Training pathologists to assess stromal tumour-infiltrating lymphocytes in breast cancer synergises efforts in clinical care and scientific research

A growing body of research supports stromal tumour-infiltrating lymphocyte (TIL) density in breast cancer to be a robust prognostic and predicive biomarker. The gold standard for stromal TIL density quantitation in breast cancer is pathologist visual assessment using haematoxylin and eosin-stained slides. Artificial intelligence/machine-learning algorithms are in development to automate the stromal TIL scoring process, and must be validated against a reference standard such as pathologist visual assessment. Visual TIL assessment may suffer from significant interobserver variability. To improve interobserver agreement, regulatory science experts at the US Food and Drug Administration partnered with academic pathologists internationally to create a freely available online continuing medical education (CME) course to train pathologists in assessing breast cancer stromal TILs using an interactive format with expert commentary. Here we describe and provide a user guide to this CME course, whose content was designed to improve pathologist accuracy in scoring breast cancer TILs. We also suggest subsequent steps to translate knowledge into clinical practice with proficiency testing.

Attention De-sparsification Matters: Inducing diversity in digital pathology representation learning

We propose DiRL, a Diversity-inducing Representation Learning technique for histopathology imaging. Self-supervised learning (SSL) techniques, such as contrastive and non-contrastive approaches, have been shown to learn rich and effective representations of digitized tissue samples with limited pathologist supervision. Our analysis of vanilla SSL-pretrained models' attention distribution reveals an insightful observation: sparsity in attention, i.e, models tends to localize most of their attention to some prominent patterns in the image. Although attention sparsity can be beneficial in natural images due to these prominent patterns being the object of interest itself, this can be sub-optimal in digital pathology; this is because, unlike natural images, digital pathology scans are not object-centric, but rather a complex phenotype of various spatially intermixed biological components. Inadequate diversification of attention in these complex images could result in crucial information loss. To address this, we leverage cell segmentation to densely extract multiple histopathology-specific representations, and then propose a prior-guided dense pretext task, designed to match the multiple corresponding representations between the views. Through this, the model learns to attend to various components more closely and evenly, thus inducing adequate diversification in attention for capturing context-rich representations. Through quantitative and qualitative analysis on multiple tasks across cancer types, we demonstrate the efficacy of our method and observe that the attention is more globally distributed.

Image-based multiplex immune profiling of cancer tissues: translational implications. A report of the International Immuno-oncology Biomarker Working Group on Breast Cancer

Recent advances in the field of immuno-oncology have brought transformative changes in the management of cancer patients. The immune profile of tumours has been found to have key value in predicting disease prognosis and treatment response in various cancers. Multiplex immunohistochemistry and immunofluorescence have emerged as potent tools for the simultaneous detection of multiple protein biomarkers in a single tissue section, thereby expanding opportunities for molecular and immune profiling while preserving tissue samples. By establishing the phenotype of individual tumour cells when distributed within a mixed cell population, the identification of clinically relevant biomarkers with high-throughput multiplex immunophenotyping of tumour samples has great potential to guide appropriate treatment choices. Moreover, the emergence of novel multi-marker imaging approaches can now provide unprecedented insights into the tumour microenvironment, including the potential interplay between various cell types. However, there are significant challenges to widespread integration of these technologies in daily research and clinical practice. This review addresses the challenges and potential solutions within a structured framework of action from a regulatory and clinical trial perspective. New developments within the field of immunophenotyping using multiplexed tissue imaging platforms and associated digital pathology are also described, with a specific focus on translational implications across different subtypes of cancer. © 2024 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Keratin 17 modulates the immune topography of pancreatic cancer

Background

The immune microenvironment impacts tumor growth, invasion, metastasis, and patient survival and may provide opportunities for therapeutic intervention in pancreatic ductal adenocarcinoma (PDAC). Although never studied as a potential modulator of the immune response in most cancers, Keratin 17 (K17), a biomarker of the most aggressive (basal) molecular subtype of PDAC, is intimately involved in the histogenesis of the immune response in psoriasis, basal cell carcinoma, and cervical squamous cell carcinoma. Thus, we hypothesized that K17 expression could also impact the immune cell response in PDAC, and that uncovering this relationship could provide insight to guide the development of immunotherapeutic opportunities to extend patient survival.

Methods

Multiplex immunohistochemistry (mIHC) and automated image analysis based on novel computational imaging technology were used to decipher the abundance and spatial distribution of T cells, macrophages, and tumor cells, relative to K17 expression in 235 PDACs.

Results

K17 expression had profound effects on the exclusion of intratumoral CD8 + T cells and was also associated with decreased numbers of peritumoral CD8 + T cells, CD16 + macrophages, and CD163 + macrophages (p < 0.0001). The differences in the intratumor and peritumoral CD8 + T cell abundance were not impacted by neoadjuvant therapy, tumor stage, grade, lymph node status, histologic subtype, nor KRAS, p53, SMAD4, or CDKN2A mutations.

Conclusions

Thus, K17 expression correlates with major differences in the immune microenvironment that are independent of any tested clinicopathologic or tumor intrinsic variables, suggesting that targeting K17-mediated immune effects on the immune system could restore the innate immunologic response to PDAC and might provide novel opportunities to restore immunotherapeutic approaches for this most deadly form of cancer.

Reproducible Reporting of the Collection and Evaluation of Annotations for Artificial Intelligence Models

This work puts forth and demonstrates the utility of a reporting framework for collecting and evaluating annotations of medical images used for training and testing artificial intelligence (AI) models in assisting detection and diagnosis. AI has unique reporting requirements, as shown by the AI extensions to the Consolidated Standards of Reporting Trials (CONSORT) and Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) checklists and the proposed AI extensions to the Standards for Reporting Diagnostic Accuracy (STARD) and Transparent Reporting of a Multivariable Prediction model for Individual Prognosis or Diagnosis (TRIPOD) checklists. AI for detection and/or diagnostic image analysis requires complete, reproducible, and transparent reporting of the annotations and metadata used in training and testing data sets. In an earlier work by other researchers, an annotation workflow and quality checklist for computational pathology annotations were proposed. In this manuscript, we operationalize this workflow into an evaluable quality checklist that applies to any reader-interpreted medical images, and we demonstrate its use for an annotation effort in digital pathology. We refer to this quality framework as the Collection and Evaluation of Annotations for Reproducible Reporting of Artificial Intelligence (CLEARR-AI).

Privacy-preserving record linkage across disparate institutions and datasets to enable a learning health system: The national COVID cohort collaborative (N3C) experience

Introduction

Research driven by real-world clinical data is increasingly vital to enabling learning health systems, but integrating such data from across disparate health systems is challenging. As part of the NCATS National COVID Cohort Collaborative (N3C), the N3C Data Enclave was established as a centralized repository of deidentified and harmonized COVID-19 patient data from institutions across the US. However, making this data most useful for research requires linking it with information such as mortality data, images, and viral variants. The objective of this project was to establish privacy-preserving record linkage (PPRL) methods to ensure that patient-level EHR data remains secure and private when governance-approved linkages with other datasets occur.

Methods

Separate agreements and approval processes govern N3C data contribution and data access. The Linkage Honest Broker (LHB), an independent neutral party (the Regenstrief Institute), ensures data linkages are robust and secure by adding an extra layer of separation between protected health information and clinical data. The LHB's PPRL methods (including algorithms, processes, and governance) match patient records using "deidentified tokens," which are hashed combinations of identifier fields that define a match across data repositories without using patients' clear-text identifiers.

Results

These methods enable three linkage functions: Deduplication, Linking Multiple Datasets, and Cohort Discovery. To date, two external repositories have been cross-linked. As of March 1, 2023, 43 sites have signed the LHB Agreement; 35 sites have sent tokens generated for 9 528 998 patients. In this initial cohort, the LHB identified 135 037 matches and 68 596 duplicates.

Conclusion

This large-scale linkage study using deidentified datasets of varying characteristics established secure methods for protecting the privacy of N3C patient data when linked for research purposes. This technology has potential for use with registries for other diseases and conditions.

