Clinically relevant orthotopic pancreatic cancer models for adoptive T cell transfer therapy

BACKGROUND

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive tumor. Prognosis is poor and survival is low in patients diagnosed with this disease, with a survival rate of ~12% at 5 years. Immunotherapy, including adoptive T cell transfer therapy, has not impacted the outcomes in patients with PDAC, due in part to the hostile tumor microenvironment (TME) which limits T cell trafficking and persistence. We posit that murine models serve as useful tools to study the fate of T cell therapy. Currently, genetically engineered mouse models (GEMMs) for PDAC are considered a "gold-standard" as they recapitulate many aspects of human disease. However, these models have limitations, including marked tumor variability across individual mice and the cost of colony maintenance.

METHODS

Using flow cytometry and immunohistochemistry, we characterized the immunological features and trafficking patterns of adoptively transferred T cells in orthotopic PDAC (C57BL/6) models using two mouse cell lines, KPC-Luc and MT-5, isolated from C57BL/6 KPC-GEMM (KrasLSL-G12D/+p53-/- and KrasLSL-G12D/+p53LSL-R172H/+, respectively).

RESULTS

The MT-5 orthotopic model best recapitulates the cellular and stromal features of the TME in the PDAC GEMM. In contrast, far more host immune cells infiltrate the KPC-Luc tumors, which have less stroma, although CD4+ and CD8+ T cells were similarly detected in the MT-5 tumors compared with KPC-GEMM in mice. Interestingly, we found that chimeric antigen receptor (CAR) T cells redirected to recognize mesothelin on these tumors that signal via CD3ζ and 41BB (Meso-41BBζ-CAR T cells) infiltrated the tumors of mice bearing stroma-devoid KPC-Luc orthotopic tumors, but not MT-5 tumors.

CONCLUSIONS

Our data establish for the first time a reproducible and realistic clinical system useful for modeling stroma-rich and stroma-devoid PDAC tumors. These models shall serve an indepth study of how to overcome barriers that limit antitumor activity of adoptively transferred T cells.

Risk of post-acute sequelae of SARS-CoV-2 infection associated with pre-coronavirus disease obstructive sleep apnea diagnoses: an electronic health record-based analysis from the RECOVER initiative

STUDY OBJECTIVES

Obstructive sleep apnea (OSA) has been associated with more severe acute coronavirus disease-2019 (COVID-19) outcomes. We assessed OSA as a potential risk factor for Post-Acute Sequelae of SARS-CoV-2 (PASC).

METHODS

We assessed the impact of preexisting OSA on the risk for probable PASC in adults and children using electronic health record data from multiple research networks. Three research networks within the REsearching COVID to Enhance Recovery initiative (PCORnet Adult, PCORnet Pediatric, and the National COVID Cohort Collaborative [N3C]) employed a harmonized analytic approach to examine the risk of probable PASC in COVID-19-positive patients with and without a diagnosis of OSA prior to pandemic onset. Unadjusted odds ratios (ORs) were calculated as well as ORs adjusted for age group, sex, race/ethnicity, hospitalization status, obesity, and preexisting comorbidities.

RESULTS

Across networks, the unadjusted OR for probable PASC associated with a preexisting OSA diagnosis in adults and children ranged from 1.41 to 3.93. Adjusted analyses found an attenuated association that remained significant among adults only. Multiple sensitivity analyses with expanded inclusion criteria and covariates yielded results consistent with the primary analysis.

CONCLUSIONS

Adults with preexisting OSA were found to have significantly elevated odds of probable PASC. This finding was consistent across data sources, approaches for identifying COVID-19-positive patients, and definitions of PASC. Patients with OSA may be at elevated risk for PASC after SARS-CoV-2 infection and should be monitored for post-acute sequelae.

Coordinated single-cell tumor microenvironment dynamics reinforce pancreatic cancer subtype

Bulk analyses of pancreatic ductal adenocarcinoma (PDAC) samples are complicated by the tumor microenvironment (TME), i.e. signals from fibroblasts, endocrine, exocrine, and immune cells. Despite this, we and others have established tumor and stroma subtypes with prognostic significance. However, understanding of underlying signals driving distinct immune and stromal landscapes is still incomplete. Here we integrate 92 single cell RNA-seq samples from seven independent studies to build a reproducible PDAC atlas with a focus on tumor-TME interdependence. Patients with activated stroma are synonymous with higher myofibroblastic and immunogenic fibroblasts, and furthermore show increased M2-like macrophages and regulatory T-cells. Contrastingly, patients with 'normal' stroma show M1-like recruitment, elevated effector and exhausted T-cells. To aid interoperability of future studies, we provide a pretrained cell type classifier and an atlas of subtype-based signaling factors that we also validate in mouse data. Ultimately, this work leverages the heterogeneity among single-cell studies to create a comprehensive view of the orchestra of signaling interactions governing PDAC.

Geographic and Temporal Trends in COVID-Associated Acute Kidney Injury in the National COVID Cohort Collaborative

BACKGROUND

AKI is associated with mortality in patients hospitalized with coronavirus disease 2019 (COVID-19); however, its incidence, geographic distribution, and temporal trends since the start of the pandemic are understudied.

METHODS

Electronic health record data were obtained from 53 health systems in the United States in the National COVID Cohort Collaborative. We selected hospitalized adults diagnosed with COVID-19 between March 6, 2020, and January 6, 2022. AKI was determined with serum creatinine and diagnosis codes. Time was divided into 16-week periods (P1-6) and geographical regions into Northeast, Midwest, South, and West. Multivariable models were used to analyze the risk factors for AKI or mortality.

RESULTS

Of a total cohort of 336,473, 129,176 (38%) patients had AKI. Fifty-six thousand three hundred and twenty-two (17%) lacked a diagnosis code but had AKI based on the change in serum creatinine. Similar to patients coded for AKI, these patients had higher mortality compared with those without AKI. The incidence of AKI was highest in P1 (47%; 23,097/48,947), lower in P2 (37%; 12,102/32,513), and relatively stable thereafter. Compared with the Midwest, the Northeast, South, and West had higher adjusted odds of AKI in P1. Subsequently, the South and West regions continued to have the highest relative AKI odds. In multivariable models, AKI defined by either serum creatinine or diagnostic code and the severity of AKI was associated with mortality.

CONCLUSIONS

The incidence and distribution of COVID-19-associated AKI changed since the first wave of the pandemic in the United States.

PODCAST

This article contains a podcast at https://dts.podtrac.com/redirect.mp3/www.asn-online.org/media/podcast/CJASN/2023_08_08_CJN0000000000000192.mp3.

FaStaNMF: a Fast and Stable Non-negative Matrix Factorization for Gene Expression

Gene expression analysis of samples with mixed cell types only provides limited insight to the characteristics of specific tissues. In silico deconvolution can be applied to extract cell type specific expression, thus avoiding prohibitively expensive techniques such as cell sorting or single-cell sequencing. Non-negative matrix factorization (NMF) is a deconvolution method shown to be useful for gene expression data, in part due to its constraint of non-negativity. Unlike other methods, NMF provides the capability to deconvolve without prior knowledge of the components of the model. However, NMF is not guaranteed to provide a globally unique solution. In this work, we present FaStaNMF, a method that balances achieving global stability of the NMF results, which is essential for inter-experiment and inter-lab reproducibility, with accuracy and speed. Results: FaStaNMF was applied to four datasets with known ground truth, created based on publicly available data or by using our simulation infrastructure, RNAGinesis. We assessed FaStaNMF on three criteria - speed, accuracy, and stability, and it favorably compared to the standard approach of achieving reproduceable results with NMF. We expect that FaStaNMF can be applied successfully to a wide array of biological data, such as different tumor/immune and other disease microenvironments.

A framework for future national pediatric pandemic respiratory disease severity triage: The HHS pediatric COVID-19 data challenge

Introduction

With persistent incidence, incomplete vaccination rates, confounding respiratory illnesses, and few therapeutic interventions available, COVID-19 continues to be a burden on the pediatric population. During a surge, it is difficult for hospitals to direct limited healthcare resources effectively. While the overwhelming majority of pediatric infections are mild, there have been life-threatening exceptions that illuminated the need to proactively identify pediatric patients at risk of severe COVID-19 and other respiratory infectious diseases. However, a nationwide capability for developing validated computational tools to identify pediatric patients at risk using real-world data does not exist.

Methods

HHS ASPR BARDA sought, through the power of competition in a challenge, to create computational models to address two clinically important questions using the National COVID Cohort Collaborative: (1) Of pediatric patients who test positive for COVID-19 in an outpatient setting, who are at risk for hospitalization? (2) Of pediatric patients who test positive for COVID-19 and are hospitalized, who are at risk for needing mechanical ventilation or cardiovascular interventions?

