(UroToday.com) The 2023 American Society of Clinical Oncology Genitourinary (ASCO GU) cancers symposium held in San Francisco, CA between February 16th and 18th was host to a prostate cancer trials in progress poster session. Dr. Rachel Flach presented the framework for the CONFIDENT-P trial, which will evaluate the clinical implementation of artificial intelligence assistance in prostate cancer pathology.
There is significant inconsistency in prostate cancer grading, which may be associated with downstream adverse effects including both under- and over-treatment of prostate cancer patients. Artificial intelligence (AI) has emerged as a promising tool to help standardize the reporting of prostate cancer pathology. AI-based algorithms have demonstrated significant clinical utility in numerous retrospective studies,1 and many such commercially available algorithms have since been approved by the Food and Drug Association (FDA) and become Conformite Europenne In Vitro Diagnostic Medical Device (CE-IVD) conformity marked. However, to date, prospective clinical implementation studies of AI in this setting are lacking. One of the main barriers to wide adoption of digital pathology remains the high implementation costs associated with such algorithms, which classically employ immunohistochemistry (IHC) strains to aid in the diagnostic pathway. In the CONFIDENT-P trial, the authors will explore the benefits of an AI-assisted pathology workflow for prostate cancer detection, while maintaining diagnostic safety standards. Specifically, this trial will focus on reducing IHC costs.
CONFIDENT-P is a Standard Protocol Items: Recommendations for Interventional Trials-Artificial Intelligence (SPIRIT-AI) compliant single center, clinical trial that will be conducted at a fully digital academic pathology laboratory (University Medical Center Utrecht). This trial will prospectively enroll 80 patients with prostate biopsy-proven prostate cancer. The pathology specimens, all assessed by a pathologist, will be randomized in a 1:1 ratio to AI assistance (Paige Prostate) versus not in a pragmatic (bi-)weekly sequential design. Patients who are redirected for a second opinion to the University Medical Center Utrecht will be excluded. In the intervention arm of AI assistance, pathologists will assess whole slide images (WSI) of the standard haematoxylin-eosin (HE)-stained sections assisted by the output of a CE-IVD approved prostate cancer detection and grading algorithm. Conversely, in the control group pathologists will assess HE WSI according to standard clinical workflow. Staining by immunohistochemistry (IHC) will only be performed if no tumour cells are identified or when the pathologist is in doubt.
The primary study outcome is the number of saved resources on IHC for detecting tumor cells, a cost-saving measure that may enhance implementation of AI-based algorithms. The proportion of IHC-use in both arms will be compared and adjusted relative risks will be calculated using a log-binomial model. The study sample size of 80 patients will provide 80% power to detect a 30% difference in IHC usage, at a one-side type 1 error rate of 0.05. Study enrolment began in November 2022.
Secondary endpoints will include:
- Accuracy of the AI-assisted pathologist in tumor detection
- Time spent on WSI analysis
- Number of IHC stains that may have safely been omitted
- Level of confidence in tumor detection
Given the nature of the study, no ethics committee approval was required, and the trial is not currently registered on clinicaltrials.gov.
Presented by: Rachel Flach, MD, Department of Oncological Urology, University Medical Center Utrecht, Utrecht, Netherlands
Written by: Rashid Sayyid, MD, MSc – Society of Urologic Oncology (SUO) Clinical Fellow at The University of Toronto, @rksayyid on Twitter during the 2023 American Society of Clinical Oncology Genitourinary (ASCO GU) Cancers Symposium, San Francisco, CA, February 16th – February 18th, 2023.
- Goldenberg SL, et al. A new era: artificial intelligence and machine learning in prostate cancer. Nat Rev Urol 2019;16:391-403.