Combined genetic mutations and DNA-methylated genes as biomarkers for endometrial cancer detection from cervical scrapings

Abstract

Background: Endometrial cancer is a common gynecologic cancer. Noninvasive molecular biomarkers for triage of high-risk patients for invasive procedures are needed. Based on the success of cytological Pap smear screening, cervical scrapings are a good source of DNA for molecular testing. In addition to genetic lesions, DNA methylation is a promising biomarker. We assessed the usefulness of combining genetic and epigenetic biomarkers from cervical scrapings to detect endometrial carcinomas.

Methods: We performed a retrospective case-control study of 96 consecutive cervical scrapings from patients with abnormal uterine bleeding who underwent surgery for diagnostic evaluation. Thirty and 16 cases were diagnosed with type I and type II endometrial cancers, respectively. The remaining non-cancer cases included normal endometrium (n = 12), benign uterine lesions (n = 20), and endometrial hyperplasia (n = 18). Quantitative methylation-specific PCR and mass spectrometry were used for DNA methylation and genetic mutation analysis. Logistic regression was used to evaluate the clinical performance of these candidate biomarkers.

Results: We tested the effectiveness of the methylation status of four genes (BHLHE22, CDO1, TBX5, and HAND2) in endometrial cancer detection. The area under the receiver operating characteristic curves ranged from 0.703 to 0.878, and panels of hypermethylated BHLHE22/CDO1/HAND2 (87.0% sensitivity and 86.0% specificity) and BHLHE22/CDO1/TBX5 (89.1% sensitivity and 80.0% specificity) showed significant differences and could distinguish benign from malignant endometrial lesions. The sensitivity and specificity in endometrial cancer detection for BHLHE22/CDO1 were 84.8% and 88.0%, respectively. Both type I and II endometrial carcinomas could be detected using a BHLHE22/CDO1-based methylation profile, suggesting that they may have common epigenomes. Moreover, PTEN and TP53 mutations were found in 63.3% of type I and 93.6% of type II endometrial cancers. Unexpectedly, PTEN and TP53 mutations were commonly found in cervical scrapings of the normal endometrium (25% and 33.3%, respectively) and in cases with benign uterine lesions (10% and 50%, respectively). Finally, combinations of any one mutation of PTEN and TP53 mutations had a sensitivity of 91.3%, but a specificity of only 42.0%.

Conclusions: Adding PTEN/TP53 mutation testing to BHLHE22/CDO1-based methylation testing did not improve the detection of endometrial cancer.

Keywords: Biomarkers; Cervical scrapings; Endometrial cancer detection; Methylation; Mutation.

Fig. 1

DNA methylation levels for four candidate genes detected by quantitative methylation-specific polymerase chain reaction (qMS-PCR) in 96 cervical scrapings. a DNA methylation levels are displayed as the difference in crossing point (ΔCp) values for each candidate gene. Dot plots indicate the distribution of ΔCp values for BHLHE22, CDO1, HAND2, and TBX5. Horizontal bars in the middle of the scattered dots indicate the average methylation levels. The lower the Cp values, the higher the gene methylation status. P values were calculated using the Kruskal–Wallis test. b Area under the receiver operating characteristic curve (AUC-ROC) for the DNA methylation status of the four candidate genes in cervical scrapings. P values were < 0.001 for all analyses, and ≤ 0.5 for the comparison of AUC-ROCs

Fig. 2

Localization of mutations in PTEN and TP53 gene sequences in the patient cohort. a The distribution and spectrum of PTEN (top) and TP53 (bottom) mutations are shown. The presence of a mutation is shown on the x-axis (lollipop), and the number of cases and their frequency of mutations are shown on the y-axis. Missense mutations are presented in green, truncating (“nonsense”) mutations in black, and in-frame mutations in brown. b The prevalence of PTEN and TP53 mutations in normal endometrium (normal), leiomyoma (Myo), endometrial hyperplasia (EH), atypical endometrial hyperplasia (AEH), type I endometrial cancer (EC), and type II EC are shown

Fig. 3

Comparison of genetic mutations (PTEN and TP53) and aberrantly DNA-methylated genes (BHLHE22, CDO1, HAND2, and TBX5) in cervical scrapings. a The distribution and frequency of PTEN and TP53 gene mutations in non-endometrial cancer (non-EC), type I EC, and type II EC are shown. PTEN mutations were more frequently observed in type I EC. b Combination of DNA methylation and any genetic mutation in EC and non-EC. The absence of PTEN and TP53 mutations and no DNA methylation of any of the four candidate genes were seen in one of the 16 type II EC samples. This sample also revealed unique copy number instability (CNI)