Results

This challenge was the first, multi-agency, coordinated computational challenge carried out by the federal government as a response to a public health emergency. Fifty-five computational models were evaluated across both tasks and two winners and three honorable mentions were selected.

Conclusion

This challenge serves as a framework for how the government, research communities, and large data repositories can be brought together to source solutions when resources are strapped during a pandemic.

De-black-boxing health AI: demonstrating reproducible machine learning computable phenotypes using the N3C-RECOVER Long COVID model in the All of Us data repository

Machine learning (ML)-driven computable phenotypes are among the most challenging to share and reproduce. Despite this difficulty, the urgent public health considerations around Long COVID make it especially important to ensure the rigor and reproducibility of Long COVID phenotyping algorithms such that they can be made available to a broad audience of researchers. As part of the NIH Researching COVID to Enhance Recovery (RECOVER) Initiative, researchers with the National COVID Cohort Collaborative (N3C) devised and trained an ML-based phenotype to identify patients highly probable to have Long COVID. Supported by RECOVER, N3C and NIH's All of Us study partnered to reproduce the output of N3C's trained model in the All of Us data enclave, demonstrating model extensibility in multiple environments. This case study in ML-based phenotype reuse illustrates how open-source software best practices and cross-site collaboration can de-black-box phenotyping algorithms, prevent unnecessary rework, and promote open science in informatics.

Long COVID risk and pre-COVID vaccination in an EHR-based cohort study from the RECOVER program

Long COVID, or complications arising from COVID-19 weeks after infection, has become a central concern for public health experts. The United States National Institutes of Health founded the RECOVER initiative to better understand long COVID. We used electronic health records available through the National COVID Cohort Collaborative to characterize the association between SARS-CoV-2 vaccination and long COVID diagnosis. Among patients with a COVID-19 infection between August 1, 2021 and January 31, 2022, we defined two cohorts using distinct definitions of long COVID-a clinical diagnosis (n = 47,404) or a previously described computational phenotype (n = 198,514)-to compare unvaccinated individuals to those with a complete vaccine series prior to infection. Evidence of long COVID was monitored through June or July of 2022, depending on patients' data availability. We found that vaccination was consistently associated with lower odds and rates of long COVID clinical diagnosis and high-confidence computationally derived diagnosis after adjusting for sex, demographics, and medical history.

A method for comparing multiple imputation techniques: A case study on the U.S. national COVID cohort collaborative

Healthcare datasets obtained from Electronic Health Records have proven to be extremely useful for assessing associations between patients' predictors and outcomes of interest. However, these datasets often suffer from missing values in a high proportion of cases, whose removal may introduce severe bias. Several multiple imputation algorithms have been proposed to attempt to recover the missing information under an assumed missingness mechanism. Each algorithm presents strengths and weaknesses, and there is currently no consensus on which multiple imputation algorithm works best in a given scenario. Furthermore, the selection of each algorithm's parameters and data-related modeling choices are also both crucial and challenging. In this paper we propose a novel framework to numerically evaluate strategies for handling missing data in the context of statistical analysis, with a particular focus on multiple imputation techniques. We demonstrate the feasibility of our approach on a large cohort of type-2 diabetes patients provided by the National COVID Cohort Collaborative (N3C) Enclave, where we explored the influence of various patient characteristics on outcomes related to COVID-19. Our analysis included classic multiple imputation techniques as well as simple complete-case Inverse Probability Weighted models. Extensive experiments show that our approach can effectively highlight the most promising and performant missing-data handling strategy for our case study. Moreover, our methodology allowed a better understanding of the behavior of the different models and of how it changed as we modified their parameters. Our method is general and can be applied to different research fields and on datasets containing heterogeneous types.

Abstract A058: Meta-analysis of single-cell RNA expression in genetically engineered mouse models of pancreatic ductal adenocarcinoma reveals inter-model heterogeneity

Background: Genetically engineered mouse models (GEMMs) are widely used in the study of pancreatic ductal adenocarcinoma (PDAC) because of their immune-competent tumor microenvironment (TME); however, the extent to which particular GEMMs recapitulate the tumor and TME observed in the patient population has not been systematically evaluated. In this study, we integrate single-cell RNA sequencing (sc-RNA-seq) data from multiple studies and multiple GEMM backgrounds to identify differences in the cellular compositions of popular PDAC GEMMs. Methods: A total of 49,191 cells were used from three studies, including normal mouse pancreas (N=2) and five different GEMM backgrounds (N=16). Data curation, integration, and analysis were based on the Seurat pipeline in R. SingelCellNet was used to train a random forest model on manually labeled human sc-RNA-seq data from 20 patients. To enable cross-species use, the classifier was trained using only genes with both human and mouse homologues. Cells classified as neoplastic were further clustered to quantify the number of classical and basal-like cells based on signature gene expression levels. The ratio of these subtypes in each GEMM and the relationship between the modified genes in each model were examined. Results: Ad-hoc clustering and a human-cell-trained single-cell classifier showed 79% agreement in an integrated data set of PDAC GEMMs. Cells identified by both methods as tumor (8,303 cells, 17% of total) were assessed for PDAC tumor subtype via subsequent clustering analysis (basal-like or classical). When comparing the ratio of differently subtyped tumor cells, we identified stark differences between GEMM genetic backgrounds. Among five different models, KIC (KrasLSL−G12D/+Ink4a/Arffl/flPtf1aCre/+), KPPCN (KrasLSL−G12D/+Trp53fl/flPdx1Cre/+Nsdhlfl/fl), and pdx1-KPC (KrasLSL−G12D/+Trp53LSL-R172H/+Pdx1Cre/+) exhibited a higher proportion of basal-like PDAC cells compared to KPfC/KPPC (KrasLSL−G12D/+Trp53fl/flPdx1Cre/+) and ptf1a-KPC (KrasLSL−G12D/+Trp53LSL-R172H/+Ptf1aCre/+). Interestingly, in the KIC model, which was harvested at early and late time points (40 or 60 days), classical PDAC was overrepresented in early models, and basal-like PDAC was more prevalent in the older tumors. While sample sizes are limited in this study, in Pdx1 driven models, we observed a bias towards basal-like phenotype in GEMMs using the Trp53LSL-R172H/+ method compared to those with Trp53fl/fl. Conclusions: In a comparison of publicly available sc-RNA-seq, we highlight potential biases in the molecular subtypes that arise from specific PDAC GEMMs. Because of the known link between tumor subtype and therapeutic response, these results suggest translational work may benefit from GEMM selection that considers transcriptomic diversity.

Citation Format: Yun Jae Yoo, Ki H Oh, Luke A. Torre-Healy, Richard A. Moffitt. Meta-analysis of single-cell RNA expression in genetically engineered mouse models of pancreatic ductal adenocarcinoma reveals inter-model heterogeneity [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 A058.

Long COVID Risk and Pre-COVID Vaccination: An EHR-Based Cohort Study from the RECOVER Program

Importance

Characterizing the effect of vaccination on long COVID allows for better healthcare recommendations.

Objective

To determine if, and to what degree, vaccination prior to COVID-19 is associated with eventual long COVID onset, among those a documented COVID-19 infection.

Design Settings and Participants

Retrospective cohort study of adults with evidence of COVID-19 between August 1, 2021 and January 31, 2022 based on electronic health records from eleven healthcare institutions taking part in the NIH Researching COVID to Enhance Recovery (RECOVER) Initiative, a project of the National Covid Cohort Collaborative (N3C).

Exposures

Pre-COVID-19 receipt of a complete vaccine series versus no pre-COVID-19 vaccination.

Main Outcomes and Measures

Two approaches to the identification of long COVID were used. In the clinical diagnosis cohort (n=47,752), ICD-10 diagnosis codes or evidence of a healthcare encounter at a long COVID clinic were used. In the model-based cohort (n=199,498), a computable phenotype was used. The association between pre-COVID vaccination and long COVID was estimated using IPTW-adjusted logistic regression and Cox proportional hazards.

Results

In both cohorts, when adjusting for demographics and medical history, pre-COVID vaccination was associated with a reduced risk of long COVID (clinic-based cohort: HR, 0.66; 95% CI, 0.55-0.80; OR, 0.69; 95% CI, 0.59-0.82; model-based cohort: HR, 0.62; 95% CI, 0.56-0.69; OR, 0.70; 95% CI, 0.65-0.75).

Conclusions and Relevance

Long COVID has become a central concern for public health experts. Prior studies have considered the effect of vaccination on the prevalence of future long COVID symptoms, but ours is the first to thoroughly characterize the association between vaccination and clinically diagnosed or computationally derived long COVID. Our results bolster the growing consensus that vaccines retain protective effects against long COVID even in breakthrough infections.

