Alternative Strategies to Provide Actionable Results When a Supply of Urinalysis Strips Is Unavailable.

CONTEXT.­: Urinalysis instrument-specific dip strips offer physicians qualitative results for actionable analytes (protein, glucose, leukocyte esterase, nitrates, hemoglobin, and ketones). OBJECTIVE.­: To explain a strategy implemented to support clinical decision-making by providing urine quantification of protein, glucose, white blood cells (WBCs), and red blood cells because of urine strip shortages. DESIGN.­: During shortages, we implemented an automated algorithm that triggered sending urine samples to the automation line for quantification of protein and glucose and ensured that urine microscopy was performed to obtain WBC and red blood cell counts. The algorithm printed 2 labels so nursing staff would collect 2 specimens. We monitored the turnaround time from the specimen being received in the laboratory to result verification, ensured that the culture reflex order was triggered, and tracked complaints by physicians regarding not having usual urinalysis results. Prior to implementation, correlation between sample types for protein and glucose measurement was found acceptable. RESULTS.­: The algorithm was put in place twice during 2022. The turnaround time of urine microscopic study was identical to that obtained when the urinalysis was done with the strips; however, the quantification of glucose and protein took approximately 30 minutes more. Urine reflex cultures were triggered correctly with the algorithm, as they were derived entirely from a WBC count higher than 10 per high-power field. During the shortage period we had only 1 complaint, by a physician wanting to have results of nitrates. CONCLUSIONS.­: During urine strip shortages, we successfully implemented a diversion algorithm that provided actionable urinalysis analytes in a timely manner with minimal provider complaints.

Regulation of antigen-specific T cell infiltration and spatial architecture in multiple myeloma and premalignancy.

Entry of antigen-specific T cells into human tumors is critical for immunotherapy, but the underlying mechanisms are poorly understood. Here, we combined high-dimensional spatial analyses with in vitro and in vivo modeling to study the mechanisms underlying immune infiltration in human multiple myeloma (MM) and its precursor monoclonal gammopathy of undetermined significance (MGUS). Clustered tumor growth was a feature of MM but not MGUS biopsies, and this growth pattern was reproduced in humanized mouse models. MM biopsies exhibited intralesional as well as spatial heterogeneity, with coexistence of T cell-rich and T cell-sparse regions and the presence of areas of T cell exclusion. In vitro studies demonstrated that T cell entry into MM clusters was regulated by agonistic signals and CD2-CD58 interactions. Upon adoptive transfer, antigen-specific T cells localized to the tumor site but required in situ DC-mediated antigen presentation for tumor entry. C-type lectin domain family 9 member A-positive (CLEC9A+) DCs appeared to mark portals of entry for gradients of T cell infiltration in MM biopsies, and their proximity to T cell factor 1-positive (TCF1+) T cells correlated with disease state and risk status. These data illustrate a role for tumor-associated DCs and in situ activation in promoting the infiltration of antigen-specific T cells in MM and provide insights into spatial alterations in tumor/immune cells with malignant evolution.

Comparative diets and prey resource competition between early-juvenile common snook Centropomus undecimalis and non-native pike killifish Belonesox belizanus.

Pike killifish Belonesox belizanus is an established non-native fish species in Florida, USA, that was first documented in south Florida in 1957 and then in Tampa Bay tributaries in 1994. Decreases in small-bodied fish abundances have been linked to the introduction of B. belizanus in both of these regions. Increases in the range and abundance of B. belizanus in the Tampa Bay area and overlap in habitat usage have led to concerns about potential competition with, and predation on, early-juvenile common snook Centropomus undecimalis [≤100 mm standard length (SL)]. Stomach contents of B. belizanus (N = 422; 14-127 mm SL) and early-juvenile C. undecimalis (N = 1132; 5-119 mm SL) were collected to examine the dietary overlap of these two species and potential differences in the diet of early-juvenile C. undecimalis from locations with and without B. belizanus co-occurring. Prey resources were collected by seine to assess prey resource limitation and prey selectivity. Stomach content analysis indicated that there was low overlap in the diet of early-juvenile C. undecimalis and B. belizanus (C ≤ 0.40). Early-juvenile C. undecimalis had a wider diet breadth, consuming many organisms that are not consumed by B. belizanus and which make up a large portion of the early-juvenile C. undecimalis diet. Analysis of prey resources indicated that some prey groups may have lower abundances in locations where B. belizanus are present, with some of these differences reflected in the diet of early-juvenile C. undecimalis. Despite these differences, there was minimal difference in the diet overlap of early-juvenile C. undecimalis from locations with and without B. belizanus co-occurring. Currently B. belizanus appear to be competing minimally with early-juvenile C. undecimalis for prey resources, with no substantial impacts being detected.

