You are currently viewing a new version of our website. To view the old version click .
MDPIMDPI
Search all articles
Current Oncology
Download PDF
  • Review
  • Open Access

6 January 2026

Machine Learning in Biomarker-Driven Precision Oncology: Automated Immunohistochemistry Scoring and Emerging Directions in Genitourinary Cancers

,
and
1
Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, BC V6T 1Z7, Canada
2
Pathology and Laboratory Medicine, BC Cancer, Vancouver, BC V5Z 4E6, Canada
*
Author to whom correspondence should be addressed.
Curr. Oncol.2026, 33(1), 31;https://doi.org/10.3390/curroncol33010031
This article belongs to the Section Genitourinary Oncology
Version Notes
Order Reprints

Simple Summary

Immunohistochemistry (IHC) is a common test used by pathologists to detect cancer biomarkers, which can help with diagnosis, prognosis, and treatment selection. However, IHC results can vary between laboratories and between observers. New digital pathology tools and artificial intelligence (AI), particularly machine learning (ML) techniques, can analyse stained tissue more consistently. This review gives an overview of how ML is being used to automate IHC scoring, first in well-studied biomarkers and then emerging biomarkers. This review then explores how these innovations can apply to genitourinary (GU) oncology, including prostate, renal, and bladder tumours, for which researchers have begun applying ML to new biomarkers that may predict outcomes or treatment response. ML use in IHC scoring is promising but requires more validation.

Abstract

Immunohistochemistry (IHC) is essential for diagnostic, prognostic, and predictive biomarker assessment in oncology, but manual interpretation is limited by subjectivity and inter-observer variability. Machine learning (ML), a computational subset of AI that allows algorithms to recognise patterns and learn from annotated datasets to make predictions or decisions, has led to advancements in digital pathology by supporting automated quantification of biomarker expression on whole-slide images (WSIs). This review evaluates the role of ML-assisted IHC scoring in the transition from validated biomarkers to the discovery of emerging prognostic and predictive IHC biomarkers for genitourinary (GU) tumours. Current applications include ML-based scoring of routinely used biomarkers such as ER/PR, HER2, mismatch repair (MMR) proteins, PD-L1, and Ki-67, demonstrating improved consistency and scalability. Emerging studies in GU cancers show that algorithms can quantify markers including androgen receptor (AR), PTEN, cytokeratins, Uroplakin II, Nectin-4 and immune checkpoint proteins, with early evidence indicating associations between ML-derived metrics and clinical outcomes. Important limitations remain, including limited availability of training datasets, variability in staining protocols, and regulatory challenges. Overall, ML-assisted IHC scoring is a reproducible and evolving approach that may support biomarker discovery and enhance precision GU oncology.

1. Introduction

Precision oncology depends on accurate biomarker detection and quantification to predict treatment response or determine patient eligibility for targeted therapies [1,2,3]. Immunohistochemistry (IHC) is traditionally assessed by visual scoring, which faces limitations in precision due to subjective threshold interpretation, inter-observer variability, and poor scalability [4,5,6,7]. Such constraints have inspired innovations in digital pathology and artificial intelligence (AI) to streamline clinical biomarker quantification [5,8].
Though liquid biopsy assays such as those detecting circulating tumour DNA are gaining traction in real-time, prognostic detection of disease burden and recurrence [9,10,11,12,13,14], their predictive role in guiding treatment decisions is limited and largely complementary to tissue-based testing [10,13,14]. IHC also provides spatial context and insights into tissue architecture and tumour–immune microenvironment interactions that correlate with certain tumour behaviours and treatment responses [15]. Moreover, as an accessible, relatively cost-effective method [6,16,17], IHC remains a clinical benchmark for diagnostic biomarker assessment [18,19,20,21,22], and is increasingly explored for prognostic and predictive applications [4,16,17,23].
Advances in whole-slide imaging, combined with machine learning (ML), now allow automated, reproducible IHC scoring, which is progressively adopted in molecular cancer research [24,25,26,27,28,29]. Deployment of automated scoring pipelines for some biomarkers has demonstrated strong concordance with expert pathologist scoring, achieving reliable clinical integration in breast [30,31,32,33,34,35,36] and lung [17,37,38,39,40,41,42] cancers. Beyond breast and lung oncology, AI-enabled IHC quantification is expanding, exploring potential applications in other common malignancies [5,43,44], notably GU cancers [9,45,46], whose heterogeneity has made prognostic and predictive biomarker panels difficult to standardise [9,45,47,48,49].
Building on recent advancements in computational pathology, this narrative review explores how automated IHC scoring can be translated into clinical routine. We summarise current applications and the state of evidence for ML use on emerging IHC biomarkers, with emphasis on GU oncology.
While numerous reviews have addressed AI in digital pathology, most have prioritised H&E-based histopathology, algorithmic methodology, or technical performance metrics. This review specifically examines ML-assisted IHC as a quantitative biomarker platform that positions GU malignancies as a logical next translational frontier, building on validated IHC scoring paradigms from other tumour types to contextualise emerging applications for GU oncology. By anchoring emerging GU applications to mature, well-studied ML–IHC frameworks, this review aims to explore the translational readiness, limitations, and future potential of automated IHC scoring in biomarker-driven GU precision oncology.

2. Methods

This narrative, non-systematic review was conducted in two stages, aiming to inductively identify themes from influential, emerging literature [Supplementary Figure S1]. The search was dynamic and iterative, using forward citation tracing to track evolving research trajectories, and was not intended to exhaustively extract or quantitatively pool data from a static snapshot of published works.

2.1. Stage 1: Exploratory Literature-Mapping to Identify Foundational ML–IHC Themes

An exploratory search was conducted primarily in PubMed, with support from Google Scholar to identify foundational literature between 2000 and 2025 (e.g., landmark articles, reviews) on digital pathology, ML, IHC, and biomarker quantification in precision oncology. Search terms included combinations of “artificial intelligence” or “AI”, “machine learning”, “immunohistochemistry” or “IHC”, “digital pathology”, “biomarker”, quantif* (quantification), cancer, tumor/tumour, malignanc*, neoplas*, “treatment response”, “therapeutic response”, “computational pathology”, “quantitative pathology”, “whole slide imaging” or “WSI”, “digital image analysis” and “automated”.
The scope of the literature search for this narrative review was limited to articles in which IHC was the primary analytical method and for which ML-based quantification was applied directly for biomarker interpretation, prognostic assessment or therapeutic decision-making, beyond technical metrics or solely image-processing objectives.
Within these bounds, 40 search results were screened by title and abstract. Articles were included if they applied AI or ML to IHC image analysis or automated biomarker quantification. Reviews relevant to digital pathology workflows were also included. Studies focused solely on non-IHC modalities, such as H&E, immunofluorescence, liquid biopsy, radiology, and transcriptomics, were excluded. Research unrelated to human solid tumours, as well as publications from non-oncology disciplines (e.g., veterinary medicine, dentistry, nutrition), was also excluded.
After full-text review, 12 core articles were retained, and reference snowballing expanded the pool to 64 papers, of which 20 involved ML-based IHC quantification. From this literature-mapping, five biomarkers (ER/PR, HER2, MMR, PD-L1, and Ki-67) emerged inductively as the most recurrently studied ML–IHC targets [Supplementary Table S1]. These biomarkers are synthesised in Section 4 as validated biomarker exemplars of automated IHC scoring. Additional studies were then identified through reference chaining and forward “cited by” tracing.

2.2. Stage 2: Targeted Narrative Exploration of ML Use for IHC Biomarkers in GU Cancers

Building on Stage 1 findings, a targeted narrative search was performed to examine the recent applications of ML-enabled IHC scoring within GU cancers. To complement database querying, we additionally used citation tracing and forward “cited by” exploration from foundational papers to identify recent (2020–2025) representative studies extending ML–IHC frameworks to GU contexts.
Study inclusion criteria were restricted to articles investigating prostate, bladder, and renal malignancies that applied digital pathology or ML workflows for the quantification of IHC biomarkers associated with prognosis or therapeutic response. Articles were excluded from the synthesis if they relied solely on other modalities, such as H&E, radiomics, or transcriptomics, without an IHC component, or if they only used manual analysis.
Furthermore, we excluded studies focused exclusively on diagnostic classification, molecular subtyping, or technical metrics lacking translational relevance to clinical outcomes. Investigations detailing purely technical ML workflows, such as nuclear segmentation without biomarker scoring, and those utilising IHC merely as a confirmatory reference rather than the primary analytical target were also omitted. Stage 2 of the narrative search prioritised studies demonstrating translational relevance, emerging biomarker development, or method validation, rather than attempting to catalogue all GU-associated markers. The objective of this search stage was to highlight domains in which ML-assisted IHC is actively evolving within GU oncology, including early, pre-validation exploratory models.

3. IHC Scoring Variability and the Need for ML-Assisted Analysis

IHC is a technique that enables visualisation of protein expression within preserved tissue architecture, providing spatial information about biomarker distribution and tumour–microenvironment interactions [4,6,16]. In clinical and research settings, interpretation typically relies on semiquantitative scoring systems that assess localisation, staining intensity, and the proportion of positive cells [7,21,50,51]. Common IHC scoring approaches include ordinal scales, proportion-based metrics, and composite scoring systems such as Allred-type and Histologic Score (H-score) methods [7,21,50,51].
Even with established scoring systems, reproducibility remains a challenge. IHC scoring cut-offs can vary across laboratories due to differences in staining protocols, tissue processing, antibody clones, scanner calibration and pathologist interpretation [50,51,52,53]. Inter-observer discordance further increases with heterogeneity and borderline cases [50,51,52,53]. These limitations highlight the need for computational approaches that improve consistency and scalability in IHC interpretation [24,26,54] (Table 1).
Table 1. Clinical Validation Status and ML Compatibility of Common IHC Scoring Frameworks.
WSI scanning technology has enabled the digitisation of entire slides at diagnostic resolution, supporting remote review, archiving, and digital image analysis (DIA) [24,25,65,66]. Early IHC quantification tools based on pixel thresholds and colour deconvolution provided approximate scoring but struggled to generalise across staining and batch variation [24,29]. With increasing digital adoption, ML methods are positioned to enhance reliability in IHC scoring [8,67]. Deep learning frameworks such as convolutional neural networks (CNNs) and multiple instance learning (MIL) can extract cellular and spatial features directly from WSIs, enabling automated detection of nuclei, tissue compartments, and chromogenic signal patterns [28,54].
In current human-in-the-loop workflows, ML algorithms function as assistive tools to pre-segment nuclei, quantify staining intensities, and flag regions of interest, prior to expert pathologist review and judgement [26,54,68]. These workflows reduce manual burden and variability for pathologists, while their expert feedback is used to further refine model performance [24,27]. ML-based scoring increasingly demonstrates agreement with expert assessment and provides a foundation for scalable, automated biomarker quantification [26,38,54,60].
Validated IHC biomarkers with established scoring frameworks and annotated datasets can thus serve as exemplars for the development of automated pipelines [26,51,52]. The next section highlights how these established biomarkers have been leveraged to demonstrate the feasibility and translational readiness of ML-based IHC quantification.