Initial interactions with the FDA on developing a validation dataset as a medical device development tool

Quantifying tumor-infiltrating lymphocytes (TILs) in breast cancer tumors is a challenging task for pathologists. With the advent of whole slide imaging that digitizes glass slides, it is possible to apply computational models to quantify TILs for pathologists. Development of computational models requires significant time, expertise, consensus, and investment. To reduce this burden, we are preparing a dataset for developers to validate their models and a proposal to the Medical Device Development Tool (MDDT) program in the Center for Devices and Radiological Health of the U.S. Food and Drug Administration (FDA). If the FDA qualifies the dataset for its submitted context of use, model developers can use it in a regulatory submission within the qualified context of use without additional documentation. Our dataset aims at reducing the regulatory burden placed on developers of models that estimate the density of TILs and will allow head-to-head comparison of multiple computational models on the same data. In this paper, we discuss the MDDT preparation and submission process, including the feedback we received from our initial interactions with the FDA and propose how a qualified MDDT validation dataset could be a mechanism for open, fair, and consistent measures of computational model performance. Our experiences will help the community understand what the FDA considers relevant and appropriate (from the perspective of the submitter), at the early stages of the MDDT submission process, for validating stromal TIL density estimation models and other potential computational models. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland. This article has been contributed to by U.S. Government employees and their work is in the public domain in the USA.

The Importance of Quality Assurance in Radiation Oncology Clinical Trials

Clinical trials have been the center of progress in modern medicine. In oncology, we are fortunate to have a structure in place through the National Clinical Trials Network (NCTN). The NCTN provides the infrastructure and a forum for scientific discussion to develop clinical concepts for trial design. The NCTN also provides a network group structure to administer trials for successful trial management and outcome analyses. There are many important aspects to trial design and conduct. Modern trials need to ensure appropriate trial conduct and secure data management processes. Of equal importance is the quality assurance of a clinical trial. If progress is to be made in oncology clinical medicine, investigators and patient care providers of service need to feel secure that trial data is complete, accurate, and well-controlled in order to be confident in trial analysis and move trial outcome results into daily practice. As our technology has matured, so has our need to apply technology in a uniform manner for appropriate interpretation of trial outcomes. In this article, we review the importance of quality assurance in clinical trials involving radiation therapy. We will include important aspects of institution and investigator credentialing for participation as well as ongoing processes to ensure that each trial is being managed in a compliant manner. We will provide examples of the importance of complete datasets to ensure study interpretation. We will describe how successful strategies for quality assurance in the past will support new initiatives moving forward.

Investigation of commonly used aortic aneurysm growth rate metrics: Comparing their suitability for clinical and research applications

An aneurysm is a pathological widening of a blood vessel. Aneurysms of the aorta are often asymptomatic until they rupture, killing approximately 10,000 Americans per year. Fortunately, rupture can be prevented through early detection and surgical repair. However, surgical risk outweighs rupture risk for small aortic aneurysms, necessitating a policy of surveillance. Understanding the growth rate of aneurysms is essential for determining appropriate surveillance windows. Further, identifying risk factors for fast growth can help identify potential interventions. However, studies in the literature have applied many different methods for defining the growth rate of abdominal aortic aneurysms. It is unclear which of these methods is most accurate and clinically meaningful, and whether these heterogeneous methodologies may have contributed to the varied results reported in the literature. To help future researchers best plan their studies and to help clinicians interpret existing studies, we compared five different models of aneurysmal growth rate. We examined their noise tolerance, temporal bias, predictive accuracy, and statistical power to detect risk factors. We found that hierarchical mixed effects models were more noise tolerant than traditional, unpooled models. We also found that linear models were sensitive to temporal bias, assigning lower growth rates to aneurysms that were detected earlier in their course. Our exponential mixed model was noise-tolerant, resistant to temporal bias, and detected the greatest number of clinical risk factors. We conclude that exponential mixed models may be optimal for large studies. Because our results suggest that choice of method can materially impact a study's findings, we recommend that future studies clearly state the method used and demonstrate its appropriateness.

Pitfalls in machine learning-based assessment of tumor-infiltrating lymphocytes in breast cancer: A report of the International Immuno-Oncology Biomarker Working Group on Breast Cancer

The clinical significance of the tumor-immune interaction in breast cancer is now established, and tumor-infiltrating lymphocytes (TILs) have emerged as predictive and prognostic biomarkers for patients with triple-negative (estrogen receptor, progesterone receptor, and HER2-negative) breast cancer and HER2-positive breast cancer. How computational assessments of TILs might complement manual TIL assessment in trial and daily practices is currently debated. Recent efforts to use machine learning (ML) to automatically evaluate TILs have shown promising results. We review state-of-the-art approaches and identify pitfalls and challenges of automated TIL evaluation by studying the root cause of ML discordances in comparison to manual TIL quantification. We categorize our findings into four main topics: (1) technical slide issues, (2) ML and image analysis aspects, (3) data challenges, and (4) validation issues. The main reason for discordant assessments is the inclusion of false-positive areas or cells identified by performance on certain tissue patterns or design choices in the computational implementation. To aid the adoption of ML for TIL assessment, we provide an in-depth discussion of ML and image analysis, including validation issues that need to be considered before reliable computational reporting of TILs can be incorporated into the trial and routine clinical management of patients with triple-negative breast cancer. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Spatial analyses of immune cell infiltration in cancer: current methods and future directions: A report of the International Immuno-Oncology Biomarker Working Group on Breast Cancer

Modern histologic imaging platforms coupled with machine learning methods have provided new opportunities to map the spatial distribution of immune cells in the tumor microenvironment. However, there exists no standardized method for describing or analyzing spatial immune cell data, and most reported spatial analyses are rudimentary. In this review, we provide an overview of two approaches for reporting and analyzing spatial data (raster versus vector-based). We then provide a compendium of spatial immune cell metrics that have been reported in the literature, summarizing prognostic associations in the context of a variety of cancers. We conclude by discussing two well-described clinical biomarkers, the breast cancer stromal tumor infiltrating lymphocytes score and the colon cancer Immunoscore, and describe investigative opportunities to improve clinical utility of these spatial biomarkers. © 2023 The Pathological Society of Great Britain and Ireland.

Effective and Efficient Active Learning for Deep Learning Based Tissue Image Analysis

MOTIVATION

Deep learning attained excellent results in Digital Pathology recently. A challenge with its use is that high quality, representative training data sets are required to build robust models. Data annotation in the domain is labor intensive and demands substantial time commitment from expert pathologists. Active Learning (AL) is a strategy to minimize annotation. The goal is to select samples from the pool of unlabeled data for annotation that improves model accuracy. However, AL is a very compute demanding approach. The benefits for model learning may vary according to the strategy used, and it may be hard for a domain specialist to fine tune the solution without an integrated interface.

RESULTS

We developed a framework that includes a friendly user interface along with run-time optimizations to reduce annotation and execution time in AL in digital pathology. Our solution implements several AL strategies along with our Diversity-Aware Data Acquisition (DADA) acquisition function, which enforces data diversity to improve the prediction performance of a model. In this work, we employed a model simplification strategy (Network Auto-Reduction (NAR)) that significantly improves AL execution time when coupled with DADA. NAR produces less compute demanding models, which replace the target models during the AL process to reduce processing demands. An evaluation with a Tumor-Infiltrating Lymphocytes (TILs) classification application shows that: (i) DADA attains superior performance compared to state-of-the-art AL strategies for different Convolutional Neural Networks (CNNs), (ii) NAR improves the AL execution time by up to 4.3 ×, and (iii) target models trained with patches/data selected by the NAR reduced versions achieve similar or superior classification quality to using target CNNs for data selection.

AVAILABILITY

Source code: https://github.com/alsmeirelles/DADA.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Quality improvements in radiation oncology clinical trials

Clinical trials have become the primary mechanism to validate process improvements in oncology clinical practice. Over the past two decades there have been considerable process improvements in the practice of radiation oncology within the structure of a modern department using advanced technology for patient care. Treatment planning is accomplished with volume definition including fusion of multiple series of diagnostic images into volumetric planning studies to optimize the definition of tumor and define the relationship of tumor to normal tissue. Daily treatment is validated by multiple tools of image guidance. Computer planning has been optimized and supported by the increasing use of artificial intelligence in treatment planning. Informatics technology has improved, and departments have become geographically transparent integrated through informatics bridges creating an economy of scale for the planning and execution of advanced technology radiation therapy. This serves to provide consistency in department habits and improve quality of patient care. Improvements in normal tissue sparing have further improved tolerance of treatment and allowed radiation oncologists to increase both daily and total dose to target. Radiation oncologists need to define a priori dose volume constraints to normal tissue as well as define how image guidance will be applied to each radiation treatment. These process improvements have enhanced the utility of radiation therapy in patient care and have made radiation therapy an attractive option for care in multiple primary disease settings. In this chapter we review how these changes have been applied to clinical practice and incorporated into clinical trials. We will discuss how the changes in clinical practice have improved the quality of clinical trials in radiation therapy. We will also identify what gaps remain and need to be addressed to offer further improvements in radiation oncology clinical trials and patient care.

