AI Algorithm-Informed Biopsy for Prostate Cancer Detection With Indeterminate and Low-Risk Prostate MRI Lesions

A Prospective Randomized Phase I/II Study of Artificial Intelligence Algorithm-Informed Biopsy for Detection of Prostate Cancer in Patients With Indeterminate and Low-risk Prostate MRI Lesions

Use of AI algorithm for PCa detection is feasible, and AI-informed biopsies (AI-targeted and perilesional biopsy) improves csPCa detection in patients with indeterminate MRI lesions and in patients with low-risk MRI lesions and high-risk clinical features.

Study Overview

• Status: Not yet recruiting
• Conditions: Prostate Cancer
• Intervention / Treatment: Device: Bi-parametric MRI-based cascaded deep-learning AI algorithm

Detailed Description

Primary Feasibility Objective:

1. Assess the acceptance rate of randomization and biopsy recommendations based on study protocol and AI algorithm results by the patients. This will be assessed in the first 10 patients who enroll during the phase I feasibility segment.

Primary Efficacy Objective:

1. Evaluate the per-patient and per-lesion csPCa detection rates of AI algorithm-informed biopsy (the intervention arm) versus contemporary biopsy (the control arm) in patients randomly allocated 1:1 to each arm. This will be evaluated in all 25 patients per arm (50 patients).

Secondary Objectives (These objectives will be satisfied using endpoint data from all 50 subjects (25/arm) enrolled):

1. Evaluate benign and clinically non-significant PCa rates (GS <7) in patients who underwent AI-algorithm informed (the intervention arm) versus contemporary (the control arm) prostate biopsies.
2. Evaluate the specificity and sensitivity of AI algorithm-informed biopsy (AI-targeted and perilesional prostate biopsy) versus contemporary biopsy in detection of csPCa.
3. Obtain and evaluate adverse events (AEs), urinary function (IPSS), sexual function (IIEF) quality of life (QOL) [ SF-12 and TMI scores] and decision regret (DRS) measures on subjects that underwent contemporary biopsy versus AI Algorithm-informed biopsy.

Exploratory Objective:

1. Collect data via genomic and transcriptomic approaches (Whole exome sequencing + Targeted RNA sequencing OR single cell RNA sequencing) in patients whose standard contemporary biopsy, perilesional biopsy and AI-targeted biopsy revealed csPCa, and compare collected data on all endpoints for differences among perilesional biopsy, AI-targeted biopsy and contemporary standard biopsy.

• Study Type: Interventional
• Enrollment (Estimated): 50
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Aaron Holley; Phone Number: 5016868274; Email: JAHolley@uams.edu

Study Locations

• United States
  • Arkansas
    • Little Rock, Arkansas, United States, 72205
      • University of Arkansas for Medical Sciences
      • Contact: Ahmet Aydin, MD; Phone Number: 501-686-8530
      • Principal Investigator: Ahmet M Aydin, MD

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

Description

Inclusion Criteria:

1. 40 years of age or older.
2. A recent pMRI performed within last 12 weeks
3. Eastern Cooperative Oncology Group (ECOG) performance status 0 - 1.
4. Any patient with PIRADS 3 lesions per pMRI, AND elevated PSA ("=> 3.0 ng/ml" for patients between 40 and 75 years old, and "=> 4.0 ng/ml" for the patients older than 75 years).

5. Patients with PIRADS 1-2 lesions per pMRI, AND elevated PSA ("=> 3.0 ng/ml" for patients between 40 and 75 years old, and "=> 4.0 ng/ml" for the patients older than 75 years), AND at least one of the following:

   1. High PSA density (0.15 ng/ml/g or higher),
   2. suspicious DRE,
   3. a positive/high-risk blood or urine biomarker test,
   4. high-risk ancestry (Black/African American),
   5. those with germline mutations that increase the risk for prostate cancer,
   6. significant personal medical history,
   7. significant family history,
   8. persistent and significant increase in PSA levels (persistently elevated PSA for at least 12 months with an increase of at least 100% or more within 24 months, last level confirmed twice).

Exclusion Criteria:

1. Patients younger than 18 years old.
2. Any patient with PIRADS 4-5 lesion per pMRI.
3. Any patient with known csPCa (GS ≥7 (3+4)) per biopsy.

4. Any patient with PCa and managed with active surveillance, surgery or radiation.

   a. (Patients who never scanned with pMRI before, had GS 6 (3+3) PCa only per systematic biopsy, and currently need confirmatory prostate biopsy will be allowed to enroll in the trial).

