Abstract 603P
Background
Early detection of gynecological malignancies can significantly improve prognosis. Liquid biopsy has shown great potential in cancer early detection. The PERformance of multi-Cancer Early-detectIon based on Various biomarkers in fEmale cancers study (PERCEIVE-I, NCT04903665) is a prospective case-control study for gynecological malignancies including ovarian, cervical, uterine, vaginal and vulvar cancers.
Methods
Blood samples were obtained prospectively from Fudan University Shanghai Cancer Center. Comprehensive assays were performed with a cfDNA methylation panel of ∼490,000 CpG sites, a mutation panel of 168 genes, and 8 protein biomarkers. Cancer cases stratified by age and clinical status were randomly split into training and test datasets at a 1:1 ratio. Age-matched non-cancer controls were selected. Finally, 298 cases and 667 controls were analyzed. Cancer detection models were developed and validated.
Results
In the test set, the specificities of the methylation, proteins, and mutation models were 99.0% (95% CI, 96.3‒99.9%), 99.4% (98.0‒99.9%) and 100% (97.8‒100%), respectively. The sensitivities were 72.4% (64.0‒79.8%), 56.8% (47.9‒65.4%) and 46.3% (38.0‒54.7%), respectively. The combination of methylation and proteins was the most efficient, with a 81.8% (74.2‒88.0%) sensitivity at a 98.5% (95.6‒99.7%) specificity as the combination of all three assays could only detect 3 more cases. The accuracy of predicted origin was 82.9% (74.3‒89.5%). Table: 603P
| The test dataset | Methylation model | Protein model | Combined model |
| Specificity (95%CI) | 99.0% (96.3‒99.9%) | 99.4% (98.0‒99.9%) | 98.5% (95.6‒99.7%) |
| Sensitivity (95%CI) | |||
| Total | 72.4% (64.0‒79.8%) | 56.8% (47.9‒65.4%) | 81.8% (74.2‒88.0%) |
| Stage I | 46.0% (31.8‒60.7%) | 26.5% (15.0‒41.1%) | 59.2% (44.2‒73.0%) |
| Stage II | 79.2% (57.9‒92.9%) | 39.1% (19.7‒61.5%) | 82.6% (61.2‒95.1%) |
| Stage III | 92.3% (79.1‒98.4%) | 87.2% (72.6‒95.7%) | 100% (91.0‒100%) |
| Stage IV | 90.0% (68.3‒98.8%) | 95.0% (75.1‒99.9%) | 100% (83.2‒100%) |
Conclusions
In this study, the methylation model was superior to mutation and proteins in identifying gynecological malignancies, especially in the early stages. The combined model of cfDNA methylation and proteins achieved a higher sensitivity. This study brings a promising approach to the early detection of gynecological malignancies. Large-scale validation studies will be conducted in the future.
Clinical trial identification
NCT04903665.
Editorial acknowledgement
Legal entity responsible for the study
Fudan University Shanghai Cancer Center.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.