In 80% of cases involving ovarian cancer patients, an artificial intelligence-based model can predict the treatment trajectory by assessing the volumetric reduction of tumor lesions accurately.
IRON (Integrated Radiogenomics for Ovarian Neoadjuvant Therapy) is a sophisticated tool that analyzes various clinical aspects of patients by examining circulating tumor DNA.
Deoxyribonucleic acid (DNA), a crucial protein, consists of two lengthy nucleotide strands coiled in a double helix structure. It carries genetic instructions essential for growth, reproduction, and overall functioning in humans and most other living organisms. While DNA is predominantly located in the cell nucleus, a small amount can be found in mitochondria, known as mitochondrial DNA (mtDNA).
The term “data-gt to translate to attributes=” is utilized in reference to converting DNA data from blood samples (liquid biopsy) into general attributes such as age and health status. CT scans are employed to detect signs of illness and tumors, providing insights into the potential success of therapy based on this analysis.
This breakthrough stems from a recent study involving 134 high-grade ovarian cancer patients, featured in Nature Communications. Professor Evis Sala, the Director of the Advanced Radiology Center at Policlinico Universitario A. Gemelli IRCCS, oversaw the research. The AI model was initially developed by Professor Sala’s team at the University of Cambridge.
Challenges in Ovarian Cancer Treatment and Diagnosis
In Italy, over 5,000 women are diagnosed with ovarian cancer annually, adding to the existing 35,000 cases. Diagnosis often occurs at advanced stages due to the absence of early symptoms. High-grade mucinous ovarian tumors, constituting 70-80% of cases, are aggressive and resistant to chemotherapy, with a current accuracy of only 50% in treatment response prediction.
Given the significant variability in this cancer type among patients, the development of an AI-based tool became imperative to accurately forecast chemotherapy responses.
Role of Indicators and AI in Tailoring Cancer Care
Professor Sala and Dr. Mireia Crispin Ortuzar from Cambridge compiled data from two independent datasets comprising 134 patients. Clinical information, treatment details, and biomarkers like CA-125 and circulating tumor DNA were collected. CT scan images of tumor sites aided in determining statistical tumor features.
By analyzing tumor mutations, treatment responses, and clinical characteristics, distinct patient groups were identified, influencing therapy outcomes. These characteristics were integrated into the AI tool’s design, which underwent rigorous training and testing to verify its effectiveness.
Potential Applications of the IRON Model in Clinical Settings
Professor Sala highlighted the tool’s ability to identify patients suitable for neoadjuvant therapy, guiding immediate medical interventions. Future studies at Policlinico Gemelli, in collaboration with Professor Giovanni Scambia’s team, aim to leverage this model to assess individual risk profiles effectively.