4. Validated Biomarkers as Models for ML-Based IHC Interpretation

Beyond diagnostics, some widely adopted biomarkers have established the prognostic and predictive utility of IHC in guiding precise treatment decisions (Table 2). These biomarkers also have well-annotated datasets and established scoring frameworks that allow ML models to be trained, benchmarked and validated [8,69,70,71,72] in ways that are still being explored for many emerging IHC biomarkers. Building from these established markers, ML aims to standardise IHC scoring and enhance the reproducibility of digital pathology for quantitative, precision oncology.
Table 2. Clinical Integration of ML-Driven Scoring Workflows for Validated IHC Biomarkers.

4.1. Hormone-Receptor IHC

Estrogen and progesterone receptors (ER/PRs) are nuclear IHC biomarkers that define hormone-receptor-positive breast cancers and determine eligibility for endocrine treatments such as tamoxifen, aromatase inhibitors, or fulvestrant [55,75]. Clinical evidence indicates that higher Allred scores can predict better outcomes and greater benefit from endocrine therapy [55,75]. For example, higher Allred scores in breast cancer patients have been associated with better five-year survival [55]. Traditional visual scoring of ER/PRs by pathologists is limited by interpretive variability, particularly in low-positive and heterogeneous cases, leading to inconsistencies in the quantification of nuclear staining [7,22,76]. ML-based scoring pipelines automating intensity thresholding and nuclear segmentation have shown close agreement with expert scoring in retrospective datasets, with improved inter-observer concordance and reduced turnaround time, despite differences in staining and scanning equipment [26,27]. Although ER/PR testing is specific to breast cancer [56], these validated frameworks provide a reference for standardising the semi-quantitative scoring of other nuclear biomarkers. For example, similar intensity-proportion scoring pipelines have been applied to nuclear biomarkers in bladder cancers for subtyping and prognostication [77,78,79,80]. These applications are revisited in Section 5.3.

4.2. Growth-Receptor IHC

ML-assisted quantification in ER/PR analysis has similarly extended to Human Epidermal Growth Factor Receptor 2 (HER2) IHC, a membranous biomarker whose scoring is used for determining eligibility for anti-HER2 therapy [57]. HER2 IHC assesses membranous expression using an ordinal scale (0, 1+, 2+, 3+) with confirmatory in situ hybridisation (ISH) for 2+ cases [56,57]. Distinguishing between HER2-negative (0) and HER2-low (1+) tumours has become increasingly important, as HER2-low cases comprise more than 50% of breast cancers and HER2-low status serves as a clinically validated, predictive biomarker of benefit from trastuzumab deruxtecan treatment [58,81]. Visually making this distinction, however, can remain challenging even for experts [57].
Now, ML models trained on curated HER2 datasets have achieved area-under-the-curve (AUC) values above 0.9 for differentiating HER2-low status from negative, highlighting the technical feasibility of reproducible assessment for weak or incomplete membranous stains [28,57]. Before routine adoption, however, such models must demonstrate robustness across antibody clones and laboratories [50,56]. The methodological principles of HER2 quantification are relevant to membranous and cytoplasmic biomarkers in GU oncology, with emerging prognostic and predictive value [77,80].

4.3. Mismatch-Repair (MMR) IHC

While HER2 and ER/PR IHC quantification is built on ordinal and continuous scoring schemes, ML-based approaches have likewise been applied to binary testing defined by loss or retention, such as MMR proteins [59,82,83]. MMR IHC is used to screen for microsatellite instability (MSI), a predictive marker of response to immune checkpoint inhibitors (ICIs) targeting immune cell Programmed Death receptor-1 (PD1), such as pembrolizumab and nivolumab [59,82,83,84]. Loss of nuclear staining for MLH1, PMS2, MSH2 or MSH6 indicates deficient mismatch repair (dMMR), whereas preserved expression of MMR defines proficiency (pMMR) [83]. MMR IHC is routinely performed on colorectal and endometrial carcinomas as part of Lynch syndrome screening and MSI assessment, with well-defined clinical thresholds [59,83,85]. Though its interpretation in clinical routine remains manual, CNN-based nuclear segmentation and classification models have shown concordance with expert identification of nuclear MMR loss vs. retention [73,86]. Validation studies have reported concordance rates above 90% in classifying tumours on WSIs as dMMR or pMMR [74,87].
ML automation of MMR loss detection highlights the potential of integrating ML-assisted workflows for high-throughput nuclear stain assessment. While MSI status is best understood as a key biomarker in colorectal and endometrial cancers, its potential role as a predictive biomarker for ICI response is now being explored within GU oncology workflows [88,89,90,91]. MMR IHC is an area of research interest in prostate cancers and upper tract urothelial carcinomas (UTUCs), for which dMMR or MSI-high status could inform hereditary cancer risk and predict response to ICI therapies [92,93,94,95]. MSI-high/dMMR status in UTUC in particular has been linked to Lynch-syndrome-related disease, supporting the use of MMR IHC in ICI selection, as well as flagging patients for germline testing [93,94,95]. Similarly, small retrospective studies have reported that dMMR prostate tumours, often with higher mutational burden, demonstrate durable response to PD-1 blockade (e.g., pembrolizumab) in select patients [92]. These applications of MSI-high/dMMR status require larger-scale validation in prostate tumours pending standardisation, with which AI-automated analysis may help, as revisited in Section 5.1.

4.4. Immune-Checkpoint IHC

While MMR IHC serves as a proxy for predicting checkpoint inhibitor response by genomic instability, IHC is also used directly on tumour-expressed Programmed Death Ligand-1 (PD-L1), which binds to the immune cell receptor PD-1 [39,59,84,96]. PD-L1 IHC is an established predictive biomarker of response to ICIs across tumour types, including non-small-cell lung cancers (NSCLCs) [64,97]. PD-L1 expression is assessed using one of three scoring systems: Tumour Proportion Score (TPS), Combined Positive Score (CPS) or Immune Cell (IC) score, with each system requiring specific spatial context and cell classification [97]. This complexity makes PD-L1 IHC challenging to quantify reproducibly by manual IHC interpretation around relevant cutoffs (e.g., CPS 1 or 10) [59,64]. ML approaches have accordingly been developed to automatically classify cell objects, segment tumour, immune and stromal compartments, detect 3,3′-Diaminobenzidine (DAB)-positive nuclei and membranes, and calculate TPS or CPS [39,64]. ML-based scoring of PD-L1 IHC has demonstrated improving agreement with expert consensus in validation studies, reporting intraclass correlation coefficients (ICCs) around the ~0.8 to 0.9 range, though agreement metrics vary by assay, tumour type, and scoring system. Inter-assay and inter-laboratory reproducibility remains a challenge for both manual and ML-based PD-L1 interpretation, though ML is likely to continue improving.
For example, a recent weakly supervised deep learning model using a 22C3 assay for PD-L1 tumour proportion scoring in NSCLC reported an ICC of 0.96 versus pathologist assessments in internal and external validation cohorts [63]. Beyond lung cancers, some immunotherapies have also been implemented in standard of care for renal and bladder cancers, in which immune infiltration and histologic heterogeneity add complexity to interpretation [64,98]. ML-based PD-L1 quantification thus highlights the potential of automated IHC scoring to extend beyond intensity thresholding toward context-aware, cell-type-specific biomarker analyses as ML models improve.

4.5. Proliferation IHC

Comparable ML-based approaches are underway for Ki-67 IHC, another established cancer biomarker [44]. Ki-67 is a nuclear marker of proliferation used for prognostication and grading in breast malignancies and neuroendocrine tumours (NETs) [8,27,52,99]. Higher Ki-67 indices correlate with more aggressive disease and worse outcomes [29,44,60,100,101,102]. In breast cancers, Ki-67 is used for the luminal A and luminal B phenotypes to guide adjuvant chemotherapy decisions [103,104]. In NETs, the Ki-67 index is part of the WHO G1–G3 grading system that directs therapy [61,105].
Manual counting of nuclear positivity on printed images is historically a common method of Ki-67 quantification, but this method can be labour-intensive and inconsistent. A reproducibility study found inter-laboratory intraclass correlations of 0.59 to 0.71 for Ki-67 visual scoring across eight laboratories [52]. In 2020, The International Ki-67 in Breast Cancer Working Group highlighted the difficulty of standardising Ki-67 IHC, noting that automated scoring could help overcome limitations [51]. Recent studies have accordingly attempted to develop and assess ML-based Ki-67 index calculation. A nuclear segmentation algorithm reported 93% accuracy in segmenting cell nuclei, with a 2.1% error margin for detecting Ki-67 positivity [62]. Another study showed that automated analysis provided greater specificity, concordance and prognostic correlation compared to manual scoring for Ki-67 in breast cancer, with 10% fewer misclassifications, 5.5% more concordance and a significantly higher prognostic value for overall survival (p = 0.044), as measured with the Likelihood Ratio Chi-square [27].
Automated pipelines built on deep learning nuclear segmentation architectures such as U-Net, Hover-Net and StarDist enable reliable nuclear identification and classification of cell positivity within region-of-interest selections, with stromal and necrotic exclusion [54,106]. ML models using deep learning nuclear segmentation can generate reproducible Ki-67 indices with strong correlation with pathologist counts (Pearson r > 0.9), while substantially reducing analysis time [8,102]. In GU malignancies, Ki-67 positivity has been associated with recurrence risk in prostate cancer and muscle invasiveness in bladder cancer [8,29,102], though clinical cut-offs remain less standardised [8,102]. Algorithmic quantification could facilitate reproducibility in multi-centre trials if integrated with harmonised protocols [8,44,85].

4.6. From Automated Scoring to Clinical Translation

Across ER/PR, HER2, MMR, PD-L1 and Ki-67, ML-based IHC quantification has achieved technical comparability with manual interpretation in some retrospective and multi-institutional validation studies. These validated IHC biomarkers illustrate the trajectory of ML in IHC: from reproducible IHC scoring towards integration of quantitative readouts into clinical decision-making. These translational exemplars provide a methodological and validatory foundation for extending automated scoring to additional biomarkers, including emerging prognostic and predictive GU biomarkers, as discussed in the following section.