Federated learning enables big data for rare cancer boundary detection

Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.

Abstract C071: Keratin 17 excludes CD8-positive T cells and recruits CD163-positive macrophages in pancreatic ductal adenocarcinoma

Background: Keratin 17 (K17) is a negative prognostic biomarker, overexpressed in the biologically most aggressive forms of pancreatic ductal adenocarcinoma (PDAC). In other anatomic sites and disease processes, K17 expression also correlates with immune cell infiltrates and could block T-cell infiltration. Thus, we hypothesized that K17 expression correlates with the inflammatory microenvironment in PDAC. In this study, we aimed to determine the relationship between the stromal immune cell infiltrates and K17 expression, using multiplexed immunohistochemistry (mIHC) and our suite of deep learning tools to quantitatively evaluate the expression of four biomarkers of T-cells and macrophages in PDAC. Methods: mIHC was performed on representative sections of 201 primary PDACs from Stony Brook University Hospital, Thomas Jefferson University Hospital, Cedars Sinai Medical Center, and from a national cohort (KYT, Pancreatic Cancer Action Network, and Perthera). Antibodies for CD4 (helper T-cells), CD8 (cytotoxic T-cells), CD16 (pan-macrophage), CD163 (M2 macrophages), pancytokeratin, and K17 were provided by Roche Diagnostics Corporation through a sponsored research agreement. mIHC was performed on a Discovery Ultra Autostainer (Roche), using horseradish peroxidase (HRP) and alkaline phosphatase (AP)-based protocols with multiple chromogens (Red: CD4, Purple: CD8, Yellow: CD16, Green: CD163, Teal: pancytokeratin, and Brown: K17) to enable multispectral imaging of diverse immune cell populations within the cancer microenvironment. A deep learning analysis workflow was used to detect and classify stromal inflammatory cells, in whole slide images (WSIs), generated using an Olympus VS120 digital microscope (Olympus, Tokyo, Japan). Pixel-wise predictions from a color auto-encoder (ColorAE) union UNET anchor UNET model were combined to create multi-class masks that were further analyzed to perform detection and classification. Results: The analysis of the inflammatory microenvironment focused on defining immune cell infiltrates located within 25 microns of the closest K17-positive versus K17-negative tumor cell in each representative section. Across the sum of K17-positive and negative zones/section, CD4 cell counts ranged from 0-10,617 (mean 2,709), CD8 cell counts ranged from 63-28,596 (mean 6,745), CD16 cell counts ranged from 4-7,797 (mean 3,024), and CD163 cell counts ranged from 35-34,696 (mean 14,968). CD4 T-helper cells, CD8 cytotoxic T cells, and CD16 macrophages were more numerous (respectively, p=0.0012; p=&lt;0.0001; p=&lt;0.0001) in K17-negative tumor zones compared to K17-positive zones. By contrast, the number of CD163 (M2) tumor-promoting macrophages was greater in K17 positive zones (p=0.0019). Conclusion: K17 expression by tumor cells impacts the chronic inflammatory microenvironment, shielding tumor cells from immune cell mediated cytotoxic responses, while recruiting tumor-promoting M2 macrophages, indicating that K17 impacts the immune response as a fundamental hallmark of aggression in PDAC.

Citation Format: Lyanne Oblein, Michael Horowitz, Mahmudul Hasan, Sruthi Babu, Mariana Torrente-Goncalves, Lucia Roa, Jaymie Oentoro, Jason Harper, Xin Yao Zheng, Wei Jiang, Andrew Hendifar, Natalie Moshayedi, Brent Larson, Veronica Placencio-Hickok, Edik Blais, Emmanuel Petricoin, Joel Saltz, Natalia D. Marchenko, Luisa F. Escobar-Hoyos, Kenneth Shroyer. Keratin 17 excludes CD8-positive T cells and recruits CD163-positive macrophages in pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer; 2022 Sep 13-16; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2022;82(22 Suppl):Abstract nr C071.

Effect of SARS-CoV-2 Infection and Infection Severity on Longer-Term Glycemic Control and Weight in People With Type 2 Diabetes

OBJECTIVE

To evaluate the association of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and severity of infection with longer-term glycemic control and weight in people with type 2 diabetes (T2D) in the U.S.

RESEARCH DESIGN AND METHODS

We conducted a retrospective cohort study using longitudinal electronic health record data of patients with SARS-CoV-2 infection from the National COVID Cohort Collaborative (N3C). Patients were ≥18 years old with an ICD-10 diagnosis of T2D and at least one HbA1c and weight measurement prior to and after an index date of their first coronavirus disease 2019 (COVID-19) diagnosis or negative SARS-CoV-2 test. We used propensity scores to identify a matched cohort balanced on demographic characteristics, comorbidities, and medications used to treat diabetes. The primary outcome was the postindex average HbA1c and postindex average weight over a 1 year time period beginning 90 days after the index date among patients who did and did not have SARS-CoV-2 infection. Secondary outcomes were postindex average HbA1c and weight in patients who required hospitalization or mechanical ventilation.

RESULTS

There was no significant difference in the postindex average HbA1c or weight in patients who had SARS-CoV-2 infection compared with control subjects. Mechanical ventilation was associated with a decrease in average HbA1c after COVID-19.

CONCLUSIONS

In a multicenter cohort of patients in the U.S. with preexisting T2D, there was no significant change in longer-term average HbA1c or weight among patients who had COVID-19. Mechanical ventilation was associated with a decrease in HbA1c after COVID-19.

Radiation Oncology: Future Vision for Quality Assurance and Data Management in Clinical Trials and Translational Science

The future of radiation oncology is exceptionally strong as we are increasingly involved in nearly all oncology disease sites due to extraordinary advances in radiation oncology treatment management platforms and improvements in treatment execution. Due to our technology and consistent accuracy, compressed radiation oncology treatment strategies are becoming more commonplace secondary to our ability to successfully treat tumor targets with increased normal tissue avoidance. In many disease sites including the central nervous system, pulmonary parenchyma, liver, and other areas, our service is redefining the standards of care. Targeting of disease has improved due to advances in tumor imaging and application of integrated imaging datasets into sophisticated planning systems which can optimize volume driven plans created by talented personnel. Treatment times have significantly decreased due to volume driven arc therapy and positioning is secured by real time imaging and optical tracking. Normal tissue exclusion has permitted compressed treatment schedules making treatment more convenient for the patient. These changes require additional study to further optimize care. Because data exchange worldwide have evolved through digital platforms and prisms, images and radiation datasets worldwide can be shared/reviewed on a same day basis using established de-identification and anonymization methods. Data storage post-trial completion can co-exist with digital pathomic and radiomic information in a single database coupled with patient specific outcome information and serve to move our translational science forward with nimble query elements and artificial intelligence to ask better questions of the data we collect and collate. This will be important moving forward to validate our process improvements at an enterprise level and support our science. We have to be thorough and complete in our data acquisition processes, however if we remain disciplined in our data management plan, our field can grow further and become more successful generating new standards of care from validated datasets.

