Reed, Katelyn A.

Abstract Objectives Traditional laboratory regulatory frameworks provide robust guidance for conventional clinical testing but lack requirements for data documentation for artificial intelligence and machine learning (AI/ML) solutions. This gap creates significant risks since AI solutions directly inherit patterns, biases, and limitations from their training data. Here, we describe the development and implementation of a comprehensive data documentation checklist with accompanying data lineage for use within an AI lifecycle framework for clinical laboratories. Methods Building on the previously established Clinical AI Readiness Evaluator (CARE) framework, we developed a comprehensive data checklist and data lineage methodology through a multiphase process, including (1) comprehensive review of existing AI/ML data documentation frameworks, (2) focused meetings with 3 institutional AI operations teams, and (3) 3 rounds of iterative refinement by our multidisciplinary team. The checklist’s effectiveness was then assessed using 3 diverse AI/ML projects moving through the CARE framework. Results The CARE Data Checklist and Data Lineage provide a structured approach to documenting critical aspects of datasets used in AI/ML projects, including data composition, demographics, collection methods, labeling processes, usage constraints, maintenance requirements, and a data readiness assessment. The checklist addresses unique data-centric challenges of AI/ML applications, facilitating transparency, reproducibility, and regulatory compliance. Conclusions The CARE Data Checklist and Data Lineage serve as both a technical guide and a communication tool bridging gaps between technical and clinical stakeholders. By working on these documents early in the AI lifecycle, laboratories can anticipate and address data-related challenges, ultimately saving time, optimizing resources, and improving the reliability of AI-augmented laboratory solutions.

Abstract Background: Recent advances in computer vision have given rise to image foundation models (FMs), large-scale neural networks pre-trained on millions of unlabeled histopathology patches, which can learn generalizable representations of tissue morphology. In many cancers, these FMs have predicted genetic mutations and prognoses directly from histology, suggesting that morphology encodes “hidden” molecular information. In lymphoma, this field is nascent but growing.General-purpose histopathology FMs have limited representation of lymphoid morphology. We developed and evaluated LymphoVision, a lymphoma-specialized image foundation model pretrained with self-supervised learning, to extract clinically meaningful features from routine hematoxylin and eosin (H&E) histology. Methods: LymphoVision was pre-trained using knowledge distillation and consistency (DINOv2) on over 37 million multi-resolution patches (5x, 10x, 20x, 40x) sampled using a density-aware strategy from 31,211 H&E whole slide images (WSIs) from 9,155 archival lymphoma and reactive lymphoproliferative cases. To ensure diverse, representative samples, our patch sampling method uses an efficient graph-based clustering algorithm and prioritizes morphologically informative regions by clustering patch embeddings and performing density sampling within each cluster. The pretrained encoder was evaluated using a clustering-constrained attention multiple instance learning (MIL) framework on three diagnostic tasks: (1) cell-of-origin (COO) classification in diffuse large B-cell lymphoma (DLBCL), distinguishing germinal center B-cell (GCB) from non-GCB subtypes based on the Hans immunohistochemistry algorithm; (2) follicular lymphoma (FL) grading, differentiating low-grade (grades 1–2) from high-grade (grade 3); and (3) multiclass lymphoma subtype classification across six categories: DLBCL, FL, marginal zone lymphoma (MZL), mantle cell lymphoma (MCL), classical Hodgkin lymphoma (CHL), and benign/reactive lymphoid conditions. WSI-level labels were derived from pathology reports, and none of the evaluation cases were included in pretraining. Models used a ViT-Giant LymphoVision backbone to generate 20x patch embeddings, followed by MIL using 50% training data and 2-fold cross-validation (25% validation, 25% test). Performance was assessed using area under the receiver operating characteristic curve (AUC), accuracy, recall (sensitivity) and F1 score. Results: For DLBCL COO classification (total n=358 cases), LymphoVision achieved a mean test AUC of 0.93 (range 0.92–0.94), accuracy of 84.5% and F1 of 0.85 across two independent test folds (n=91 per fold, 50% GCB). These results outperformed general-purpose FMs on the same task and cases, including Virchow-v2 (AUC 0.85) and UNI-v2 (AUC 0.88), despite those models being pretrained on 10–100 times more images. For FL grading (total n=660 cases), the test set in each fold included 166 cases (18% grade 3), yielding a mean test AUC of 0.88 (range 0.87–0.90), accuracy of 89.2% and F1 of 0.80. In multiclass lymphoma subtype classification (total n=1,176 cases), the test set included 296 cases per fold. LymphoVision achieved a mean test AUC of 0.98 (range 0.97-0.99), mean test accuracy of 87.1% and mean F1 of 0.84. Class-wise recall was 92.0% for DLBCL (n=112), 88.2% for FL (n=72), 77.2% for MZL (n=29), 70.8% for MCL (n=35), 87.5% for CHL (n=24), and 95.8% for benign/reactive lymphoid conditions (n=24). Conclusion: LymphoVision is a lymphoma-specialized FM trained using self-supervision that achieves high diagnostic accuracy across multiple clinically meaningful tasks using weak supervision and standard H&E images, suggesting the encoder has learned fine-grained lymphoma morphology. Based on historical benchmarks from the 1997 International Lymphoma Study Group classification project, expert hematopathologists achieved diagnostic agreement rates of 55% to 84% using H&E morphology alone for the non-Hodgkin lymphoma (NHL) entities included in our multiclass task. LymphoVision approached or exceeded these benchmarks. These findings underscore the value of our approach using disease-specialized pretraining and intelligent patch sampling for data-efficient histopathology learning. LymphoVision has the potential to augment diagnostics, enable scalable AI-powered classification in real-world workflows, and serve as a foundation for prognostic and theragnostic applications. External validation and multimodal integration are ongoing.