5. Emerging Translational Applications of ML-Based IHC Scoring: An Evolving Frontier in GU Oncology

Applications of ML-assisted IHC quantification in GU oncology are a growing and evolving research focus [49]. GU tumours affecting the bladder, prostate, kidney, and related organs are highly heterogeneous, which has previously driven heavier research emphasis on diagnostic markers and molecular subtyping [49,80]. IHC biomarkers are central to routine diagnostic practice in GU malignancies, but robust prognostic and predictive panels have yet to be established [107,108]. Recent reviews have detailed the growing use of AI and digital pathology in GU malignancies, but many of these efforts have focused primarily on H&E-based histopathology and subtype classification [80,109].
Few studies have specifically examined the prognostic and predictive potential of ML-based IHC quantification across the broader spectrum of GU malignancies. Research has begun shifting from demonstrating quantitative accuracy to establishing meaningful thresholds for clinical utility. The emerging biomarkers discussed in this section describe how digital IHC is evolving from assisted quantification to prognostication and treatment response prediction, with implications in GU oncology (Table 3).
Table 3. ML Applications to Emerging Prognostic and Predictive GU IHC Biomarkers.

5.1. Prostate Cancer IHC

In prostate cancers, several biomarkers are under investigation for ML-based IHC interpretation. Automated scoring of androgen receptor (AR) expression is being explored for its potential association with lineage plasticity and treatment resistance. Phosphatase and tensin homolog (PTEN) loss assessed by binary IHC scoring, which has prognostic implications and correlates with phosphatidylinositol 3-kinase (PI3K) signalling pathway activation, is currently being evaluated using ML-based nuclear segmentation [53]. p63, an established basal cell marker used diagnostically, is also increasingly being examined in digital workflows for its potential subtype relevance [77]. Prostate-specific membrane antigen (PSMA), which is highly expressed in prostate tumours and underlies radioligand therapy, is being digitally quantified in exploratory ML-based workflows to assess expression patterns relevant to therapeutic targeting. B7 Homolog 3 (B7-H3), an immune checkpoint protein explored as a tumour-associated target in prostate cancer, is likewise being evaluated in early ML-enabled quantification approaches [113,117]. Early ML pipelines show improving performance in analyses with membranous and cytoplasmic staining patterns [121].
In addition, ML-based automation of MMR IHC binary classification represents a logical next step to explore and validate the prognostic and predictive applications of dMMR or MSI-high status in prostate cancers. Recent published AI studies on MSI status in prostate tumours have primarily focused on predicting MMR status from H&E slides, with manual IHC as the reference standard. But ML-driven workflows aimed at standardising the confirmatory binary IHC test itself have already been validated in colorectal cancers [74,86,87]. In principle, similar pipelines can be adapted to MMR IHC in prostate cancers. Specifically, automated single-cell nuclear detection of MLH1, PMS2, MSH2 and MSH6 loss has achieved near-perfect concordance with expert assessment (AUC ≈ 0.98) in colorectal tumours [73,87]. Adapting validated models trained on colorectal histology can steer AI research beyond H&E to standardising binary IHC for prostate pathology, enabling reproducible outputs for robust prognostic and predictive clinical correlations.

5.2. Renal Cancer IHC

Renal cancers have seen some progress in ML-based IHC quantification, with emerging prognostic applications [112,114]. Quantitative IHC of immune markers in Renal Cell Carcinoma (RCC) specimens is currently being integrated into exploratory prognostic models of immunotherapy response [114]. ML-based frameworks for WSI IHC analysis are being developed for the automated detection of BRCA1-Associated Protein 1 (BAP1) loss, which has retrospectively been associated with higher tumour grade, increased metastatic potential, and poorer clinical outcomes in clear cell RCCs [113]. V-domain immunoglobulin suppressor of T cell activation (VISTA), another immune checkpoint protein expressed in subsets of RCC and studied for immunotherapy resistance phenomena, is emerging as a potential, exploratory ML-quantifiable IHC biomarker [112,114,115]. These prognostic associations in RCC IHC, however, are based on surrogate endpoints and lack prospective clinical validation [80]. Building on frameworks developed for PD-L1, other immune checkpoint markers such as B7-H3, T-cell immunoglobulin and mucin-domain containing-3 (TIM-3) and Lymphocyte Activation Gene 3 (LAG-3), along with VISTA, are also under active research as potential predictive biomarkers for immunotherapy response in renal cancers, as well as bladder cancers [115,116,117,122,123].

5.3. Bladder Cancer IHC

Among GU malignancies, urothelial carcinoma of the bladder has received considerable attention in computational AI and IHC pathology research, which investigates IHC biomarkers with emerging prognostic and predictive utility [44,48].
Cytokeratin (CK) profiling is an important element of molecular subtyping in bladder cancers. CK5/6 and CK14, which are often co-expressed with the basal/squamous nuclear marker p63, define basal-like subtypes associated with aggressive behaviour and increased sensitivity to platinum-based chemotherapy [49,77,80]. CK20, similar to GATA3, is a marker of luminal differentiation, which is typically linked to more favourable prognosis and response to immunotherapy [49,109].
ML-based quantification of CK expression through automated nuclear and cytoplasmic segmentation is being used in subtyping [77]. Multi-institutional studies employing tissue microarrays constructed from retrospective cohort specimens have shown that ML-quantified CK5/6 and CK20 expression strongly correlates with transcriptomic classifications and clinical outcomes. Similar ML-based workflows have been applied to predict neoadjuvant chemotherapy (NAC) response in muscle-invasive bladder cancer (MIBC), by associating ML-quantified CK and immune markers with clinical variables [124].
Uroplakin II (UPK II), another bladder-specific marker, is a highly specific marker of urothelial differentiation with diagnostic use in determining tumour tissue-of-origin, and is being explored for its emerging prognostic and predictive value [9]. Digital quantification of UPK II staining intensity and distribution is currently aimed at distinguishing between primary bladder carcinomas and metastases while refining molecular subtype assignments, but is also under investigation for potential correlations with prognosis and treatment response in retrospective analyses [80].
Some therapeutically actionable surface markers are also currently being studied in bladder cancers. Trophoblast cell-surface antigen 2 (Trop-2), a transmembrane glycoprotein, and Nectin-4, a cellular adhesion molecule abundant in urothelial carcinomas, are being explored as candidate predictive markers to investigate associations with risk stratification and response to antibody-drug conjugates (ADCs), e.g., sacituzumab govitecan and enfortumab vedotin, respectively [118]. For Nectin-4 in particular, quantitative Nectin-4 IHC has shown retrospective associations with clinical outcomes, though predictive validation for treatment response remains to be established [9,118,119,120]. While high Nectin-4 expression is frequently assessed in the context of enfortumab vedotin treatment, thresholding practices vary and are not yet globally standardised [9,118]. Greater Nectin-4 expression, however, has been reported in aggressive urothelial carcinoma tissue variants, suggesting potential prognostic relevance [49,113,120].
Independent studies have found that greater Nectin-4 expression correlates with papillary morphology, lower tumour grade, and earlier pathological stage, with more favourable outcomes associated with predominantly membranous staining [120]. Recently, a QuPath-based ML workflow using a Random Trees classifier for tumour-cell classification and automated H-score computation was applied to quantify Nectin-4 intensity and assess spatial distribution and prognostic correlations in MIBC [119]. This study reported that tumour-front-enriched expression was associated with more favourable disease-free survival (DFS), while tumour-centre downregulation correlated with higher recurrence risk [119]. These findings support hypothesis generation but require validation in prospective, treatment-stratified cohorts. Although the QuPath workflow was not designed to validate Nectin-4 as a prognostic or predictive biomarker directly, it highlights how ML-assisted IHC quantification can offer reproducibility and standardisation to support digital pathology workflows requiring robust quantification.
This application of ML demonstrates that automated IHC scoring can support efforts to capture the spatial biology of tumours, which may have prognostic associations with clinical surrogate endpoints such as DFS [119]. Nectin-4 expression also aligns with luminal differentiation markers and shows inverse associations with PD-L1 expression on tumour-infiltrating immune cells, suggesting its integration within broader immune-microenvironment contexts [120]. Specifically, ML-assisted quantitative IHC pipelines are being explored in Cluster of Differentiation (CD) marker analysis to support immune profiling and to map immune infiltration patterns within bladder cancer tissue [44,49,64]. ML-enhanced quantitative IHC can improve the reliability of immune infiltration metrics, which are currently being investigated for potential associations with outcomes in MIBC patients receiving trimodal therapies aimed at bladder preservation [44,49,125].

5.4. Toward Clinical Translation and Therapeutic Action in GU Precision Oncology

As summarised in Table 3, most associations with clinical outcomes and evidence for ML applications in GU IHC remain retrospective and exploratory, with only few applications in select biomarkers approaching early clinical evaluation. Even for therapeutically actionable target markers, existing ML-based IHC workflows primarily support hypothesis generation rather than treatment selection.
The current state of evidence demonstrates a translational gap between technical feasibility and clinical implementation, highlighting the potential of ML-based methods to support future integration of reproducible GU IHC biomarker scoring into clinical decision-making. Notable biological heterogeneity across GU tumour sites, histologic variants and treatment contexts also further challenges model generalisability, underscoring the need for tumour-specific validation.
Current efforts at automating GU cancer IHC biomarker quantification support the feasibility of large-scale retrospective and prospective validation studies to improve discrimination between treatment responders and non-responders. As ML-based IHC quantification matures, emerging efforts in GU oncology are positioned to accelerate translation, following the roadmap established by frameworks in breast, lung, and gastrointestinal oncology. The emerging and growing emphasis on GU tumours suggests that ML-based IHC quantification can move beyond reproducing manual reads toward prognostication and treatment response monitoring. Realistic next steps toward clinical translation would involve multi-centre reproducibility studies and consensus on scoring thresholds, particularly for borderline or continuous biomarkers, to demonstrate robustness across laboratories, scanning systems and staining platforms.

