Artificial intelligence (AI) algorithms leveraging digital pathology slides are currently transforming the way urological cancers are diagnosed and graded, and they add additional prognostic, predictive and molecular subtyping information beyond traditional pathological risk stratification. This review explores recent advances in histopathology-based AI systems for prostate cancer. We examine how these algorithms perform relative to pathologists for tumour diagnosis and grading, and the ways in which they surpass pathologists with respect to reducing inter-observer variability and providing quantified tumour metrics. We particularly focus on prognostic algorithms that have been benchmarked against 'gold standard' patient outcomes such as metastasis or death, and we highlight the emerging role of digital pathology-enabled AI for predicting response to therapy or underlying tumour molecular alteration status. Finally, we touch on the advantages of, and barriers to, implementation of digital pathology and histopathology-based AI algorithms in clinical practice. Through this synthesis of current literature, we underscore the emerging potential of AI for standardising pathological assessment, guiding clinical management, and improving patient outcomes in prostate cancer.

Inflammatory myofibroblastic tumor (IMT) and sarcomatoid urothelial carcinoma (SarUC) can have striking histologic overlap but have significantly different prognoses and clinical management paradigms. Loss of methylthioadenosine phosphorylase (MTAP) protein expression by immunohistochemistry (IHC) serves as a useful surrogate for homozygous 9p21 deletion, a recurrent genomic alteration in urothelial carcinoma (UC). We analyzed MTAP expression by IHC in 65 SarUCs and 27 urinary tract IMTs to evaluate its utility in navigating this challenging differential diagnosis. Overall, MTAP loss was significantly more frequent in SarUC (55%) compared with IMT (4%) (P < .0001). Among 46 biphasic SarUCs with independently evaluable epithelial and mesenchymal components, divergent expression patterns were frequent. The most common pattern was retention of MTAP staining in both epithelial and mesenchymal components (19/46; 41% of cases), followed by selective retention of MTAP in the epithelial component and loss in the mesenchymal component (16/46; 35% of cases). MTAP loss was observed in both the epithelial and mesenchymal components in 11 out of 46 (24%) SarUC cases. None of the 46 biphasic SarUC cases showed selective MTAP loss in the epithelial component but retention in the mesenchymal component. MTAP IHC was also particularly valuable in assessing clonal relationships in 2 challenging biphasic cases in which the differential diagnosis included a collision between a noninvasive low-grade papillary UC and an IMT versus a subtle IMT-like SarUC arising in association with an overlying noninvasive low-grade papillary UC. Next-generation sequencing on a subset of cases (n = 11) was useful for confirming 9p deletion in cases with MTAP loss by IHC, and for demonstrating molecular hallmarks of urothelial neoplasia thereby providing additional diagnostic support for morphologically challenging SarUC cases with IMT-like morphology. Therefore, MTAP IHC can be useful in evaluating spindle cell lesions of the urinary tract, as loss is significantly more common in SarUC than in IMT, and enriched in the mesenchymal component of biphasic SarUC. However, MTAP loss can be seen in both entities, and the diagnosis of IMT-like spindle cell tumors in the urinary tract requires careful integration of morphologic, immunohistochemical, and molecular data.

