents a promising step towards EC screening in high-risk asymptomatic or presymptomatic populations.

Conclusion

In summary, we demonstrated that DNA methylation-based EC detection from easy-to-access cervical Pap brush samples is feasible. It may have the potential for large-scale application.

Abbreviations

ACTB

beta-actin

AH

atypical hyperplasia

ANOVA

analysis of variance

AUC

area under the curve

CIN

cervical intraepithelial neoplasia

EC

endometrial cancer

FFPE

formalin-fixed paraffin-embedded

MRI

magnetic resonance imaging

NE

normal endometrium

Pap

Papanicolaou

qMS-PCR

quantitative methylation-sensitive PCR

ROC

receiver operating characteristic.