6. Future Directions

ML-assisted IHC is moving beyond proof-of-concept studies toward translational adoption [110,126,127]. Future progress will depend on achieving reproducible IHC quantification methods for various experts across institutions that use varying staining platforms and scanners. Developing harmonised validation criteria and regulatory guidelines for software as a medical device will be essential to ensure reliability, transparency, and safe integration into clinical routine [71,111].
A key barrier to widespread implementation is the limited availability of large, diverse, and well-annotated IHC datasets across tumour types [72,110,127]. Broader data sharing and federated learning approaches, together with the creation of public benchmark datasets similar to CAMELYON in digital breast pathology, could accelerate evaluation and regulatory review [72,110,127]. This pathway of ML model validation for clinical use has been demonstrated in colorectal oncology as well. Specifically, the success of the Immunoscore showed that spatial immune metrics can achieve clinical use [128,129,130] with multi-centre and prognostic validation, suggesting that similar ML-driven spatial IHC analysis may be valuable to GU oncology [44,48,112].
On this path forward, research will likely prioritise interpretability and calibrated confidence outputs to support clinical trust and decision-making [43,110,126,127]. Progress is expected toward multimodal frameworks that integrate IHC-derived spatial features with genomic, transcriptomic, and clinical data [17,70,127]. For example, evidence from the IMvigor010 and IMvigor011 trials suggested that combined assessment of dynamic biomarkers such as ctDNA with histologic features may refine adjuvant therapy selection after cystectomy [131,132]. Under appropriate regulatory oversight, such composite approaches to integrating automation into IHC pathology could ultimately support more robust prognostic and predictive applications of AI for precision oncology.

7. Conclusions

This review demonstrates that although ML-based IHC scoring has gained traction in breast, lung, and gastrointestinal cancer research, its application in GU oncology remains largely exploratory. By outlining persistent barriers such as interobserver scoring variability, limited benchmark datasets, and the absence of harmonised cut-offs, we identify where ML-driven quantification could improve reproducibility, enable finer spatial and phenotypic analysis than manual methods, and support the optimisation or discovery of clinical IHC biomarkers. The novelty of this review lies in positioning GU malignancies as the next logical frontier for ML applications in IHC pathology, drawing on validated precedent frameworks to demonstrate how AI automation can enhance research efforts at targeted therapy, with continued development of multi-centre validation and reproducibility standards. As GU biomarker research accelerates, ML-driven IHC analysis offers a feasible pathway toward standardisation and scalability in tissue-based testing, with the potential to bridge current gaps between histopathology and molecular profiling in the digital era of precision oncology.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/curroncol33010031/s1, Figure S1. Overview of the Two-Stage Thematic Literature Search Strategy for this Narrative Synthesis; Table S1. AI/ML-based IHC Biomarker Studies Identified in “Stage 1” of the Literature Search.

Author Contributions

Conceptualisation, M.Y.; methodology, M.Y. and I.-M.M.; data curation, M.Y.; writing—original draft preparation, M.Y.; writing—review and editing, I.-M.M. and G.W.; supervision, I.-M.M. and G.W. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Data Availability Statement

No new data were created or analysed in this study. Data sharing is not applicable to this article.

Acknowledgments

M.Y. was supported by the University of British Columbia (UBC) Department of Pathology and Laboratory Medicine Summer Student Fellowship Grant.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
AIArtificial Intelligence
ARAndrogen Receptor
ASCOAmerican Society of Clinical Oncology
AUCArea Under the Curve
BAP1BRCA1-Associated Protein 1
B7-H3B7 Homolog 3
CAPCollege of American Pathologists
CKCytokeratin
CPSCombined Positive Score
CNNConvolutional Neural Network
ctDNACirculating Tumour DNA
DAB3,3′-Diaminobenzidine
DFSDisease-Free Survival
dMMRDeficient Mismatch Repair
ER/PREstrogen Receptor/Progesterone Receptor
FFPEFormalin-Fixed Paraffin-Embedded
GUGenitourinary
HER2Human Epidermal Growth Factor Receptor 2
H&EHematoxylin and Eosin
H-scoreHistologic/Histochemical Score
ICCIntraclass Correlation Coefficient
ICIImmune Checkpoint Inhibitor
IHCImmunohistochemistry
ISHIn Situ Hybridisation
Ki-67Nuclear proliferation biomarker
LAG-3Lymphocyte Activation Gene 3
MILMultiple Instance Learning
MLMachine Learning
MLH1, PMS2, MSH2, MSH6(Mismatch repair proteins)
MMRMismatch Repair
MIBCMuscle-Invasive Bladder Cancer
MSIMicrosatellite Instability
NACNeoadjuvant Chemotherapy
NETsNeuroendocrine Tumours
NSCLCNon–Small Cell Lung Cancer
PD-1Programmed Death Receptor 1
PD-L1Programmed Death-Ligand 1
PI3KPhosphatidylinositol 3-Kinase
pMMRProficient Mismatch Repair
PSMAProstate-Specific Membrane Antigen
PTENPhosphatase and Tensin Homolog
RCCRenal Cell Carcinoma
TIM-3T-cell immunoglobulin and mucin-domain containing-3
TPSTumour Proportion Score
Trop-2Trophoblast cell-surface antigen 2
UPKIIUroplakin II
UTUCUpper Tract Urothelial Carcinoma
VISTAV-domain Immunoglobulin Suppressor of T-cell Activation
WSIWhole-Slide Imaging

