Background/Aim:Lung cancer is one of the leading causes of cancer deaths. While low-dose computed tomography (CT) screening improves survival, radiological detection is increasingly challenged by a shortage of radiologists. This study aimed to develop and evaluate a novel, precise, and computationally efficient AI-based algorithm for lung cancer diagnosis using chest CT scans. Patients and Methods:A total of 156 patient chest CT scans were utilized to form Databases I and II. We then conducted extensive feature extraction [statistics, histograms, Fast Fourier Transform (FFT), Discrete Cosine Transform (DCT), Walsh-Hadamard Transform (WHT)] and optimized classifiers [Multi Layer Perceptron (MLP), Generalized Feed Forward Neural Network (GFF-NN), Modular Neural Network (MNN), Support Vector Machine (SVM)] with genetic algorithms. Performance evaluation measures employed were classification accuracy, Mean Squared Error (MSE), Area under the ROC curve (AUC), and computational efficiency. Results:The MNN (Topology II) classifier employing FFT-based features with momentum learning achieved 100% classification accuracy during cross-validation for both Database I and Database II, consistently yielding perfect average classification accuracy across both datasets. Conclusion:The genetically optimized MNN (Topology II) classifier shows remarkable performance in lung cancer diagnosis from CT scan images. Its ability to achieve perfect classification accuracy suggests strong potential for clinical application, offering both diagnostic precision, acting as a triage, and workload reduction in healthcare settings.

e24019 Background: Asian Cardio Oncology Society (ACOS) was started in 2020 to promote awareness regarding cardio-oncology in south-eastern Asia. This study describes our baseline survey on knowledge, attitude and practice (KAP) of cardio-oncology amongst cardiologists & oncologists of India and south-east Asia. Methods: A 38-question survey was administered online via google forms using email and watsapp to contact members of ACOS and other cardiologists/oncologists in south-east Asia from December 2020 to January 2021. Results: Total 215 responses were collected. Maximum respondents were from India (90%) followed by Bangladesh (4%). Majority were medical oncologists (41%) followed by cardiologists (23%) & radiation oncologists (20%). Female participation was 18%. Cardiologists encountered cancer diagnosis in 5% of their patients while oncologists encountered average 30% patients with cardiac disease. Risk of cancer therapy related cardiac dysfunction (CTRCD) was estimated < 10% by 39%, 10-20% by 25% and > 20% by 35%. Regarding knowledge of cardio-oncology and familiarity with guidelines, 29% rated themselves poor, 48% average and 23% good. 85% did not have dedicated cardiologist at their centre. Echocardiography was not performed by trained cardiologists at 33% centres. 68% believed in baseline cardiac evaluation for all oncology patients. Cardiology referral was opined for all patients before starting anticancer therapy by 24%, while 10% believed in referral only for cardiac symptoms/toxicity and 66% believed in referral only before starting cardio-toxic therapy. Only 21% were confident of using potentially cardio-toxic treatment in patients with high risk of cardiotoxicity or suboptimal cardiac function. 67% expressed need for separate guidelines for CTRCD in lower-middle-income countries (LMIC) and 70% strongly agreed for separate cardio-oncology clinic to improve patient outcomes. Conclusions: Despite significant growth in field of cardio-oncology in West, there are significant gaps in KAP among health care providers in south-east Asia. Collaboration between cardiologists and oncologists and separate cardio-oncology guidelines will improve KAP of cardio-oncology and consequentially the patient outcomes, for south-east Asian region and LMIC.

