Michele Catellani

Artificial intelligence is highly regarded as the most promising future technology that will have a great impact on healthcare across all specialties. Its subsets, machine learning, deep learning, and artificial neural networks, are able to automatically learn from massive amounts of data and can improve the prediction algorithms to enhance their performance. This area is still under development, but the latest evidence shows great potential in the diagnosis, prognosis, and treatment of urological diseases, including bladder cancer, which are currently using old prediction tools and historical nomograms. This review focuses on highly significant and comprehensive literature evidence of artificial intelligence in the management of bladder cancer and investigates the near introduction in clinical practice.

Objectives: to investigate the accuracy of transurethral resection of bladder tumours (TURBT) in detecting histological variants (BHV) at radical cystectomy (RC) and to evaluate the impact of TURBT before cystectomy on oncological outcomes. Methods: Data of 410 consecutive RCs were assessed. Positive and negative predictive values were used to assess the accuracy of TURBT in detecting BHV. Cohen’s Kappa coefficient was used to calculate the agreement grade. Logistic regression analysis predicted features based on the presence of BHV at TURBT. Multivariable backward conditional Cox regression analysis was used to estimate oncological outcomes. Results: A total of 73 patients (17.8%) showed BHV at TURBT as compared to 108 (26.3%) at RC. A moderate agreement in histological diagnosis was found between TURBT and RC (0.58). However, sensitivity and specificity in detecting BHV were 56% and 96%, respectively. Furthermore, positive predictive value (PPV) was 84.7% and negative predictive value (NPV) was 84.6%. Presence of BHV at TURBT was an independent predictor for pathologic upstage, albeit not a predictor for positive nodes or positive surgical margins. However, at multivariable analysis adjusted for all confounders, presence of BHV at TURBT was an independent predictor for recurrence after RC, but not for survival. Conversely, the presence of BHV at RC was an independent predictor for both recurrence and survival. Conclusion: There was a moderate agreement between TURBT and RC histopathological findings. TURBT, alone, could not provide an accurate and definitive histological diagnosis. Detection of BHV in TURBT specimens is not an independent predictor of oncological outcomes; indeed, only pathological features at RC are associated with worse survival. However, BHV presence in cystectomy specimens resulted as an independent predictor of both cancer-specific and overall mortality.

Background To test clinically significant prostate cancer (csPCa) rates during follow-up in biopsy naïve patients that underwent two different diagnostic pathways: (1) SB GROUP (n = 354): systematic random biopsies (SB) vs. (2) TB GROUP (n = 264): multiparametric magnetic resonance imaging (mpMRI) and only targeted biopsies (TB) of PI-RADS ≥ 3 lesions. Patients with PI-RADS ≤ 2 score avoided prostate biopsies. Methods Retrospective single centre study of 618 biopsy naive patients (2015–2018). Two different definitions of csPCa were used: (1) csPCa ISUP GG ≥ 2 (ISUP grade group [GG] ≥ 2) and (2) csPCa ISUP GG ≥ 3. Kaplan–Meier plots and univariable Cox regression models tested rates over time of csPCa ISUP GG ≥ 2 and caPCa ISUP GG ≥ 3 in SB GROUP vs. TB GROUP. Results At initial biopsy, TB achieved higher rates of csPCa ISUP GG ≥ 2 (35.3 vs. 18.9%; p < 0.001) and csPCa ISUP GG ≥ 3 (12.6 vs. 6.2%; p = 0.04), relative to SB. After a median time follow-up of 36 months, the rates of csPCa ISUP GG ≥ 2 (6.1 vs. 4.4%; p = 0.6) and csPCa ISUP GG ≥ 3 (3.3 vs. 1.1%; p = 0.2) were similar in SB GROUP vs. TB GROUP. Moreover, in TB GROUP patients that avoided prostate biopsies because of negative baseline mpMRI (n = 145), only 4.1% exhibited csPCa ISUP GG ≥ 2 during follow-up. Moreover, none of these patients (PI-RADS ≤ 2) had csPCa ISUP GG ≥ 3. Conclusions In biopsy naïve setting, a diagnostic pathway including pre-biopsy mpMRI and TB of only PCa suspicious lesions is not associated with higher rates of csPCa during follow-up, relative to a diagnostic pathway of SB. Moreover, patients with negative baseline mpMRI could safely avoid prostate biopsies and could be followed with repeated PSA testing, since only a small proportion of them would harbor csPCa.