Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Aug 26;5(10):986-997.
doi: 10.1002/bco2.421. eCollection 2024 Oct.

Extracapsular extension risk assessment using an artificial intelligence prostate cancer mapping algorithm

Alan Priester  1   2 , Sakina Mohammed Mota  1 , Kyla P Grunden  2 , Joshua Shubert  1 , Shannon Richardson  2 , Anthony Sisk  3 , Ely R Felker  4 , James Sayre  5 , Leonard S Marks  2 , Shyam Natarajan  1   2 , Wayne G Brisbane  2
Affiliations

Extracapsular extension risk assessment using an artificial intelligence prostate cancer mapping algorithm

Alan Priester et al. BJUI Compass. .

Erratum in

  • Erratum.
    [No authors listed] [No authors listed] BJUI Compass. 2024 Dec 30;5(12):1324-1329. doi: 10.1002/bco2.482. eCollection 2024 Dec. BJUI Compass. 2024. PMID: 39744071 Free PMC article.

Abstract

Objective: The objective of this study is to compare detection rates of extracapsular extension (ECE) of prostate cancer (PCa) using artificial intelligence (AI)-generated cancer maps versus MRI and conventional nomograms.

Materials and methods: We retrospectively analysed data from 147 patients who received MRI-targeted biopsy and subsequent radical prostatectomy between September 2016 and May 2022. AI-based software cleared by the United States Food and Drug Administration (Unfold AI, Avenda Health) was used to map 3D cancer probability and estimate ECE risk. Conventional ECE predictors including MRI Likert scores, capsular contact length of MRI-visible lesions, PSMA T stage, Partin tables, and the "PRedicting ExtraCapsular Extension" nomogram were used for comparison.Postsurgical specimens were processed using whole-mount histopathology sectioning, and a genitourinary pathologist assessed each quadrant for ECE presence. ECE predictors were then evaluated on the patient (Unfold AI versus all comparators) and quadrant level (Unfold AI versus MRI Likert score). Receiver operator characteristic curves were generated and compared using DeLong's test.

Results: Unfold AI had a significantly higher area under the curve (AUC = 0.81) than other predictors for patient-level ECE prediction. Unfold AI achieved 68% sensitivity, 78% specificity, 71% positive predictive value, and 75% negative predictive value. At the quadrant level, Unfold AI exceeded the AUC of MRI Likert scores for posterior (0.89 versus 0.82, p = 0.003), anterior (0.84 versus 0.80, p = 0.34), and all quadrants (0.89 versus 0.82, p = 0.002). The false negative rate of Unfold AI was lower than MRI in both the anterior (-60%) and posterior prostate (-40%).

Conclusions: Unfold AI accurately predicted ECE risk, outperforming conventional methodologies. It notably improved ECE prediction over MRI in posterior quadrants, with the potential to inform nerve-spare technique and prevent positive margins. By enhancing PCa staging and risk stratification, AI-based cancer mapping may lead to better oncological and functional outcomes for patients.

Keywords: MRI; artificial intelligence; extracapsular extension; fusion biopsy; prostate cancer.

PubMed Disclaimer

Conflict of interest statement

Unfold AI is commercially available cancer mapping software provided by Avenda Health. Dr. Brisbane receives no financial compensation from Avenda Health. He had complete control of the data and supervised manuscript preparation. Dr. Sayre performed all statistical analyses. Dr. Priester and Dr. Mota were responsible for deriving AI‐based ECE risk prediction. Dr. Priester, Dr. Mota, and Mr. Shubert are employees at Avenda Health. Dr. Natarajan and Dr. Marks are co‐founders of Avenda Health.

Figures

FIGURE 1
FIGURE 1
Two exemplary cases with similar MRI regions of interest, showing (A, D) T2‐weighted MRI, (B, E) Unfold AI ECE risk assessment, and (C, F) whole‐mount histopathology. The first case (A‐C) had low ECE risk on Unfold AI, no ECE on histopathology, and negative surgical margins. The second case had high ECE risk on Unfold AI, ECE on histopathology, and focally positive surgical margins. It is plausible that Unfold AI could have helped prevent positive margins for the second case.
FIGURE 2
FIGURE 2
Flowchart illustrating dataset selection for this study. The final column describes the conventional ECE predictors and the number of patients available for comparison with Unfold AI.
FIGURE 3
FIGURE 3
Receiver operating characteristic curves for patient‐level ECE prediction, computed using the data subset available for each metric (sample sizes vary; see Table 3).
FIGURE 4
FIGURE 4
Receiver operating characteristic curves for quadrant‐level ECE prediction, analysing (A) all quadrants in aggregate, N = 542, and (B) quadrants stratified into anterior and posterior subgroups, N = 271.
FIGURE 5
FIGURE 5
Example cases for which MRI failed (false negative) but Unfold AI succeeded (true positive) to predict posterior‐quadrant ECE. The left column shows MR images and MRI Likert Scores for the posterior quadrants, with the prostate outlined in white, the ROI outlined in red, and prostate midline annotated with a grey dotted line. The right column shows Unfold AI cancer estimation maps and histopathology ECE ground truth for posterior quadrants, and the region of highest ECE risk annotated with a black dotted line. Example cases include (A‐B) a case where MRI predicted ECE (Likert = 3) only in the right posterior, but Unfold AI successfully predicted bilateral ECE; (C‐D) a case where MRI predicted no ECE (Likert = 1), but Unfold AI successfully predicted ECE in the left posterior; (E‐F) a case where MRI predicted no ECE (Likert = 2), but Unfold AI successfully predicted bilateral ECE; and (G‐H) a case where MRI predicted no ECE (Likert = 2), but Unfold AI successfully predicted ECE in the right posterior.
FIGURE B1
FIGURE B1
Unfold AI ECE risk metric values for an illustrative case, which can be used for prospective ECE risk assessment.
See this image and copyright information in PMC

References

    1. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin. 2024;74(1):12–49. 10.3322/caac.21820 - DOI - PubMed
    1. Ennis RD, Hu L, Ryemon SN, Lin J, Mazumdar M. Brachytherapy‐based radiotherapy and radical prostatectomy are associated with similar survival in high‐risk localized prostate cancer. J Clin Oncol [Internet]. 2018;36(12):1192–1198. Available from: 10.1200/JCO.2017.75.9134 - DOI - PubMed
    1. Preisser F, Cooperberg MR, Crook J, Feng F, Graefen M, Karakiewicz PI, et al. Intermediate‐risk prostate cancer: stratification and management. Eur Urol Oncol [Internet]. 2020;3(3):270–280. Available from: https://www.sciencedirect.com/science/article/pii/S2588931120300444, 10.1016/j.euo.2020.03.002 - DOI - PubMed
    1. Godoy G, Tareen BU, Lepor H. Site of positive surgical margins influences biochemical recurrence after radical prostatectomy. BJU Int. 2009;104(11):1610–1614. 10.1111/j.1464-410X.2009.08688.x - DOI - PubMed
    1. Diamand R, Roche JB, Lievore E, Lacetera V, Chiacchio G, Beatrici V, et al. External validation of models for prediction of side‐specific extracapsular extension in prostate cancer patients undergoing radical prostatectomy. Eur Urol Focus. 2023;9(2):309–316. 10.1016/j.euf.2022.09.006 - DOI - PubMed