Abstract Prostate adenocarcinoma (PRAD) is a multifocal and highly heterogeneous malignancy, marked by the coexistence of multiple Gleason grade glands within the same tumor microenvironment (TME). Spatial Transcriptomics (ST) has recently emerged as a powerful technology for characterizing the TME, yet its application is largely limited to two-dimensional (2D) histological sections, which typically capture less than 0.1% of the three-dimensional (3D) tumor context available upon resection. While promising, recently proposed in-situ approaches for 3D ST remain confined to reduced patient cohorts due to lengthy processing times and substantial costs. To enable scalable 3D morphomolecular analysis of PRAD TME, we devise an AI framework that obtains 3D spatial molecular maps in a cost-effective and efficient manner. Our framework, VORTEX (Volumetrically Resolved Transcriptomics EXpression), leverages 3D high-resolution tissue morphology from 3D pathology imaging modalities, and minimal 2D ST to predict 3D ST. By pretraining on diverse 3D morphology-transcriptomic pairs from heterogeneous tissue samples and then fine-tuning on minimal 2D ST data from a specific volume of interest, VORTEX captures both generic tissue-related and sample-specific morphological correlates of gene expression. Our framework leverages pathology and single-cell foundation models, and integrates a cross-modal registration pipeline for accurate learning of morphomolecular links. To evaluate our approach, we apply VORTEX to a cohort of 23 3D pathology volumetric images of PRAD specimens across 17 patients, acquired with micro computed tomography (microCT, 11 volumes) and open-top light-sheet microscopy (OTLS, 12 volumes). We additionally collect 88 sections of Visium ST across these samples and public cohorts, resulting in 243,682 spots with corresponding morphology. We demonstrate that VORTEX accurately predicts 3D ST and we identify two major trends: incorporating 3D morphology enhances the ability to learn morphomolecular links compared to 2D morphology alone, and including 2D ST data from the volume of interest further improves the performance by accounting for inter-patient heterogeneity. We observe that VORTEX captures intra-tumoral and inter-tumoral heterogeneity, with genes such as AZGP1 or GLO1 showing different expression profiles across patients and Gleason Grades. Furthermore, by analyzing 3D ST from multiple genes through Hallmark's pathways, we identify hidden structures in 2D views, including the 3D invasive tumor front. In summary, VORTEX, by leveraging AI, 3D tissue morphology and 2D ST, generates 3D spatial molecular maps in a reliable, efficient and scalable manner. The combined morphomolecular analysis of the 3D tumor context can provide a novel perspective of the TME for improved characterization of PRAD heterogeneity. Citation Format: Cristina Almagro-Pérez, Andrew Song, Luca Weishaupt, Ahrong Kim, Guillaume Jaume, Konstantin Hemker, Drew F.K. Williamson, Stephanie Pei Tung Yiu, Qinghua Han, Renao Yan, Elena Baraznenok, Long Phi Le, Alexander S. Baras, Ali Bashashati, Sizun Jiang, Jonathan T.C. Liu, Faisal Mahmood. AI-driven 3D spatial transcriptomics for 3D tumor microenvironment mapping in prostate adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 77.

Introduction Detection of genomic alterations and the presence of molecular residual disease (MRD) through the assessment of peripheral blood circulating tumor DNA (ctDNA) has important clinical relevance in muscle-invasive (MIBC) and advanced bladder cancer patients. Similarly, analysis of tumor DNA within patient urine specimens (utDNA) presents an intriguing option to optimize clinical care and investigate tumor biology in NMIBC patients. We investigated the utility of utDNA to characterize the genomic landscape and monitor MRD status in longitudinally collected urine samples from NMIBC patients treated with durvalumab containing regimens in the previously reported HCRN GU16-243: ADAPT-BLADDER Trial (Hahn N et al, Eur Urol 2023). Methods Baseline peripheral blood mononuclear cell and longitudinal urine samples at baseline, 3-month, 6-month, tumor recurrence (if observed), and 24-month (if available) time points were collected from BCG-unresponsive NMIBC patients treated with durvalumab monotherapy (D), durvalumab plus BCG (D + BCG), or durvalumab plus external beam radiation therapy (D + EBRT) in the HCRN 16-243: ADAPT-BLADDER Trial. utDNA mutations, copy number aberrations, tumor mutational burden (TMB), and MRD status were assessed by the PredicineWES+ and PredicineBEACON MRD assays respectively. Mutational, copy number aberrations, and TMB data were summarized in graphical form. Associations between post-treatment utDNA MRD status and clinical outcomes were assessed by generalized estimating equation (GEE) regression testing. Results Sixty-six utDNA samples from 26 patients were collected. Sixty-five (98.5%) samples were sufficient for analysis (Baseline n=25; 3-month n=19; 6-month n=16; recurrence n=1; 24-month n=4). Baseline utDNA was detectable in all (100%) patients with a median of 75 genomic alterations per sample, a median TMB of 1.4 muts/Mb (IQR 0.6 – 3.5), and a median tumor fraction (TF) of 21.0% (IQR 9.9 – 37.0). Frequent alterations in TERT, TP53, KMT2D, ARID1A, ATM, RB1, ERBB2, KDM6A, CDKN2A, and MDM2 were detected. Following treatment, MRD remained detectable in 36 (90%) samples (3-month – 17/19 (90%); 6-month – 15/16 (94%); recurrence – 1/1 (100%); 24-month – 3/4 (75%)) with 4 samples (10%) demonstrating undetectable MRD. In a multivariate GEE analysis, TF declined from baseline to cycle 4 (p=0.057), and to cycle 8 (p=0.039). TF was significantly lower in patients with RFS durations greater than 12 months, p < 0.001. Collection of utDNA specimens in additional treatment cohorts is ongoing. Conclusions Detection and monitoring of utDNA MRD proved feasible in this initial investigation in BCG-unresponsive NMIBC patients treated with durvalumab containing regimens. Post-treatment utDNA MRD was detected in most samples. Post-treatment reduction of utDNA tumor fraction was associated with durable clinical benefit. Validation of the utility of utDNA MRD status as a novel biomarker of response in NMIBC patients is ongoing in larger randomized prospective trials.