An Automated Pipeline for Differential Cell Counts on Whole-Slide Bone Marrow Aspirate Smears.

The pathologic diagnosis of bone marrow disorders relies in part on the microscopic analysis of bone marrow aspirate (BMA) smears and the manual counting of marrow nucleated cells to obtain a differential cell count (DCC). This manual process has significant limitations, including the analysis of only a small subset of optimal slide areas and nucleated cells, as well as interobserver variability due to differences in cell selection and classification. To address these shortcomings, we developed an automated machine learning-based pipeline for obtaining 11-component DCCs on whole-slide BMAs. This pipeline uses a sequential process of identifying optimal BMA regions with high proportions of marrow nucleated cells, detecting individual cells within these optimal areas, and classifying these cells into 1 of 11 DCC components. Convolutional neural network models were trained on 396,048 BMA region, 28,914 cell boundary, and 1,510,976 cell class images from manual annotations. The resulting automated pipeline produced 11-component DCCs that demonstrated a high statistical and diagnostic concordance with manual DCCs among a heterogeneous group of testing BMA slides with varying pathologies and cellularities. Additionally, we demonstrated that an automated analysis can reduce the intraslide variance in DCCs by analyzing the whole slide and marrow nucleated cells within all optimal regions. Finally, the pipeline outputs of region classification, cell detection, and cell classification can be visualized using whole-slide image analysis software. This study demonstrates the feasibility of a fully automated pipeline for generating DCCs on scanned whole-slide BMA images, with the potential for improving the current standard of practice for utilizing BMA smears in the laboratory analysis of hematologic disorders.

Lightweight, open source, easy-use algorithm and web service for paraprotein screening using spatial frequency domain analysis of electrophoresis studies.

Introduction: Serum protein electrophoresis (SPEP) is commonly used to detect monoclonal paraproteins to meet laboratory diagnostic criteria for plasma cell neoplasms. We propose an automated screening method for paraprotein detection that uses minimal computational resources for training and deployment. Methods: A model screening for paraproteins based on the presence of high-frequency components in the spatial frequency spectrum of the SPEP densitometry curve was calibrated on a set of 330 samples, and evaluated on representative (n=110) and external (n=1,321) test sets. The model takes as input a patient's serum densitometry curve and a standardized control curve and outputs a prediction of whether a paraprotein is present. We built an interactive web application allowing users to easily perform paraprotein screening given inputs for densitometry curves, as well as a macro-enabled spreadsheet for easy automated screening. Results: When tuned to maximize likelihood ratio with minimum sensitivity 0.90, the model achieved AUC 0.90, sensitivity 0.90, positive-predictive value 0.64, specificity 0.55, and accuracy 0.72 in the representative test set. In the external test set, the model achieved AUC 0.90, sensitivity 0.97, positive-predictive value 0.42, specificity 0.29, and accuracy 0.52. A subset analysis showed sensitivities of 0.90, 0.96, and 1.0 in detecting low (0.1-0.5 g/dL), medium (0.5-3.0 g/dL), and high paraprotein levels (≥3.0 g/dL), respectively. We have released a web service allowing users to score their own SPEP data, and also released the algorithm and application programming interface in an open-source package allowing users to customize the model to their needs. Conclusions: We developed a proof of concept for an automated method for paraprotein screening using only the characteristics of the SPEP curve. Future work should focus on testing the method with other laboratory data including immunofixation gels, as well as incorporation of outside data sources including clinical data.

Deep Learning-Based Pathology Image Analysis Enhances Magee Feature Correlation With Oncotype DX Breast Recurrence Score.

