Christopher J. Weight

Prostate cancer (CaP) remains a major cause of cancer death in men, underscoring the need for mechanistically novel therapies. We previously identified the action of the mitotic kinase citron (CIT) kinase as a key driver of CaP progression and promising therapeutic target, but selective CIT inhibitors are not available. We therefore asked whether a better understanding of the action of CIT substrates, which execute CIT kinase action in CaP cells, can lead to novel approaches and/or alternative targets to indirectly interfere with CIT’s activity during CaP progression. In view of the enrichment in functions in alternative splicing among CIT substrates, we focus on the RNA splicing regulator THRAP3 as a representative substrate. Direct THRAP3 phosphorylation by CIT determined THRAP3’s RNA binding to CIT-controlled pre–messenger RNA targets. Cross-linking and immunoprecipitation sequencing identified a subset of CIT-dependent THRAP3-bound genes that were preferentially involved in cell proliferation. THRAP3 RNA binding was enriched in genes that undergo alternative splicing during CaP progression. CIT-dependent THRAP3-bound transcripts stimulated the growth of CaP cell lines, xenografts, patient-derived organoids, and patient-derived xenograft (PDX)–derived organoids, whereas CIT-independent THRAP3-bound transcripts from the same gene did not. An antisense oligo that blocks THRAP3’s recruitment to a CIT-dependent RNA binding site prevented the production of CIT- and THRAP3-dependent transcripts and inhibited the growth of cell lines and PDX-derived organoids representing different CaP stages and lineages but not of benign cells. Thus, CIT-mediated alternative splicing drives CaP progression, and its inhibition is a new therapeutic strategy in treatment-resistant CaP.

Objective: Histological grading is a key predictor of progression in non‐muscle‐invasive bladder cancer (NMIBC). The WHO 1973 three‐tier system is limited by heterogeneity within the G2 category, whereas the WHO 2004/2016 binary classification reduces prognostic resolution by grouping biologically diverse tumours into broad low‐grade (LG) and high‐grade (HG) categories. We aimed to evaluate the prognostic performance of hybrid grading systems compared with WHO 2004/2016, while further characterizing the histomorphological continuum of non‐invasive papillary urothelial carcinoma. Methods: Non‐invasive papillary urothelial carcinomas (2012–2023; n = 472) were reviewed and reclassified by two urological pathologists according to WHO 2004/2016, hybrid three‐tier and four‐tier systems, and an investigational five‐tier framework (Grades I, IIA, IIB, IIIA, IIIB) integrating architectural, cytological and proliferative features. Primary endpoints were recurrence, grade progression and T1+ stage progression, assessed using ROC analysis, Cox proportional hazards regression and model‐performance metrics (AIC, BIC, C‐index). Results: Increasing morphological atypia was associated with larger tumour size, greater multifocality and progressively higher progression risk, with events concentrated in Grades IIIA and IIIB. All systems showed limited recurrence discrimination (C‐index 0.54–0.56). For T1+ progression, discrimination improved with increasing stratification (AUC 0.754 for WHO 2004/2016 vs. 0.815, 0.828 and 0.827 for the hybrid three‐tier, four‐tier and five‐tier systems, respectively; all P < 0.001). The Hybrid 3‐Tier model achieved the most favourable AIC and BIC, while the five‐tier model yielded the highest C‐index (0.818). Conclusions: Hybrid grading systems improve progression‐risk stratification in non‐invasive papillary urothelial carcinoma compared with WHO 2004/2016, with most benefit achieved by the Hybrid 3‐Tier framework. These findings support the concept of a histomorphological continuum of progression risk and favour continued refinement of biologically informed hybrid grading for NMIBC risk stratification.

Background and Objective: Surgical complexity for renal tumors has traditionally been assessed using manual nephrometry scores, which require unreimbursed physician effort and are subject to interobserver variability. This study introduces an objective, fully automated alternative derived from decades of experience at a large academic center. Methods: We trained a CT classification model to predict whether a patient would ultimately undergo Partial or Radical Nephrectomy (PN or RN). We hypothesized that the model’s confidence in RN (termed the CLARITY score) would serve as a surrogate for the difficulty of nephron-sparing approaches and thus for tumor complexity. This hypothesis was tested using multivariate logistic regression for failure to achieve trifecta, estimated blood loss (EBL) ≥ 500 mL, and length of stay ≥ 3 d. CLARITY was compared with tumor size and R.E.N.A.L. score. External validation in a geographically distinct cohort was performed. Key Findings and Limitations: For predicting RN, CLARITY achieved an AUROC of 0.899 internally and 0.898 externally. In the external PN subgroup, it outperformed tumor size and R.E.N.A.L. score in predicting failure to achieve trifecta (AUROC 0.613), EBL ≥ 500 mL (0.727), and length of stay ≥ 3 d (0.673). In multivariable analysis, CLARITY remained associated with each outcome, whereas R.E.N.A.L. and size were not. This study is limited by its retrospective design. Conclusions and Clinical Implications: CLARITY is an automated CT-derived marker that quantifies renal tumor complexity more effectively than tumor size and R.E.N.A.L. score and may support scalable, objective preoperative complexity assessment. To support reproducibility and external validation, we have released a public inference pipeline and web-based DICOM upload portal for research use.

As artificial intelligence advances in medical image analysis, its environmental impact remains largely overlooked. This study analyzes the energy demands of AI workflows for medical image segmentation using the popular Kidney Tumor Segmentation-2019 (KiTS-19) dataset. It examines how training and inference differ in energy consumption, focusing on factors that influence resource usage, such as computational complexity, memory access, and I/O operations. To address these aspects, we evaluated three variants of convolution—Standard Convolution, Depthwise Convolution, and Group Convolution—combined with optimization techniques such as Mixed Precision and Gradient Accumulation. While training is energy-intensive, the recurring nature of inference often results in significantly higher cumulative energy consumption over a model’s life cycle. Depthwise Convolution with Mixed Precision achieves the lowest energy consumption during training while maintaining strong performance, making it the most energy-efficient configuration among those tested. In contrast, Group Convolution fails to achieve energy efficiency due to significant input/output overhead. These findings emphasize the need for GPU-centric strategies and energy-conscious AI practices, offering actionable guidance for designing scalable, sustainable innovation in medical image analysis.