Fig. 2

Marker discovery in The Cancer Genome Atlas-Cervical Squamous Cell Carcinoma and Endocervical Adenocarcinoma (TCGA-CESC) and Uterine Corpus Endometrial Carcinoma (TCGA-UCEC) databases. A Principal component analysis of 485,000 probes shows a clear visual separation of cervical cancer (orange/red) and normal (green/pink) tissue samples. B, C Histogram plots of cumulative β-methylation in the indicated numbers of carcinomas and normal samples are shown (B) for 14 markers and, C for the final 5 markers in two histological subtypes of cervical carcinoma squamous cell carcinoma (SCC, N = 254) and adenocarcinoma (AC, N = 53). Also, in (C), Mann–Whitney box plots show cumulative β-methylation in SCC and AC samples for the five-marker panel indicating a lower, but not statistically significant difference in methylation (P = 0.139) between the two histological subtypes. In the next panel, results of receiver operator curve area under the curve (ROC AUC) analysis are shown. The sensitivity and specificity were based on the 95th percentile of cumulative β-methylation in normal samples (dotted line, histogram). Also tabulated in (C) is the average array beta methylation for each gene for TCGA CESC tumor (N = 307), CESC normal cervix (N = 3) and UCEC normal uterus (N = 45). Figure S1 contains additional details of the marker selection process. Tables S2 and S3 provide additional probe information. Abbreviations: SCC, squamous cell carcinoma; AC, adenocarcinoma