Pilot study to evaluate tools to collect pathologist annotations for validating machine learning algorithms

Purpose: Validation of artificial intelligence (AI) algorithms in digital pathology with a reference standard is necessary before widespread clinical use, but few examples focus on creating a reference standard based on pathologist annotations. This work assesses the results of a pilot study that collects density estimates of stromal tumor-infiltrating lymphocytes (sTILs) in breast cancer biopsy specimens. This work will inform the creation of a validation dataset for the evaluation of AI algorithms fit for a regulatory purpose. Approach: Collaborators and crowdsourced pathologists contributed glass slides, digital images, and annotations. Here, "annotations" refer to any marks, segmentations, measurements, or labels a pathologist adds to a report, image, region of interest (ROI), or biological feature. Pathologists estimated sTILs density in 640 ROIs from hematoxylin and eosin stained slides of 64 patients via two modalities: an optical light microscope and two digital image viewing platforms. Results: The pilot study generated 7373 sTILs density estimates from 29 pathologists. Analysis of annotations found the variability of density estimates per ROI increases with the mean; the root mean square differences were 4.46, 14.25, and 26.25 as the mean density ranged from 0% to 10%, 11% to 40%, and 41% to 100%, respectively. The pilot study informs three areas of improvement for future work: technical workflows, annotation platforms, and agreement analysis methods. Upgrades to the workflows and platforms will improve operability and increase annotation speed and consistency. Conclusions: Exploratory data analysis demonstrates the need to develop new statistical approaches for agreement. The pilot study dataset and analysis methods are publicly available to allow community feedback. The development and results of the validation dataset will be publicly available to serve as an instructive tool that can be replicated by developers and researchers.

Effective active learning in digital pathology: A case study in tumor infiltrating lymphocytes

BACKGROUND AND OBJECTIVE

Deep learning methods have demonstrated remarkable performance in pathology image analysis, but they require a large amount of annotated training data from expert pathologists. The aim of this study is to minimize the data annotation need in these analyses.

METHODS

Active learning (AL) is an iterative approach to training deep learning models. It was used in our context with a Tumor Infiltrating Lymphocytes (TIL) classification task to minimize annotation. State-of-the-art AL methods were evaluated with the TIL application and we have proposed and evaluated a more efficient and effective AL acquisition method. The proposed method uses data grouping based on imaging features and model prediction uncertainty to select meaningful training samples (image patches).

RESULTS

An experimental evaluation with a collection of cancer tissue images shows that: (i) Our approach reduces the number of patches required to attain a given AUC as compared to other approaches, and (ii) our optimization (subpooling) leads to AL execution time improvement of about 2.12×.

CONCLUSIONS

This strategy enabled TIL based deep learning analyses using smaller annotation demand. We expect this approach may be used to build other analyses in digital pathology with fewer training samples.

Spatial Characterization of Tumor-Infiltrating Lymphocytes and Breast Cancer Progression

Tumor-infiltrating lymphocytes (TILs) have been established as a robust prognostic biomarker in breast cancer, with emerging utility in predicting treatment response in the adjuvant and neoadjuvant settings. In this study, the role of TILs in predicting overall survival and progression-free interval was evaluated in two independent cohorts of breast cancer from the Cancer Genome Atlas (TCGA BRCA) and the Carolina Breast Cancer Study (UNC CBCS). We utilized machine learning and computer vision algorithms to characterize TIL infiltrates in digital whole-slide images (WSIs) of breast cancer stained with hematoxylin and eosin (H&E). Multiple parameters were used to characterize the global abundance and spatial features of TIL infiltrates. Univariate and multivariate analyses show that large aggregates of peritumoral and intratumoral TILs (forests) were associated with longer survival, whereas the absence of intratumoral TILs (deserts) is associated with increased risk of recurrence. Patients with two or more high-risk spatial features were associated with significantly shorter progression-free interval (PFI). This study demonstrates the practical utility of Pathomics in evaluating the clinical significance of the abundance and spatial patterns of distribution of TIL infiltrates as important biomarkers in breast cancer.

Harmonizing units and values of quantitative data elements in a very large nationally pooled electronic health record (EHR) dataset

Objective

The goals of this study were to harmonize data from electronic health records (EHRs) into common units, and impute units that were missing.

Materials and Methods

The National COVID Cohort Collaborative (N3C) table of laboratory measurement data—over 3.1 billion patient records and over 19 000 unique measurement concepts in the Observational Medical Outcomes Partnership (OMOP) common-data-model format from 55 data partners. We grouped ontologically similar OMOP concepts together for 52 variables relevant to COVID-19 research, and developed a unit-harmonization pipeline comprised of (1) selecting a canonical unit for each measurement variable, (2) arriving at a formula for conversion, (3) obtaining clinical review of each formula, (4) applying the formula to convert data values in each unit into the target canonical unit, and (5) removing any harmonized value that fell outside of accepted value ranges for the variable. For data with missing units for all the results within a lab test for a data partner, we compared values with pooled values of all data partners, using the Kolmogorov-Smirnov test.

Results

Of the concepts without missing values, we harmonized 88.1% of the values, and imputed units for 78.2% of records where units were absent (41% of contributors’ records lacked units).

Discussion

The harmonization and inference methods developed herein can serve as a resource for initiatives aiming to extract insight from heterogeneous EHR collections. Unique properties of centralized data are harnessed to enable unit inference.

Conclusion

The pipeline we developed for the pooled N3C data enables use of measurements that would otherwise be unavailable for analysis.

Deep Learning-Based Mapping of Tumor Infiltrating Lymphocytes in Whole Slide Images of 23 Types of Cancer

The role of tumor infiltrating lymphocytes (TILs) as a biomarker to predict disease progression and clinical outcomes has generated tremendous interest in translational cancer research. We present an updated and enhanced deep learning workflow to classify 50x50 um tiled image patches (100x100 pixels at 20x magnification) as TIL positive or negative based on the presence of 2 or more TILs in gigapixel whole slide images (WSIs) from the Cancer Genome Atlas (TCGA). This workflow generates TIL maps to study the abundance and spatial distribution of TILs in 23 different types of cancer. We trained three state-of-the-art, popular convolutional neural network (CNN) architectures (namely VGG16, Inception-V4, and ResNet-34) with a large volume of training data, which combined manual annotations from pathologists (strong annotations) and computer-generated labels from our previously reported first-generation TIL model for 13 cancer types (model-generated annotations). Specifically, this training dataset contains TIL positive and negative patches from cancers in additional organ sites and curated data to help improve algorithmic performance by decreasing known false positives and false negatives. Our new TIL workflow also incorporates automated thresholding to convert model predictions into binary classifications to generate TIL maps. The new TIL models all achieve better performance with improvements of up to 13% in accuracy and 15% in F-score. We report these new TIL models and a curated dataset of TIL maps, referred to as TIL-Maps-23, for 7983 WSIs spanning 23 types of cancer with complex and diverse visual appearances, which will be publicly available along with the code to evaluate performance.

Code Available at:https://github.com/ShahiraAbousamra/til_classification.

Glycemic Control and Clinical Outcomes in U.S. Patients With COVID-19: Data From the National COVID Cohort Collaborative (N3C) Database

OBJECTIVE

The purpose of the study is to evaluate the relationship between HbA1c and severity of coronavirus disease 2019 (COVID-19) outcomes in patients with type 2 diabetes (T2D) with acute COVID-19 infection.

RESEARCH DESIGN AND METHODS

We conducted a retrospective study using observational data from the National COVID Cohort Collaborative (N3C), a longitudinal, multicenter U.S. cohort of patients with COVID-19 infection. Patients were ≥18 years old with T2D and confirmed COVID-19 infection by laboratory testing or diagnosis code. The primary outcome was 30-day mortality following the date of COVID-19 diagnosis. Secondary outcomes included need for invasive ventilation or extracorporeal membrane oxygenation (ECMO), hospitalization within 7 days before or 30 days after COVID-19 diagnosis, and length of stay (LOS) for patients who were hospitalized.

RESULTS

The study included 39,616 patients (50.9% female, 55.4% White, 26.4% Black or African American, and 16.1% Hispanic or Latino, with mean ± SD age 62.1 ± 13.9 years and mean ± SD HbA1c 7.6% ± 2.0). There was an increasing risk of hospitalization with incrementally higher HbA1c levels, but risk of death plateaued at HbA1c >8%, and risk of invasive ventilation or ECMO plateaued >9%. There was no significant difference in LOS across HbA1c levels.

CONCLUSIONS

In a large, multicenter cohort of patients in the U.S. with T2D and COVID-19 infection, risk of hospitalization increased with incrementally higher HbA1c levels. Risk of death and invasive ventilation also increased but plateaued at different levels of glycemic control.