MSK fits. Cox and neural net model fits were mean normalized and averaged over 10 K-folds. TMB distributions shown as rug plots.

Simulated step function metrics. Log-likelihood and C-indexes for the test folds of a simulated step function dataset for either a Cox model, a FCN neural network, and a neural network comprised of a single neuron with sigmoid activation.

Model variability across folds. A, mean-normalized model fits for each training fold for an FCN model with simulated linear data. B, mean-normalized model fits for each training fold for an FCN model with simulated non-monotonic data.

A comprehensive three-dimensional (3D) map of tissue architecture and gene expression is crucial for illuminating the complexity and heterogeneity of tissues across diverse biomedical applications. However, most spatial transcriptomics (ST) approaches remain limited to two-dimensional (2D) sections of tissue. Although current 3D ST methods hold promise, they typically require extensive tissue sectioning, are complex, are not compatible with non-destructive 3D tissue imaging technologies, and often lack scalability. Here, we present VOlumetrically Resolved Transcriptomics EXpression (VORTEX), an AI framework that leverages 3D tissue morphology and minimal 2D ST to predict volumetric 3D ST. By pretraining on diverse 3D morphology-transcriptomic pairs from heterogeneous tissue samples and then fine-tuning on minimal 2D ST data from a specific volume of interest, VORTEX learns both generic tissue-related and sample-specific morphological correlates of gene expression. This approach enables dense, high-throughput, and fast 3D ST, scaling seamlessly to large tissue volumes far beyond the reach of existing 3D ST techniques. By offering a cost-effective and minimally destructive route to obtaining volumetric molecular insights, we anticipate that VORTEX will accelerate biomarker discovery and our understanding of morphomolecular associations and cell states in complex tissues. Interactive 3D ST volumes can be viewed at https://vortex-demo.github.io/

667 Background: A critical need persists to develop treatments for BCG-unresponsive (BCG-U) non-muscle invasive bladder cancer (NMIBC) patients (pts). The intravesical gemcitabine plus docetaxel (Gem/Doc) doublet and intravenous agents targeting the PD-(L)1 immune checkpoint have both demonstrated complete responses (CRs) in BCG-U NMIBC investigations. With this knowledge, we aimed to assess the clinical efficacy and safety of durvalumab (D) in combination with intravesical Gem/Doc. Methods: The multi-arm, multi-stage ADAPT-BLADDER trial design has been previously described (Hahn NM et al, Eur Urol 2023). Here, we report outcomes from the D + Gem/Doc (cohort 4) phase 1 and phase 2 expansion arms. In phase 1, BCG-U NMIBC patients were enrolled in a 6 + 3 + 3 fashion to establish safety. In phase 2, additional patients were enrolled to evaluate the primary endpoint of CR rate in the total study population and to provide a CR rate estimate within the subset of patients with CIS. Per protocol, phase 1 and 2 efficacy analyses were combined. Enrollment of pure papillary patients was capped to ensure at least 20 patients with CIS. Patients received D 1500 mg iv on day 1 of each 4-week cycle for up to 6 cycles. In addition, they received intravesical Gem 1000 mg + Doc 37.5 mg weekly for the first 6 weeks. Patients achieving a CR were encouraged, but not required, to receive Gem/Doc monthly maintenance therapy. Cystoscopic and urine cytology assessments were performed every three months in year one with a mandatory biopsy at 12-months in responding patients. Toxicity rates were reported per CTCAE v5.0. Results: Between 1/2022-10/2024, 40 pts enrolled (12 phase 1, 28 phase 2) from 6 sites. The study completed its full planned accrual. Demographics included: median age 69 years; 83% male; CIS (8 pts), high-grade (HG) T1 + CIS (7 pts), HG Ta + CIS (6 pts), HG Ta (13 pts) and HG T1 (6 pts). No dose limiting toxicities were observed in the phase 1 portion. Among 27 patients (15 CIS, 12 papillary) evaluable for response at the August 2024 data lock, a CR was observed in 24 patients (89%) (CIS – 13/15 (87%); Papillary – 11/12 (92%)). Evaluation of pts still receiving study treatment for CR and durability of response is ongoing. One pt (4%) progressed to muscle invasion on study treatment. Highest grade treatment-related adverse events observed were grade 1 – 12 (36%) pts, grade 2 – 11 (33%) pts, grade 3 – 2 (6%) pts (sepsis n=1, pneumonitis n=1), and grade 4 – 1 (3%) pt (cough n=1) respectively. One on-study death due to retroperitoneal bleed unrelated to study therapy was observed. Conclusions: Combination treatment with durvalumab plus intravesical gemcitabine and docetaxel demonstrates promising clinical efficacy with a high complete response rate. Observed adverse event type, frequency, and severity were consistent with prior durvalumab trial experiences. Clinical trial information: NCT03317158 .