Key Points

Question: Does vaccination prior to COVID-19 onset change the risk of long COVID diagnosis?Findings: Four observational analyses of EHRs showed a statistically significant reduction in long COVID risk associated with pre-COVID vaccination (first cohort: HR, 0.66; 95% CI, 0.55-0.80; OR, 0.69; 95% CI, 0.59-0.82; second cohort: HR, 0.62; 95% CI, 0.56-0.69; OR, 0.70; 95% CI, 0.65-0.75).Meaning: Vaccination prior to COVID onset has a protective association with long COVID even in the case of breakthrough infections.

COVID-19-associated AKI in hospitalized US patients: incidence, temporal trends, geographical distribution, risk factors and mortality

Background

Acute kidney injury (AKI) is associated with mortality in patients hospitalized with COVID-19, however, its incidence, geographic distribution, and temporal trends since the start of the pandemic are understudied.

Methods

Electronic health record data were obtained from 53 health systems in the United States (US) in the National COVID Cohort Collaborative (N3C). We selected hospitalized adults diagnosed with COVID-19 between March 6th, 2020, and January 6th, 2022. AKI was determined with serum creatinine (SCr) and diagnosis codes. Time were divided into 16-weeks (P1-6) periods and geographical regions into Northeast, Midwest, South, and West. Multivariable models were used to analyze the risk factors for AKI or mortality.

Results

Out of a total cohort of 306,061, 126,478 (41.0 %) patients had AKI. Among these, 17.9% lacked a diagnosis code but had AKI based on the change in SCr. Similar to patients coded for AKI, these patients had higher mortality compared to those without AKI. The incidence of AKI was highest in P1 (49.3%), reduced in P2 (40.6%), and relatively stable thereafter. Compared to the Midwest, the Northeast, South, and West had higher adjusted AKI incidence in P1, subsequently, the South and West regions continued to have the highest relative incidence. In multivariable models, AKI defined by either SCr or diagnostic code, and the severity of AKI was associated with mortality.

Conclusions

Uncoded cases of COVID-19-associated AKI are common and associated with mortality. The incidence and distribution of COVID-19-associated AKI have changed since the first wave of the pandemic in the US.

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.

Synergies between centralized and federated approaches to data quality: a report from the national COVID cohort collaborative

OBJECTIVE

In response to COVID-19, the informatics community united to aggregate as much clinical data as possible to characterize this new disease and reduce its impact through collaborative analytics. The National COVID Cohort Collaborative (N3C) is now the largest publicly available HIPAA limited dataset in US history with over 6.4 million patients and is a testament to a partnership of over 100 organizations.

MATERIALS AND METHODS

We developed a pipeline for ingesting, harmonizing, and centralizing data from 56 contributing data partners using 4 federated Common Data Models. N3C data quality (DQ) review involves both automated and manual procedures. In the process, several DQ heuristics were discovered in our centralized context, both within the pipeline and during downstream project-based analysis. Feedback to the sites led to many local and centralized DQ improvements.

RESULTS

Beyond well-recognized DQ findings, we discovered 15 heuristics relating to source Common Data Model conformance, demographics, COVID tests, conditions, encounters, measurements, observations, coding completeness, and fitness for use. Of 56 sites, 37 sites (66%) demonstrated issues through these heuristics. These 37 sites demonstrated improvement after receiving feedback.

DISCUSSION

We encountered site-to-site differences in DQ which would have been challenging to discover using federated checks alone. We have demonstrated that centralized DQ benchmarking reveals unique opportunities for DQ improvement that will support improved research analytics locally and in aggregate.

CONCLUSION

By combining rapid, continual assessment of DQ with a large volume of multisite data, it is possible to support more nuanced scientific questions with the scale and rigor that they require.

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.

Clinical Characterization and Prediction of Clinical Severity of SARS-CoV-2 Infection Among US Adults Using Data From the US National COVID Cohort Collaborative

Importance

The National COVID Cohort Collaborative (N3C) is a centralized, harmonized, high-granularity electronic health record repository that is the largest, most representative COVID-19 cohort to date. This multicenter data set can support robust evidence-based development of predictive and diagnostic tools and inform clinical care and policy.

Objectives

To evaluate COVID-19 severity and risk factors over time and assess the use of machine learning to predict clinical severity.

Design, Setting, and Participants

In a retrospective cohort study of 1 926 526 US adults with SARS-CoV-2 infection (polymerase chain reaction >99% or antigen <1%) and adult patients without SARS-CoV-2 infection who served as controls from 34 medical centers nationwide between January 1, 2020, and December 7, 2020, patients were stratified using a World Health Organization COVID-19 severity scale and demographic characteristics. Differences between groups over time were evaluated using multivariable logistic regression. Random forest and XGBoost models were used to predict severe clinical course (death, discharge to hospice, invasive ventilatory support, or extracorporeal membrane oxygenation).

Main Outcomes and Measures

Patient demographic characteristics and COVID-19 severity using the World Health Organization COVID-19 severity scale and differences between groups over time using multivariable logistic regression.

Results

The cohort included 174 568 adults who tested positive for SARS-CoV-2 (mean [SD] age, 44.4 [18.6] years; 53.2% female) and 1 133 848 adult controls who tested negative for SARS-CoV-2 (mean [SD] age, 49.5 [19.2] years; 57.1% female). Of the 174 568 adults with SARS-CoV-2, 32 472 (18.6%) were hospitalized, and 6565 (20.2%) of those had a severe clinical course (invasive ventilatory support, extracorporeal membrane oxygenation, death, or discharge to hospice). Of the hospitalized patients, mortality was 11.6% overall and decreased from 16.4% in March to April 2020 to 8.6% in September to October 2020 (P = .002 for monthly trend). Using 64 inputs available on the first hospital day, this study predicted a severe clinical course using random forest and XGBoost models (area under the receiver operating curve = 0.87 for both) that were stable over time. The factor most strongly associated with clinical severity was pH; this result was consistent across machine learning methods. In a separate multivariable logistic regression model built for inference, age (odds ratio [OR], 1.03 per year; 95% CI, 1.03-1.04), male sex (OR, 1.60; 95% CI, 1.51-1.69), liver disease (OR, 1.20; 95% CI, 1.08-1.34), dementia (OR, 1.26; 95% CI, 1.13-1.41), African American (OR, 1.12; 95% CI, 1.05-1.20) and Asian (OR, 1.33; 95% CI, 1.12-1.57) race, and obesity (OR, 1.36; 95% CI, 1.27-1.46) were independently associated with higher clinical severity.

Conclusions and Relevance

This cohort study found that COVID-19 mortality decreased over time during 2020 and that patient demographic characteristics and comorbidities were associated with higher clinical severity. The machine learning models accurately predicted ultimate clinical severity using commonly collected clinical data from the first 24 hours of a hospital admission.

Association Between Glucagon-Like Peptide 1 Receptor Agonist and Sodium-Glucose Cotransporter 2 Inhibitor Use and COVID-19 Outcomes

OBJECTIVE

To determine the respective associations of premorbid glucagon-like peptide-1 receptor agonist (GLP1-RA) and sodium-glucose cotransporter 2 inhibitor (SGLT2i) use, compared with premorbid dipeptidyl peptidase 4 inhibitor (DPP4i) use, with severity of outcomes in the setting of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection.

RESEARCH DESIGN AND METHODS

We analyzed observational data from SARS-CoV-2-positive adults in the National COVID Cohort Collaborative (N3C), a multicenter, longitudinal U.S. cohort (January 2018-February 2021), with a prescription for GLP1-RA, SGLT2i, or DPP4i within 24 months of positive SARS-CoV-2 PCR test. The primary outcome was 60-day mortality, measured from positive SARS-CoV-2 test date. Secondary outcomes were total mortality during the observation period and emergency room visits, hospitalization, and mechanical ventilation within 14 days. Associations were quantified with odds ratios (ORs) estimated with targeted maximum likelihood estimation using a super learner approach, accounting for baseline characteristics.

RESULTS

The study included 12,446 individuals (53.4% female, 62.5% White, mean ± SD age 58.6 ± 13.1 years). The 60-day mortality was 3.11% (387 of 12,446), with 2.06% (138 of 6,692) for GLP1-RA use, 2.32% (85 of 3,665) for SGLT2i use, and 5.67% (199 of 3,511) for DPP4i use. Both GLP1-RA and SGLT2i use were associated with lower 60-day mortality compared with DPP4i use (OR 0.54 [95% CI 0.37-0.80] and 0.66 [0.50-0.86], respectively). Use of both medications was also associated with decreased total mortality, emergency room visits, and hospitalizations.

