Prostate cancer ranks as the second most prevalent cancer among men, with nearly 300,000 new diagnoses occurring annually in the United States. Researchers at Mass General Brigham have developed a reliable technique for estimating the size of prostate cancer, enhancing clinicians’ ability to make informed decisions regarding treatment. They achieved this by training and validating an AI model using MRI scans from over 700 prostate cancer patients. Impressively, this model could accurately identify and outline the boundaries of 85% of the most aggressive lesions detected on radiological images.
According to the AI model’s assessments, larger tumors were linked to an increased likelihood of treatment failure and metastasis, regardless of other commonly used risk factors. Moreover, for patients undergoing radiation therapy, the volume of the tumor indicated by the AI outperformed traditional methods in predicting metastasis. Researchers are optimistic that this tool could assist clinicians in determining a tumor’s severity, leading to more tailored treatment strategies and guiding radiation therapy. The findings of the study are detailed in the journal Radiology.
“AI-derived tumor volume could greatly enhance precision medicine for prostate cancer patients by refining our understanding of cancer aggressiveness, thus enabling optimal treatment recommendations,” remarked the study’s lead author, Dr. David D. Yang, from the Department of Radiation Oncology at Brigham and Women’s Hospital, part of the Mass General Brigham healthcare system.
MRI technology has significantly advanced prostate cancer diagnosis and is now a routine component of treatment protocols. While human physicians can estimate tumor sizes using MRI results, these assessments can be subjective and inconsistent.
To establish a more uniform method for measuring tumor size, researchers utilized an AI model trained on MRI images from 732 patients receiving treatment at a single clinical center. They further explored whether the AI’s size assessments correlated with treatment outcomes in the subsequent 5 to 10 years after diagnosis.
The AI model successfully identified and measured nearly 85% of prostate tumors that scored a PI-RADS (Prostate Imaging Reporting and Data System) 5, indicating a high probability of significant prostate cancer. The model’s size estimates proved to be a potential prognostic indicator: larger tumors were linked to a greater risk of recurrence and metastasis, as indicated by prostate-specific antigen (PSA) levels, whether the patients were treated surgically or with radiation therapy.
“The AI measurement provides additional insights regarding patient outcomes,” stated Dr. Martin King, the study’s senior author from the Department of Radiation Oncology at the Brigham. “For patients, this can offer clarity on their chances of being cured and the likelihood of their cancer recurring or spreading.”
Beyond aiding clinicians and patients in understanding the tumor’s severity, the AI model could also assist radiation oncologists by localizing the tumor’s precise area for targeted treatment. It also offers a faster evaluation compared to existing approaches for assessing prostate cancer aggressiveness, which can take two weeks or longer for results. By utilizing AI, patients could potentially begin their treatment sooner.
Cancer research is a cornerstone of the care provided by Mass General Brigham to their patients. The integration of research with the system’s strengths in innovation and community involvement enables Mass General Brigham Cancer to offer comprehensive cancer care while prioritizing health equity. Their vision is to create a seamless, integrated, and research-driven cancer care experience that supports patients throughout their journey—encompassing prevention, early detection, treatment, and survivorship.
Looking to the future, researchers aim to validate their model with a broader, multi-institutional dataset.
“Our goal is to confirm our results by collaborating with other institutions and studying patient groups with varying disease characteristics to ensure that this method is applicable to all patients,” Dr. Yang explained.
