2023-11-10 16:21:49
Artificial intelligence can predict the success of treatments in patients with ovarian cancer. The AI-based tool is called ‘Iron’, the results of which are described in a study published in ‘Nature Communications’, which also involves experts from the Catholic University, Rome campus and Policlinico A. Gemelli Irccs. It proved capable of predicting the response to treatment in 80% of patients with ovarian tumors, understood as the volumetric reduction of tumor lesions, with an accuracy of 80%, which is far superior to that of the methods currently used in the clinical setting. . Ovarian cancer affects more than 5 thousand women in Italy every year, in addition to the over 30 thousand undergoing treatment. Since it does not give specific symptoms in the early stages, its diagnosis often occurs when the disease is already at a rather advanced stage.
The tool analyzes various clinical characteristics of the patient, from tumor DNA circulating in the blood (liquid biopsy) to general characteristics (age, state of health, etc.) to tumor markers and images of the disease acquired with CT scan, and on the basis of they express a prediction on the chances of success of the therapy. The study, carried out on 134 patients with high-grade ovarian cancer, coordinated by Evis Sala, Professor of Diagnostic Imaging and Radiotherapy at the Catholic University and director of the Advanced Gemini Radiology Center of Rome, and carried out by the University of Cambridge.
High-grade serous ovarian cancer is one of the most aggressive forms and represents approximately 70-80% of ovarian tumors: it often presents resistance to chemotherapy drugs, but to date the response to therapies can be predicted with a maximum accuracy of 50%. Furthermore, for this form of the tumor, very few clinically usable biomarkers are known due to the high degree of heterogeneity of the disease, which varies greatly from patient to patient. Hence the idea of developing a tool based on artificial intelligence capable of predicting with high accuracy the patients who will respond to chemotherapy was born.
“We put together two independent datasets with a total of 134 patients – explain Evis Sala and Mireia Crispin Ortuzar of Cambridge -. For all patients at the beginning (before treatment), we obtained clinical data, including demographic data and details of their treatment, as well as biomarkers present in the blood such as Ca-125 and circulating tumor DNA (ctDNA), as well as quantitative tumor characteristics deduced from CT images of all primary and metastatic tumor sites”.
“From a clinical perspective, the proposed framework addresses the unmet need to early identify patients who are likely to not respond to neoadjuvant therapy and may be referred for immediate surgery,” Sala emphasizes. “The tool could be applied to stratify the risk of the individual patient also in future clinical research that we are carrying out at the Gemelli Polyclinic with the collaboration of the group of Professor Giovanni Scambia, Professor of Gynecology and Obstetrics and scientific director of the Irccs Polyclinic Gemelli Foundation”.
#Ovarian #cancer #artificial #intelligence #predicts #therapy #success