Abstract Objective Despite exponential growth in artificial intelligence (AI) research for laboratory medicine and pathology, a significant gap exists between model development and clinical AI implementation. This article introduces a structured framework, the Clinical AI Readiness Evaluator (CARE), to bridge this gap and support the responsible adoption of AI in clinical laboratory settings. Methods Building upon the Machine Learning Technology Readiness Levels framework, we developed CARE specifically for the clinical laboratory environment by incorporating health care–specific requirements, regulatory considerations, and workflow integration needs. This framework was iteratively refined through practical application across diverse AI use cases within laboratory medicine and pathology. Results The CARE framework provides a systematic approach to AI development and implementation through 8 component workstreams: clinical use case, data, data pipeline, code, clinical user experience, clinical technology infrastructure, clinical orchestration, and regulatory compliance. Unlike generic AI frameworks, CARE distinguishes itself by emphasizing both health care and laboratory workflow integration, regulatory requirements, ethical considerations, and comprehensive validation for clinical contexts. The framework accommodates both internally developed models and commercial AI solutions, providing clear guidance through technology readiness levels and structured review processes. Conclusions The CARE framework addresses the unique challenges of implementing AI in laboratory medicine and pathology by providing a comprehensive roadmap from initial concepts through clinical deployment and maintenance. This article, the first in a series of 4, establishes the foundational AI lifecycle framework, while subsequent articles will explore data documentation, ethical AI considerations, and governance structures. By adopting this structured approach, laboratories can responsibly harness AI’s potential to enhance diagnostic accuracy and operational efficiencies and, ultimately, improve patient care.