CONCLUSIONS

Among SARS-CoV-2-positive adults, premorbid GLP1-RA and SGLT2i use, compared with DPP4i use, was associated with lower odds of mortality and other adverse outcomes, although DPP4i users were older and generally sicker.

Challenges in defining Long COVID: Striking differences across literature, Electronic Health Records, and patient-reported information

Since late 2019, the novel coronavirus SARS-CoV-2 has introduced a wide array of health challenges globally. In addition to a complex acute presentation that can affect multiple organ systems, increasing evidence points to long-term sequelae being common and impactful. The worldwide scientific community is forging ahead to characterize a wide range of outcomes associated with SARS-CoV-2 infection; however the underlying assumptions in these studies have varied so widely that the resulting data are difficult to compareFormal definitions are needed in order to design robust and consistent studies of Long COVID that consistently capture variation in long-term outcomes. Even the condition itself goes by three terms, most widely "Long COVID", but also "COVID-19 syndrome (PACS)" or, "post-acute sequelae of SARS-CoV-2 infection (PASC)". In the present study, we investigate the definitions used in the literature published to date and compare them against data available from electronic health records and patient-reported information collected via surveys. Long COVID holds the potential to produce a second public health crisis on the heels of the pandemic itself. Proactive efforts to identify the characteristics of this heterogeneous condition are imperative for a rigorous scientific effort to investigate and mitigate this threat.

A distinct microglial subset at the tumor-stroma interface of glioma

The characterization of the tumor microenvironment (TME) in high grade gliomas (HGG) has generated significant interest in an effort to understand how neoplastic lesions in the central nervous system (CNS) are supported and to devise novel therapeutic targets. The TME of the CNS contains unique and specialized cells, including the resident myeloid cells, microglia. Myeloid involvement in HGG, such as glioblastoma, is associated with poor outcomes. Glioma-associated microglia and infiltrating monocytes/macrophages (GAM) accumulate within the neoplastic lesion where they facilitate tumor growth and drive immunosuppression. However, it has been difficult to differentiate whether microglia and macrophages have similar or distinct roles in pathology, and if the spatial organization of these cells informs outcomes. Here, we characterize the tumor-stroma border and identify peritumoral GAM (PGAM) as a unique subpopulation of GAM. Using data mining and analyses of samples derived from both murine and human sources we show that PGAM exhibit a pro-inflammatory and chemotactic phenotype that is associated with peripheral monocyte recruitment, and decreased overall survival. PGAM act as a unique subset of GAM at the tumor-stroma interface. We define a novel gene signature to identify these cells and suggest that PGAM constitute a cellular target of the TME.

The National COVID Cohort Collaborative: Clinical Characterization and Early Severity Prediction

Background:

The majority of U.S. reports of COVID-19 clinical characteristics, disease course, and treatments are from single health systems or focused on one domain. Here we report the creation of the National COVID Cohort Collaborative (N3C), a centralized, harmonized, high-granularity electronic health record repository that is the largest, most representative U.S. cohort of COVID-19 cases and controls to date. This multi-center dataset supports robust evidence-based development of predictive and diagnostic tools and informs critical care and policy.

Methods and Findings:

In a retrospective cohort study of 1,926,526 patients from 34 medical centers nationwide, we stratified patients using a World Health Organization COVID-19 severity scale and demographics; we then evaluated differences between groups over time using multivariable logistic regression. We established vital signs and laboratory values among COVID-19 patients with different severities, providing the foundation for predictive analytics. The cohort included 174,568 adults with severe acute respiratory syndrome associated with SARS-CoV-2 (PCR >99% or antigen <1%) as well as 1,133,848 adult patients that served as lab-negative controls. Among 32,472 hospitalized patients, mortality was 11.6% overall and decreased from 16.4% in March/April 2020 to 8.6% in September/October 2020 (p = 0.002 monthly trend). In a multivariable logistic regression model, age, male sex, liver disease, dementia, African-American and Asian race, and obesity were independently associated with higher clinical severity. To demonstrate the utility of the N3C cohort for analytics, we used machine learning (ML) to predict clinical severity and risk factors over time. Using 64 inputs available on the first hospital day, we predicted a severe clinical course (death, discharge to hospice, invasive ventilation, or extracorporeal membrane oxygenation) using random forest and XGBoost models (AUROC 0.86 and 0.87 respectively) that were stable over time. The most powerful predictors in these models are patient age and widely available vital sign and laboratory values. The established expected trajectories for many vital signs and laboratory values among patients with different clinical severities validates observations from smaller studies, and provides comprehensive insight into COVID-19 characterization in U.S. patients.

Conclusions:

This is the first description of an ongoing longitudinal observational study of patients seen in diverse clinical settings and geographical regions and is the largest COVID-19 cohort in the United States. Such data are the foundation for ML models that can be the basis for generalizable clinical decision support tools. The N3C Data Enclave is unique in providing transparent, reproducible, easily shared, versioned, and fully auditable data and analytic provenance for national-scale patient-level EHR data. The N3C is built for intensive ML analyses by academic, industry, and citizen scientists internationally. Many observational correlations can inform trial designs and care guidelines for this new disease.

Cholesterol Pathway Inhibition Induces TGF-β Signaling to Promote Basal Differentiation in Pancreatic Cancer

Oncogenic transformation alters lipid metabolism to sustain tumor growth. We define a mechanism by which cholesterol metabolism controls the development and differentiation of pancreatic ductal adenocarcinoma (PDAC). Disruption of distal cholesterol biosynthesis by conditional inactivation of the rate-limiting enzyme Nsdhl or treatment with cholesterol-lowering statins switches glandular pancreatic carcinomas to a basal (mesenchymal) phenotype in mouse models driven by Kras^G12D expression and homozygous Trp53 loss. Consistently, PDACs in patients receiving statins show enhanced mesenchymal features. Mechanistically, statins and NSDHL loss induce SREBP1 activation, which promotes the expression of Tgfb1, enabling epithelial-mesenchymal transition. Evidence from patient samples in this study suggests that activation of transforming growth factor β signaling and epithelial-mesenchymal transition by cholesterol-lowering statins may promote the basal type of PDAC, conferring poor outcomes in patients.

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.

Irreversible JNK1-JUN inhibition by JNK-IN-8 sensitizes pancreatic cancer to 5-FU/FOLFOX chemotherapy

Over 55,000 people in the United States are diagnosed with pancreatic ductal adenocarcinoma (PDAC) yearly, and fewer than 20% of these patients survive a year beyond diagnosis. Chemotherapies are considered or used in nearly every PDAC case, but there is limited understanding of the complex signaling responses underlying resistance to these common treatments. Here, we take an unbiased approach to study protein kinase network changes following chemotherapies in patient-derived xenograft (PDX) models of PDAC to facilitate design of rational drug combinations. Proteomics profiling following chemotherapy regimens reveals that activation of JNK-JUN signaling occurs after 5-fluorouracil plus leucovorin (5-FU + LEU) and FOLFOX (5-FU + LEU plus oxaliplatin [OX]), but not after OX alone or gemcitabine. Cell and tumor growth assays with the irreversible inhibitor JNK-IN-8 and genetic manipulations demonstrate that JNK and JUN each contribute to chemoresistance and cancer cell survival after FOLFOX. Active JNK1 and JUN are specifically implicated in these effects, and synergy with JNK-IN-8 is linked to FOLFOX-mediated JUN activation, cell cycle dysregulation, and DNA damage response. This study highlights the potential for JNK-IN-8 as a biological tool and potential combination therapy with FOLFOX in PDAC and reinforces the need to tailor treatment to functional characteristics of individual tumors.