Background: Oncotype DX Recurrence Score (RS) has been widely used to predict chemotherapy benefits in patients with estrogen receptor-positive breast cancer. Studies showed that the features used in Magee equations correlate with RS. We aimed to examine whether deep learning (DL)-based histology image analyses can enhance such correlations. Methods: We retrieved 382 cases with RS diagnosed between 2011 and 2015 from the Emory University and the Ohio State University. All patients received surgery. DL models were developed to detect nuclei of tumor cells and tumor-infiltrating lymphocytes (TILs) and segment tumor cell nuclei in hematoxylin and eosin (H&E) stained histopathology whole slide images (WSIs). Based on the DL-based analysis, we derived image features from WSIs, such as tumor cell number, TIL number variance, and nuclear grades. The entire patient cohorts were divided into one training set (125 cases) and two validation sets (82 and 175 cases) based on the data sources and WSI resolutions. The training set was used to train the linear regression models to predict RS. For prediction performance comparison, we used independent variables from Magee features alone or the combination of WSI-derived image and Magee features. Results: The Pearson's correlation coefficients between the actual RS and predicted RS by DL-based analysis were 0.7058 (p-value = 1.32 × 10-13) and 0.5041 (p-value = 1.15 × 10-12) for the validation sets 1 and 2, respectively. The adjusted R 2 values using Magee features alone are 0.3442 and 0.2167 in the two validation sets, respectively. In contrast, the adjusted R 2 values were enhanced to 0.4431 and 0.2182 when WSI-derived imaging features were jointly used with Magee features. Conclusion: Our results suggest that DL-based digital pathological features can enhance Magee feature correlation with RS.

PathologyBERT - Pre-trained Vs. A New Transformer Language Model for Pathology Domain.

Pathology text mining is a challenging task given the reporting variability and constant new findings in cancer sub-type definitions. However, successful text mining of a large pathology database can play a critical role to advance 'big data' cancer research like similarity-based treatment selection, case identification, prognostication, surveillance, clinical trial screening, risk stratification, and many others. While there is a growing interest in developing language models for more specific clinical domains, no pathology-specific language space exist to support the rapid data-mining development in pathology space. In literature, a few approaches fine-tuned general transformer models on specialized corpora while maintaining the original tokenizer, but in fields requiring specialized terminology, these models often fail to perform adequately. We propose PathologyBERT - a pre-trained masked language model which was trained on 347,173 histopathology specimen reports and publicly released in the Huggingface1 repository2. Our comprehensive experiments demonstrate that pre-training of transformer model on pathology corpora yields performance improvements on Natural Language Understanding (NLU) and Breast Cancer Diagnose Classification when compared to nonspecific language models.

Rapid Generation of Neutralizing Antibody Responses in COVID-19 Patients.

SARS-CoV-2, the virus responsible for COVID-19, is causing a devastating worldwide pandemic, and there is a pressing need to understand the development, specificity, and neutralizing potency of humoral immune responses during acute infection. We report a cross-sectional study of antibody responses to the receptor-binding domain (RBD) of the spike protein and virus neutralization activity in a cohort of 44 hospitalized COVID-19 patients. RBD-specific IgG responses are detectable in all patients 6 days after PCR confirmation. Isotype switching to IgG occurs rapidly, primarily to IgG1 and IgG3. Using a clinical SARS-CoV-2 isolate, neutralizing antibody titers are detectable in all patients by 6 days after PCR confirmation and correlate with RBD-specific binding IgG titers. The RBD-specific binding data were further validated in a clinical setting with 231 PCR-confirmed COVID-19 patient samples. These findings have implications for understanding protective immunity against SARS-CoV-2, therapeutic use of immune plasma, and development of much-needed vaccines.

A simple electronic tool for assessing amino acid sequence polymorphisms within exon-2 of HLA-DPB1 alleles.

