“Artificial Intelligence in NAFLD Diagnosis: A New Frontier in Managing Metastatic Prostate Cancer Risk”

To the Editor: A recent study by Hongyi Zhang et al. reveals that patients with non-alcoholic fatty liver disease (NAFLD) typically exhibit an elevated risk of biochemical recurrence in patients with metastatic prostate cancer. Utilizing computed tomography (CT) scans, the research estimated liver fat levels and employed Cox models to examine the association between NAFLD and biochemical recurrence, while accounting for variables such as body mass index, adipose tissue, hypertension, and diabetes it was determined that NAFLD functions as an autonomous risk factor for biochemical recurrence in individuals with advanced-grade metastatic prostate cancer [1]. This study signifies the urgent need for new methods to accurately diagnose and treat NAFLD to lower the recurrence of metastatic prostate cancer. Artificial intelligence (AI) has emerged as a promising tool in NAFLD detection, as demonstrated by a systematic review of 25 studies showing AI-enhanced ultrasonography achieved an impressive sensitivity of 0.97 and specificity of 0.98, surpassing AI-assisted clinical databases. Likewise, the incorporation of AI in clinical datasets exhibited notable levels of sensitivity and specificity in identifying NASH and diagnosing liver fibrosis stages F1–F4. Overall, AI-driven systems show considerable promise in improving diagnostic precision for individuals with NAFLD, NASH, and liver fibrosis [2]. Liver biopsy is still the most effective method for the diagnosis of NAFLD; however, it is an invasive process that comes with the potential for serious complications like pain, infection, and bleeding, so it is not feasible for all suspected persons. Conversely, ultrasound assessment represents a noninvasive and convenient method widely employed in the clinical evaluation of NAFLD; nonetheless, the diagnostic accuracy may be affected by various factors, notably the subjective judgment of the examiner. Recent studies [3] have incorporated biomarkers such as adiponectin and caspase-cleaved cytokeratin 18 fragment (M30) into predictive models for NAFLD activity score (NAS). These models exhibit potential in monitoring disease progression, fluctuations in body mass, and differentiation between NASH and NAFLD, demonstrating an AUROC range of 0.70–0.73. Although many studies demonstrate elevated diagnostic efficacy, the practical deployment of these findings within clinical settings is still nascent. Furthermore, while AI significantly augments histological and imaging-based evaluations of NAFLD, its capacity to forecast long-term clinical outcomes—including its interaction with the progression of prostate cancer—necessitates validation through rigorously designed prospective trials. Presently, initiatives such as the Liver Investigation: Testing Marker Utility in Steatohepatitis (LITMUS) consortium [4] and the Non-Invasive Biomarkers of Metabolic Liver Disease (NIMBLE) project [5] are scrutinizing noninvasive indicators, including AI-augmented imaging techniques, for the detection and progression monitoring of NAFLD. Additional research is imperative to evaluate whether the incorporation of AI into established prostate cancer screening protocols can enhance risk stratification and inform treatment strategies. Moreover, the variability of AI methodologies across different studies raises significant concerns regarding reproducibility and generalizability, thereby necessitating the establishment of standardized validation frameworks. By leveraging AI-driven diagnostic models, we can potentially bridge the gap between NAFLD and metastatic prostate cancer recurrence, enabling earlier identification of high-risk patients and personalized treatment strategies. Considering the emerging role of NAFLD as an autonomous risk factor for biochemical recurrence, the incorporation of AI-augmented hepatic evaluations into oncological protocols may enhance prognostic precision and therapeutic decision-making for individuals with prostate cancer. Subsequent investigations ought to prioritize the validation of AI's predictive efficacy across heterogeneous populations, as well as the examination of its potential in alleviating NAFLD-associated cancer progression. Addressing these deficiencies will be paramount in transitioning AI from a potentially transformative innovation into a clinically essential instrument for optimizing outcomes in both hepatic and oncological disease management. The authors declare no conflicts of interest. No new data were generated for this research.