Revolutionizing Prostate Cancer Care: AI-Based PTEN Assessment of Digital Pathology Images as a Game-Changer in Predicting Metastasis - Beyond the Abstract

PTEN loss has been recognized as a clinically meaningful biomarker with prognostic importance in identifying patients at higher risk of early biochemical recurrence. In the previous work using commercial image analysis software, the team demonstrated the significance of quantitative PTEN loss assessment compared to a conventional, manual scoring method that is routinely performed in pathology practices. In their previous paper (Jamaspishvili et al. 2020), the team discovered that not all subclonal PTEN loss patients have a bad prognosis, as was known in earlier retrospective cohort studies. The team found that a certain degree of PTEN protein loss in human prostate cancer tissues is important for predicting biochemical recurrence more accurately.

This clinically significant amount of PTEN can only be estimated using more accurate quantification methods such as digital image analysis (DIA) approaches. However, this type of image analysis still requires substantial manual effort, including identifying the regions of interest (ex: areas with cancer, and areas with PTEN loss) and, therefore, is not fully automated. Building upon these findings, the team developed a more robust and streamlined, fully automated in-house artificial intelligence (AI)-based assessment of PTEN loss that goes beyond visual and dichotomous assessment (Harmon et al. 2021). In this new work, the group showed how this system could be beneficial not only streamlining PTEN loss detection in prostate cancer tissues but also for computing clinically significant cut-offs for predicting disease recurrence and early metastasis in prostate cancer patients who are surgically treated (Figure 1).

 
Figure 1. Image Analysis Pipeline. (A) Core-level Analysis: Each TMA core is assessed independently, undergoing cancer detection and PTEN loss identification (classifier followed by pixel-based segmentation). The final output is an area estimate for PTEN loss area and cancer area. A representative core is shown with partial PTEN loss, i.e. not all cancer areas exhibiting loss. An expert pathologist performs qualitative assessment for each core. (B) Patient-level Analysis: All cores from an individual patient are aggregated and selected by total tumor content based on cross-validation (CV) analysis before a patient-level estimate of PTEN loss relative to cancer burden (AI-qPTEN) is calculated. Clinically relevant cut-points for an association of AI-qPTEN with patient outcomes are identified through a modified CV partitioned by the institution/center. Conventional assessment (cPTEN) is determined by qualitative review across all cores.

Briefly, the entire AI-qPTEN workflow consists of the following steps: AI-based stain normalization, prostate cancer detection, and statistical modeling. The research group showed that the AI algorithm exhibited balanced sensitivity and specificity through rigorous validation across multiple centers in the CANARY retrospective cohort, which outperformed conventional PTEN loss assessment. Moreover, the AI-qPTEN approach remained statistically predictive of both metastasis-free survival and recurrence-free survival, even in cases classified as low-risk by post-surgical CAPRA-S criteria.

Figure 2. Decision curve analysis assessing the added benefit of including High-Risk AI-qPTEN (>0.04) in patients classified as low risk by CAPRA-S. Net increase in the proportion of men appropriately identified for therapeutic intervention post-surgery is shown on the y-axis and the associated probabilities on the x-axis. The green line represents a strategy of treating no men, assuming none of them experience metastasis. The orange line on the other hand assumes that all men will develop metastasis and therefore all men are treated. Multivariable models of CAPRA-S with and without High-Risk AI-qPTEN are shown as teal and purple lines, respectively. A combined model involving CAPRA-S and High-Risk AI-qPTEN showed greater net benefit across all threshold probabilities, signifying the ability to appropriately identify patients initially diagnosed with low-risk PCa for additional therapeutic interventions after surgery.

Why this study is important?

The study showcases the clinical value of an affordable and fully automated AI-powered PTEN assessment, which can evaluate the risk of metastasis or disease recurrence after radical prostatectomy. Integrating the AI-qPTEN assessment workflow with conventional risk stratification tools can inform physicians about patients who may require intensified surveillance or post-operative interventions, particularly in low-risk cases (Figure 2). In light of the promising results of this study, AI-qPTEN should be evaluated in prospective clinical trials to evaluate the benefits of post-surgical intensified therapies for prostate cancer patients and demonstrate the value of biomarker-guided trials. AI-qPTEN can improve outcomes by enabling physicians to make more informed decisions and tailor treatments to individual patients. Further, the workflow's cost-effectiveness makes it a viable alternative to commercial molecular tests, enabling personalized cancer care in low-income countries.

Written by: Tamara Jamaspishvili, MD, PhD, Department of Pathology and Molecular Medicine, Queen's University, Kingston, Ontario, Canada; Department of Pathology and Molecular Medicine, SUNY Upstate Medical University, Syracuse, New York

1. Jamaspishvili, T. et al. Risk Stratification of Prostate Cancer Through Quantitative Assessment of PTEN Loss (qPTEN). J Natl Cancer Inst 112, 1098–1104 (2020).
2. Harmon, S. A. et al. High throughput assessment of biomarkers in tissue microarrays using artificial intelligence: PTEN loss as a proof-of-principle in multi-center prostate cancer cohorts. Mod Pathol 34, 478–489 (2021).