Characteristics, Outcomes, and Severity Risk Factors Associated With SARS-CoV-2 Infection Among Children in the US National COVID Cohort Collaborative

Importance

Understanding of SARS-CoV-2 infection in US children has been limited by the lack of large, multicenter studies with granular data.

Objective

To examine the characteristics, changes over time, outcomes, and severity risk factors of children with SARS-CoV-2 within the National COVID Cohort Collaborative (N3C).

Design, Setting, and Participants

A prospective cohort study of encounters with end dates before September 24, 2021, was conducted at 56 N3C facilities throughout the US. Participants included children younger than 19 years at initial SARS-CoV-2 testing.

Main Outcomes and Measures

Case incidence and severity over time, demographic and comorbidity severity risk factors, vital sign and laboratory trajectories, clinical outcomes, and acute COVID-19 vs multisystem inflammatory syndrome in children (MIS-C), and Delta vs pre-Delta variant differences for children with SARS-CoV-2.

Results

A total of 1 068 410 children were tested for SARS-CoV-2 and 167 262 test results (15.6%) were positive (82 882 [49.6%] girls; median age, 11.9 [IQR, 6.0-16.1] years). Among the 10 245 children (6.1%) who were hospitalized, 1423 (13.9%) met the criteria for severe disease: mechanical ventilation (796 [7.8%]), vasopressor-inotropic support (868 [8.5%]), extracorporeal membrane oxygenation (42 [0.4%]), or death (131 [1.3%]). Male sex (odds ratio [OR], 1.37; 95% CI, 1.21-1.56), Black/African American race (OR, 1.25; 95% CI, 1.06-1.47), obesity (OR, 1.19; 95% CI, 1.01-1.41), and several pediatric complex chronic condition (PCCC) subcategories were associated with higher severity disease. Vital signs and many laboratory test values from the day of admission were predictive of peak disease severity. Variables associated with increased odds for MIS-C vs acute COVID-19 included male sex (OR, 1.59; 95% CI, 1.33-1.90), Black/African American race (OR, 1.44; 95% CI, 1.17-1.77), younger than 12 years (OR, 1.81; 95% CI, 1.51-2.18), obesity (OR, 1.76; 95% CI, 1.40-2.22), and not having a pediatric complex chronic condition (OR, 0.72; 95% CI, 0.65-0.80). The children with MIS-C had a more inflammatory laboratory profile and severe clinical phenotype, with higher rates of invasive ventilation (117 of 707 [16.5%] vs 514 of 8241 [6.2%]; P < .001) and need for vasoactive-inotropic support (191 of 707 [27.0%] vs 426 of 8241 [5.2%]; P < .001) compared with those who had acute COVID-19. Comparing children during the Delta vs pre-Delta eras, there was no significant change in hospitalization rate (1738 [6.0%] vs 8507 [6.2%]; P = .18) and lower odds for severe disease (179 [10.3%] vs 1242 [14.6%]) (decreased by a factor of 0.67; 95% CI, 0.57-0.79; P < .001).

Conclusions and Relevance

In this cohort study of US children with SARS-CoV-2, there were observed differences in demographic characteristics, preexisting comorbidities, and initial vital sign and laboratory values between severity subgroups. Taken together, these results suggest that early identification of children likely to progress to severe disease could be achieved using readily available data elements from the day of admission. Further work is needed to translate this knowledge into improved outcomes.

Robust Image Population Based Stain Color Normalization: How Many Reference Slides Are Enough?

Histopathologic evaluation of Hematoxylin & Eosin (H&E) stained slides is essential for disease diagnosis, revealing tissue morphology, structure, and cellular composition. Variations in staining protocols and equipment result in images with color nonconformity. Although pathologists compensate for color variations, these disparities introduce inaccuracies in computational whole slide image (WSI) analysis, accentuating data domain shift and degrading generalization. Current state-of-the-art normalization methods employ a single WSI as reference, but selecting a single WSI representative of a complete WSI-cohort is infeasible, inadvertently introducing normalization bias. We seek the optimal number of slides to construct a more representative reference based on composite/aggregate of multiple H&E density histograms and stain-vectors, obtained from a randomly selected WSI population (WSI-Cohort-Subset). We utilized 1,864 IvyGAP WSIs as a WSI-cohort, and built 200 WSI-Cohort-Subsets varying in size (from 1 to 200 WSI-pairs) using randomly selected WSIs. The WSI-pairs' mean Wasserstein Distances and WSI-Cohort-Subsets' standard deviations were calculated. The Pareto Principle defined the optimal WSI-Cohort-Subset size. The WSI-cohort underwent structure-preserving color normalization using the optimal WSI-Cohort-Subset histogram and stain-vector aggregates. Numerous normalization permutations support WSI-Cohort-Subset aggregates as representative of a WSI-cohort through WSI-cohort CIELAB color space swift convergence, as a result of the law of large numbers and shown as a power law distribution. We show normalization at the optimal (Pareto Principle) WSI-Cohort-Subset size and corresponding CIELAB convergence: a) Quantitatively, using 500 WSI-cohorts; b) Quantitatively, using 8,100 WSI-regions; c) Qualitatively, using 30 cellular tumor normalization permutations. Aggregate-based stain normalization may contribute in increasing computational pathology robustness, reproducibility, and integrity.

A Pathologist-Annotated Dataset for Validating Artificial Intelligence: A Project Description and Pilot Study

Purpose:

Validating artificial intelligence algorithms for clinical use in medical images is a challenging endeavor due to a lack of standard reference data (ground truth). This topic typically occupies a small portion of the discussion in research papers since most of the efforts are focused on developing novel algorithms. In this work, we present a collaboration to create a validation dataset of pathologist annotations for algorithms that process whole slide images. We focus on data collection and evaluation of algorithm performance in the context of estimating the density of stromal tumor-infiltrating lymphocytes (sTILs) in breast cancer.

Methods:

We digitized 64 glass slides of hematoxylin- and eosin-stained invasive ductal carcinoma core biopsies prepared at a single clinical site. A collaborating pathologist selected 10 regions of interest (ROIs) per slide for evaluation. We created training materials and workflows to crowdsource pathologist image annotations on two modes: an optical microscope and two digital platforms. The microscope platform allows the same ROIs to be evaluated in both modes. The workflows collect the ROI type, a decision on whether the ROI is appropriate for estimating the density of sTILs, and if appropriate, the sTIL density value for that ROI.

Results:

In total, 19 pathologists made 1645 ROI evaluations during a data collection event and the following 2 weeks. The pilot study yielded an abundant number of cases with nominal sTIL infiltration. Furthermore, we found that the sTIL densities are correlated within a case, and there is notable pathologist variability. Consequently, we outline plans to improve our ROI and case sampling methods. We also outline statistical methods to account for ROI correlations within a case and pathologist variability when validating an algorithm.

Conclusion:

We have built workflows for efficient data collection and tested them in a pilot study. As we prepare for pivotal studies, we will investigate methods to use the dataset as an external validation tool for algorithms. We will also consider what it will take for the dataset to be fit for a regulatory purpose: study size, patient population, and pathologist training and qualifications. To this end, we will elicit feedback from the Food and Drug Administration via the Medical Device Development Tool program and from the broader digital pathology and AI community. Ultimately, we intend to share the dataset, statistical methods, and lessons learned.

Characterizing Long COVID: Deep Phenotype of a Complex Condition

BACKGROUND

Numerous publications describe the clinical manifestations of post-acute sequelae of SARS-CoV-2 (PASC or "long COVID"), but they are difficult to integrate because of heterogeneous methods and the lack of a standard for denoting the many phenotypic manifestations. Patient-led studies are of particular importance for understanding the natural history of COVID-19, but integration is hampered because they often use different terms to describe the same symptom or condition. This significant disparity in patient versus clinical characterization motivated the proposed ontological approach to specifying manifestations, which will improve capture and integration of future long COVID studies.

METHODS

The Human Phenotype Ontology (HPO) is a widely used standard for exchange and analysis of phenotypic abnormalities in human disease but has not yet been applied to the analysis of COVID-19.