Multiple Instance Learning (MIL) has shown potential for analyzing Whole Slide Images (WSIs) in digital pathology, but it faces challenges related to redundant information learning and generalization due to limited supervision and the computational complexity of Gigapixel WSIs. Many MIL-based methods apply a small weight matrix to all WSI patches. In this study, we focus on developing computationally efficient models that improve MIL-based WSI classification by processing fewer patches while improving performance. We propose an attention-based approach using knowledge distillation, where a compute-intensive “instructor” model analyzes all WSI patches to train a resourceefficient “learner” model, which considers only a subset of patches. Comprehensive evaluations on four cancer subtype datasets—TCGA-BRCA, TCGA-NSCLC, TCGA-RCC, and PANDA—demonstrate that an “observe-everything” instructor can effectively train an “observe-minimally” learner network. Overall, our proposed learner network enhances performance by 4% compared to the state-of-the-art, while reducing inference time by 45% and FLOPs by approximately 88%.

150 Background: Current guidelines recommend NGS testing for all patients with metastatic prostate cancer, and yet this remains underutilized. In addition, there is uncertainty on when patients should undergo NGS testing. This study aims to determine the timing and characteristics of patients undergoing NGS testing at a single academic institution. Methods: Patients with metastatic prostate cancer diagnosed between January 2020 and January 2024 who underwent NGS testing were identified. We recorded patient and disease characteristics at time of diagnosis and time of metastatic disease. We used a time-varying cox hazard model to investigate the associations of disease factors (M stage, Gleason grade, castration resistant prostate cancer), social characteristics (race, smoking history, area deprivation index (ADI)) of the participants and the timing of NGS testing. Additionally, we analyzed median overall survival from date of metastatic disease comparing those with early (pre castration resistant prostate cancer (CRPC)) NGS testing versus late NGS (after CRPC) testing. Results: 205 patients were included. Median age at testing was 70 years [range 50-94]. 68% of participants identified as White, 26% as Black and 84% were non-Hispanic (15% unknown). At diagnosis, 56% had metastatic disease, 57% had Gleason grade group 5, 35% had T3/4 disease. 74% had ECOG 0-1 at time of metastasis. The most common source of tissue for testing was the prostate (67%) followed by blood (13%). 44% of patients developed CRPC by the last follow up date. Median time from metastatic disease to NGS testing was 2.92 months in Black patients, 2.25 in White and 1.54 in those of other races. Of the deceased (n = 66), 6.06% had testing within 3 months of death and 24.24% within 6 months of death. In multivariable time-varying analyses, higher Gleason group (HR range: 5.41-7.27, p < 0.03) and metastasis at diagnosis (12.42 [6.22, 24.78], p < 0.001) were significantly associated with increased hazard of NGS testing and thus implies a higher likelihood of earlier testing. Age at diagnosis (1.05 [1.02,1.08], p < 0.001) was also significant with each additional year increasing the hazard of NGS testing by 5%. For overall survival, the median time for early testing patients was 58.7 months versus 27.2 months for late testing patients ( p < 0.001). The relationship between NGS testing and ADI is undergoing analysis and will be shared at the meeting. Conclusions: Older age, Gleason grade group, and metastasis at diagnosis were associated with earlier NGS testing among men with metastatic prostate cancer. Nearly a quarter of men who died had testing within the last 6 months of life. As targeted therapies move earlier into treatment of metastatic prostate cancer, facilitating additional tissue acquisition and understanding of use and limitations of liquid biopsies are needed to facilitate earlier NGS testing in metastatic prostate cancer.