BACKGROUND:Effective risk stratification is essential for guiding treatment decisions in patients with metastatic renal cell carcinoma (mRCC). The Meet-URO score is a novel prognostic model that integrates the International Metastatic RCC Database Consortium (IMDC) criteria with neutrophil-to-lymphocyte ratio (NLR) and the presence of bone metastases. Developed in the immunotherapy era, it has demonstrated superior prognostic accuracy compared to the IMDC score across various clinical settings and treatment strategies. Its validation in the context of first-line immune-based combinations has been awaited. METHODS:External validation of Meet-URO was performed using a large retrospective real-world cohort of mRCC patients treated with first-line immune-based combinations. Secondary analyses included a comparison with the IMDC score for predicting overall survival (OS) and progression-free survival (PFS). Additionally, restricted mean survival time (RMST) was assessed. RESULTS:A total of 1,418 patients were included in the analysis: 54% received ICI-ICI regimen (nivolumab plus ipilimumab), while 46% received the ICI-TKI combination. At baseline, 52.5% of patients had an NLR ≥ 3.2, and 32% had bone metastases. After a median follow-up of 26.8 months, the median OS and median PFS were 34.7 and 11.3 months, respectively. Meet-URO demonstrated effective prognostic stratification, identifying patient groups with markedly different outcomes (median OS 11.5-51.4 months; 3-year OS 26-66%; RMST 20.0-42.8 months). Compared to IMDC, Meet-URO showed a significantly better OS (c-index 0.675 vs 0.643; Δc = 0.032, P < .001) and PFS (c-index 0.60 vs 0.58; P < .001) prediction performance. CONCLUSIONS:Meet-URO demonstrated robust prognostic accuracy. Its integration into routine clinical practice and use as a stratification factor in clinical trials may support more personalized treatment strategies and enhance clinical trial design.

Background: Prognostic assessment in metastatic castration-resistant prostate cancer (mCRPC) remains important, but metastatic burden is not routinely captured as a simple combined measure. This exploratory study developed the Metastatic Integrated Risk Assessment (MIRA) score using disease burden and routine variables. Methods: Anonymised placebo-arm individual patient data from D4320C00014/ENTHUSE-M1 were analysed. Baseline variables were assessed for association with overall survival (OS) using Cox regression. A hierarchical composite tumour-burden variable combined bone-metastasis categories with RECIST target-lesion presence and sum of longest diameters (SLD). Selected variables formed an additive score, stratified into three risk groups, and were compared with the reconstructed Halabi classification using Harrell’s C-index. The final multivariate Cox model underwent internal validation using 1000 bootstrap resamples. Results: Among 266 placebo-treated patients, 133 deaths occurred, and median OS was 22.21 months. Composite tumour burden showed the strongest prognostic association and remained independently associated with OS. MIRA risk groups showed clear OS separation. In this derivation cohort, MIRA showed higher apparent discrimination than the reconstructed Halabi classification, with C-indices of 0.755 and 0.646, respectively; the paired difference was statistically significant (p < 0.001). For the final multivariate Cox model, the apparent C-index was 0.779, and the optimism-corrected C-index was 0.760; the optimism-corrected calibration slope was 0.841. Conclusions: These exploratory findings support metastatic tumour burden as a relevant prognostic factor in mCRPC and suggest that a simple combined measure of skeletal and measurable soft-tissue disease may have value for risk stratification. Refinement and independent external validation of MIRA are required before clinical use.

1085 Background: Cyclin-dependent kinase 4/6 inhibitors (CDK4/6i), in combination with endocrine therapy, are the mainstay for HR+/HER- metastatic breast cancer (MBC). However, ethnically diverse populations remain underrepresented in randomised clinical trials, limiting the ability to stratify by ethnicity. This meta-analysis utilises real-world evidence to compare treatment endpoints i.e., overall survival (OS), progression-free survival (PFS), objective response rate (ORR) and adverse effects (AE) across ethnic groups. Methods: 116 studies from MEDLINE and Embase were included (Palbociclib: 67, Abemaciclib: 29, Ribociclib: 20). 31580 HR+/HER- MBC patients (Asian [AS]:22067; White [WH]: 8993; Black [BL]: 510) were included. Median survival and response rates were calculated using a random effects model to account for between-study variability. Parentheses represent 95% confidence intervals (CI). Results: Pooled analysis shows disparities in efficacy and significant variation in haematological toxicity. Maximum mOS benefit for AS was on palbociclib (61.8mo) whereas for WH was ribociclib (63.9mo), primarily driven by long-term follow-up in MONALEESA. mPFS benefit was longest in BL patients on abemaciclib (17mo), WH patients on palbociclib (27.7mo), AS patients on ribociclib (25.2mo). ORR was highest for ribociclib across all ethnic groups (Table). Meta-regression indicated ethnicity was not a significant independent moderator of AE risk (p > 0.05). AS cohorts demonstrated the highest incidence of Grade 3+ neutropenia across all agents: Palbociclib (72.0%; 49.5–87.1), Abemaciclib (52.8%; 46.1-59.4), and Ribociclib (45.8%; 39.1-52.5; p < 0.0001). In contrast, WH and BL patients exhibited significantly lower G3+ neutropenia rates, ranging from 12.0% to 33.8% (p < 0.0001). AS patients also showed the highest rates of leukopenia and anaemia across all CDK4/6i. Conclusions: Our large-scale meta-analysis findings establish a novel, evidence-based ethnicity-informed CDK4/6i selection framework to optimize treatment endpoints in HR+/HER2- MBC patients. This highlights the urgent need for prospective trials focused on underrepresented populations to close the survival gap. Metric Subgroup Palbociclib Abemaciclib Ribociclib ORR (%) AS 34.0 (27-42) 28.7 (16.1-41.3) 52.4 (46.4-58.5) WH 34.0 (28-65) 41.9 (14.9-68.9) 44.9 (32.1-57.7) BL 25.0 (17-23) N/A 34.6 mPFS (mo) AS 21.0 13.0 25.2 WH 27.7 21.8 24.1 BL 14.1 17.0 10.8 mOS (mo) AS 61.8 25.2 58.7 WH 60.5 37.6 63.9