In November 2014, the OPTN/UNOS Board of Directors mandated that HLA-DPB1 typing be performed for all deceased donors. Currently, there are over 1,000 known HLA DPB1 alleles, yet fewer than 30 are represented on commonly used single antigen bead (SAB) solid phase antibody assays. Moreover, the official World Health Organization (WHO) nomenclature for the DPB1 locus does not permit assessment of structural relationships between alleles based on their names. Thus, for donor DPB1 alleles lacking a corresponding SAB, determining the compatibility between a donor-recipient pair when the recipient possesses DPB1 antibodies currently requires the use of manual sequence alignments. Multiple studies have reported that DPB1 alleles can be classified into serological-defined categories based on shared protein sequence motifs residing in distinct hypervariable regions. To date, six such motifs have been recognized. To address this problem, we developed a computer-assisted tool to compare donor and recipient DPB1 allele sequences, specifically those defined by DPB1 hypervariable region motifs located in exon 2 (http://dpreport.hlatools.org). This tool quickly identifies mismatched DPB1 motifs, and easily permits the identification of motif-based donor-specific antibodies (DSA) to DPB1.

Genomic copy number variation correlates with survival outcomes in WHO grade IV glioma.

Allele-specific copy number analysis of tumors (ASCAT) assesses copy number variations (CNV) while accounting for aberrant cell fraction and tumor ploidy. We evaluated if ASCAT-assessed CNV are associated with survival outcomes in 56 patients with WHO grade IV gliomas. Tumor data analyzed by Affymetrix OncoScan FFPE Assay yielded the log ratio (R) and B-allele frequency (BAF). Input into ASCAT quantified CNV using the segmentation function to measure copy number inflection points throughout the genome. Quantified CNV was reported as log R and BAF segment counts. Results were confirmed on The Cancer Genome Atlas (TCGA) glioblastoma dataset. 25 (44.6%) patients had MGMT hyper-methylated tumors, 6 (10.7%) were IDH1 mutated. Median follow-up was 36.4 months. Higher log R segment counts were associate with longer progression-free survival (PFS) [hazard ratio (HR) 0.32, p < 0.001], and overall survival (OS) [HR 0.45, p = 0.01], and was an independent predictor of PFS and OS on multivariable analysis. Higher BAF segment counts were linked to longer PFS (HR 0.49, p = 0.022) and OS (HR 0.49, p = 0.052). In the TCGA confirmation cohort, longer 12-month OS was seen in patients with higher BAF segment counts (62.3% vs. 51.9%, p = 0.0129) and higher log R (63.6% vs. 55.2%, p = 0.0696). Genomic CNV may be a novel prognostic biomarker for WHO grade IV glioma patient outcomes.

Is a 500-Cell Count Necessary for Bone Marrow Differentials?: A Proposed Analytical Method for Validating a Lower Cutoff.

OBJECTIVES: By convention, 500 cells are counted for bone marrow aspirate differentials. Evidence supporting such a cutoff is lacking. We hypothesized that 300-cell counts could be sufficient. METHODS: Cell count results from 165 cases, for which values were recorded at 300 and 500 cells, were analyzed. We tested for statistical differences and changes in diagnostic classification between the two cutoffs. RESULTS: Three hundred cell counts did not produce diagnostically different results, particularly for myeloblasts and plasma cells, where cell percentages are critical for disease classification. Method comparison analysis did not reach statistical significance for any cell type when comparing the two methods. Bias plots showed narrow, even spread about the mean bias. Contingency table analysis yielded no significant diagnostic discrepancies. CONCLUSIONS: Performing differential counts on 300 cells would produce clinically and statistically similar results to 500 cells. Reducing the cell number counted has potential cost/labor reductions without affecting quality of care.

Validation of a Customized Bioinformatics Pipeline for a Clinical Next-Generation Sequencing Test Targeting Solid Tumor-Associated Variants.

Bioinformatic analysis is an integral and critical part of clinical next-generation sequencing. It is especially challenging for some pipelines to consistently identify insertions and deletions. We present the validation of an open source tumor amplicon pipeline (OTA-pipeline) for clinical next-generation sequencing targeting solid tumor-associated variants. Raw data generated from 557 TruSight Tumor 26 samples and in silico data were analyzed by the OTA-pipeline and legacy pipeline and compared. Discrepant results were confirmed by orthogonal methods. The OTA-pipeline reported 22 variants that were not detected by the previously validated pipeline, including seven synonymous or intronic single-nucleotide variants, five single-nucleotide variants at frequency <5%, one insertion, and nine deletions. Variant allele frequencies reported by the two pipelines were highly concordant, although a few significant discrepancies were present. Analysis of in silico FASTQ files demonstrated a higher sensitivity of detecting complex insertions and deletions with the OTA-pipeline. The higher sensitivity came at a cost, because false-positive calls were increased in difficult-to-sequence regions. However, these calls were all flagged by our strand bias filter, distinguishing them from true variants. Our validation process provides a model for laboratories that want to establish an in-house bioinformatics pipeline for clinical next-generation sequencing.