Several reports describing a rare primary liver tumor with histologic features reminiscent of follicular thyroid neoplasms have been published under a variety of descriptive terms including thyroid-like, solid tubulocystic, and cholangioblastic cholangiocarcinoma. Although these tumors are considered to represent histologic variants, they lack classic features of cholangiocarcinoma and have unique characteristics, namely immunoreactivity for inhibin and NIPBL::NACC1 fusions. The purpose of this study is to present clinicopathologic and molecular data for a large series of these tumors to better understand their pathogenesis. We identified 11 hepatic tumors with these features. Immunohistochemical and NACC1 and NIPBL fluorescence in situ hybridization assays were performed on all cases. Four cases had available material for whole-genome sequencing (WGS) analysis. Most patients were adult women (mean age: 42 y) who presented with abdominal pain and large hepatic masses (mean size: 14 cm). Ten patients had no known liver disease. Of the patients with follow-up information, 3/9 (33%) pursued aggressive behavior. All tumors were composed of bland cuboidal cells with follicular and solid/trabecular growth patterns in various combinations, were immunoreactive for inhibin, showed albumin mRNA by in situ hybridization, and harbored the NIPBL::NACC1 fusion by fluorescence in situ hybridization. WGS corroborated the presence of the fusion in all 4 tested cases, high tumor mutational burden in 2 cases, and over 30 structural variants per case in 3 sequenced tumors. The cases lacked mutations typical of conventional intrahepatic cholangiocarcinoma. In this report, we describe the largest series of primary inhibin–positive hepatic neoplasms harboring a NIPBL::NACC1 fusion and the first WGS analysis of these tumors. We propose to name this neoplasm NIPBL:NACC1 fusion hepatic carcinoma.

AbstractBackgroundMT‐RNR1 variants are a well‐known cause of aminoglycoside‐induced hearing loss (AIHL). Individuals with cystic fibrosis (CF) routinely receive aminoglycosides and are at high risk of AIHL. However, genetic testing before treatment is not routinely performed due to perceived rarity of risk, and cost ineffectiveness with traditional technologies.AimAssess the utility of large‐scale screening for AIHL risk in the CF population, using digital droplet polymerase chain reaction (ddPCR), a novel and scalable low‐cost molecular technique.MethodsUsing a clinically validated ddPCR assay, we performed retrospective testing on 122 and prospective testing on 32 individuals with CF for the two most common pathogenic variants associated with AIHL, MT‐RNR1 m.1555 A > G and m.1494 C > T. Our study screened the largest known cohort of pediatric cases of CF (94/154) for these specific alterations.ResultsWe identified two individuals positive for MT‐RNR1 m.1555 A > G and no positives for m.1494 C > T. Of 32 prospective cases, 17 had aminoglycoside exposure. The positive case in our prospective group recently began inhaled tobramycin and denied hearing issues. The clinician adjusted to care for both the patient and sibling with CF (not included in cohort) who is presumed positive for m.1555 A > G due to the nature of mitochondrial inheritance.ConclusionOur findings demonstrate the utility of pretreatment screening in the cystic fibrosis population for AIHL risk using ddPCR, a scalable and robust testing methodology at a fraction of the cost as compared to other sequencing‐based methods. Therefore, the use of large‐scale screening for AIHL risk in the cystic fibrosis community should be re‐visited.

Objective The objective of this study is to determine the association of proliferation indices and pathologic biomarkers on overall and recurrence/metastasis‐free survival (OS and RMFS) in patients with sinonasal mucosal melanoma (SNMM) and to assess the genetic mutational landscape of SNMM. Methods This is a retrospective cohort study of 45 SNMM patients without neoadjuvant therapy who underwent surgical therapy with curative intent and had tumor tissue available for histopathologic review, molecular analysis, and genetic mutational assessment. The OS and RMFS were assessed for associations with numerous tumor and patient‐related factors. Results Among proliferative indices, higher Ki67 and mitotic rates were associated with worsened OS and RMFS (Ki67: p = 0.0007 and p < 0.0001; mitotic rate: p = 0.005 and p = 0.0009, respectively). The presence of brisk tumor‐infiltrating lymphocytes (TILs) was associated with improved RMFS ( p = 0.007) and the presence of lymphovascular invasion was associated with worsened OS and RMFS ( p = 0.02 and p = 0.04, respectively). Patients with amelanotic tumors were more likely to have higher T‐stage ( p = 0.046), less likely to have brisk TILs ( p = 0.02) and had worsened RMFS ( p = 0.03). Patients on immunotherapy with tumor Ki67 < 40% had better 3‐year OS compared to those with higher Ki67 index ( p = 0.004). Actionable genetic mutations such as BRAF V600E are rare and present in only 1 of 20 patients tested. Conclusion In SNMM patients, pathologic and proliferation markers such as Ki67, mitotic rate and brisk TILs are associated with survival and may be considered in future staging systems. Clinical response to immunotherapy appears to correlate with the Ki67 index. Given the distinct genetic profile of SNMM, targeted therapies against the MAPK kinase pathway have limited utility. Level of Evidence 3 Laryngoscope , 132:2350–2358, 2022