References

  1. La Thangue, N.B.; Kerr, D.J. Predictive biomarkers: A paradigm shift towards personalized cancer medicine. Nat. Rev. Clin. Oncol. 2011, 8, 587–596. [Google Scholar] [CrossRef]
  2. Schmidt, K.T.; Chau, C.H.; Price, D.K.; Figg, W.D. Precision Oncology Medicine: The Clinical Relevance of Patient-Specific Biomarkers Used to Optimize Cancer Treatment. J. Clin. Pharmacol. 2016, 56, 1484–1499. [Google Scholar] [CrossRef]
  3. Duffy, M.J.; O’DOnovan, N.; McDermott, E.; Crown, J. Validated biomarkers: The key to precision treatment in patients with breast cancer. Breast 2016, 29, 192–201. [Google Scholar] [CrossRef]
  4. Kohale, M.G.; Dhobale, A.V.; Bankar, N.J.; Noman, O.; Hatgaonkar, K.; Mishra, V. Immunohistochemistry in pathology: A review. J. Cell. Biotechnol. 2023, 9, 131–138. [Google Scholar] [CrossRef]
  5. Wasinger, G.; Koeller, M.C.; Compérat, E. Pathology in the artificial intelligence era: Practical insights for immunohistochemistry and molecular pathology. Diagn. Histopathol. 2025, 31, 416–423. [Google Scholar] [CrossRef]
  6. Matos, L.L.; Trufelli, D.C.; De Matos, M.G.L.; da Silva Pinhal, M.A. Immunohistochemistry as an Important Tool in Biomarkers Detection and Clinical Practice. Biomark. Insights 2010, 5, 9–20. [Google Scholar] [CrossRef]
  7. Kim, S.-W.; Roh, J.; Park, C.-S. Immunohistochemistry for Pathologists: Protocols, Pitfalls, and Tips. J. Pathol. Transl. Med. 2016, 50, 411–418. [Google Scholar] [CrossRef]
  8. Acs, B.; Rantalainen, M.; Hartman, J. Artificial intelligence as the next step towards precision pathology. J. Intern. Med. 2020, 288, 62–81. [Google Scholar] [CrossRef]
  9. Mihai, I.M.; Wang, G. Biomarkers for predicting bladder cancer therapy response. Oncol. Res. Featur. Preclin. Clin. Cancer Ther. 2025, 33, 533–547. [Google Scholar] [CrossRef]
  10. Pantel, K.; Alix-Panabières, C. Minimal residual disease as a target for liquid biopsy in patients with solid tumours. Nat. Rev. Clin. Oncol. 2024, 22, 65–77. [Google Scholar] [CrossRef]
  11. Peng, Y.; Mei, W.; Ma, K.; Zeng, C. Circulating Tumor DNA and Minimal Residual Disease (MRD) in Solid Tumors: Current Horizons and Future Perspectives. Front. Oncol. 2021, 11, 763790. [Google Scholar] [CrossRef] [PubMed]
  12. Chin, R.-I.; Chen, K.; Usmani, A.; Chua, C.; Harris, P.K.; Binkley, M.S.; Azad, T.D.; Dudley, J.C.; Chaudhuri, A.A. Detection of Solid Tumor Molecular Residual Disease (MRD) Using Circulating Tumor DNA (ctDNA). Mol. Diagn. Ther. 2019, 23, 311–331. [Google Scholar] [CrossRef]
  13. Zhu, L.; Xu, R.; Yang, L.; Shi, W.; Zhang, Y.; Liu, J.; Li, X.; Zhou, J.; Bing, P. Minimal residual disease (MRD) detection in solid tumors using circulating tumor DNA: A systematic review. Front. Genet. 2023, 14, 1172108. [Google Scholar] [CrossRef]
  14. Hitchen, N.; Shahnam, A.; Tie, J. Circulating Tumor DNA: A Pan-Cancer Biomarker in Solid Tumors with Prognostic and Predictive Value. Annu. Rev. Med. 2025, 76, 207–223. [Google Scholar] [CrossRef]
  15. Hsieh, W.-C.; Budiarto, B.R.; Wang, Y.-F.; Lin, C.-Y.; Gwo, M.-C.; So, D.K.; Tzeng, Y.-S.; Chen, S.-Y. Spatial multi-omics analyses of the tumor immune microenvironment. J. Biomed. Sci. 2022, 29, 96. [Google Scholar] [CrossRef]
  16. Dunstan, R.W.; Wharton, K.A., Jr.; Quigley, C.; Lowe, A. The Use of Immunohistochemistry for Biomarker Assessment—Can It Compete with Other Technologies? Toxicol. Pathol. 2011, 39, 988–1002. [Google Scholar] [CrossRef] [PubMed]
  17. Poalelungi, D.G.; Neagu, A.I.; Fulga, A.; Neagu, M.; Tutunaru, D.; Nechita, A.; Fulga, I. Revolutionizing Pathology with Artificial Intelligence: Innovations in Immunohistochemistry. J. Pers. Med. 2024, 14, 693. [Google Scholar] [CrossRef] [PubMed]
  18. Xie, S.; Wang, Y.; Gong, Z.; Li, Y.; Yang, W.; Liu, G.; Li, J.; Hu, X.; Wang, Y.; Tong, Y.; et al. Liquid Biopsy and Tissue Biopsy Comparison with Digital PCR and IHC/FISH for HER2 Amplification Detection in Breast Cancer Patients. J. Cancer 2022, 13, 744–751. [Google Scholar] [CrossRef]
  19. Kim, Y.-G.; Lee, B.; Ha, C.; Lee, C.; Jung, H.A.; Sun, J.-M.; Lee, S.-H.; Ahn, M.-J.; Choi, Y.-L.; Park, S.; et al. Clinical utility of circulating tumor DNA profiling in detecting targetable fusions in non-small cell lung cancer. Front. Oncol. 2024, 14, 1463341. [Google Scholar] [CrossRef]
  20. Sheffield, B.S. Immunohistochemistry as a Practical Tool in Molecular Pathology. Arch. Pathol. Lab. Med. 2016, 140, 766–769. [Google Scholar] [CrossRef] [PubMed]
  21. Magaki, S.; Hojat, S.A.; Wei, B.; So, A.; Yong, W.H. An Introduction to the Performance of Immunohistochemistry. In Methods in Molecular Biology; Clifton, N.J., Ed.; Humana Press: New York, NY, USA, 2018; Volume 1897, pp. 289–298. [Google Scholar] [CrossRef]
  22. Taylor, C.R.; Shi, S.R.; Barr, N.J.; Wu, N. Techniques of Immunohistochemistry: Principles, Pitfalls and Standardization. In Diagnostic Immunohistochemistry; Churchill Livingstone: London, UK, 2006; pp. 1–42. ISBN 9780443066528. Available online: https://linkinghub.elsevier.com/retrieve/pii/B9780443066528500077 (accessed on 21 November 2025).
  23. Sukswai, N.; Khoury, J.D. Immunohistochemistry Innovations for Diagnosis and Tissue-Based Biomarker Detection. Curr. Hematol. Malig. Rep. 2019, 14, 368–375. [Google Scholar] [CrossRef]
  24. Aeffner, F.; Zarella, M.D.; Buchbinder, N.; Bui, M.M.; Goodman, M.R.; Hartman, D.J.; Lujan, G.M.; Molani, M.A.; Parwani, A.V.; Lillard, K.; et al. Introduction to Digital Image Analysis in Whole-slide Imaging: A White Paper from the Digital Pathology Association. J. Pathol. Inform. 2019, 10, 9. [Google Scholar] [CrossRef] [PubMed]
  25. Hanna, M.G.; Parwani, A.M.; Sirintrapun, S.J. Whole Slide Imaging: Technology and Applications. Adv. Anat. Pathol. 2020, 27, 251–259. [Google Scholar] [CrossRef]
  26. Shafi, S.; Kellough, D.A.; Lujan, G.; Satturwar, S.; Parwani, A.V.; Li, Z. Integrating and validating automated digital imaging analysis of estrogen receptor immunohistochemistry in a fully digital workflow for clinical use. J. Pathol. Inform. 2022, 13, 100122. [Google Scholar] [CrossRef]
  27. Stålhammar, G.; Martinez, N.F.; Lippert, M.; Tobin, N.P.; Mølholm, I.; Kis, L.; Rosin, G.; Rantalainen, M.; Pedersen, L.; Bergh, J.; et al. Digital image analysis outperforms manual biomarker assessment in breast cancer. Mod. Pathol. 2016, 29, 318–329. [Google Scholar] [CrossRef] [PubMed]
  28. Priego-Torres, B.M.; Lobato-Delgado, B.; Atienza-Cuevas, L.; Sanchez-Morillo, D. Deep learning-based instance segmentation for the precise automated quantification of digital breast cancer immunohistochemistry images. Expert Syst. Appl. 2022, 193, 116471. [Google Scholar] [CrossRef]
  29. Mulrane, L.; Rexhepaj, E.; Penney, S.; Callanan, J.J.; Gallagher, W.M. Automated image analysis in histopathology: A valuable tool in medical diagnostics. Expert Rev. Mol. Diagn. 2008, 8, 707–725. [Google Scholar] [CrossRef] [PubMed]
  30. Vandenberghe, M.E.; Scott, M.L.J.; Scorer, P.W.; Söderberg, M.; Balcerzak, D.; Barker, C. Relevance of deep learning to facilitate the diagnosis of HER2 status in breast cancer. Sci. Rep. 2017, 7, 45938. [Google Scholar] [CrossRef]
  31. Howat, W.J.; Blows, F.M.; Provenzano, E.; Brook, M.N.; Morris, L.; Gazinska, P.; Johnson, N.; McDuffus, L.-A.; Miller, J.; Sawyer, E.J.; et al. Performance of automated scoring of ER, PR, HER2, CK5/6 and EGFR in breast cancer tissue microarrays in the Breast Cancer Association Consortium. J. Pathol. Clin. Res. 2014, 1, 18–32. [Google Scholar] [CrossRef]
  32. Lagree, A.; Shiner, A.; Alera, M.A.; Fleshner, L.; Law, E.; Law, B.; Lu, F.-I.; Dodington, D.; Gandhi, S.; Slodkowska, E.A.; et al. Assessment of Digital Pathology Imaging Biomarkers Associated with Breast Cancer Histologic Grade. Curr. Oncol. 2021, 28, 4298–4316. [Google Scholar] [CrossRef]
  33. Jaroensri, R.; Wulczyn, E.; Hegde, N.; Brown, T.; Flament-Auvigne, I.; Tan, F.; Cai, Y.; Nagpal, K.; Rakha, E.A.; Dabbs, D.J.; et al. Deep learning models for histologic grading of breast cancer and association with disease prognosis. NPJ Breast Cancer 2022, 8, 113. [Google Scholar] [CrossRef]
  34. Ivanova, M.; Pescia, C.; Trapani, D.; Venetis, K.; Frascarelli, C.; Mane, E.; Cursano, G.; Sajjadi, E.; Scatena, C.; Cerbelli, B.; et al. Early Breast Cancer Risk Assessment: Integrating Histopathology with Artificial Intelligence. Cancers 2024, 16, 1981. [Google Scholar] [CrossRef]
  35. Ivanov, V.; Khalid, U.; Gurung, J.; Dimov, R.; Chonov, V.; Uchikov, P.; Kostov, G.; Ivanov, S. Use of AI Histopathology in Breast Cancer Diagnosis. Medicina 2025, 61, 1878. [Google Scholar] [CrossRef]
  36. Ivanova, M.; Porta, F.M.; D’eRcole, M.; Pescia, C.; Sajjadi, E.; Cursano, G.; De Camilli, E.; Pala, O.; Mazzarol, G.; Venetis, K.; et al. Standardized pathology report for HER2 testing in compliance with 2023 ASCO/CAP updates and 2023 ESMO consensus statements on HER2-low breast cancer. Virchows Arch. 2023, 484, 3–14. [Google Scholar] [CrossRef] [PubMed]