Development of CD3 cell quantitation algorithms for renal allograft biopsy rejection assessment utilizing open source image analysis software.

Renal allograft rejection diagnosis depends on assessment of parameters such as interstitial inflammation; however, studies have shown interobserver variability regarding interstitial inflammation assessment. Since automated image analysis quantitation can be reproducible, we devised customized analysis methods for CD3+ T-cell staining density as a measure of rejection severity and compared them with established commercial methods along with visual assessment. Renal biopsy CD3 immunohistochemistry slides (n = 45), including renal allografts with various degrees of acute cellular rejection (ACR) were scanned for whole slide images (WSIs). Inflammation was quantitated in the WSIs using pathologist visual assessment, commercial algorithms (Aperio nuclear algorithm for CD3+ cells/mm2 and Aperio positive pixel count algorithm), and customized open source algorithms developed in ImageJ with thresholding/positive pixel counting (custom CD3+%) and identification of pixels fulfilling "maxima" criteria for CD3 expression (custom CD3+ cells/mm2). Based on visual inspections of "markup" images, CD3 quantitation algorithms produced adequate accuracy. Additionally, CD3 quantitation algorithms correlated between each other and also with visual assessment in a statistically significant manner (r = 0.44 to 0.94, p = 0.003 to < 0.0001). Methods for assessing inflammation suggested a progression through the tubulointerstitial ACR grades, with statistically different results in borderline versus other ACR types, in all but the custom methods. Assessment of CD3-stained slides using various open source image analysis algorithms presents salient correlations with established methods of CD3 quantitation. These analysis techniques are promising and highly customizable, providing a form of on-slide "flow cytometry" that can facilitate additional diagnostic accuracy in tissue-based assessments.

Targeting adhesion signaling in KRAS, LKB1 mutant lung adenocarcinoma.

Loss of LKB1 activity is prevalent in KRAS mutant lung adenocarcinoma and promotes aggressive and treatment-resistant tumors. Previous studies have shown that LKB1 is a negative regulator of the focal adhesion kinase (FAK), but in vivo studies testing the efficacy of FAK inhibition in LKB1 mutant cancers are lacking. Here, we took a pharmacologic approach to show that FAK inhibition is an effective early-treatment strategy for this high-risk molecular subtype. We established a lenti-Cre-induced Kras and Lkb1 mutant genetically engineered mouse model (KLLenti) that develops 100% lung adenocarcinoma and showed that high spatiotemporal FAK activation occurs in collective invasive cells that are surrounded by high levels of collagen. Modeling invasion in 3D, loss of Lkb1, but not p53, was sufficient to drive collective invasion and collagen alignment that was highly sensitive to FAK inhibition. Treatment of early, stage-matched KLLenti tumors with FAK inhibitor monotherapy resulted in a striking effect on tumor progression, invasion, and tumor-associated collagen. Chronic treatment extended survival and impeded local lymph node spread. Lastly, we identified focally upregulated FAK and collagen-associated collective invasion in KRAS and LKB1 comutated human lung adenocarcinoma patients. Our results suggest that patients with LKB1 mutant tumors should be stratified for early treatment with FAK inhibitors.

Whole Slide Imaging for Analytical Anatomic Pathology and Telepathology: Practical Applications Today, Promises, and Perils.