BPC fits. Cox and neural net model fits were mean normalized and averaged over 10 K-folds. TMB distributions shown as rug plots.

Multimodal survival methods combining gigapixel histology whole-slide images (WSIs) and transcriptomic profiles are particularly promising for patient prognostication and stratification. Current approaches involve tokenizing the WSIs into smaller patches (>10,000 patches) and transcriptomics into gene groups, which are then integrated using a Transformer for predicting outcomes. However, this process generates many tokens, which leads to high memory requirements for computing attention and complicates post-hoc interpretability analyses. Instead, we hypothesize that we can: (1) effectively summarize the morphological content of a WSI by condensing its constituting tokens using morphological prototypes, achieving more than 300x compression; and (2) accurately characterize cellular functions by encoding the transcriptomic profile with biological pathway prototypes, all in an unsupervised fashion. The resulting multimodal tokens are then processed by a fusion network, either with a Transformer or an optimal transport cross-alignment, which now operates with a small and fixed number of tokens without approximations. Extensive evaluation on six cancer types shows that our framework outperforms state-of-the-art methods with much less computation while unlocking new interpretability analyses.

An integral stage in typical digital pathology workflows involves deriving specific features from tiles extracted from a tessellated whole-slide image. Notably, various computer vision neural network architectures, particularly the ImageNet pretrained, have been extensively used in this domain. This study critically analyzes multiple strategies for encoding tiles to understand the extent of transfer learning and identify the most effective approach. The study categorizes neural network performance into 3 weight initialization methods: random, ImageNet-based, and self-supervised learning. Additionally, we propose a framework based on task-specific self-supervised learning, which introduces a shallow feature extraction method, employing a spatial-channel attention block to glean distinctive features optimized for histopathology intricacies. Across 2 different downstream classification tasks (patch classification and weakly supervised whole-slide image classification) with diverse classification data sets, including colorectal cancer histology, Patch Camelyon, prostate cancer detection, The Cancer Genome Atlas, and CIFAR-10, our task-specific self-supervised encoding approach consistently outperforms other convolutional neural network-based encoders. The better performances highlight the potential of task-specific attention-based self-supervised training in tailoring feature extraction for histopathology, indicating a shift from using pretrained models originating outside the histopathology domain. Our study supports the idea that task-specific self-supervised learning allows domain-specific feature extraction, encouraging a more focused analysis.

Advancements in imaging technologies have revolutionized our ability to deeply profile pathological tissue architectures, generating large volumes of imaging data with unparalleled spatial resolution. This type of data collection, namely, spatial proteomics, offers invaluable insights into various human diseases. Simultaneously, computational algorithms have evolved to manage the increasing dimensionality of spatial proteomics inherent in this progress. Numerous imaging-based computational frameworks, such as computational pathology, have been proposed for research and clinical applications. However, the development of these fields demands diverse domain expertise, creating barriers to their integration and further application. This review seeks to bridge this divide by presenting a comprehensive guideline. We consolidate prevailing computational methods and outline a roadmap from image processing to data-driven, statistics-informed biomarker discovery. Additionally, we explore future perspectives as the field moves toward interfacing with other quantitative domains, holding significant promise for precision care in immuno-oncology.

BACKGROUND: Sarcomatoid urothelial cancer of the bladder (SBC) is a rare, but aggressive histological subtype for which novel treatments are needed. OBJECTIVE: We evaluated the clinical activity and safety of neoadjuvant cisplatin plus gemcitabine plus docetaxel (CGD) in muscle-invasive patients with SBC and assessed SBC tumor biology by whole transcriptome RNA sequencing. METHODS: A single-institution, retrospective analysis of muscle-invasive SBC patients treated with neoadjuvant CGD with molecular analysis. Patients received cisplatin 35 mg/m2 + gemcitabine 800 mg/m2 + docetaxel 35 mg/m2 intravenously on days 1 and 8 + pegfilgrastim 6 mg subcutaneously on day 9 every 3 weeks for 4 cycles followed by cystectomy. The primary endpoint was pathologic complete response (ypCR) rate. RESULTS: Sixteen patients with SBC received neoadjuvant CGD with a ypCR rate of 38% and a < ypT2 rate of 50%. Grade 3 and 4 toxicity occurred in 80% and 40% of patients, but was manageable with 81% of patients completing > 3 CGD cycles. Whole transcriptome RNA sequencing demonstrates co-clustering of SBC with conventional urothelial tumors. SBC tumors are characterized by basal-squamous and stroma rich gene signatures with frequent increased expression of immune checkpoint ( CD274 (PD-L1)), chemokine ( CXCL9), and T-cell ( CD8A) genes. CONCLUSIONS: SBC is a chemosensitive subtype, with ypCR rate similar to urothelial bladder cancer following CGD neoadjuvant therapy. Whole transcriptome tissue analyses demonstrate increased expression of immune checkpoint and T-cell genes with therapeutic implications.