Abstract A27: Deep learning for analysis of tumor-lymphocyte interactions in pancreatic ductal adenocarcinoma

The interaction of tumor, stroma, and immune cells in pancreatic ductal adenocarcinoma (PDAC) is complex and difficult to quantify in patient samples. Recently, deep learning algorithms have shown successes in identifying tumor and lymphocytes regions on whole-slide images derived from routinely collected histopathologic specimens. The Cancer Genome Atlas (TCGA) in particular has generated whole-slide images as well as paired molecular data, thus allowing for combined spatial and molecular analyses of tumor-lymphocyte interactions. We have previously highlighted this resource by computationally mapping tumor-infiltrating lymphocytes (TILs) on digital images across 13 tumor types. To achieve this, convolutional neural networks were trained on lymphocyte images annotated by expert pathologists and then used to detect spatial TIL patterns. This led to identification of four qualitative TIL pattern categories, which varied depending on tumor type as well as molecular immune subtype, demonstrating the potential of these spatial structures to provide further insights into tumor microenvironments and their relationship to overall survival. We have now extended this deep learning pipeline to include identification of tumor regions in PDAC, allowing study of TIL patterns in the context of their relative spatial localization to tumors. Using the deep learning algorithm to define the tumor region, we applied erosion and dilation operations to further capture the peritumoral region, the outer and inner regions of the tumor, as well as desmoplasia far from the tumor cells. We thus defined lymphocytes by their spatial localization as being internal, tumoral, peritumoral, or outer with these masks. We then used nearest-neighbor and density-based approaches to quantify TIL infiltration patterns with respect to tumor. These features vary significantly across the previously identified TIL patterns and may serve as additional parameters to define the microenvironment conditions in patient samples. Here we demonstrate that features extracted using our pipeline recapitulate canonical histologic properties. Using immune cell abundance estimates from gene expression generated by CIBERSORT, we find that samples with tumor TIL densities above median have more M1 macrophages, while those below median have more M2 macrophages. We also observe that slides with a higher peritumoral TIL density relative to tumoral TIL density have higher Treg fractions. Ongoing work on improving the resolution and cell specificity of our pipeline will allow us to ask more specific questions and permit higher granularity in linking clinical outcomes to spatial immune phenotypes.

Citation Format: Soma Kobayashi, Han Le, Lucie Chrastecka, Rajarsi Gupta, Le Hou, Shahira Abousamra, Danielle Fassler, Kenneth R. Shroyer, Dimitris Samaras, Tahsin Kurc, Richard A. Moffitt, Joel H. Saltz. Deep learning for analysis of tumor-lymphocyte interactions in pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Advances in Science and Clinical Care; 2019 Sept 6-9; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2019;79(24 Suppl):Abstract nr A27.

Abstract B41: Compartment deconvolution in pancreatic cancer with biologic and clinical implications

Pancreatic ductal adenocarcinoma (PDAC) is characterized by relatively low tumor purity and an abundant tumor microenvironment. To dissect the contribution of the biologic components, we developed DECODER, which performs de novo compartment deconvolution and weight estimation of tumor samples. DECODER is a sophisticated framework that integrates runs of non-negative matrix factorization (NMF) and non-negative least square (NNLS) algorithms and can be applied to any non-negative matrices without the need to know the number of resultant factors or compartments. DECODER was used to deconvolve the TCGA pancreatic adenocarcinoma (PAAD) RNA-seq dataset, which resulted in the identification of 7 major compartments (basal tumor, classical tumor, activated stroma, normal stroma, immune, endocrine, and exocrine), confirming prior manual NMF-based solutions. These results were then used for single-sample based weight estimation in the COMPASS trial and ICGC PACA-AU RNA-seq dataset. We saw a significant positive correlation between DECODER immune weight and leukocyte fraction (r = 0.757, p &lt; 0.001) or ESTIMATE immune score (r = 0.773, p &lt; 0.001). Samples with high immune weights corresponded to immune infiltration by histology. A significant correlation was found between the sum of basal and classical tumor weights, and tumor purity based on both ABSOLUTE (r = 0.699, p &lt; 0.001) and methylation (r = 0.71, p &lt; 0.001). Similarly, the sum of activated and normal stroma weights correlated with ESTIMATE stromal score (r = 0.729, p &lt; 0.001). Interestingly, we found that the ratio between the basal and classical compartment (bcRatio) was significantly associated with survival outcome (p = 0.049 in TCGA and 0.008 in ICGC) in all patients and treatment response in basal-like patients (r = 0.884, p &lt; 0.001 in COMPASS trial), suggesting that bcRatio may help explain the molecular basis for tumor behavior in PDAC. DECODER was also applied for de novo deconvolution for all the cancer types in TCGA RNA-seq dataset and identified the cancer type specific compartments. Results from DECODER can then be used for single-sample weight estimation of new samples for any cancer type. In addition, we applied DECODER on the PanCan ATAC-seq dataset containing 23 cancer types in a combined fashion, and identified compartments associated with cancer types or organ systems. This proves that DECODER is highly feasible to data of multiple platforms. In summary, we present an automated method for de novo deconvolution that may be used across tumor and data types. With deconvolved results as the reference, DECODER enables the single-sample weight estimation for a new sample, which is plausible in the clinical setting.

Citation Format: Xianlu L. Peng, Richard A. Moffitt, Robert J. Torphy, Keith E. Volmar, Jen Jen Yeh. Compartment deconvolution in pancreatic cancer with biologic and clinical implications [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Advances in Science and Clinical Care; 2019 Sept 6-9; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2019;79(24 Suppl):Abstract nr B41.

Abstract C39: A novel rewired pathway of nucleotide metabolism drives chemoresistance in pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) is characterized by two molecular subtypes, of which the basal-like subtype is associated with the worst survival and is highly resistant to the currently available first-line chemotherapy. Our laboratory has identified that keratin 17 (K17) is a novel negative prognostic biomarker, as accurate as molecular subtyping in predicting patient survival. Patient-derived data analysis suggests that K17 expression correlates with increased resistance to chemotherapeutic agents. The goal of this study is to determine the role of K17 in chemoresistance, and to identify novel therapeutic approaches for around 50% of PDAC patients with tumors that express high levels of K17. In multiple in vivo and in vitro models of PDAC, spanning human and murine PDAC cells, patient-derived organoids, and orthotopic xenograft models, we determined that K17 expression causes more than two-fold increase in resistance to gemcitabine (Gem) and 5-fluorouracil (5-FU), key components of the current standard-of-care chemotherapeutic regimens. To uncover the mechanism associated to K17-induced chemoresistance, we performed unbiased metabolomic studies in isogenic PDAC cell lines and found that K17 reprograms several key metabolic pathways. In particular, K17 increases pyrimidine biosynthesis, a pathway has been linked to chemoresistance. Rescue experiments showed that deoxycytidine (dC) was sufficient to promote Gem (dC analogue) resistance in K17-nonexpressing PDAC cells, suggesting that upregulation of pyrimidine synthesis by K17 underlies resistance to chemotherapeutic agents. Through unbiased RNA-sequencing studies, we identified that gene expression of enzymes involved in pyrimidine biosynthesis was increased specifically in high K17-expressing cells. Previous reports from our group and others suggest that nuclear K17 regulates cell-cycle progression and gene expression. Through domain-prediction analyses, we discovered a novel domain on K17 involved in transcriptional regulation that is required for metabolic reprogramming. Currently, we are testing the role of this domain in metabolic reprograming. In addition, are pursuing two approaches to determine the “druggability” of these findings. First, we are testing if interrupting K17-mediated nucleotide metabolism, by means of small-molecule inhibitors, resensitizes tumor cells to pyrimidine analogues. Second, we are validating the results of a large-scale small-molecule inhibitor screen of FDA-approved, pharma-developed tools to identify compounds that target DNA metabolism and transcription in K17-expressing PDAC cells. In summary, we identified a novel and potentially druggable pathway of chemoresistance that could ultimately result in developing novel therapeutic strategies to enhance patient survival.

Citation Format: Chun-Hao Pan, Cindy V. Leiton, Lucia Roa-Peña, Ryan R. Kawalerski, Richard A. Moffitt, Jiang Zhao, Timothy Spicer, Peter Bailey, David K. Chang, Andrew Biankin, Tim Duong, Pankaj K. Singh, Kenneth R. Shroyer, Luisa F. Escobar-Hoyos. A novel rewired pathway of nucleotide metabolism drives chemoresistance in pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Advances in Science and Clinical Care; 2019 Sept 6-9; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2019;79(24 Suppl):Abstract nr C39.

Abstract A38: Defining heterogeneity of molecular subtypes in human PDAC with scRNA-Seq

Introduction: Due to the widespread stromal involvement and sparse neoplastic populations in pancreatic ductal adenocarcinoma (PDAC) tumor biopsies, the discovery of novel biomarkers, subtypes, and therapeutic targets has been challenging. Previous work by our lab in the study of bulk tumors has established prognostic gene signatures of basal-like and classical tumor subtypes, and also normal and activated stromal subtypes. Leveraging the advances of microfluidics and single-cell RNA sequencing to digitally separate components of the tumor microenvironment (TME), we built a comprehensive single-cell atlas of human PDAC to better characterize the composition, pathology, and interactome of PDAC cells.

Materials and Methods: Freshly resected human PDAC tissue is acquired immediately following resection. Bulk tissue is rapidly dissociated into single-cell suspension in order to reduce transcriptional artifacts of handling and time. Without manual phenotype preselection for specific cell types, suspension is processed and sequenced through 10X Chromium scRNA-seq and Illumina platforms. Seurat v3.0 and custom R packages were used for bioinformatic analysis.