Metastatic renal cell carcinoma remains clinically challenging because of heterogeneous outcomes and limited predictive biomarkers for immunotherapy. We performed an explainable machine learning analysis using data from the multicenter retrospective Meet-URO 15 study, including 571 patients with metastatic renal cell carcinoma treated with second-line or later nivolumab. Clinical and inflammatory variables were used to develop classification models for disease control rate, progression-free survival at 3 and 9 months, and overall survival at 6, 18 and 24 months, as well as survival models for continuous progression-free and overall survival. Model performance was assessed using weighted F1-score for classification and concordance index for survival analysis, with interpretability provided through Shapley additive explanations. The best classification performance was observed for 6-month overall survival using a support vector machine model combined with minimum redundancy maximum relevance feature selection, achieving an F1-score of 0.81 on the test set and 0.77 in external validation. In survival analysis, random survival forest achieved a test-set concordance index of 0.68 for overall survival. Inflammatory indices, IMDC score, hemoglobin, lymphocytes and platelets consistently emerged as relevant prognostic features. These findings support explainable machine learning as a transparent approach to refine outcome prediction in immunotherapy-treated metastatic renal cell carcinoma.

The management of advanced prostate cancer has evolved significantly with the advent of androgen receptor-targeted agents (ARTAs) and docetaxel, which now form the cornerstone of first-line systemic therapy in advanced disease. However, as increasing numbers of patients progress after exposure to both an ARTA and docetaxel, optimal treatment sequencing in the post-ARTA, post-docetaxel metastatic castration-resistant prostate cancer (mCRPC) setting remains an area of clinical uncertainty, with limited comparative data to guide decisions. This narrative review synthesises the current evidence surrounding treatment strategies for mCRPC in this later-line setting. Available therapeutic options include radioligand therapy (e.g. lutetium-177–PSMA-617), cabazitaxel, PARP inhibitors (particularly for patients with DNA repair gene alterations), second-line ARTAs, further chemotherapy and immunotherapeutic approaches. However, direct head-to-head trials comparing these modalities are sparse, and much of the available data comes from retrospective or subgroup analyses, necessitating a nuanced interpretation of outcomes and limitations. Beyond efficacy, this article emphasises the critical role of practical and individualised considerations in therapy selection. These include toxicity profiles, patient comorbidities and preferences, biomarker-driven stratification (e.g. BRCA status, PSMA expression, sites of metastases), and health system factors such as drug availability and reimbursement policies, which may significantly influence access to optimal care. Finally, the review provides an overview of promising emerging therapies poised to expand the treatment landscape for mCRPC. These include bispecific T cell engagers, androgen receptor degraders, and antibody–drug conjugates, which are currently under investigation in clinical trials and may soon offer new avenues for treatment beyond traditional mechanisms.

