Optimized Identification of High-Grade Prostate Cancer by Combining Different PSA Molecular Forms and PSA Density in a Deep Learning Model

Abstract

After skin cancer, prostate cancer (PC) is the most common cancer among men. The gold standard for PC diagnosis is based on the PSA (prostate-specific antigen) test. Based on this preliminary screening, the physician decides whether to proceed with further tests, typically prostate biopsy, to confirm cancer and evaluate its aggressiveness. Nevertheless, the specificity of the PSA test is suboptimal and, as a result, about 75% of men who undergo a prostate biopsy do not have cancer even if they have elevated PSA levels. Overdiagnosis leads to unnecessary overtreatment of prostate cancer with undesirable side effects, such as incontinence, erectile dysfunction, infections, and pain. Here, we used artificial neuronal networks to develop models that can diagnose PC efficiently. The model receives as an input a panel of 4 clinical variables (total PSA, free PSA, p2PSA, and PSA density) plus age. The output of the model is an estimate of the Gleason score of the patient. After training on a dataset of 190 samples and optimization of the variables, the model achieved values of sensitivity as high as 86% and 89% specificity. The efficiency of the method can be improved even further by training the model on larger datasets.

The Pathogenesis of Prostate Cancer

“…However, prostate cancer screening has been surrounded by controversy and can be seen as an ethical dilemma. Studies have shown that distinguishing between indolent pathology and aggressive tumors using PSA levels alone is insufficient as it is not cancer-specific (7). It is also a challenge as high PSA levels do not directly correlate with pathogenicity and requires further investigations through multiparametric magnetic resonance imaging (MRI) and histology.…”

“…It is also a challenge as high PSA levels do not directly correlate with pathogenicity and requires further investigations through multiparametric magnetic resonance imaging (MRI) and histology. Therefore, early diagnosis often leads to harm to patients via overdiagnosis and unnecessary treatment (radical surgery, radiotherapy and chemical castration) leading to reduced quality of life (QoL) and psychological trauma knowing they have a cancer diagnosis (7). This chapter aims to address the developmental processes involved in prostate cancer including hereditary and molecular components.…”

Prostate Health Index and Multiparametric MRI: Partners in Crime Fighting Overdiagnosis and Overtreatment in Prostate Cancer

“…We recently demonstrated that by using artificial neuronal networks (ANNs), it is possible to develop models combining different PSA-derivatives (including total PSA, free PSA, p2PSA and PSA density) optimizing high-grade PCa recognition [ 39 ]. Taking into account that AI is revolutionizing PCa clinical management [ 40 , 41 ] and ameliorates accuracy in the detection of csPCa when applied to MRI [ 42 ], it is plausible to hypothesize that the sequential or the combined use of novel biomarkers and MRI based on machine learning approach may produce models able to minimize overdiagnosis, without missing csPCa, providing the clinicians a tool to match tumor aggressiveness and treatment invasiveness.…”

Reply to Jue et al. Value of MRI to Improve Deep Learning Model That Identifies High-Grade Prostate Cancer. Comment on “Gentile et al. Optimized Identification of High-Grade Prostate Cancer by Combining Different PSA Molecular Forms and PSA Density in a Deep Learning Model. Diagnostics 2021, 11, 335”

“…In this respect, our group struggled with the use of additional biomarkers for identification of clinically significant prostate cancer (PCa) and recognised a potential of PSA molecular forms that should not be dismissed [ 1 ]. We also noted that the model combining different biomarkers offers the potential for better diagnostic performance [ 2 , 3 ]. Thus, as clinical studies will confirm the relevance of other variables for PCa diagnosis, we will be able to include them in this model template, increasing even further sensitivity, specificity, and accuracy.…”