  37. Phillips, T.; Simmons, P.; Inzunza, H.D.; Cogswell, J.; Novotny, J., Jr.; Taylor, C.; Zhang, X. Development of an Automated PD-L1 Immunohistochemistry (IHC) Assay for Non-Small Cell Lung Cancer. Appl. Immunohistochem. Mol. Morphol. 2015, 23, 541–549. [Google Scholar] [CrossRef] [PubMed]
  38. Naso, J.R.; Povshedna, T.; Wang, G.; Banyi, N.; MacAulay, C.; Ionescu, D.N.; Zhou, C. Automated PD-L1 Scoring for Non-Small Cell Lung Carcinoma Using Open-Source Software. Pathol. Oncol. Res. 2021, 27, 609717. [Google Scholar] [CrossRef]
  39. Baxi, V.; Lee, G.; Duan, C.; Pandya, D.; Cohen, D.N.; Edwards, R.; Chang, H.; Li, J.; Elliott, H.; Pokkalla, H.; et al. Association of artificial intelligence-powered and manual quantification of programmed death-ligand 1 (PD-L1) expression with outcomes in patients treated with nivolumab ± ipilimumab. Mod. Pathol. 2022, 35, 1529–1539. [Google Scholar] [CrossRef]
  40. Taylor, C.R.; Jadhav, A.P.; Gholap, A.; Kamble, G.; Huang, J.; Gown, A.; Doshi, I.; Rimm, D.L. A Multi-Institutional Study to Evaluate Automated Whole Slide Scoring of Immunohistochemistry for Assessment of Programmed Death-Ligand 1 (PD-L1) Expression in Non–Small Cell Lung Cancer. Appl. Immunohistochem. Mol. Morphol. 2019, 27, 263–269. [Google Scholar] [CrossRef]
  41. Widmaier, M.; Wiestler, T.; Walker, J.; Barker, C.; Scott, M.L.; Sekhavati, F.; Budco, A.; Schneider, K.; Segerer, F.J.; Steele, K.; et al. Comparison of continuous measures across diagnostic PD-L1 assays in non-small cell lung cancer using automated image analysis. Mod. Pathol. 2019, 33, 380–390. [Google Scholar] [CrossRef]
  42. Cheng, G.; Zhang, F.; Xing, Y.; Hu, X.; Zhang, H.; Chen, S.; Li, M.; Peng, C.; Ding, G.; Zhang, D.; et al. Artificial Intelligence-Assisted Score Analysis for Predicting the Expression of the Immunotherapy Biomarker PD-L1 in Lung Cancer. Front. Immunol. 2022, 13, 893198. [Google Scholar] [CrossRef] [PubMed]
  43. Moxley-Wyles, B.; Colling, R.; Verrill, C. Artificial intelligence in pathology: An overview. Diagn. Histopathol. 2020, 26, 513–520. [Google Scholar] [CrossRef]
  44. Khoraminia, F.; Fuster, S.; Kanwal, N.; Olislagers, M.; Engan, K.; van Leenders, G.J.L.H.; Stubbs, A.P.; Akram, F.; Zuiverloon, T.C.M. Artificial Intelligence in Digital Pathology for Bladder Cancer: Hype or Hope? A Systematic Review. Cancers 2023, 15, 4518. [Google Scholar] [CrossRef] [PubMed]
  45. Zheng, Q.; Yang, R.; Ni, X.; Yang, S.; Xiong, L.; Yan, D.; Xia, L.; Yuan, J.; Wang, J.; Jiao, P.; et al. Accurate Diagnosis and Survival Prediction of Bladder Cancer Using Deep Learning on Histological Slides. Cancers 2022, 14, 5807. [Google Scholar] [CrossRef]
  46. Kim, H.; Kim, J.; Yeon, S.Y.; You, S. Machine learning approaches for spatial omics data analysis in digital pathology: Tools and applications in genitourinary oncology. Front. Oncol. 2024, 14, 1465098. [Google Scholar] [CrossRef]
  47. Vlachostergios, P.J.; Faltas, B.M. The molecular limitations of biomarker research in bladder cancer. World J. Urol. 2018, 37, 837–848. [Google Scholar] [CrossRef] [PubMed]
  48. Lobo, J.; Zein-Sabatto, B.; Lal, P.; Netto, G.J. Digital and Computational Pathology Applications in Bladder Cancer: Novel Tools Addressing Clinically Pressing Needs. Mod. Pathol. 2024, 38, 100631. [Google Scholar] [CrossRef] [PubMed]
  49. Attanasio, G.; Failla, M.; Poidomani, S.; Buzzanca, T.; Salzano, S.; Zizzo, M.; Palicelli, A.; Zanelli, M.; Koufopoulos, N.; Russo, G.I.; et al. Histological and immuno-histochemical approaches to molecular subtyping in muscle-invasive bladder cancer. Front. Oncol. 2025, 15, 1546160. [Google Scholar] [CrossRef]
  50. Meyerholz, D.K.; Beck, A.P. Principles and approaches for reproducible scoring of tissue stains in research. Mod. Pathol. 2018, 98, 844–855. [Google Scholar] [CrossRef]
  51. Nielsen, T.O.; Leung, S.C.Y.; Rimm, D.L.; Dodson, A.; Acs, B.; Badve, S.; Denkert, C.; Ellis, M.J.; Fineberg, S.; Flowers, M.; et al. Assessment of Ki67 in Breast Cancer: Updated Recommendations from the International Ki67 in Breast Cancer Working Group. JNCI J. Natl. Cancer Inst. 2020, 113, 808–819. [Google Scholar] [CrossRef]
  52. Polley, M.-Y.C.; Leung, S.C.Y.; McShane, L.M.; Gao, D.; Hugh, J.C.; Mastropasqua, M.G.; Viale, G.; Zabaglo, L.A.; Penault-Llorca, F.; Bartlett, J.M.; et al. An International Ki67 Reproducibility Study. JNCI J. Natl. Cancer Inst. 2013, 105, 1897–1906. [Google Scholar] [CrossRef] [PubMed]
  53. Hommerding, O.; Bernhardt, M.; Kreft, T.; Scherping, A.; Abbas, M.; Baretton, G.; Bräsen, J.H.; Breyer, J.; Darr, C.; Dressler, F.F.; et al. High interobserver variability of PTEN immunohistochemistry defining PTEN status in low- to intermediate-risk prostate cancer: Results of the first German ring trial. Virchows Arch. 2024, 487, 87–96. [Google Scholar] [CrossRef]
  54. Wen, Z.; Luo, D.; Wang, S.; Rong, R.; Evers, B.M.; Jia, L.; Fang, Y.; Daoud, E.V.; Yang, S.; Gu, Z.; et al. Deep Learning–Based H-Score Quantification of Immunohistochemistry-Stained Images. Mod. Pathol. 2023, 37, 100398. [Google Scholar] [CrossRef] [PubMed]
  55. Allison, K.H. Prognostic and predictive parameters in breast pathology: A pathologist’s primer. Mod. Pathol. 2021, 34, 94–106. [Google Scholar] [CrossRef] [PubMed]
  56. Walker, R.A. Immunohistochemical markers as predictive tools for breast cancer. J. Clin. Pathol. 2007, 61, 689–696. [Google Scholar] [CrossRef]
  57. Wu, S.; Yue, M.; Zhang, J.; Li, X.; Li, Z.; Zhang, H.; Wang, X.; Han, X.; Cai, L.; Shang, J.; et al. The Role of Artificial Intelligence in Accurate Interpre-tation of HER2 Immunohistochemical Scores 0 and 1+ in Breast Cancer. Mod. Pathol. 2023, 36, 100054. [Google Scholar] [CrossRef]
  58. Tozbikian, G.; Bui, M.M.; Hicks, D.G.; Jaffer, S.; Khoury, T.; Wen, H.Y.; Krishnamurthy, S.; Wei, S. Best practices for achieving consensus in HER2-low expression in breast cancer: Current perspectives from practising pathologists. Histopathology 2024, 85, 489–502. [Google Scholar] [CrossRef]
  59. Rüschoff, J.; Schildhaus, H.-U.; Rüschoff, J.H.; Jöhrens, K.; Edmonston, T.B.; Dietmaier, W.; Bläker, H.; Baretton, G.; Horst, D.; Dietel, M.; et al. Testing for deficient mismatch repair and microsatellite instability. Die Pathol. 2023, 44, 61–70. [Google Scholar] [CrossRef]
  60. Blessin, N.C.; Yang, C.; Mandelkow, T.; Raedler, J.B.; Li, W.; Bady, E.; Simon, R.; Vettorazzi, E.; Lennartz, M.; Bernreuther, C.; et al. Automated Ki-67 labeling index assessment in prostate cancer using artificial intelligence and multiplex fluorescence immunohistochemistry. J. Pathol. 2023, 260, 5–16. [Google Scholar] [CrossRef]
  61. Abukhiran, I.; Neyaz, A.; Kop, M.; Baroudi, I.; Christensen, D.; Baki, M.-N.A.; Surakji, H.; Shaker, N.; Bedell, M.L.; Jasser, J.; et al. Optimal Approaches to Grading Enteropancreatic Neuroendocrine Tumors Using Ki-67 Proliferation Index: Hotspot and Whole-Slide Digital Quantitative Analysis. Mod. Pathol. 2025, 38, 100780. [Google Scholar] [CrossRef] [PubMed]
  62. Lakshmi, S.; Vijayasenan, D.; Sumam, D.S.; Sreeram, S.; Suresh, P.K. An Integrated Deep Learning Approach towards Automatic Evaluation of Ki-67 Labeling Index. In Proceedings of the TENCON 2019—2019 IEEE Region 10 Conference (TENCON), Kochi, India, 17–20 October 2019; pp. 2310–2314. [Google Scholar]
  63. Jiao, F.; Shang, Z.; Lu, H.; Chen, P.; Chen, S.; Xiao, J.; Zhang, F.; Zhang, D.; Lv, C.; Han, Y. A weakly supervised deep learning framework for automated PD-L1 expression analysis in lung cancer. Front. Immunol. 2025, 16, 1540087. [Google Scholar] [CrossRef]
  64. Rodrigues, A.; Nogueira, C.; Marinho, L.C.; Velozo, G.; Sousa, J.; Silva, P.G.; Tavora, F. Computer-assisted tumor grading, validation of PD-L1 scoring, and quantification of CD8-positive immune cell density in urothelial carcinoma, a visual guide for pathologists using QuPath. Surg. Exp. Pathol. 2022, 5, 12. [Google Scholar] [CrossRef]
  65. Jain, E.; Patel, A.; Parwani, A.V.; Shafi, S.; Brar, Z.; Sharma, S.; Mohanty, S.K. Whole Slide Imaging Technology and Its Applications: Current and Emerging Perspectives. Int. J. Surg. Pathol. 2023, 32, 433–448. [Google Scholar] [CrossRef]
  66. Kumar, N.; Gupta, R.; Gupta, S. Whole Slide Imaging (WSI) in Pathology: Current Perspectives and Future Directions. J. Digit. Imaging 2020, 33, 1034–1040. [Google Scholar] [CrossRef]
  67. Corti, C.; Cobanaj, M.; Dee, E.C.; Criscitiello, C.; Tolaney, S.M.; Celi, L.A.; Curigliano, G. Artificial intelligence in cancer research and precision medicine: Applications, limitations and priorities to drive transformation in the delivery of equitable and unbiased care. Cancer Treat. Rev. 2022, 112, 102498. [Google Scholar] [CrossRef]
  68. Bodén, A.C.S.; Molin, J.; Garvin, S.; West, R.A.; Lundström, C.; Treanor, D. The human-in-the-loop: An evaluation of pathologists’ interaction with artificial intelligence in clinical practice. Histopathology 2021, 79, 210–218. [Google Scholar] [CrossRef] [PubMed]
  69. Shaktah, L.A.; Carrero, Z.I.; Hewitt, K.J.; Gustav, M.; Cecchini, M.; Foersch, S.; Berezowska, S.; Kather, J.N. Application of artificial intel-ligence and digital tools in cancer pathology. Lancet Digit. Health 2025, 7, 100933. [Google Scholar] [CrossRef]
