The CONFIDENT-P trial: Clinical implementation of artificial intelligence assistance in prostate cancer pathology.

Abstract

TPS405 Background: Prostate cancer grading has been subject to variation, putting patients at risk for over- and under-treatment (Flach et al., 2022). As a response, development of artificial intelligence (AI) prostate cancer algorithms has been on the rise in the field of pathology. Despite promising results in retrospective studies, and several FDA-approved and CE-IVD certified algorithms on the market, prospective clinical implementation studies of AI are lacking. Moreover, uptake of digital pathology is currently insufficient due to high implementation costs (Ho et al., 2014). In this trial, we will explore the benefits of an AI-assisted pathology workflow in prostate cancer detection, while maintaining diagnostic safety standards. We will focus on reducing costly immunohistochemistry stains (IHC), which are currently used to aid in the diagnosis of prostate cancer. Methods: CONFIDENT-P is a SPIRIT-AI compliant single-centre, clinical trial, in a fully digital academic pathology laboratory. We will prospectively enroll 80 prostate cancer patients who undergo prostate needle biopsies. The pathology specimens will be pseudo-randomized to be assessed by a pathologist with- or without AI-assistance in a pragmatic (bi-)weekly sequential design, in a 1:1 allocation ratio. Patients are excluded when they are redirected for a second opinion to the UMC Utrecht. In the intervention group, 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. In the control group, pathologists will assess HE WSI according to the current clinical workflow. If no tumour cells are identified or when the pathologist is in doubt, staining by immunohistochemistry (IHC) will be performed. Primary endpoint is the number of saved resources on IHC for detecting tumour cells, since this will clarify tangible cost savings that will help to build the business case for AI. We will compare the proportion of IHC-use in both arms, and calculate adjusted relative risks, using a log-binomial model. The sample size gives at least 80% power to detect a 30% difference in IHC usage, using a one-sided significance level of 5%. Enrolment is set to begin in November 2022. The ethics committee (MREC NedMec) waived the need of official ethical approval, as participants are not subjected to procedures and as they are not required to follow rules. Furthermore, they are not at risk of an inferior diagnosis. Trial registration is therefore applicable nor suitable.

Ähnliche Arbeiten