FUNDING

We identified 303 articles published before April 29, 2021, curated 59 relevant manuscripts that described clinical manifestations in 81 cohorts three weeks or more following acute COVID-19, and mapped 287 unique clinical findings to HPO terms. We present layperson synonyms and definitions that can be used to link patient self-report questionnaires to standard medical terminology. Long COVID clinical manifestations are not assessed consistently across studies, and most manifestations have been reported with a wide range of synonyms by different authors. Across at least 10 cohorts, authors reported 31 unique clinical features corresponding to HPO terms; the most commonly reported feature was Fatigue (median 45.1%) and the least commonly reported was Nausea (median 3.9%), but the reported percentages varied widely between studies.

INTERPRETATION

Translating long COVID manifestations into computable HPO terms will improve analysis, data capture, and classification of long COVID patients. If researchers, clinicians, and patients share a common language, then studies can be compared/pooled more effectively. Furthermore, mapping lay terminology to HPO will help patients assist clinicians and researchers in creating phenotypic characterizations that are computationally accessible, thereby improving the stratification, diagnosis, and treatment of long COVID.

FUNDING

U24TR002306; UL1TR001439; P30AG024832; GBMF4552; R01HG010067; UL1TR002535; K23HL128909; UL1TR002389; K99GM145411.

Short- and Long-Term Recovery after Moderate/Severe AKI in Patients with and without COVID-19

Background

Severe AKI is strongly associated with poor outcomes in coronavirus disease 2019 (COVID-19), but data on renal recovery are lacking.

Methods

We retrospectively analyzed these associations in 3299 hospitalized patients (1338 with COVID-19 and 1961 with acute respiratory illness but who tested negative for COVID-19). Uni- and multivariable analyses were used to study mortality and recovery after Kidney Disease Improving Global Outcomes Stages 2 and 3 AKI (AKI-2/3), and Machine Learning was used to predict AKI and recovery using admission data. Long-term renal function and other outcomes were studied in a subgroup of AKI-2/3 survivors.

Results

Among the 172 COVID-19-negative patients with AKI-2/3, 74% had partial and 44% complete renal recovery, whereas 12% died. Among 255 COVID-19 positive patients with AKI-2/3, lower recovery and higher mortality were noted (51% partial renal recovery, 25% complete renal recovery, 24% died). On multivariable analysis, intensive care unit admission and acute respiratory distress syndrome were associated with nonrecovery, and recovery was significantly associated with survival in COVID-19-positive patients. With Machine Learning, we were able to predict recovery from COVID-19-associated AKI-2/3 with an average precision of 0.62, and the strongest predictors of recovery were initial arterial partial pressure of oxygen and carbon dioxide, serum creatinine, potassium, lymphocyte count, and creatine phosphokinase. At 12-month follow-up, among 52 survivors with AKI-2/3, 26% COVID-19-positive and 24% COVID-19-negative patients had incident or progressive CKD.

Conclusions

Recovery from COVID-19-associated moderate/severe AKI can be predicted using admission data and is associated with severity of respiratory disease and in-hospital death. The risk of CKD might be similar between COVID-19-positive and -negative patients.

Quantitative Imaging Informatics for Cancer Research

PURPOSE

We summarize Quantitative Imaging Informatics for Cancer Research (QIICR; U24 CA180918), one of the first projects funded by the National Cancer Institute (NCI) Informatics Technology for Cancer Research program.

METHODS

QIICR was motivated by the 3 use cases from the NCI Quantitative Imaging Network. 3D Slicer was selected as the platform for implementation of open-source quantitative imaging (QI) tools. Digital Imaging and Communications in Medicine (DICOM) was chosen for standardization of QI analysis outputs. Support of improved integration with community repositories focused on The Cancer Imaging Archive (TCIA). Priorities included improved capabilities of the standard, toolkits and tools, reference datasets, collaborations, and training and outreach.

RESULTS

Fourteen new tools to support head and neck cancer, glioblastoma, and prostate cancer QI research were introduced and downloaded over 100,000 times. DICOM was amended, with over 40 correction proposals addressing QI needs. Reference implementations of the standard in a popular toolkit and standalone tools were introduced. Eight datasets exemplifying the application of the standard and tools were contributed. An open demonstration/connectathon was organized, attracting the participation of academic groups and commercial vendors. Integration of tools with TCIA was improved by implementing programmatic communication interface and by refining best practices for QI analysis results curation.

CONCLUSION

Tools, capabilities of the DICOM standard, and datasets we introduced found adoption and utility within the cancer imaging community. A collaborative approach is critical to addressing challenges in imaging informatics at the national and international levels. Numerous challenges remain in establishing and maintaining the infrastructure of analysis tools and standardized datasets for the imaging community. Ideas and technology developed by the QIICR project are contributing to the NCI Imaging Data Commons currently being developed.

Children with SARS-CoV-2 in the National COVID Cohort Collaborative (N3C)

IMPORTANCE

SARS-CoV-2.

OBJECTIVE

To determine the characteristics, changes over time, outcomes, and severity risk factors of SARS-CoV-2 affected children within the National COVID Cohort Collaborative (N3C).

DESIGN

Prospective cohort study of patient encounters with end dates before May 27th, 2021.

SETTING

45 N3C institutions.

PARTICIPANTS

Children <19-years-old at initial SARS-CoV-2 testing.

MAIN OUTCOMES AND MEASURES

Case incidence and severity over time, demographic and comorbidity severity risk factors, vital sign and laboratory trajectories, clinical outcomes, and acute COVID-19 vs MIS-C contrasts for children infected with SARS-CoV-2.

RESULTS

728,047 children in the N3C were tested for SARS-CoV-2; of these, 91,865 (12.6%) were positive. Among the 5,213 (6%) hospitalized children, 685 (13%) met criteria for severe disease: mechanical ventilation (7%), vasopressor/inotropic support (7%), ECMO (0.6%), or death/discharge to hospice (1.1%). Male gender, African American race, older age, and several pediatric complex chronic condition (PCCC) subcategories were associated with higher clinical severity (p ≤ 0.05). Vital signs (all p≤0.002) and many laboratory tests from the first day of hospitalization were predictive of peak disease severity. Children with severe (vs moderate) disease were more likely to receive antimicrobials (71% vs 32%, p<0.001) and immunomodulatory medications (53% vs 16%, p<0.001). Compared to those with acute COVID-19, children with MIS-C were more likely to be male, Black/African American, 1-to-12-years-old, and less likely to have asthma, diabetes, or a PCCC (p < 0.04). MIS-C cases demonstrated a more inflammatory laboratory profile and more severe clinical phenotype with higher rates of invasive ventilation (12% vs 6%) and need for vasoactive-inotropic support (31% vs 6%) compared to acute COVID-19 cases, respectively (p<0.03).

CONCLUSIONS

In the largest U.S. SARS-CoV-2-positive pediatric cohort to date, we observed differences in demographics, pre-existing comorbidities, and initial vital sign and laboratory test values between severity subgroups. Taken together, these results suggest that early identification of children likely to progress to severe disease could be achieved using readily available data elements from the day of admission. Further work is needed to translate this knowledge into improved outcomes.

Quality assurance in radiation oncology

The Children's Oncology Group (COG) has a strong quality assurance (QA) program managed by the Imaging and Radiation Oncology Core (IROC). This program consists of credentialing centers and providing real-time management of each case for protocol compliant target definition and radiation delivery. In the International Society of Pediatric Oncology (SIOP), the lack of an available, reliable online data platform has been a challenge and the European Society for Paediatric Oncology (SIOPE) quality and excellence in radiotherapy and imaging for children and adolescents with cancer across Europe in clinical trials (QUARTET) program currently provides QA review for prospective clinical trials. The COG and SIOP are fully committed to a QA program that ensures uniform execution of protocol treatments and provides validity of the clinical data used for analysis.

Cancer Informatics for Cancer Centers (CI4CC): Building a Community Focused on Sharing Ideas and Best Practices to Improve Cancer Care and Patient Outcomes

Cancer Informatics for Cancer Centers (CI4CC) is a grassroots, nonprofit 501c3 organization intended to provide a focused national forum for engagement of senior cancer informatics leaders, primarily aimed at academic cancer centers anywhere in the world but with a special emphasis on the 70 National Cancer Institute-funded cancer centers. Although each of the participating cancer centers is structured differently, and leaders' titles vary, we know firsthand there are similarities in both the issues we face and the solutions we achieve. As a consortium, we have initiated a dedicated listserv, an open-initiatives program, and targeted biannual face-to-face meetings. These meetings are a place to review our priorities and initiatives, providing a forum for discussion of the strategic and pragmatic issues we, as informatics leaders, individually face at our respective institutions and cancer centers. Here we provide a brief history of the CI4CC organization and meeting highlights from the latest CI4CC meeting that took place in Napa, California from October 14-16, 2019. The focus of this meeting was "intersections between informatics, data science, and population science." We conclude with a discussion on "hot topics" on the horizon for cancer informatics.