e21512 Background: Immunotherapy is the mainstay for metastatic melanoma (MM), yet ethnic differences in treatment outcomes remain underexplored courtesy due to inconsistent reporting. Our systematic review and meta-analysis assessed overall survival (OS), progression-free survival (PFS), and objective response rate (ORR) among BRAF wild-type/undetermined MM patients receiving anti-PD1 monotherapy (nivolumab or pembrolizumab) or dual immune checkpoint inhibition (nivolumab plus ipilimumab), focusing on ethnicity based benchmarking. Methods: PubMed, Embase, and Cochrane Central were searched for real world studies reporting outcomes in adults with BRAF wild-type or undetermined MM treated with first-line anti-PD1 monotherapy or dual checkpoint inhibition. Meta-analysis using a random-effects model was conducted for the aforesaid treatment endpoints. Results: 2,659 patients from 8 studies were included for analysis. Median ages ranged from 62 to 75 years. Majority of patients had good performance status (ECOG 0-1). OS and PFS were summarised qualitatively due to limited hazard ratio data. Median OS for anti-PD1 monotherapy (n=595) was 31.5 months, with Asians demonstrating lower survival (28.1 months) compared with White patients (32.5 months). Dual checkpoint inhibition (n=417) was associated with a longer median OS (58.3 months), although markedly shorter among Asians (12 months) relative to Whites (59.2 months). Median PFS followed a similar pattern, with anti-PD1 monotherapy yielding 7.6 months overall (Asians 6.1; Whites 8.2) and dual therapy 17.8 months overall (Asians 8.2; Whites 18.8). Meta-analysis of ORR included 1,757 patients, demonstrating a higher pooled response for dual therapy (52%) than for monotherapy (36%). Among White patients, dual therapy achieved higher ORR (57.2%) (n=682) than monotherapy (36.9%) (n=902), whereas Asians had comparable responses to either regimen [(37.5% for dual (n=40); 35.3% for monotherapy (n=144)]. Differences between ethnic subgroups did not reach statistical significance. No eligible studies reported outcomes for Black or mixed-ethnicity populations, precluding quantitative or qualitative analyses in these groups. Conclusions: Dual checkpoint inhibition confers superior ORR compared with anti-PD1 monotherapy, particularly in White patients, whereas response among Asian patients appears similar across treatments. Across included studies, dual therapy was associated with a longer median OS of 58.3 months versus a median OS of 31.5 months with anti-PD1 monotherapy, dual therapy also demonstrates improved PFS of 17.8 months as compared to 7.6 months shown with monotherapy. Limited data on Black and Mixed ethnicities highlights the need for larger, ethnically diverse studies to guide optimised, personalised treatment.

e13673 Background: Multi-Disciplinary Team Meetings (MDTMs) are “gold standard” in the UK Cancer Care continuum. Multiple, fragmented systems complicate MDTM workflows due to lack of data integration and system coordination. A proof-of-concept exploratory, comparative analysis of a multiple Large Language Model (LLM) benchmarked Oncology Intelligence Platform was conducted prospectively for treatment decision making support in the Breast Cancer MDTM. The aim was to identify the most sustainable and scalable, real-world LLM, i.e., with maximal accuracy, minimal hallucination rate, and manageable GPU (Graphics Processing Unit) usage. Methods: A retrospective validation dataset of 225 matched National Health Service England (NHSE) radiology and histopathology text reports from randomised, heterogenous, confirmed and/or suspicious breast cancer patients with a wide range of disease stage (I to IV/early to advanced) and objective disease characteristic findings was curated from 2018 to 2024. This was a mixture of primarily unstructured, structured, unimodal, multi-source data. 125/225 were radiology consisting of Ultrasound, Mammogram, MRI Breast, CT and Bone Scans whereas the rest 100 were histopathology and/or supplementary reports. OncoflowTM, an Artificial Intelligence (AI) powered Cancer MDTM Coordinator CoPilot tool, was implemented. This software platform deployed multiple LLMs synchronously for treatment matching, i.e., mapping the objective output of data extracted from reports to recommendations from clinician preferred knowledge sources.These involved clinical practice guidelines namely European Society of Medical Oncology (ESMO), National Institute for Health and Care Excellence (NICE), Breast Systemic Anti-Cancer Therapy (SACT) Protocols from the Clatterbridge NHS Cancer Centre. Results: 6 LLMs were benchmark tested in a sandbox environment. LLM 1 and 4 were Base/Foundational models, Domain Specific (Medical) and Proprietary whereas the rest were Instruction tuned, Open Source with general purpose and multilinguality. LLM 4 was Autoencoding (Encoder only) whereas the rest were Autoregressive (Decoder only). All these LLMs were fine-tuned and a proprietary data processing strategy was used to enhance the models towards task suitability. LLM 2 showed Guideline (ESMO/NICE) Accuracy rate of 92%, SACT Protocol Accuracy Rate of 100%, hallucination rate of 3.4% and GPU usage of 18GB, outperforming others as the most preferred. Conclusions: The rising cancer incidence and rapidly evolving evidence-based treatment decision making needs have resulted in unsurmountable MDTM pressures. The 10 year health plan envisions to “make the NHS the most AI-enabled health system” “with AI seamlessly integrated into clinical pathways”. This will in turn invite incorporation into real world clinical workflows via electronic health record integration.