Whole slide imaging (WSI) offers a convenient, tractable platform for measuring features of routine and special-stain histology or in immunohistochemistry staining by using digital image analysis (IA). We now routinely use IA for quantitative and qualitative analysis of theranostic markers such as human epidermal growth factor 2 (HER2/neu), estrogen and progesterone receptors, and Ki-67. Quantitative IA requires extensive validation, however, and may not always be the best approach, with pancreatic neuroendocrine tumors being one example in which a semiautomated approach may be preferable for patient care. We find that IA has great utility for objective assessment of gastrointestinal tract dysplasia, microvessel density in hepatocellular carcinoma, hepatic fibrosis and steatosis, renal fibrosis, and general quality analysis/quality control, although the applications of these to daily practice are still in development. Collaborations with bioinformatics specialists have explored novel applications to gliomas, including in silico approaches for mining histologic data and correlating with molecular and radiologic findings. We and many others are using WSI for rapid, remote-access slide reviews (telepathology), though technical factors currently limit its utility for routine, high-volume diagnostics. In our experience, the greatest current practical impact of WSI lies in facilitating long-term storage and retrieval of images while obviating the need to keep slides on site. Once the existing barriers of capital cost, validation, operator training, software design, and storage/back-up concerns are overcome, these technologies appear destined to be a cornerstone of precision medicine and personalized patient care, and to become a routine part of pathology practice.

Multi-Institutional FASTQ File Exchange as a Means of Proficiency Testing for Next-Generation Sequencing Bioinformatics and Variant Interpretation.

Next-generation sequencing is becoming increasingly common in clinical laboratories worldwide and is revolutionizing clinical molecular testing. However, the large amounts of raw data produced by next-generation sequencing assays and the need for complex bioinformatics analyses present unique challenges. Proficiency testing in clinical laboratories has traditionally been designed to evaluate assays in their entirety; however, it can be alternatively applied to separate assay components. We developed and implemented a multi-institutional proficiency testing approach to directly assess custom bioinformatics and variant interpretation processes. Six clinical laboratories, all of which use the same commercial library preparation kit for next-generation sequencing analysis of tumor specimens, each submitted raw data (FASTQ files) from four samples. These 24 file sets were then deidentified and redistributed to five of the institutions for analysis and interpretation according to their clinically validated approach. Among the laboratories, there was a high rate of concordance in the calling of single-nucleotide variants, in particular those we considered clinically significant (100% concordance). However, there was significant discordance in the calling of clinically significant insertions/deletions, with only two of seven being called by all participating laboratories. Missed calls were addressed by each laboratory to improve their bioinformatics processes. Thus, through our alternative proficiency testing approach, we identified the bioinformatic detection of insertions/deletions as an area of particular concern for clinical laboratories performing next-generation sequencing testing.

Clinical Validation and Implementation of a Targeted Next-Generation Sequencing Assay to Detect Somatic Variants in Non-Small Cell Lung, Melanoma, and Gastrointestinal Malignancies.

We tested and clinically validated a targeted next-generation sequencing (NGS) mutation panel using 80 formalin-fixed, paraffin-embedded (FFPE) tumor samples. Forty non-small cell lung carcinoma (NSCLC), 30 melanoma, and 30 gastrointestinal (12 colonic, 10 gastric, and 8 pancreatic adenocarcinoma) FFPE samples were selected from laboratory archives. After appropriate specimen and nucleic acid quality control, 80 NGS libraries were prepared using the Illumina TruSight tumor (TST) kit and sequenced on the Illumina MiSeq. Sequence alignment, variant calling, and sequencing quality control were performed using vendor software and laboratory-developed analysis workflows. TST generated ≥500× coverage for 98.4% of the 13,952 targeted bases. Reproducible and accurate variant calling was achieved at ≥5% variant allele frequency with 8 to 12 multiplexed samples per MiSeq flow cell. TST detected 112 variants overall, and confirmed all known single-nucleotide variants (n = 27), deletions (n = 5), insertions (n = 3), and multinucleotide variants (n = 3). TST detected at least one variant in 85.0% (68/80), and two or more variants in 36.2% (29/80), of samples. TP53 was the most frequently mutated gene in NSCLC (13 variants; 13/32 samples), gastrointestinal malignancies (15 variants; 13/25 samples), and overall (30 variants; 28/80 samples). BRAF mutations were most common in melanoma (nine variants; 9/23 samples). Clinically relevant NGS data can be obtained from routine clinical FFPE solid tumor specimens using TST, benchtop instruments, and vendor-supplied bioinformatics pipelines.