KEY TAKEAWAYS
- This study aimed to identify candidate biomarkers using ML and AI for early EC diagnosis and identifying pts at low or high risk for cancer progression and recurrence.
- The study has concluded its initial phase, concentrating on creating and validating diagnostic and prognostic models for early EC diagnosis and patient stratification based on cancer progression and recurrence risk.
This prospective observational case-control study was conducted across six medical centers in Europe. Plasma samples from women diagnosed with endometrial carcinoma (EC) and controls were subjected to non-targeted/targeted metabolomic and semi-quantitative immune-based proteomic analyses. These analyses, along with clinical and epidemiological data, were processed using advanced artificial intelligence (AI) and machine learning (ML) techniques to create new models for early EC diagnosis and identify patients (pts) at low or high risk for cancer progression and recurrence.
BioEndoCar enrolled over 440 pts and maintained stringent procedures for sample collection, processing, and storage. The AI/ML-based diagnostic and prognostic models, utilizing all available data, displayed promising characteristics, with a repeated k-fold cross-validation AUC > 0.8. Further validation will be performed using statistical (AI/ML) approaches to determine which subset of proteomic and metabolomic data can serve as diagnostic and prognostic biomarkers in endometrial cancer.
The BioEndoCar study has completed its initial phase, focusing on the identification and validation of diagnostic and prognostic models for early EC diagnosis and the stratification of pts based on cancer progression and recurrence risk. If validated, these models, incorporating selected proteomic and metabolomic data, have the potential to form the basis for valuable non-invasive tools for diagnosing and prognosticating EC.
Source: https://www.emma.events/site/programme/?sessiondetail=4534565&trackid=0&a=esgo2023#!
Clinical Trial: https://classic.clinicaltrials.gov/ct2/show/NCT03553589
Kokol, M., Romano, A., Werner, E., Smrkolj, S., Roškar, L., Pirš, B., Semczuk, A., Kaminska, A., Adamiak-Godlewska, A., Fishman, D., Vilo, J., Lowy, C., Griesbeck, A., Schroeder, C., Tokarz, J., Adamski, J., Weinberger, V., Bednaríková, M., Vinklerova, P., Ferrero, S., Barra, F., Takac, I., Sobocan, M., Knez, J., Rižner, T.L. BioEndoCar: Identifying Candidate Biomarkers For Diagnosis And Prognosis Of Endometrial Carcinoma Using Machine Learning And Artificial Intelligence.