Results: Our single-cell human PDAC datasets recapitulate the highly desmoplastic composition expected from a PDAC tissue biopsy. Fibroblasts were most abundant alongside populations of preneoplastic cells, and neoplastic epithelial cells of both basal and classical subtypes. Both tumor subtype signatures were expressed within the same patient’s biopsy, raising further questions about the dichotomy and origin of PDAC subtypes. We also analyzed transcriptional heterogeneity in endocrine, exocrine, myeloid, and immune cells to establish a comprehensive transcriptomic picture of human PDAC TME. Several cancer-associated fibroblasts (CAFs) subtypes have been described. However, we observe enrichment of these gene signatures only in a subset of our fibroblasts, suggesting a broader landscape of fibroblast activity and differentiation may exist. Lastly, ligand and receptor interactions were mapped between cell types to highlight potential paracrine and autocrine mechanisms that potentiate the tumor microenvironment growth and maintenance.

Conclusion: We present a comprehensive single-cell atlas of human PDAC with a focus on molecular subtypes in the TME at an unprecedented resolution. This offers novel insight into cell type heterogeneity, pathologic signaling, lineage trajectories, and specialized functions of TME components including macrophages, fibroblasts, and neoplastic cells leading to multi-axis pathways and interactions. Notably, we have found intrapatient heterogeneity in both the stromal and neoplastic tumor cell subtypes. Resolving the transcriptional heterogeneity of cells and the network of pathologic interactions within the TME allows for further revelations about cancer biology and potentially opens doors to novel combination therapies in the treatment of pancreatic cancer.

Citation Format: Ki Oh, Manisha Rao, Mei Gao, Aaron Sasson, Georgios Georgakis, Joseph Kim, R. Scott Powers, Richard A Moffitt. Defining heterogeneity of molecular subtypes in human PDAC with scRNA-Seq [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Advances in Science and Clinical Care; 2019 Sept 6-9; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2019;79(24 Suppl):Abstract nr A38.

Purity Independent Subtyping of Tumors (PurIST), A Clinically Robust, Single-sample Classifier for Tumor Subtyping in Pancreatic Cancer

PURPOSE

Molecular subtyping for pancreatic cancer has made substantial progress in recent years, facilitating the optimization of existing therapeutic approaches to improve clinical outcomes in pancreatic cancer. With advances in treatment combinations and choices, it is becoming increasingly important to determine ways to place patients on the best therapies upfront. Although various molecular subtyping systems for pancreatic cancer have been proposed, consensus regarding proposed subtypes, as well as their relative clinical utility, remains largely unknown and presents a natural barrier to wider clinical adoption.

EXPERIMENTAL DESIGN

We assess three major subtype classification schemas in the context of results from two clinical trials and by meta-analysis of publicly available expression data to assess statistical criteria of subtype robustness and overall clinical relevance. We then developed a single-sample classifier (SSC) using penalized logistic regression based on the most robust and replicable schema.

RESULTS

We demonstrate that a tumor-intrinsic two-subtype schema is most robust, replicable, and clinically relevant. We developed Purity Independent Subtyping of Tumors (PurIST), a SSC with robust and highly replicable performance on a wide range of platforms and sample types. We show that PurIST subtypes have meaningful associations with patient prognosis and have significant implications for treatment response to FOLIFIRNOX.

CONCLUSIONS

The flexibility and utility of PurIST on low-input samples such as tumor biopsies allows it to be used at the time of diagnosis to facilitate the choice of effective therapies for patients with pancreatic ductal adenocarcinoma and should be considered in the context of future clinical trials.

Deep Learning on Electronic Health Records to Improve Disease Coding Accuracy

Characterization of a patient's clinical phenotype is central to biomedical informatics. ICD codes, assigned to inpatient encounters by coders, is important for population health and cohort discovery when clinical information is limited. While ICD codes are assigned to patients by professionals trained and certified in coding there is substantial variability in coding. We present a methodology that uses deep learning methods to model coder decision making and that predicts ICD codes. Our approach predicts codes based on demographics, lab results, and medications, as well as codes from previous encounters. We are able to predict existing codes with high accuracy for all three of the test cases we investigated: diabetes, acute renal failure, and chronic kidney disease. We employed a panel of clinicians, in a blinded manner, to assess ground truth and compared the predictions of coders, model and clinicians. When disparities between the model prediction and coder assigned codes were reviewed, our model outperformed coder assigned ICD codes.

Expression of neuropilin-1 is linked to glioma associated microglia and macrophages and correlates with unfavorable prognosis in high grade gliomas

High grade gliomas, including glioblastoma (GB), are devastating malignancies with very poor prognosis. Over the course of the last decade, there has been a failure to develop new treatments for GB. Reasons for this failure include the lack of validation of novel molecular targets, which are often characterized in animal models and directly transposed to human trials. Here we build on our previous findings, which describe how the multi-functional co-receptor Neuropilin-1 (NRP1) signals through glioma associated microglia/macrophages (GAMS) to promote murine glioma, and investigate NRP1 expression in human glioma. Clinical and gene expression data were obtained via The Cancer Genome Atlas (TCGA), and analyzed using R statistical software. Additionally, CIBERSORT in silico deconvolution was used to determine fractions of immune cell sub-populations within the gene expression datasets. We find that NRP1 expression is correlated with poor prognosis, glioma grade, and associates with the mesenchymal GB subtype. In human GB, NRP1 expression is highly correlated with markers of monocytes/macrophages, as well as genes that contribute to the pro-tumorigenic phenotype of these cells.

Direct therapeutic targeting of immune checkpoint PD-1 in pancreatic cancer

BACKGROUND

Pancreatic cancer (PC) hijacks innate cellular processes to promote cancer growth. We hypothesized that PC exploits PD-1/PD-L1 not only to avoid immune responses, but to directly enhance growth. We also hypothesized that immune checkpoint inhibitors (ICIs) have direct cytotoxicity in PC. We sought to elucidate therapeutic targeting of PD-1/PD-L1.

METHODS

PD-1 was assessed in PC cells, patient-derived organoids (PDOs), and clinical tissues. Then, PC cells were exposed to PD-L1 to evaluate proliferation. To test PD-1/PD-L1 signaling, cells were exposed to PD-L1 and MAPK was examined. Radio-immunoconjugates with anti-PD-1 drugs were developed to test uptake in patient-derived tumor xenografts (PDTXs). Next, PD-1 function was assessed by xenografting PD-1-knockdown cells. Finally, PC models were exposed to ICIs.

RESULTS

PD-1 expression was demonstrated in PCs. PD-L1 exposure increased proliferation and activated MAPK. Imaging PDTXs revealed uptake of radio-immunoconjugates. PD-1 knockdown in vivo revealed 67% smaller volumes than controls. Finally, ICI treatment of both PDOs/PDTXs demonstrated cytotoxicity and anti-MEK1/2 combined with anti-PD-1 drugs produced highest cytotoxicity in PDOs/PDTXs.

CONCLUSIONS

Our data reveal PCs innately express PD-1 and activate druggable oncogenic pathways supporting PDAC growth. Strategies directly targeting PC with novel ICI regimens may work with adaptive immune responses for optimal cytotoxicity.

Real-time Genomic Characterization of Advanced Pancreatic Cancer to Enable Precision Medicine

Clinically relevant subtypes exist for pancreatic ductal adenocarcinoma (PDAC), but molecular characterization is not yet standard in clinical care. We implemented a biopsy protocol to perform time-sensitive whole-exome sequencing and RNA sequencing for patients with advanced PDAC. Therapeutically relevant genomic alterations were identified in 48% (34/71) and pathogenic/likely pathogenic germline alterations in 18% (13/71) of patients. Overall, 30% (21/71) of enrolled patients experienced a change in clinical management as a result of genomic data. Twenty-six patients had germline and/or somatic alterations in DNA-damage repair genes, and 5 additional patients had mutational signatures of homologous recombination deficiency but no identified causal genomic alteration. Two patients had oncogenic in-frame BRAF deletions, and we report the first clinical evidence that this alteration confers sensitivity to MAPK pathway inhibition. Moreover, we identified tumor/stroma gene expression signatures with clinical relevance. Collectively, these data demonstrate the feasibility and value of real-time genomic characterization of advanced PDAC.Significance: Molecular analyses of metastatic PDAC tumors are challenging due to the heterogeneous cellular composition of biopsy specimens and rapid progression of the disease. Using an integrated multidisciplinary biopsy program, we demonstrate that real-time genomic characterization of advanced PDAC can identify clinically relevant alterations that inform management of this difficult disease. Cancer Discov; 8(9); 1096-111. ©2018 AACR.See related commentary by Collisson, p. 1062This article is highlighted in the In This Issue feature, p. 1047.

