Jesus Gonzalez Bosquet, MD, PhD, highlights test for earlier ovarian cancer detection

In a recent interview with Contemporary OB/GYN, Jesus Gonzalez Bosquet, MD, PhD associate professor of obstetrics and gynecology at The University of Iowa, highlighted the efficacy of a machine learning model using blood-based biomarkers to identify ovarian cancer.

Ovarian cancer remains one of the most difficult gynecologic cancers to detect early, and this significantly impacts patient survival outcomes. According to Gonzalez Bosquet, the primary challenge is the lack of effective early screening or detection methods. Current tools, such as ultrasound and tumor markers, have been tested in multiple studies both in the United States and abroad, but none have demonstrated consistent accuracy.

As a result, most patients are diagnosed at advanced stages, where treatment options are limited and survival rates drop significantly. Gonzalez Bosquet noted that 5-year survival rates for early-stage ovarian cancer reach approximately 95%, but this falls between 40% and 50% for patients diagnosed at later stages, underscoring the need for earlier detection strategies.

To address this gap, Gonzalez Bosquet and his team have explored the use of machine learning models to identify ovarian cancer through blood-based biomarkers. Their work focuses on detecting methylated DNA fragments shed by tumors into the bloodstream. Starting with a platform that examined over 850,000 methylation probes, the researchers used machine learning to narrow the list down to 9 probes that showed strong potential for accurately identifying ovarian cancer. While promising, these findings are still in the pilot phase and require further validation before they can be used in clinical settings.

According to Gonzalez Gosquet, the next step in this research is to confirm whether these methylated probes can be reliably detected in blood samples, not just in surgical tissue specimens. He that his team, in collaboration with Mayo Clinic, has submitted a project aimed at this validation step. If successful, the model will then need to be tested in large-scale screening studies involving both patients with and without ovarian cancer to determine its true diagnostic accuracy and clinical value.

Looking ahead, Gonzalez Bosquet envisions practical applications of these AI models that would make them accessible to clinicians worldwide. Once fully developed, the models could be hosted on web-based platforms, allowing physicians to upload data such as imaging scans or blood test results. The system would then provide a risk assessment, ranging from low to high probability of ovarian cancer. This streamlined, user-friendly approach could integrate seamlessly into clinical workflows, offering physicians a powerful tool for earlier and more accurate cancer detection.

Ultimately, Gonzalez Bosquet emphasized that while the technology is still under development, AI-driven models hold great promise in transforming ovarian cancer diagnosis, potentially improving survival rates by identifying the disease before it reaches advanced stages.

No relevant disclosures.

Reference

Gonzalez Bosquet J, Wagner VM, Russo D, et al. Identifying ovarian cancer with machine learning DNA methylation pattern analysis. Scientific Reports. 2025;15. doi:10.1038/s41598-025-05460-9