e13669 Background: Cancer Multi-Disciplinary Team Meetings (MDTMs) are central to UK Cancer Pathways irrespective of patient case complexities. A major bottleneck for the MDT is time-consuming, laborious manual review of clinical summaries and investigation reports to prepare MDTM cases. Large Language Models (LLMs) can extract critical, structured information from this vast, complex unstructured text reservoir. After a multiple LLM benchmark testing model in an internal sandbox environment as part of an Oncology Intelligence Platform, the most optimal LLM (maximal accuracy, minimal hallucinations and Graphics Processing Unit usage) was deployed for a Data Extraction task to prepare cases for the Breast Cancer MDTM. This was a prospective, external validation experience to highlight LLM performance. Methods: A retrospective dataset consisting of structured and unstructured radiology investigation text reports of confirmed breast cancer patients from the Barts Health NHS Trust Data Platform was obtained from 2018 to 2024. These reports were multisource including regional (Mammogram, Ultrasound and MRI Breast) and systemic scans (Staging/Response Assessment CT Chest Abdomen Pelvis and Bone Scans). An Artificial Intelligence (AI) powered Cancer MDTM CoPilot software platform (OncoflowTM) was used on this data to perform strategic extraction to a set of defined objective parameters, including clinical TNM (tumour-node-metastasis) classification points. Results: 165 aforementioned reports of varying disease stages (I to IV) were prospectively processed by OncoFlow’s fine tuned, cancer data extraction task specific LLM. This LLM was an open source, domain-specific, multilingual, instruction-tuned (having undergone distillation and reinforcement learning), autoregressive transformer model. There were 21 extraction features. These were divided into 3 Tiers based on data types - T1a (continuous numeric), T1b (discrete ordinal), T2 (categorical with intrinsic order), T3 (free text) comprising 2, 4, 3, 12 parameters respectively. Performance metrics for T1 features used Mean Absolute Error (MAE), which ranged from 96 to 99% for T1a and 74 to 93% for T1b. T2 being multi-class, used F1 scores, i.e., Micro-F1 (model performance on whole dataset/all classes) ranging 0.8 - 0.9 and Macro-F1 (average model performance across each class) ranging 0.6 - 0.8. Token-level F1 score, measuring precision and recall, was used in model performance for T3 parameters. This ranged from 78 to 95%. Exact match accuracy for the aforesaid was 62 to 93%. Conclusions: The LLM achieved robust, clinically relevant accuracy scores across all data tiers. The reliable scores showcase the model’s readiness to streamline and standardise MDTM case preparations.

Bladder cancer remains a major global health challenge, characterized by diagnostic uncertainty, substantial treatment costs and high recurrence rates. Current diagnostic and treatment modalities, including cystoscopy, transurethral resection of bladder tumour and standard histopathology, have limitations, including the inability to detect flat lesions, frequent understaging and interobserver variability, highlighting a crucial need for improved approaches. Advances in artificial intelligence (AI), blue-light cystoscopy, narrow-band imaging, cytology and urinary markers show promise in enhancing early detection and diagnosis. Developments in multiparametric MRI, radiomics, genomics and AI-driven algorithms for histopathological analyses have demonstrated considerable improvements in staging and risk stratification of bladder tumours, enabling personalized therapy selection and prognostication. Despite these promising developments, challenges remain regarding standardization, external validation, cost-effectiveness and ethical considerations in clinical implementation. Future research should prioritize addressing these barriers through collaborative, multi-institutional studies and robust validation frameworks. Ultimately, adopting a comprehensive multimodal strategy, such as proposed, novel, multimodal decision-making frameworks in which these advances and technologies are integrated, promises to considerably advance precision oncology in bladder cancer, improving patient outcomes and reducing health care burdens. In this Review, the authors describe and discuss how advances in artificial intelligence, genomics, radiomics and cytology can be integrated into decision-making processes to improve the management of bladder cancer.