  70. Cheng, C.H.; Shi, S.-S. Artificial intelligence in cancer: Applications, challenges, and future perspectives. Mol. Cancer 2025, 24, 274. [Google Scholar] [CrossRef] [PubMed]
  71. Hanna, M.G.; Pantanowitz, L.; Dash, R.; Harrison, J.H.; Deebajah, M.; Pantanowitz, J.; Rashidi, H.H. Future of Artificial Intelligence—Machine Learning Trends in Pathology and Medicine. Mod. Pathol. 2025, 38, 100705. [Google Scholar] [CrossRef]
  72. Zarella, M.D.; McClintock, D.S.; Batra, H.; Gullapalli, R.R.; Valante, M.; Tan, V.O.; Dayal, S.; Oh, K.S.; Lara, H.; Garcia, C.A.; et al. Artificial intelligence and digital pathology: Clinical promise and deployment considerations. J. Med. Imaging 2023, 10, 051802. [Google Scholar] [CrossRef] [PubMed]
  73. Nowak, M.; Jabbar, F.; Rodewald, A.-K.; Gneo, L.; Tomasevic, T.; Harkin, A.; Iveson, T.; Saunders, M.; Kerr, R.; Oein, K.; et al. Single-cell AI-based detection and prognostic and predictive value of DNA mismatch repair deficiency in colorectal cancer. Cell Rep. Med. 2024, 5, 101727. [Google Scholar] [CrossRef] [PubMed]
  74. Echle, A.; Grabsch, H.I.; Quirke, P.; van den Brandt, P.A.; West, N.P.; Hutchins, G.G.A.; Heij, L.R.; Tan, X.; Richman, S.D.; Krause, J.; et al. Clinical-grade Detection of Microsatellite Instability in Colorectal Tumors by Deep Learning. Gastroenterology 2020, 159, 1406–1416.e11. [Google Scholar] [CrossRef] [PubMed]
  75. Ma, C.X.; Bose, R.; Ellis, M.J. Prognostic and Predictive Biomarkers of Endocrine Responsiveness for Estrogen Receptor Positive Breast Cancer. In Novel Biomarkers in the Continuum of Breast Cancer; Stearns, V., Ed.; Springer International Publishing: Cham, Switzerland, 2016; pp. 125–154. [Google Scholar] [CrossRef]
  76. Leong, A.S.Y. Pitfalls in Diagnostic Immunohistology. Adv. Anat. Pathol. 2004, 11, 86. [Google Scholar] [CrossRef] [PubMed]
  77. Yassen, N.N.; Elsharkawy, S.L.; Abbas, N.F.; Shabana, M.E. Immunohistochemical expression of GATA3, CK5/6 and CK20 in molecular subtypes of bladder carcinoma: Correlation with clinicopathological features. Bull. Natl. Res. Cent. 2024, 48, 83. [Google Scholar] [CrossRef]
  78. Weyerer, V.; Stoehr, R.; Bertz, S.; Lange, F.; Geppert, C.I.; Wach, S.; Taubert, H.; Sikic, D.; Wullich, B.; Hartmann, A.; et al. Prognostic impact of molecular muscle-invasive bladder cancer subtyping approaches and correlations with variant histology in a population-based mono-institutional cystectomy cohort. World J. Urol. 2021, 39, 4011–4019. [Google Scholar] [CrossRef]
  79. Guo, C.C.; Bondaruk, J.; Yao, H.; Wang, Z.; Zhang, L.; Lee, S.; Lee, J.-G.; Cogdell, D.; Zhang, M.; Yang, G.; et al. Assessment of Luminal and Basal Phenotypes in Bladder Cancer. Sci. Rep. 2020, 10, 9743. [Google Scholar] [CrossRef] [PubMed]
  80. Cano Barbadilla, T.; Álvarez Pérez, M.; Prieto Cuadra, J.D.; Dawid de Vera, M.T.; Alberca-del Arco, F.; García Muñoz, I.; Santos-Pérez de la Blanca, R.; Herrera-Imbroda, B.; Matas-Rico, E.; Hierro Martín, M.I. The Role of Immunohistochemistry as a Surrogate Marker in Molecular Subtyping and Classifi-cation of Bladder Cancer. Diagnostics 2024, 14, 2501. [Google Scholar] [CrossRef]
  81. Fraggetta, F.; L’imperio, V.; Ameisen, D.; Carvalho, R.; Leh, S.; Kiehl, T.-R.; Serbanescu, M.; Racoceanu, D.; Della Mea, V.; Polonia, A.; et al. Best Practice Recommendations for the Implementation of a Digital Pathology Workflow in the Anatomic Pathology Laboratory by the European Society of Digital and Integrative Pathology (ESDIP). Diagnostics 2021, 11, 2167. [Google Scholar] [CrossRef]
  82. Fan, W.-X.; Su, F.; Zhang, Y.; Zhang, X.-L.; Du, Y.-Y.; Gao, Y.-J.; Li, W.-L.; Hu, W.-Q.; Zhao, J. Oncological characteristics, treatments and prognostic outcomes in MMR-deficient colorectal cancer. Biomark. Res. 2024, 12, 89. [Google Scholar] [CrossRef]
  83. Loughrey, M.B.; McGrath, J.; Coleman, H.G.; Bankhead, P.; Maxwell, P.; McGready, C.; Bingham, V.; Humphries, M.P.; Craig, S.G.; McQuaid, S.; et al. Identifying mismatch repair-deficient colon cancer: Near-perfect concordance between immunohistochemistry and microsatellite instability testing in a large, population-based series. Histopathology 2020, 78, 401–413. [Google Scholar] [CrossRef]
  84. Meng, X.; Huang, Z.; Teng, F.; Xing, L.; Yu, J. Predictive biomarkers in PD-1/PD-L1 checkpoint blockade im-munotherapy. Cancer Treat. Rev. 2015, 41, 868–876. [Google Scholar] [CrossRef]
  85. Cheng, J.; Han, Y.; Yuan, Y.; Huang, S.; Xiao, B.; Kong, Y.; Xue, W.; Yuan, R.; Liu, H.; Lan, P.; et al. Automated multi-regional IHC scoring enhances prognostication in colorectal cancer. J. Pathol. Clin. Res. 2025, 11, e70047. [Google Scholar] [CrossRef] [PubMed]
  86. Sinicrope, F.; Nelson, G.; Ardestani, B.S.; Park, G.; Lim, Y.; Yan, D.; Shanmugam, K.; Alberts, S.; Shi, Q.; Ock, C.-Y.; et al. 569P Artificial intelligence for detection of mismatch repair deficiency in colon carcinomas (alliance). Ann. Oncol. 2023, 34, S420. [Google Scholar] [CrossRef]
  87. Li, H.; Qin, J.; Li, Z.; Ouyang, R.; Chen, Z.; Huang, S.; Qin, S.; Huang, Q. Systematic review and meta-analysis of deep learning for MSI-H in colorectal cancer whole slide images. npj Digit. Med. 2025, 8, 456. [Google Scholar] [CrossRef]
  88. Yang, Y.; Jain, R.K.; Glenn, S.T.; Xu, B.; Singh, P.K.; Wei, L.; Hu, Q.; Long, M.; Hutson, N.; Wang, J.; et al. Complete response to anti-PD-L1 antibody in a metastatic bladder cancer associated with novel MSH4 mutation and microsatellite instability. J. Immunother. Cancer 2020, 8, e000128. [Google Scholar] [CrossRef]
  89. Zhang, H.; Yang, X.; Xie, J.; Cheng, X.; Chen, J.; Shen, M.; Ding, W.; Wang, S.; Zhang, Z.; Wang, C.; et al. Clinicopathological and molecular analysis of microsatellite instability in prostate cancer: A multi-institutional study in China. Front. Oncol. 2023, 13, 1277233. [Google Scholar] [CrossRef]
  90. Gabriel, P.; Cancel-Tassin, G.; Audenet, F.; Masson-Lecomte, A.; Allory, Y.; Roumiguié, M.; Pradère, B.; Loriot, Y.; Léon, P.; Traxer, O.; et al. A collaborative review of the microsatellite instability/deficient mismatch repair phenotype in patients with upper tract urothelial carcinoma. BJU Int. 2024, 134, 723–735. [Google Scholar] [CrossRef]
  91. Lopez-Beltran, A.; Cimadamore, A.; Blanca, A.; Massari, F.; Vau, N.; Scarpelli, M.; Cheng, L.; Montironi, R. Immune Checkpoint Inhibitors for the Treatment of Bladder Cancer. Cancers 2021, 13, 131. [Google Scholar] [CrossRef]
  92. Boiarsky, D.; Gulhan, D.C.; Savignano, H.; Lakshminarayanan, G.; McClure, H.M.; Silver, R.; Hirsch, M.S.; Sholl, L.M.; Choudhury, A.D.; Ananda, G.; et al. A Panel-Based Mutational Signature of Mismatch Repair Deficiency is Associated with Durable Response to Pembrolizumab in Metastatic Castration-Resistant Prostate Cancer. Clin. Genitourin. Cancer 2024, 22, 558–568.e3. [Google Scholar] [CrossRef] [PubMed]
  93. Tanabe, K.; Nakanishi, Y.; Okubo, N.; Matsumoto, S.; Umino, Y.; Kataoka, M.; Yajima, S.; Yoshida, T.; Miyazaki, S.; Kuwata, T.; et al. Prevalence and characteristics of patients with upper urinary tract urothelial carcinoma having potential Lynch syndrome identified by immunohistochemical universal screening and Amsterdam criteria II. BMC Cancer 2023, 23, 940. [Google Scholar] [CrossRef]
  94. Guan, B.; Wang, J.; Li, X.; Lin, L.; Fang, D.; Kong, W.; Tian, C.; Li, J.; Yang, K.; Han, G.; et al. Identification of Germline Mutations in Upper Tract Urothelial Carcinoma with Suspected Lynch Syndrome. Front. Oncol. 2022, 12, 774202. [Google Scholar] [CrossRef]
  95. Ju, J.Y.; Mills, A.M.; Mahadevan, M.S.; Fan, J.; Culp, S.H.; Thomas, M.H.; Cathro, H.P. Universal Lynch Syndrome Screening Should be Performed in All Upper Tract Urothelial Carcinomas. Am. J. Surg. Pathol. 2018, 42, 1549–1555. [Google Scholar] [CrossRef]
  96. Vranic, S.; Gatalica, Z. PD-L1 testing by immunohistochemistry in immuno-oncology. Biomol. Biomed. 2023, 23, 15–25. [Google Scholar] [CrossRef]
  97. Haragan, A.; Gosney, J.R. Immunohistochemistry for prediction of response to immunotherapy. Diagn. Histopathol. 2020, 26, 506–512. [Google Scholar] [CrossRef]
  98. Rüschoff, J.; Kumar, G.; Badve, S.; Jasani, B.; Krause, E.; Rioux-Leclercq, N.; Rojo, F.; Martini, M.; Cheng, L.; Tretiakova, M.; et al. Scoring PD-L1 Expression in Urothelial Carcinoma: An International Multi-Institutional Study on Comparison of Manual and Artificial Intelligence Measurement Model (AIM-PD-L1) Pathology Assessments. Virchows Arch. 2024, 484, 597–608. [Google Scholar] [CrossRef]
  99. Klöppel, G.; La Rosa, S. Ki67 labeling index: Assessment and prognostic role in gastroenteropancreatic neuroendocrine neoplasms. Virchows Arch. 2017, 472, 341–349. [Google Scholar] [CrossRef]
  100. Wang, Y.; Dai, M.; Chen, X. Prognostic and clinicopathological value of Ki-67 in patients with oesophageal squamous cell carcinoma: A systematic review and meta-analysis. BMJ Open 2024, 14, e083637. [Google Scholar] [CrossRef] [PubMed]