Genomics-Driven Precision Medicine for Advanced Pancreatic Cancer: Early Results from the COMPASS Trial

Purpose: To perform real-time whole genome sequencing (WGS) and RNA sequencing (RNASeq) of advanced pancreatic ductal adenocarcinoma (PDAC) to identify predictive mutational and transcriptional features for better treatment selection.Experimental Design: Patients with advanced PDAC were prospectively recruited prior to first-line combination chemotherapy. Fresh tumor tissue was acquired by image-guided percutaneous core biopsy for WGS and RNASeq. Laser capture microdissection was performed for all cases. Primary endpoint was feasibility to report WGS results prior to first disease assessment CT scan at 8 weeks. The main secondary endpoint was discovery of patient subsets with predictive mutational and transcriptional signatures.Results: Sixty-three patients underwent a tumor biopsy between December 2015 and June 2017. WGS and RNASeq were successful in 62 (98%) and 60 (95%), respectively. Genomic results were reported at a median of 35 days (range, 19-52 days) from biopsy, meeting the primary feasibility endpoint. Objective responses to first-line chemotherapy were significantly better in patients with the classical PDAC RNA subtype compared with those with the basal-like subtype (P = 0.004). The best progression-free survival was observed in those with classical subtype treated with m-FOLFIRINOX. GATA6 expression in tumor measured by RNA in situ hybridization was found to be a robust surrogate biomarker for differentiating classical and basal-like PDAC subtypes. Potentially actionable genetic alterations were found in 30% of patients.Conclusions: Prospective genomic profiling of advanced PDAC is feasible, and our early data indicate that chemotherapy response differs among patients with different genomic/transcriptomic subtypes. Clin Cancer Res; 24(6); 1344-54. ©2017 AACR.

MYC activation cooperates with Vhl and Ink4a/Arf loss to induce clear cell renal cell carcinoma

Renal carcinoma is a common and aggressive malignancy whose histopathogenesis is incompletely understood and that is largely resistant to cytotoxic chemotherapy. We present two mouse models of kidney cancer that recapitulate the genomic alterations found in human papillary (pRCC) and clear cell RCC (ccRCC), the most common RCC subtypes. MYC activation results in highly penetrant pRCC tumours (MYC), while MYC activation, when combined with Vhl and Cdkn2a (Ink4a/Arf) deletion (VIM), produce kidney tumours that approximate human ccRCC. RNAseq of the mouse tumours demonstrate that MYC tumours resemble Type 2 pRCC, which are known to harbour MYC activation. Furthermore, VIM tumours more closely simulate human ccRCC. Based on their high penetrance, short latency, and histologic fidelity, these models of papillary and clear cell RCC should be significant contributions to the field of kidney cancer research.

Renal cell carcinoma (RCC) is a common and aggressive malignancy. Here, the authors generate two mouse models of the most common RCC subtypes: the human papillary RCC through MYC activation and clear cell RCC through MYC activation combined with Vhl and Cdkn2a deletion.

Abstract B21: Integrated genomic characterization of pancreatic ductal adenocarcinoma and the confounding role of purity

We have utilized integrated genomic, transcriptomic and proteomic profiling to perform comprehensive molecular characterization of 150 pancreatic ductal adenocarcinoma (PAAD) specimens. PAAD demonstrates a characteristic biology notable for low neoplastic cellularity and prominent fibrotic stroma. Prior genome sequencing studies have focused on tumors with high neoplastic cellularity or have used physical techniques to enrich for tumor cell purity. Consequently, tumors with low neoplastic cellularity have been significantly underrepresented in existing genome sequencing efforts, even though they represent the majority of surgically resected PAAD samples. In a collaborative effort through The Cancer Genome Atlas (TCGA) network, we comprehensively analyzed 150 clinically annotated pancreatic cancers with a minimum of 1% neoplastic cellularity. We evaluated somatic DNA mutation and copy number alterations, mRNA and miRNA sequencing, and promoter methylation in all samples as well as proteomic profiles in a subset of tumors.

We have overcome technical challenges of low neoplastic cellularity in primary tumors through deep whole exome and targeted gene panel sequencing to identify high-confidence somatic mutations in all samples, including 20 samples with less than 15% cellularity as assessed by ABSOLUTE. We observe recurrent somatic mutations in genes previously reported to be common in PAAD, including KRAS, TP53, SMAD4, and CDKN2A. We have observed copy number alterations in loci harboring the MYC, AKT2, and KRAS oncogenes and the CDKN2A tumor suppressor gene, among others. Moreover, we demonstrate two distinct classes of KRAS wild-type tumors based on the presence or absence of a RAS pathway genomic alteration.

While low neoplastic cellularity was addressed through increased coverage with exome sequencing, other platforms were heavily influenced by purity, including mRNA, miRNA, RPPA and methylation. We find that, in our data, previously defined expression subtypes based on NMF-consensus clustering are highly confounded by purity, whereas recently published tumor-specific subtypes are purity-independent. Furthermore, we find that naïve approaches to defining molecular subtypes across each of our platforms resulted in defining groups of samples which were confounded with purity. Therefore, we divided our dataset by the median ABSOLUTE tumor purity, creating both a high- and low-purity classifications, and employed a two-stage approach to classifying the data. We first performed unsupervised clustering on higher-purity tumors and derived molecular classifiers that are more likely to reflect tumor cell biology, rather than that of the admixed stromal, immune and normal pancreas cells. We then projected these classifiers onto the larger dataset including both the high- and low-purity samples. We found that by pre-defining these tumor-specific discriminatory features that we were able to help mitigate the tendency of low purity samples to form their own groups, and instead define subtypes of disease which were not statistically significantly associated with tumor purity. Thus, we have used quantitative purity information to create and interpret methylation, copy number, mRNA and miRNA expression signatures as well as reverse phase protein array-based proteomic profiles which also have prognostic significance in an multivariate model. Finally, we observe multiple instances of predicted and observed methylation and miRNA regulation of RNA which help describe subtypes of disease. In summary, we have used our data, including miRNA and protein expression, to expand and help understand previous genomic and transcriptomic descriptions of PAAD. These data describe the molecular landscape of pancreatic cancer and may serve as a roadmap for future study of this disease.

Citation Format: Richard A. Moffitt, on behalf of the Cancer Genome Atlas Research Network, The Cancer Genome Atlas Research Network.{Authors}. Integrated genomic characterization of pancreatic ductal adenocarcinoma and the confounding role of purity. [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Advances in Science and Clinical Care; 2016 May 12-15; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2016;76(24 Suppl):Abstract nr B21.

Comprehensive Pan-Genomic Characterization of Adrenocortical Carcinoma

We describe a comprehensive genomic characterization of adrenocortical carcinoma (ACC). Using this dataset, we expand the catalogue of known ACC driver genes to include PRKAR1A, RPL22, TERF2, CCNE1, and NF1. Genome wide DNA copy-number analysis revealed frequent occurrence of massive DNA loss followed by whole-genome doubling (WGD), which was associated with aggressive clinical course, suggesting WGD is a hallmark of disease progression. Corroborating this hypothesis were increased TERT expression, decreased telomere length, and activation of cell-cycle programs. Integrated subtype analysis identified three ACC subtypes with distinct clinical outcome and molecular alterations which could be captured by a 68-CpG probe DNA-methylation signature, proposing a strategy for clinical stratification of patients based on molecular markers.

Local iontophoretic administration of cytotoxic therapies to solid tumors

Parenteral and oral routes have been the traditional methods of administering cytotoxic agents to cancer patients. Unfortunately, the maximum potential effect of these cytotoxic agents has been limited because of systemic toxicity and poor tumor perfusion. In an attempt to improve the efficacy of cytotoxic agents while mitigating their side effects, we have developed modalities for the localized iontophoretic delivery of cytotoxic agents. These iontophoretic devices were designed to be implanted proximal to the tumor with external control of power and drug flow. Three distinct orthotopic mouse models of cancer and a canine model were evaluated for device efficacy and toxicity. Orthotopic patient-derived pancreatic cancer xenografts treated biweekly with gemcitabine via the device for 7 weeks experienced a mean log2 fold change in tumor volume of -0.8 compared to a mean log2 fold change in tumor volume of 1.1 for intravenous (IV) gemcitabine, 3.0 for IV saline, and 2.6 for device saline groups. The weekly coadministration of systemic cisplatin therapy and transdermal device cisplatin therapy significantly increased tumor growth inhibition and doubled the survival in two aggressive orthotopic models of breast cancer. The addition of radiotherapy to this treatment further extended survival. Device delivery of gemcitabine in dogs resulted in more than 7-fold difference in local drug concentrations and 25-fold lower systemic drug levels than the IV treatment. Overall, these devices have potential paradigm shifting implications for the treatment of pancreatic, breast, and other solid tumors.