While neoadjuvant chemotherapy and radical surgery represent mainstay treatments for muscle-invasive urothelial carcinoma (MIUC), recurrence with lethal metastasis remains high and highlights the need for adjuvant therapies in MIUC, like immunotherapy, already in use for metastatic UC with favourable results. This review provides an overview of the key clinical trials investigating adjuvant immune checkpoint inhibitors (ICIs) for MIUC and their clinical implications; in particular, examining factors that may be relevant in guiding adjuvant therapy to patients, such as ICI therapy choice, tumour subtype and the clinical utility of biomarkers, specifically PD-L1 status and circulating tumour DNA (ctDNA). Of the three key trials CheckMate-274, IMvigor010 and AMBASSADOR, anti-PD1 inhibitors nivolumab and pembrolizumab have shown statistically significant improvements in disease-free survival (DFS) compared to controls, highlighting their promising use as seen with nivolumab's approval into clinical practice. Results are conflicting on the association of PD-L1 tumour expression with treatment outcome, yet ctDNA has emerged as a key biomarker from IMvigor010, showing not only prognostic value but also association between its clearance and atezolizumab treatment benefit derived. Notably, it remains uncertain across the trials whether adjuvant treatment efficacy differs by tumour origin (upper tract or lower tract disease), and larger subgroup numbers are needed for future trials in order to adequately assess statistical significance of speculative associations. Altogether, study findings support increasing clinical incorporation of adjuvant nivolumab and pembrolizumab into the MIUC setting, and emphasise ctDNA's utility as a biomarker for MIUC, demonstrating roles in both prognostication and prediction of treatment benefit.

e12568 Background: Accurate assessment of breast cancer biomarkers (HR-ER/PR, HER2, Ki-67 index) is essential for prognostication and treatment selection but is limited by inter-observer variability and subjective interpretation. The emergence of HER2-low disease and quantitative thresholds has highlighted limitations in conventional pathology. Artificial Intelligence (AI) has potential to improve consistency and infer molecular phenotypes, however performance relative to human pathologists remains unclear. We conducted a meta-analysis to compare AI and human performance in biomarker assessment. Methods: 670 MEDLINE records evaluating AI-based diagnostic frameworks in breast cancer pathology were identified (PROSPERO ID: CRD420251161868). 9/52 studies directly comparing AI with human pathologists in biomarker assessment were included. Data extracted included study characteristics, biomarkers assessed, AI methodology, comparator performance, and diagnostic metrics, including accuracy, sensitivity, specificity, and area under the curve (AUC). Results: Analysis of nine studies encompassing 2,575 patients reveals that AI-driven biomarker assessment significantly enhances diagnostic precision in breast oncology. Evaluating Hormone Receptors, HER2, and Ki-67, AI frameworks achieved a mean diagnostic accuracy of 92.0% (range: 85.0%–97.8%), consistently outperforming human benchmarks, which yielded a mean accuracy of 82.2% (range: 60.0%–85.3%). Head-to-head comparisons underscored AI's capacity to mitigate inter-observer variability, particularly in HER2 scoring. In this domain, AI achieved 92.1% accuracy against consensus standards, exceeding manual reads (85.3%) and demonstrating superior inter-rater reliability (Cohen's κ = 0.84) compared to human evaluators (0.675). In Ki-67 quantification, AI demonstrated high expert concordance (Spearman's ρ = 0.745-0.861), providing crucial standardisation at clinical thresholds. The relationship between sample size (n = 12 to 1,341) and performance remained remarkably robust; mean accuracy varied by only ±3.5% across heterogeneous cohorts. While AUC reporting was limited to 0.71-0.72, the data confirm that AI frameworks reliably resolve "borderline" cases, such as the HER2 0 vs. 1+ distinction, effectively matching or exceeding traditional pathology standards. Conclusions: AI-based pathology frameworks demonstrate consistently high diagnostic performance for breast cancer molecular biomarker assessment compared with human pathologists. However, heterogeneity in reported metrics and limited use of molecular reference standards highlight the need for standardised validation and reporting frameworks to support clinical translation.