Detecting Miscoded Diabetes Diagnosis Codes in Electronic Health Records for Quality Improvement: Temporal Deep Learning Approach

BACKGROUND

Diabetes affects more than 30 million patients across the United States. With such a large disease burden, even a small error in classification can be significant. Currently billing codes, assigned at the time of a medical encounter, are the "gold standard" reflecting the actual diseases present in an individual, and thus in aggregate reflect disease prevalence in the population. These codes are generated by highly trained coders and by health care providers but are not always accurate.

OBJECTIVE

This work provides a scalable deep learning methodology to more accurately classify individuals with diabetes across multiple health care systems.

METHODS

We leveraged a long short-term memory-dense neural network (LSTM-DNN) model to identify patients with or without diabetes using data from 5 acute care facilities with 187,187 patients and 275,407 encounters, incorporating data elements including laboratory test results, diagnostic/procedure codes, medications, demographic data, and admission information. Furthermore, a blinded physician panel reviewed discordant cases, providing an estimate of the total impact on the population.

RESULTS

When predicting the documented diagnosis of diabetes, our model achieved an 84% F1 score, 96% area under the curve-receiver operating characteristic curve, and 91% average precision on a heterogeneous data set from 5 distinct health facilities. However, in 81% of cases where the model disagreed with the documented phenotype, a blinded physician panel agreed with the model. Taken together, this suggests that 4.3% of our studied population have either missing or improper diabetes diagnosis.

CONCLUSIONS

This study demonstrates that deep learning methods can improve clinical phenotyping even when patient data are noisy, sparse, and heterogeneous.

Association of Proteinuria and Hematuria with Acute Kidney Injury and Mortality in Hospitalized Patients with COVID-19

INTRODUCTION

Acute kidney injury (AKI) is strongly associated with poor outcomes in hospitalized patients with coronavirus disease 2019 (COVID-19), but data on the association of proteinuria and hematuria are limited to non-US populations. In addition, admission and in-hospital measures for kidney abnormalities have not been studied separately.

METHODS

This retrospective cohort study aimed to analyze these associations in 321 patients sequentially admitted between March 7, 2020 and April 1, 2020 at Stony Brook University Medical Center, New York. We investigated the association of proteinuria, hematuria, and AKI with outcomes of inflammation, intensive care unit (ICU) admission, invasive mechanical ventilation (IMV), and in-hospital death. We used ANOVA, t test, χ2 test, and Fisher's exact test for bivariate analyses and logistic regression for multivariable analysis.

RESULTS

Three hundred patients met the inclusion criteria for the study cohort. Multivariable analysis demonstrated that admission proteinuria was significantly associated with risk of in-hospital AKI (OR 4.71, 95% CI 1.28-17.38), while admission hematuria was associated with ICU admission (OR 4.56, 95% CI 1.12-18.64), IMV (OR 8.79, 95% CI 2.08-37.00), and death (OR 18.03, 95% CI 2.84-114.57). During hospitalization, de novo proteinuria was significantly associated with increased risk of death (OR 8.94, 95% CI 1.19-114.4, p = 0.04). In-hospital AKI increased (OR 27.14, 95% CI 4.44-240.17) while recovery from in-hospital AKI decreased the risk of death (OR 0.001, 95% CI 0.001-0.06).

CONCLUSION

Proteinuria and hematuria both at the time of admission and during hospitalization are associated with adverse clinical outcomes in hospitalized patients with COVID-19.

Geospatial Distribution and Predictors of Mortality in Hospitalized Patients With COVID-19: A Cohort Study

BACKGROUND

The global coronavirus disease 2019 (COVID-19) pandemic offers the opportunity to assess how hospitals manage the care of hospitalized patients with varying demographics and clinical presentations. The goal of this study was to demonstrate the impact of densely populated residential areas on hospitalization and to identify predictors of length of stay and mortality in hospitalized patients with COVID-19 in one of the hardest hit counties internationally.

METHODS

This was a single-center cohort study of 1325 sequentially hospitalized patients with COVID-19 in New York between March 2, 2020, to May 11, 2020. Geospatial distribution of study patients' residences relative to population density in the region were mapped, and data analysis included hospital length of stay, need and duration of invasive mechanical ventilation (IMV), and mortality. Logistic regression models were constructed to predict discharge dispositions in the remaining active study patients.

RESULTS

The median age of the study cohort (interquartile range [IQR]) was 62 (49-75) years, and more than half were male (57%) with history of hypertension (60%), obesity (41%), and diabetes (42%). Geographic residence of the study patients was disproportionately associated with areas of higher population density (r s = 0.235; P = .004), with noted "hot spots" in the region. Study patients were predominantly hypertensive (MAP > 90 mmHg; 670, 51%) on presentation with lymphopenia (590, 55%), hyponatremia (411, 31%), and kidney dysfunction (estimated glomerular filtration rate < 60 mL/min/1.73 m2; 381, 29%). Of the patients with a disposition (1188/1325), 15% (182/1188) required IMV and 21% (250/1188) developed acute kidney injury. In patients on IMV, the median (IQR) hospital length of stay in survivors (22 [16.5-29.5] days) was significantly longer than that of nonsurvivors (15 [10-23.75] days), but this was not due to prolonged time on the ventilator. The overall mortality in all hospitalized patients was 15%, and in patients receiving IMV it was 48%, which is predicted to minimally rise from 48% to 49% based on logistic regression models constructed to project disposition in the remaining patients on ventilators. Acute kidney injury during hospitalization (odds ratioE, 3.23) was the strongest predictor of mortality in patients requiring IMV.

CONCLUSIONS

This is the first study to collectively utilize the demographics, clinical characteristics, and hospital course of COVID-19 patients to identify predictors of poor outcomes that can be used for resource allocation in future waves of the pandemic.

ImageBox 2 - Efficient and Rapid Access of Image Tiles from Whole-Slide Images Using Serverless HTTP Range Requests

BACKGROUND

Whole-slide images (WSI) are produced by a high-resolution scanning of pathology glass slides. There are a large number of whole-slide imaging scanners, and the resulting images are frequently larger than 100,000 × 100,000 pixels which typically image 100,000 to one million cells, ranging from several hundred megabytes to many gigabytes in size.

AIMS AND OBJECTIVES

Provide HTTP access over the web to Whole Slide Image tiles that do not have localized tiling servers but only basic HTTP access. Move all image decode and tiling functions to calling agent (ImageBox).

METHODS

Current software systems require tiling image servers to be installed on systems providing local disk access to these images. ImageBox2 breaks this requirement by accessing tiles from remote HTTP source via byte-level HTTP range requests. This method does not require changing the client software as the operation is relegated to the ImageBox2 server which is local (or remote) to the client and can access tiles from remote images that have no server of their own such as Amazon S3 hosted images. That is, it provides a data service [on a server that does not need to be managed], the definition of serverless execution model increasingly favored by cloud computing infrastructure.

CONCLUSIONS

The specific methodology described and assessed in this report preserves normal client connection semantics by enabling cloud-friendly tiling, promoting a web of http connected whole-slide images from a wide-ranging number of sources, and providing tiling where local tiling servers would have been otherwise unavailable.

Deep learning-based image analysis methods for brightfield-acquired multiplex immunohistochemistry images

BACKGROUND

Multiplex immunohistochemistry (mIHC) permits the labeling of six or more distinct cell types within a single histologic tissue section. The classification of each cell type requires detection of the unique colored chromogens localized to cells expressing biomarkers of interest. The most comprehensive and reproducible method to evaluate such slides is to employ digital pathology and image analysis pipelines to whole-slide images (WSIs). Our suite of deep learning tools quantitatively evaluates the expression of six biomarkers in mIHC WSIs. These methods address the current lack of readily available methods to evaluate more than four biomarkers and circumvent the need for specialized instrumentation to spectrally separate different colors. The use case application for our methods is a study that investigates tumor immune interactions in pancreatic ductal adenocarcinoma (PDAC) with a customized mIHC panel.