  101. Erdian, D.N.; Ham, M.F.; Khoirunnisa, D.; Harahap, A.S. High Ki-67 labeling index correlates with aggressive clinicopathological features in papillary thyroid carcinoma: A retrospective study. Thyroid. Res. 2025, 18, 54. [Google Scholar] [CrossRef] [PubMed]
  102. Maia, R.; DOS Santos, G.A.; Reis, S.; Viana, N.I.; Pimenta, R.; Guimarães, V.R.; Recuero, S.; Romão, P.; Leite, K.R.M.; Srougi, M.; et al. Can we use Ki67 expression to predict prostate cancer aggressiveness? Rev. Colégio Bras. Cir. 2022, 49, e20223200. [Google Scholar] [CrossRef]
  103. Höller, A.; Nguyen-Sträuli, B.D.; Frauchiger-Heuer, H.; Ring, A. Diagnostic and Prognostic Biomarkers of Luminal Breast Cancer: Where are We Now? Breast Cancer Targets Ther. 2023, 15, 525–540. [Google Scholar] [CrossRef]
  104. Dębska-Szmich, S.; Potemski, P. Adjuvant systemic treatment in luminal breast cancer—What else apart from hormone therapy? Oncol. Clin. Pract. 2025. [Google Scholar] [CrossRef]
  105. Cavalcanti, M.S.; Gönen, M.; Klimstra, D.S. The ENETS/WHO Grading System for Neuroendocrine Neoplasms of the Gastroenteropancreatic System: A Review of the Current State, Limitations and Proposals for Modifications. Int. J. Endocr. Oncol. 2016, 3, 203–219. [Google Scholar] [CrossRef]
  106. Li, W.; Ye, S.; Jin, Z.; Chen, L.; Chao, Y.; Wei, G.; Huang, Q.; Tu, H.; Wang, Q. Artificial Intelligence in digital pathology of breast cancer, new era of practice? Int. J. Surg. 2025, 111, 8270–8283. [Google Scholar] [CrossRef]
  107. Merae Alshahrani, M. A glance at the emerging diagnostic biomarkers in the most prevalent genitouri-nary cancers. Saudi J. Biol. Sci. 2022, 29, 2072–2084. [Google Scholar] [CrossRef]
  108. Sanguedolce, F.; Cormio, A.; Zanelli, M.; Zizzo, M.; Palicelli, A.; Falagario, U.G.; Milanese, G.; Galosi, A.B.; Mazzucchelli, R.; Cormio, L.; et al. Diagnostic, Prognostic, and Predictive Tissue Biomarkers in Urothelial Carcinoma In Situ: A Narrative Review. Diagnostics 2025, 15, 2163. [Google Scholar] [CrossRef]
  109. Sanguedolce, F.; Zanelli, M.; Palicelli, A.; Ascani, S.; Zizzo, M.; Cocco, G.; Björnebo, L.; Lantz, A.; Falagario, U.G.; Cormio, L.; et al. Are We Ready to Implement Molecular Subtyping of Bladder Cancer in Clinical Practice? Part 1: General Issues and Marker Expression. Int. J. Mol. Sci. 2022, 23, 7819. [Google Scholar] [CrossRef]
  110. Aggarwal, A.; Bharadwaj, S.; Corredor, G.; Pathak, T.; Badve, S.; Madabhushi, A. Artificial intelligence in digital pathology—time for a reality check. Nat. Rev. Clin. Oncol. 2025, 22, 283–291. [Google Scholar] [CrossRef]
  111. Baweja, B.; Vats, P.; Kausik, S.; Singh, J.; Nema, R.; Yadav, P. Challenges and Limitations of Computational Methods in Oncology. In Advances in Cancer Detection, Prediction, and Prognosis Using Artificial Intelligence and Machine Learning; Nema, R., Kumar, A., Saini, D.K., Eds.; Springer Nature: Singapore, 2025; pp. 307–324. Available online: https://doi.org/10.1007/978-981-96-9346-7_13 (accessed on 23 November 2025). [CrossRef]
  112. Distante, A.; Marandino, L.; Bertolo, R.; Ingels, A.; Pavan, N.; Pecoraro, A.; Marchioni, M.; Carbonara, U.; Erdem, S.; Amparore, D.; et al. Artificial Intelligence in Renal Cell Carcinoma Histopathology: Current Applications and Future Perspectives. Diagnostics 2023, 13, 2294. [Google Scholar] [CrossRef]
  113. Cimadamore, A.; Gasparrini, S.; Santoni, M.; Cheng, L.; Lopez-Beltran, A.; Battelli, N.; Massari, F.; Giunchi, F.; Fiorentino, M.; Scarpelli, M.; et al. Biomarkers of aggressiveness in genitourinary tumors with emphasis on kidney, bladder, and prostate cancer. Expert Rev. Mol. Diagn. 2018, 18, 645–655. [Google Scholar] [CrossRef]
  114. Fernandes-Pontes, F.; Lobo, J.; Jeronimo, C.; Henrique, R. Identification of novel biomarkers in renal cell carcinoma. Expert Rev. Mol. Diagn. 2025, 25, 465–477. [Google Scholar] [CrossRef]
  115. Martin, A.S.; Molloy, M.; Ugolkov, A.; von Roemeling, R.W.; Noelle, R.J.; Lewis, L.D.; Johnson, M.; Radvanyi, L.; Martell, R.E. VISTA expression and patient selection for immune-based anticancer therapy. Front. Immunol. 2023, 14, 1086102. [Google Scholar] [CrossRef]
  116. Li, W.; Liu, Z.; Jin, K.; Shao, F.; Zeng, H.; Wang, Y.; Zhu, Y.; Xu, L.; Wang, Z.; Chang, Y.; et al. Immune inactivation by VISTA predicts clinical outcome and therapeutic benefit in muscle-invasive bladder cancer. BMC Cancer 2023, 23, 661. [Google Scholar] [CrossRef]
  117. Zhao, S.; Zhang, H.; Shang, G. Research progress of B7-H3 in malignant tumors. Front. Immunol. 2025, 16, 1586759. [Google Scholar] [CrossRef]
  118. Yajima, S.; Masuda, H. Immune Checkpoint Inhibitors and Antibody-Drug Conjugates in Urothelial Carcinoma: Current Landscape and Future Directions. Cancers 2025, 17, 1594. [Google Scholar] [CrossRef]
  119. Dernbach, G.; Eich, M.-L.; Dragomir, M.P.; Anders, P.; Jurczok, N.; Stief, C.; Jurmeister, P.; Schlomm, T.; Klauschen, F.; Horst, D.; et al. Spatial Expression of HER2, NECTIN4, and TROP-2 in Muscle-Invasive Bladder Cancer and Metastases: Implications for Pathological and Clinical Management. Mod. Pathol. 2025, 38, 100753. [Google Scholar] [CrossRef]
  120. Kobayashi, G.; Sekino, Y.; Hayashi, T.; Nakahara, H.; Yukihiro, K.; Kobatake, K.; Kitano, H.; Goto, K.; Niitsu, H.; Hinoi, T.; et al. Nectin-4 expression in upper and lower tract urothelial carcinoma: Correlation with early-stage disease and prognostic relevance. Virchows Arch. 2025, 487, 1–16. [Google Scholar] [CrossRef] [PubMed]
  121. Bjartell, A.; Krzyzanowska, A.; Liu, V.Y.; Tierney, M.; Royce, T.J.; Sjöström, M.; Palominos-Rivera, M.M.; Chen, E.; Kraft, A.; Esteva, A.; et al. Validation of a Digital Pathology–Based Multimodal Artificial Intelligence Biomarker in a Prospective, Real-World Prostate Cancer Cohort Treated with Prostatectomy. Clin. Cancer Res. 2025, 31, 1546–1553. [Google Scholar] [CrossRef]
  122. Yoshida, T.; Nakamoto, T.; Atsumi, N.; Ohe, C.; Sano, T.; Yasukochi, Y.; Tsuta, K.; Kinoshita, H. Impact of LAG-3/FGL1 pathway on immune evasive contexture and clinical outcomes in advanced urothelial carcinoma. J. Immunother. Cancer 2024, 12, e009358–14. [Google Scholar] [CrossRef]
  123. Cai, L.; Li, Y.; Tan, J.; Xu, L.; Li, Y. Targeting LAG-3, TIM-3, and TIGIT for cancer immunotherapy. J. Hematol. Oncol. 2023, 16, 101. [Google Scholar] [CrossRef]
  124. Meng, L.; Khorasanchi, A.; Jain, R. Advancing Bladder Cancer Management: The Role of Neoadjuvant and Adjuvant Therapies and Biomarkers in Muscle Invasive Bladder Cancer. Curr. Treat. Options Oncol. 2025, 26, 929–942. [Google Scholar] [CrossRef]
  125. Efstathiou, J.A.; Mouw, K.W.; Gibb, E.A.; Liu, Y.; Wu, C.-L.; Drumm, M.R.; da Costa, J.B.; du Plessis, M.; Wang, N.Q.; Davicioni, E.; et al. Impact of Immune and Stromal Infiltration on Outcomes Following Bladder-Sparing Trimodality Therapy for Muscle-Invasive Bladder Cancer. Eur. Urol. 2019, 76, 59–68. [Google Scholar] [CrossRef] [PubMed]
  126. Colling, R.; Pitman, H.; Oien, K.; Rajpoot, N.; Macklin, P.; CM-Path AI in Histopathology Working Group; Snead, D.; Sackville, T.; Verrill, C. Artificial intelligence in digital pathology: A roadmap to routine use in clinical practice. J. Pathol. 2019, 249, 143–150. [Google Scholar] [CrossRef]
  127. van der Laak, J.; Litjens, G.; Ciompi, F. Deep learning in histopathology: The path to the clinic. Nat. Med. 2021, 27, 775–784. [Google Scholar] [CrossRef]
  128. Marliot, F.; Lafontaine, L.; Galon, J. Immunoscore assay for the immune classification of solid tumors: Technical aspects, improvements and clinical perspectives. Methods Enzymol. 2020, 636, 109–128. [Google Scholar]
  129. El Sissy, C.; Kirilovsky, A.; Zeitoun, G.; Marliot, F.; Haicheur, N.; Lagorce-Pagès, C.; Galon, J.; Pagès, F. Therapeutic Implica-tions of the Immunoscore in Patients with Colorectal Cancer. Cancers 2021, 13, 1281. [Google Scholar] [CrossRef]
  130. Marliot, F.; Chen, X.; Kirilovsky, A.; Sbarrato, T.; El Sissy, C.; Batista, L.; Van den Eynde, M.; Haicheur-Adjouri, N.; Anitei, M.G.; Musina, A.M.; et al. Analytical validation of the Im-munoscore and its associated prognostic value in patients with colon cancer. J. Immunother. Cancer 2020, 8, e000272. [Google Scholar] [CrossRef]
  131. Powles, T.; Assaf, Z.J.; Degaonkar, V.; Grivas, P.; Hussain, M.; Oudard, S.; Gschwend, J.E.; Albers, P.; Castellano, D.; Nishiyama, H.; et al. Updated Overall Survival by Circulating Tumor DNA Status from the Phase 3 IMvigor010 Trial: Adjuvant Atezolizumab Versus Observation in Muscle-invasive Urothelial Carcinoma. Eur. Urol. 2023, 85, 114–122. [Google Scholar] [CrossRef]
  132. Powles, T.; Kann, A.G.; Castellano, D.; Gross-Goupil, M.; Nishiyama, H.; Bracarda, S.; Jensen, J.B.; Makaroff, L.; Jiang, S.; Ku, J.H.; et al. ctDNA-Guided Adjuvant Atezolizumab in Muscle-Invasive Bladder Cancer. N. Engl. J. Med. 2025, 393, 2395–2408. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Journal Statistics
Multiple requests from the same IP address are counted as one view.
Download PDF
Improving Prognostic Prediction in Head and Neck Cancer Through a Combined Systemic Immune-Inflammation Index and Prognostic Nutritional Index Score
Previous
Hepatocellular Carcinoma Around the Clock
Next