Introduction
When it comes to treating localized prostate cancer, it's crucial to accurately predict the risk of side-specific extraprostatic extension (ssEPE). This helps surgeons plan nerve-sparing surgeries effectively and reduce potential side effects like impotence and incontinence. Thanks to recent advancements in artificial intelligence (AI), we now have a powerful tool called SEPERA (Side-specific Extra-Prostatic Extension Risk Assessment) that can provide personalized and reliable predictions for ssEPE during radical prostatectomy.
Methodology
To develop SEPERA, researchers collected data from 2468 patients, including a total of 4936 cases (prostatic lobes). The AI model was trained using information from 1022 cases at Trillium Health Partners in Mississauga, Canada, between 2010 and 2020. The model's performance was then tested and validated using data from three academic centers: Princess Margaret Cancer Centre (Toronto, Canada) from 2008 to 2020, L'Institut Mutualiste Montsouris (Paris, France) from 2010 to 2020, and Jules Bordet Institute (Brussels, Belgium) from 2015 to 2020. The researchers assessed SEPERA's accuracy using metrics such as the area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), calibration, and net benefit. They also compared SEPERA against other prediction models and nomograms.
Findings
SEPERA demonstrated excellent accuracy and outperformed other models in all validation cohorts. Its performance was measured using the AUROC and AUPRC, and the results showed that SEPERA had a strong predictive ability. For instance, when comparing cases where patients had ssEPE despite benign ipsilateral biopsies, SEPERA correctly predicted ssEPE in 68% of cases (72 out of 106), surpassing the performance of other models. SEPERA also provided a higher net benefit, meaning it allowed more patients to undergo nerve-sparing surgery safely. The researchers conducted an algorithmic audit to ensure the model was fair and unbiased, and the results showed that SEPERA was not influenced by factors such as race, biopsy year, age, biopsy type, biopsy location, or D'Amico risk group. The most common errors made by the model were false positives, particularly in older patients with high-risk disease. Importantly, SEPERA did not miss any aggressive tumors.
Implications
SEPERA has significant implications for surgical planning and patient counseling in cases of localized prostate cancer. Its accurate predictions of ssEPE enable surgeons to personalize their approach and balance oncological control with minimizing postoperative functional decline. SEPERA outperforms existing prediction models, especially in challenging scenarios such as predicting contralateral ssEPE in unilateral high-risk disease or ssEPE when benign ipsilateral biopsies are present. The reliability, accuracy, and clinical usefulness of SEPERA make it a valuable tool for real-world clinical practice.
Conclusion
The use of artificial intelligence, exemplified by the SEPERA tool, holds great promise for improving nerve-sparing surgery in patients with localized prostate cancer. By providing personalized and reliable predictions of side-specific extraprostatic extension, SEPERA empowers surgeons to make informed decisions, manage patient expectations, and minimize treatment-related side effects. Ongoing research and application of AI in this field have the potential to revolutionize the quality of care for prostate cancer patients.
Reference:
KWONG, Jethro CC, et al. Development, multi-institutional external validation, and algorithmic audit of an artificial intelligence-based Side-specific Extra-Prostatic Extension Risk Assessment tool (SEPERA) for patients undergoing radical prostatectomy: a retrospective cohort study. The Lancet Digital Health, 2023.
Comments