METHODS

Six different colored chromogens were utilized to label T-cells (CD3, CD4, CD8), B-cells (CD20), macrophages (CD16), and tumor cells (K17) in formalin-fixed paraffin-embedded (FFPE) PDAC tissue sections. We leveraged pathologist annotations to develop complementary deep learning-based methods: (1) ColorAE is a deep autoencoder which segments stained objects based on color; (2) U-Net is a convolutional neural network (CNN) trained to segment cells based on color, texture and shape; and ensemble methods that employ both ColorAE and U-Net, collectively referred to as (3) ColorAE:U-Net. We assessed the performance of our methods using: structural similarity and DICE score to evaluate segmentation results of ColorAE against traditional color deconvolution; F1 score, sensitivity, positive predictive value, and DICE score to evaluate the predictions from ColorAE, U-Net, and ColorAE:U-Net ensemble methods against pathologist-generated ground truth. We then used prediction results for spatial analysis (nearest neighbor).

RESULTS

We observed that (1) the performance of ColorAE is comparable to traditional color deconvolution for single-stain IHC images (note: traditional color deconvolution cannot be used for mIHC); (2) ColorAE and U-Net are complementary methods that detect 6 different classes of cells with comparable performance; (3) combinations of ColorAE and U-Net into ensemble methods outperform using either ColorAE and U-Net alone; and (4) ColorAE:U-Net ensemble methods can be employed for detailed analysis of the tumor microenvironment (TME). We developed a suite of scalable deep learning methods to analyze 6 distinctly labeled cell populations in mIHC WSIs. We evaluated our methods and found that they reliably detected and classified cells in the PDAC tumor microenvironment. We also present a use case, wherein we apply the ColorAE:U-Net ensemble method across 3 mIHC WSIs and use the predictions to quantify all stained cell populations and perform nearest neighbor spatial analysis. Thus, we provide proof of concept that these methods can be employed to quantitatively describe the spatial distribution immune cells within the tumor microenvironment. These complementary deep learning methods are readily deployable for use in clinical research studies.

Utilizing Automated Breast Cancer Detection to Identify Spatial Distributions of Tumor-Infiltrating Lymphocytes in Invasive Breast Cancer

Quantitative assessment of spatial relations between tumor and tumor-infiltrating lymphocytes (TIL) is increasingly important in both basic science and clinical aspects of breast cancer research. We have developed and evaluated convolutional neural network analysis pipelines to generate combined maps of cancer regions and TILs in routine diagnostic breast cancer whole slide tissue images. The combined maps provide insight about the structural patterns and spatial distribution of lymphocytic infiltrates and facilitate improved quantification of TILs. Both tumor and TIL analyses were evaluated by using three convolutional neural network networks (34-layer ResNet, 16-layer VGG, and Inception v4); the results compared favorably with those obtained by using the best published methods. We have produced open-source tools and a public data set consisting of tumor/TIL maps for 1090 invasive breast cancer images from The Cancer Genome Atlas. The maps can be downloaded for further downstream analyses.

Pitfalls in assessing stromal tumor infiltrating lymphocytes (sTILs) in breast cancer

Stromal tumor-infiltrating lymphocytes (sTILs) are important prognostic and predictive biomarkers in triple-negative (TNBC) and HER2-positive breast cancer. Incorporating sTILs into clinical practice necessitates reproducible assessment. Previously developed standardized scoring guidelines have been widely embraced by the clinical and research communities. We evaluated sources of variability in sTIL assessment by pathologists in three previous sTIL ring studies. We identify common challenges and evaluate impact of discrepancies on outcome estimates in early TNBC using a newly-developed prognostic tool. Discordant sTIL assessment is driven by heterogeneity in lymphocyte distribution. Additional factors include: technical slide-related issues; scoring outside the tumor boundary; tumors with minimal assessable stroma; including lymphocytes associated with other structures; and including other inflammatory cells. Small variations in sTIL assessment modestly alter risk estimation in early TNBC but have the potential to affect treatment selection if cutpoints are employed. Scoring and averaging multiple areas, as well as use of reference images, improve consistency of sTIL evaluation. Moreover, to assist in avoiding the pitfalls identified in this analysis, we developed an educational resource available at www.tilsinbreastcancer.org/pitfalls.

Report on computational assessment of Tumor Infiltrating Lymphocytes from the International Immuno-Oncology Biomarker Working Group

Assessment of tumor-infiltrating lymphocytes (TILs) is increasingly recognized as an integral part of the prognostic workflow in triple-negative (TNBC) and HER2-positive breast cancer, as well as many other solid tumors. This recognition has come about thanks to standardized visual reporting guidelines, which helped to reduce inter-reader variability. Now, there are ripe opportunities to employ computational methods that extract spatio-morphologic predictive features, enabling computer-aided diagnostics. We detail the benefits of computational TILs assessment, the readiness of TILs scoring for computational assessment, and outline considerations for overcoming key barriers to clinical translation in this arena. Specifically, we discuss: 1. ensuring computational workflows closely capture visual guidelines and standards; 2. challenges and thoughts standards for assessment of algorithms including training, preanalytical, analytical, and clinical validation; 3. perspectives on how to realize the potential of machine learning models and to overcome the perceptual and practical limits of visual scoring.

Machine Learning Based Opioid Overdose Prediction Using Electronic Health Records

Opioid addiction in the United States has come to national attention as opioid overdose (OD) related deaths have risen at alarming rates. Combating opioid epidemic becomes a high priority for not only governments but also healthcare providers. This depends on critical knowledge to understand the risk of opioid overdose of patients. In this paper, we present our work on building machine learning based prediction models to predict opioid overdose of patients based on the history of patients' electronic health records (EHR). We performed two studies using New York State claims data (SPARCS) with 440,000 patients and Cerner's Health Facts database with 110,000 patients. Our experiments demonstrated that EHR based prediction can achieve best recall with random forest method (precision: 95.3%, recall: 85.7%, F1 score: 90.3%), best precision with deep learning (precision: 99.2%, recall: 77.8%, F1 score: 87.2%). We also discovered that clinical events are among critical features for the predictions.

Segmentation and Classification in Digital Pathology for Glioma Research: Challenges and Deep Learning Approaches

Biomedical imaging Is an important source of information in cancer research. Characterizations of cancer morphology at onset, progression, and in response to treatment provide complementary information to that gleaned from genomics and clinical data. Accurate extraction and classification of both visual and latent image features Is an increasingly complex challenge due to the increased complexity and resolution of biomedical image data. In this paper, we present four deep learning-based image analysis methods from the Computational Precision Medicine (CPM) satellite event of the 21st International Medical Image Computing and Computer Assisted Intervention (MICCAI 2018) conference. One method Is a segmentation method designed to segment nuclei in whole slide tissue images (WSIs) of adult diffuse glioma cases. It achieved a Dice similarity coefficient of 0.868 with the CPM challenge datasets. Three methods are classification methods developed to categorize adult diffuse glioma cases into oligodendroglioma and astrocytoma classes using radiographic and histologic image data. These methods achieved accuracy values of 0.75, 0.80, and 0.90, measured as the ratio of the number of correct classifications to the number of total cases, with the challenge datasets. The evaluations of the four methods indicate that (1) carefully constructed deep learning algorithms are able to produce high accuracy in the analysis of biomedical image data and (2) the combination of radiographic with histologic image information improves classification performance.

Open access image repositories: high-quality data to enable machine learning research

Originally motivated by the need for research reproducibility and data reuse, large-scale, open access information repositories have become key resources for training and testing of advanced machine learning applications in biomedical and clinical research. To be of value, such repositories must provide large, high-quality data sets, where quality is defined as minimising variance due to data collection protocols and data misrepresentations. Curation is the key to quality. We have constructed a large public access image repository, The Cancer Imaging Archive, dedicated to the promotion of open science to advance the global effort to diagnose and treat cancer. Drawing on this experience and our experience in applying machine learning techniques to the analysis of radiology and pathology image data, we will review the requirements placed on such information repositories by state-of-the-art machine learning applications and how these requirements can be met.