Prostate cancer, a leading cause of cancer-related mortality among men, often presents challenges in accurate diagnosis and effective monitoring. This systematic review explores the potential of exosomal biomolecules as noninvasive biomarkers for the diagnosis, prognosis, and treatment response of prostate cancer. A thorough systematic literature search through online public databases (Medline via PubMed, Scopus, and Web of science) using structured search terms and screening using predefined eligibility criteria resulted in 137 studies that we analyzed in this systematic review. We evaluated the findings from these clinical studies, revealing that the load of exosomes in the blood and urine of prostate cancer patients, which includes microRNAs (miRNAs), proteins, and lipids, demonstrates disease-specific changes. It also shows that some exosomal markers can differentiate between malignant and benign hyperplasia of the prostate, predict disease aggressiveness, and monitor treatment efficacy. Notably, miRNA emerged as the most frequently studied biomolecule, demonstrating superior diagnostic potential compared to traditional methods like prostate-specific antigen (PSA) testing. The analysis also highlights the pressing need for a standardised analytic approach through multi-centre studies to validate the full potential of exosomal biomarkers for the diagnosis and monitoring of prostate cancer.

Urological cancers, including prostate, bladder, renal, and testicular cancers, present significant challenges in terms of incidence, mortality, and treatment resistance. Immunotherapy, particularly bispecific T-cell engagers (BiTEs), has emerged as a promising therapeutic strategy, targeting tumor-specific antigens to activate T cells and enhance anti-tumor immunity. BiTEs, such as pasotuxizumab for prostate cancer, and CD3 ×B7-H3 BiTE for bladder cancer, demonstrate potential in overcoming the limitations of traditional therapies and immune checkpoint inhibitors. This review explores the application of BiTEs in urological cancers, highlighting their clinical outcomes, challenges, and future prospects. Although BiTEs offer significant advantages, including selective T-cell activation and low-dose efficacy, obstacles such as on-target off-tumor toxicity, the immunosuppressive tumour microenvironment (TME), and immune-related adverse effects need to be addressed for broader clinical success. Combination therapies with immune checkpoint inhibitors and oncolytic viruses, as well as advancements in BiTE technology, are essential to improving treatment efficacy. The future of BiTEs in uro-oncology lies in overcoming current limitations, optimising therapeutic strategies, and expanding clinical trials to solidify their role in cancer immunotherapy.

Multiparametric MRI (mpMRI) is the cornerstone for diagnosing clinically significant prostate cancer (csPC), that is, International Society of Urological Pathology (ISUP) Grade ≥2. The European Association of Urology (EAU) recommends its upfront implementation in biopsy-naive patients to guide prostate biopsy decision making (PMID 38614820). However, significant inter- and intra-observer reporting variability affects patient outcomes. Our objective was to systematically review and evaluate the comparative performance of AI vs radiologists in diagnosing prostate cancer from pre-biopsy prostate mpMRI. Our systematic review [PROSPERO ID: CRD420251037432] included MEDLINE, PMC, EMBASE, SCOPUS, and COCHRANE databases. Searching primary research published in 2010 and beyond yielded 6,389 records. After screening 3,747 articles, 137 underwent full-text review, with 3 meeting inclusion criteria—2 from database searches and 1 from grey literature. A meta-analysis using R assessed the diagnostic performance of the AI models, with area under the curve (AUC) as the primary outcome. Three multi-center, multi-vendor intervention studies on prostate mpMRI compared AI characterization of csPC vs standard of care (SOC), that is, ≥2 radiologists performing Prostate Imaging-Reporting and Data Systems (PI-RADS) version ≥2 scoring. A meta-analysis, including 552 pre-biopsy patients, yielded pooled sensitivity 0.884 (95% CI, 0.75-0.98), specificity 0.681 (95% CI, 0.51-0.80), and AUC 0.837 (95% CI, 0.690-0.950) for the AI—principally supervised machine learning (ML) models. Model 1 (PMID 40016318) had a 95% sensitivity, 67% specificity and was non-inferior to SOC (AUC 0.91 vs 0.95; P = .044). Model 2 (PMID 37345961) reported 86%-91% sensitivity, 64%-75% specificity and was also non-inferior to SOC (comparable AUCs 0.82-0.86). Model 3 (PMID 33671533) showed 89% sensitivity and superiority over SOC (AUC 0.75 vs 0.47). Models 1 and 2 exhibited strong generalizability, with Model 2 aligning closely with PI-RADSv2 lesion characterization. The mpMRI-directed prostate biopsy pathway increases csPC detection and decreases PC negative biopsy rates. Adopting this implies a significant time and labor-intensive radiology workforce pressure. Our meta-analysis demonstrates how AI PC diagnostic accuracy is comparable to radiologists. This aids standardization, reduces diagnostic variability and radiologist workload.