JCO Precis Oncol · 2026;10:e250108 · Comparative Study · China

Self- vs Physician-Collected Methylation for EC Detection

Diagnostic Accuracy and Concordance of Self-Collected Cervical CDO1 and CELF4 Methylation Testing Compared With Physician Sampling for Endometrial Cancer Detection
142 women (33 EC + 109 non-EC) · self AUC 0.930 vs physician 0.896 · CDO1 R=0.79, CELF4 R=0.76 · Bland-Altman minimal bias
He P, Wang Y, Xie N, Liu C, Liu P, Wang L, Yin S, Zhou S, Cai H, Fu C
Second Xiangya Hospital, Central South University · JCO Precis Oncol 2026;10:e250108 · doi:10.1200/PO-25-01081
142
paired samples
0.930
AUC · self-collected
0.896
AUC · physician
R 0.79
CDO1 concordance

1Background & Objective

  • Gap: self-sampling proven for cervical cancer; EC self-sampling methylation lacks robust evidence.
  • Objective: compare self- vs physician-collected CDO1/CELF4 methylation for EC detection.

2Study Design & Cohort

33
EC
109
non-EC
2024.3–25.2
enrollment
  • Design: prospective cross-sectional; paired Evalyn Brush self + physician cervical samples.
  • Assay: qPCR CDO1/CELF4 methylation (ΔCt); hysteroscopy histology = gold standard.
  • Agreement: Spearman correlation + Bland-Altman plots.
  • Exclusion: menstruation · pregnancy · hormonal therapy · IUD · recent procedures.
  • Inclusion: adults scheduled for hysteroscopy (AUB / TVS abnormal / clinical concern).
Evalyn
self-sampling brush
qPCR
methylation assay
n=102 est.
sample size
0.930
AUC self
0.896
AUC physician
R 0.79/0.76
concordance
EC vs non-EC ΔCt
P<0.001
  • Pathology spectrum: EC · EH · EAH · EP · EM — broad comparison.
  • Home value: reduces specialist visits; patient-centred care.
  • Sample size: estimated n=102 (AUC 0.85, α 0.05, half-width 0.05); 142 enrolled.
  • Analysis: Wilcoxon · χ² · Spearman · Bland-Altman · ROC/AUC.
  • Sensitivity: robust findings accounting for missing data.
  • Limits: single-center; symptomatic/referred cohort — not population screening.
STROBE-compliant; robust to missing-data sensitivity analysis; pre/postmenopausal subgroups stable; self AUC even slightly higher.

3ROC — Diagnostic Accuracy

ROC curves self vs physician sampling
Fig. ROC — self-sampling AUC 0.930 vs physician 0.896 (equivalent accuracy).
0.930
AUC · self
0.896
AUC · physician

4Concordance & Key Metrics

R 0.79
CDO1 agreement
R 0.76
CELF4 agreement
minimal bias
Bland-Altman
P<0.001
EC vs non-EC ΔCt
Self AUC
0.930
Physician AUC
0.896
  • No systematic bias between sampling methods — home collection is reliable.
  • Stable subgroups: high discrimination pre- & postmenopausal.

5Clinical Value

33
EC cases
109
non-EC
7
pathology groups
  • Patient-centric: at-home self-collection expands access to EC triage.
  • Workflow fit: triage before hysteroscopy — fewer invasive procedures.
  • Broad spectrum: EC · EH · EAH · EP · EM all compared.
  • Next step: multicenter validation in larger populations.
pre / post
menopause subgroups
high AUC
both groups
0.930
self AUC
  • Compliance: self-collection acceptable & convenient for women.
  • Cost: reduces clinic visits & invasive procedure burden.
  • Telehealth: pairs with remote counselling & referral triage.
Equivalent diagnostic accuracy (self 0.930 vs physician 0.896) supports a home-based EC triage pathway; reduces specialist visits & improves patient experience.
Clinical Significance