5. Medically unfit for anesthesia.
6. Any history of allergic reactions attributed to contrast agents, or other compounds of similar chemical compositions.
7. Any medical history preventing pMRI or prostate biopsy.
8. Any medical condition distorting quality of pMRI such as artificial hip prosthesis, and excessive rectal gas.
9. Any other condition that, in the opinion of the investigator, might interfere with the safe conduct of the study.

Inclusion of Women and Minorities: All participants will be men without previous diagnosis for PCa. Men of all ethnic groups and races are eligible for the study. Thus, women will not be included in this study.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Diagnostic
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: None (Open Label)

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: Bi-parametric MRI-based cascaded deep-learning AI algorithm; The AI model inputs biparametric DICOM sequences (T2-weighted images, high-b-value diffusion-weighted images, and apparent diffusion coefficient maps), and the outputs include binary prostate organ and intraprostatic lesion segmentations. This study will assess a recently developed and both internally and externally validated AI algorithm for PCa detection capability in patients with equivocal lesions (PI-RADS 3 lesions) and negative lesions (PI-RADS 1-2 lesions) with higher clinical risk features such as high PSA density.	Device: Bi-parametric MRI-based cascaded deep-learning AI algorithm; Artificial intelligence system used in medical imaging, primarily for the automated detection and classification of lesions (such as prostate cancer) using only specific types of magnetic resonance imaging (MRI) data.
No Intervention: Perilesional prostate biopsy; Standard of care prostate biopsy which is a systematic template biopsy (with 12 biopsy cores) + MRI-targeted biopsy (for PI-RADS category 3 lesions only, with 3 biopsy cores), consistent with current NCCN guideline recommendations

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Acceptance rate of randomization and biopsy recommendations based on study protocol and AI algorithm results by the patients	Number and percent of the first 10 enrolled and randomized subjects who agree to undergo the prostate biopsy procedure to which they were randomized and accept the biopsy recommendations based on study protocol and AI algorithm results.	4 months
Per-patient and per-lesion csPCa detection rates of AI algorithm-informed biopsy (the intervention arm) versus contemporary biopsy (the control arm) in patients randomly allocated 1:1 to each arm	The percent of csPCa detected per-patient and per-lesion in the biopsy cores obtained from each study arm. We expect at least 42% for the csPCa detection rate on the AI algorithm-informed-biopsy arm, which would be a 27% increase relative to the current csPCa detection rate (15%) expected on the contemporary prostate-biopsy arm.	4 months

Secondary Outcome Measures

Outcome Measure	Time Frame
Percentage of benign and clinically non-significant PCa detected in patients who underwent AI Algorithm-informed or contemporary prostate biopsies.	4 months
Percentage of true positive, false positive, true negative and false negative findings for csPCa in all patients who enrolled in both study arms.versus contemporary biopsy in detection of csPCa	4 months
Evaluation of adverse events on subjects that underwent contemporary biopsy versus AI Algorithm-informed biopsy graded by CTCAE v5.0	4 months
Evaluation of urinary function (IPSS) on subjects that underwent contemporary biopsy versus AI Algorithm-informed biopsy	4 months
Evaluation of sexual function (Form IIEF) on subjects that underwent contemporary biopsy versus AI Algorithm-informed biopsy	4 months
Evaluation of quality of life using Form TMI and Form SF12 scores on subjects that underwent contemporary biopsy versus AI Algorithm-informed biopsy	4 months
Evaluation of decision regret using the Decision Regret Scale (Form DRS) to measure on subjects that underwent contemporary biopsy versus AI Algorithm-informed biopsy	4 months

Other Outcome Measures

Outcome Measure	Time Frame
Expression levels and/or expression profiles of genomic and transcriptomic signatures in csPCa diagnosed from perilesional biopsy, AI-targeted biopsy and contemporary standard biopsy	4 months

Collaborators and Investigators

• Sponsor: University of Arkansas
• Investigators: Principal Investigator: Ahmet M Aydin, MD, University of Arkansas

Study record dates

Study Major Dates

• Study Start (Estimated): May 1, 2026
• Primary Completion (Estimated): January 1, 2028
• Study Completion (Estimated): January 1, 2029

Study Registration Dates

• First Submitted: November 13, 2025
• First Submitted That Met QC Criteria: November 14, 2025
• First Posted (Actual): November 17, 2025

Study Record Updates

• Last Update Posted (Actual): May 19, 2026
• Last Update Submitted That Met QC Criteria: May 15, 2026
• Last Verified: May 1, 2026

More Information

Terms related to this study

• Additional Relevant MeSH Terms: Urogenital Diseases; Genital Diseases; Genital Neoplasms, Male; Urogenital Neoplasms; Neoplasms by Site; Neoplasms; Genital Diseases, Male; Prostatic Diseases; Male Urogenital Diseases; Prostatic Neoplasms
• Other Study ID Numbers: 299514

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No