Rapid and accurate diagnostic testing is essential to bring the ongoing COVID-19 pandemic to an end. As the demand for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) testing continues to increase amid supply shortages, many laboratories have investigated the use of sources other than nasopharyngeal (NP) swabs.

Candida auris is an emerging fungal pathogen with cases reported in countries around the world and in 19 states within the United States as of August 2020. The CDC has recommended that hospitals perform active surveillance upon admission for patients with the appropriate risk factors. Currently, active surveillance requires that local hospitals send surveillance swabs to a public health laboratory for analysis. In this work, a real-time PCR assay was developed for the specific detection of C. auris from surveillance swabs, blood, and urine to enable rapid detection of this pathogen. The assay uses commercially available primers and reporter probes and it was verified on the LightCycler 480 PCR platform. Contrived specimens and prospectively collected composite groin/axilla surveillance swabs were used to validate the assay. The performance of the PCR assay on surveillance swabs was also compared to a second PCR assay targeting C. auris that was performed at the Minnesota Department of Health–Public Health Laboratory (MDH-PHL). Our PCR assay is able to detect and differentiate C. auris from closely related Candida species such as C. duobushaemulonii, C. haemulonii, and C. pseudohaemulonii on the basis of melting curve temperature differences.

Abstract Background Rapid blood culture diagnostics are of unclear benefit for patients with gram-negative bacilli (GNB) bloodstream infections (BSIs). We conducted a multicenter, randomized, controlled trial comparing outcomes of patients with GNB BSIs who had blood culture testing with standard-of-care (SOC) culture and antimicrobial susceptibility testing (AST) vs rapid organism identification (ID) and phenotypic AST using the Accelerate Pheno System (RAPID). Methods Patients with positive blood cultures with Gram stains showing GNB were randomized to SOC testing with antimicrobial stewardship (AS) review or RAPID with AS. The primary outcome was time to first antibiotic modification within 72 hours of randomization. Results Of 500 randomized patients, 448 were included (226 SOC, 222 RAPID). Mean (standard deviation) time to results was faster for RAPID than SOC for organism ID (2.7 [1.2] vs 11.7 [10.5] hours; P &lt; .001) and AST (13.5 [56] vs 44.9 [12.1] hours; P &lt; .001). Median (interquartile range [IQR]) time to first antibiotic modification was faster in the RAPID arm vs the SOC arm for overall antibiotics (8.6 [2.6–27.6] vs 14.9 [3.3–41.1] hours; P = .02) and gram-negative antibiotics (17.3 [4.9–72] vs 42.1 [10.1–72] hours; P &lt; .001). Median (IQR) time to antibiotic escalation was faster in the RAPID arm vs the SOC arm for antimicrobial-resistant BSIs (18.4 [5.8–72] vs 61.7 [30.4–72] hours; P = .01). There were no differences between the arms in patient outcomes. Conclusions Rapid organism ID and phenotypic AST led to faster changes in antibiotic therapy for gram-negative BSIs. Clinical Trials Registration NCT03218397.