Human tissue, which is inherently three-dimensional (3D), is traditionally examined through standard-of-care histopathology as limited two-dimensional (2D) cross-sections that can insufficiently represent the tissue due to sampling bias. To holistically characterize histomorphology, 3D imaging modalities have been developed, but clinical translation is hampered by complex manual evaluation and lack of computational platforms to distill clinical insights from large, high-resolution datasets. We present TriPath, a deep-learning platform for processing tissue volumes and efficiently predicting clinical outcomes based on 3D morphological features. Recurrence risk-stratification models were trained on prostate cancer specimens imaged with open-top light-sheet microscopy or microcomputed tomography. By comprehensively capturing 3D morphologies, 3D volume-based prognostication achieves superior performance to traditional 2D slice-based approaches, including clinical/histopathological baselines from six certified genitourinary pathologists. Incorporating greater tissue volume improves prognostic performance and mitigates risk prediction variability from sampling bias, further emphasizing the value of capturing larger extents of heterogeneous morphology.

ABSTRACT Potential clinical biomarkers are often assessed with Cox regressions or their ability to differentiate two groups of patients based on a single cutoff. However, both of these approaches assume a monotonic relationship between the potential biomarker and survival. Tumor mutational burden (TMB) is currently being studied as a predictive biomarker for immunotherapy, and a single cutoff is often used to divide patients. In this study we introduce a two-cutoff approach that allows splitting of patients when a non-monotonic relationship is present, and explore the use of neural networks to model more complex relationships of TMB to outcome data. Using real-world data we find that while in most cases the true relationship between TMB and survival appears monotonic, that is not always the case and researchers should be made aware of this possibility. Significance When a non-monotonic relationship to survival is present it is not possible to divide patients by a single value of a predictor. Neural networks allow for complex transformations and can be used to correctly split patients when a non-monotonic relationship is present.

Pathologic response is an endpoint in many ongoing clinical trials for neoadjuvant regimens, including immune checkpoint blockade and chemotherapy. Whole-slide scanning of glass slides generates high-resolution digital images and allows for remote review and potential measurement with image analysis tools, but concordance of pathologic response assessment on digital scans compared with that on glass slides has yet to be evaluated. Such a validation goes beyond previous concordance studies, which focused on establishing surgical pathology diagnoses, as it requires quantitative assessment of tumor, necrosis, and regression. Further, as pathologic response assessment is being used as an endpoint, such concordance studies have regulatory implications. The purpose of this study was 2-fold, which was as follows: first, to determine the concordance between pathologic response assessed on glass slides and that assessed on digital scans, and second, to determine if pathologists benefited from using measurement tools when determining pathologic response. To that end, hematoxylin and eosin-stained glass slides from 64 non-small cell lung carcinoma specimens were visually assessed for percent residual viable tumor (%RVT). The sensitivity and specificity for digital vs glass reads of pathologic complete response (0% RVT) and major pathologic response (≤10% RVT) were all >95%. When %RVT was considered as a continuous variable, the intraclass correlation coefficient of digital vs glass reads was 0.94. The visual assessments of pathologic response were supported by pathologist annotations of residual tumor and tumor bed areas. In a separate subset of hematoxylin and eosin-stained glass slides, several measurement approaches to quantifying %RVT were performed. Pathologist estimates strongly reflected measured %RVT. This study demonstrates the high level of concordance between glass slides evaluated using light microscopy and digital whole-slide images for pathologic response assessments. Pathologists did not require measurement tools to generate robust %RVT values from slide annotations. These findings have broad implications for improving clinical workflows and multisite clinical trials.