Abstract 2976: Comprehensive Pan-Genomic characterization of adrenocortical carcinoma

Adrenocortical carcinoma (ACC) is a rare neoplasm with a heterogeneous outcome and limited treatment options. To understand its molecular and genomic landscape as a part of The Cancer Genome Atlas (TCGA) project, we performed the genomic, transcriptomic, epigenomic and proteomic profiling of 91 ACCs.

We identified potential driving alterations including amplifications (TERT, TERF2 and CDK4), deletions (ZNRF3, CDKN2A and RB1) and point mutations in genes unknown to participate in adrenal disease (RPL22) and genes known to initiate familial syndromes that occasionally include adrenocortical neoplasms (TP53, CTNNB1, PRKAR1A, MEN1). The finding of PRKAR1A expands the catalogue of pathogenic pathways underlying ACC, suggesting of the protein kinase alpha signaling pathway as a potential target for molecular interventions. Novel transcript fusions potentially leading to overactive kinases included EXOSC10-MTOR and PPP1CB-BRE.

DNA copy number analysis unveiled prevalent DNA losses leading to hypodiploidy as well as whole genome doubling (WGD) in 51% of ACC. The similar penetrance of loss of heterozygosity before and after WGD suggests a sequential development from hypodiploidy to polyploidy along the doubling in a subset of ACCs, which was endorsed by the worse outcome for WGD samples relative to nonWGD ACCs. An association between TERT expression and WGD was observed, suggesting a role for telomere regulation. These findings present ACC as a model disease for studies of WGD which is a frequent event in many tumor types.

Unsupervised clustering of DNA methylation, copy number, gene expression, miRNA expression and protein abundance converged into three classes with specific biological characteristics and a respective median event free survival of 8, 38 and &gt;100 months (p-value 1.7e-13). Comparison of the subtypes suggested additional drivers such as protein kinase C (PKC) phosphorylation and upregulation of a miRNA cluster at chromosome Xq27.3, which complemented the genomic alterations identified in these subtypes.

To gain more insights into this rare cancer type, we placed ACC in a broader context of cancer genomic profiles including an array of other cancer types. These analyses revealed interesting shared features, including beta-catenin activation with a subset of endometroid cancer, DNA mismatch repair gene mutational signature with gastrointestinal cancers and a smoking signature with lung cancer. These findings highlight the commonalities between ACC and other lineages of cancer.

Taken together, we found Wnt signaling pathway and p53/Rb signaling pathway were the most frequently altered pathways in ACC. Meanwhile, new players surfaced from our analyses including the PKA and PKC pathways. Our results present a comprehensive genomic landscape and refined molecular classification of ACC improve our understanding of its pathogenesis, and will ultimately improve the care of patients.

Citation Format: Siyuan Zheng, Andrew D. Cherniack, Ninad Dewal, Richard A. Moffitt, Ludmila Danilova, Bradley A. Murray, Antonio M. Lerario, Tobias Else, Theo A. Knijnenburg, Giovanni Ciriello, Seungchan Kim, Guillaume Assie, Olena Morozova, Rehan Akbani, Juliann Shih, Katherine A. Hoadley, Toni K. Choueiri, Jens Waldmann, Ozgur Mete, Gordon A. Robertson, Matthew Meyerson, Michael J. Demeure, Felix Beuschlein, Anthony Gill, Ana C. Latronico, Maria C. Fragosa, Leslie Cope, Electron Kebebew, Mouhammed A. Habra, Timothy G. Whitsett, Kimberly J. Bussey, William E. Rainey, Sylvia Asa, Jérôme Bertherat, Martin Fassnacht, David A. Wheeler, The Cancer Genome Atlas Research Network, Gary D. Hammer, Thomas J. Giordano, Roel Verhaak. Comprehensive Pan-Genomic characterization of adrenocortical carcinoma. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2976. doi:10.1158/1538-7445.AM2015-2976

Abstract A82: Virtual microdissection reveals tumor specific heterogeneity in pancreatic cancer

Using matrix factorization, we have removed confounding normal tissue gene expression from profiles of primary and metastatic tumors, facilitating the study of underlying tumor biology without the need for laser capture microdissection.

Understanding molecular mechanisms of metastasis in pancreatic cancer has the potential to yield rationally designed therapies. Previous work has used gene expression arrays and next generation sequencing to look for potential targets, focusing on identifying differences between metastatic lesions and primary tumors. However, such analyses are hampered by the low cellularity of malignant epithelium in patient samples. Since profiles of primary and metastatic tumors often contain contributions from adjacent normal tissues from the tumor site, comparison of these samples are confounded by the differences between normal pancreas and normal distant organs. We seek to avoid this pitfall by leveraging the blind source separation technique, nonnegative matrix factorization (NMF)[1], to the analysis of gene expression data.

Microarray data were obtained from matched primary (n=8) and metastatic tumors (n=31) as well as adjacent normal tissues (n=85) for both local and distant sites of patients who died of metastatic pancreatic cancer[2]. Pathological analysis showed that tumor cellularity varied with a mean of 48% and a standard deviation of 31% across all sites. We applied NMF to our cohort, thus computationally microdissecting tumors into the source tissues composing our samples. 200 iterations of 5-fold resampling were performed to achieve stable NMF results. Genes with expression ranked in the top 50 of any factor together were recorded in a consensus matrix. This consensus matrix was then used as the basis of a hierarchical clustering as to yield k gene clusters. These k gene-clusters were used to seed a supervised NMF using the MATLAB NMF multiplicative update solver until completion.

Using an unbiased approach, we identified distinct molecular signatures associated with adjacent normal and tumor tissue in primary and metastatic pancreatic cancer samples. Specifically, we identified confounding liver, lung, lymph, and muscle tissue gene expression from our metastatic samples as well as both endocrine and exocrine pancreas gene expression from our primary tumors. Furthermore, we estimated the relative composition of each of our samples in terms of a weighted sum of these tissue-specific signatures. Both primary tumors and metastases appeared to be admixtures of pancreatic tumor epithelial expression and adjacent normal tissue. For example, in liver tissues histological assessment of tumor content correlated well with NMF predictions of tumor content, (n=39, R2=0.72, p&lt;0.01). Prior to NMF, unsupervised clustering of all data caused samples to cluster by site of harvest. After digital subtraction of confounding tissue factors identified by NMF, samples grouped by patient of origin, suggesting intrinsic similarity among tumor sites within a patient.

We have demonstrated a method to virtually microdissect tissues, thereby identifying tumor-specific gene expression data in pancreatic cancer. By applying a fresh computational approach to a large cohort of data, we can generate new insight into the complex nature of low-cellularity tumors such as pancreatic cancer, and facilitate the study of inter- and intra-patient heterogeneity. This approach may be further leveraged to study the role of tumor and stroma signatures in pancreatic cancer.

1. Brunet, J.P., et al., Metagenes and molecular pattern discovery using matrix factorization. Proceedings of the National Academy of Sciences, 2004. 101(12): p. 4164-4169.

2. Stratford, J.K., et al., A six-gene signature predicts survival of patients with localized pancreatic ductal adenocarcinoma. PLoS medicine, 2010. 7(7): p. e1000307.

Citation Format: Richard A. Moffitt, Keith A. Volmar, Judy M. Anderson, Michael A. Hollingsworth, Jen Jen Yeh. Virtual microdissection reveals tumor specific heterogeneity in pancreatic cancer. [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Innovations in Research and Treatment; May 18-21, 2014; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2015;75(13 Suppl):Abstract nr A82.

Circulating tumor cells as a biomarker of response to treatment in patient-derived xenograft mouse models of pancreatic adenocarcinoma

Circulating tumor cells (CTCs) are cells shed from solid tumors into circulation and have been shown to be prognostic in the setting of metastatic disease. These cells are obtained through a routine blood draw and may serve as an easily accessible marker for monitoring treatment effectiveness. Because of the rapid progression of pancreatic ductal adenocarcinoma (PDAC), early insight into treatment effectiveness may allow for necessary and timely changes in treatment regimens. The objective of this study was to evaluate CTC burden as a biomarker of response to treatment with a oral phosphatidylinositol-3-kinase inhibitor, BKM120, in patient-derived xenograft (PDX) mouse models of PDAC. PDX mice were randomized to receive vehicle or BKM120 treatment for 28 days and CTCs were enumerated from whole blood before and after treatment using a microfluidic chip that selected for EpCAM (epithelial cell adhesion molecule) positive cells. This microfluidic device allowed for the release of captured CTCs and enumeration of these cells via their electrical impedance signatures. Median CTC counts significantly decreased in the BKM120 group from pre- to post-treatment (26.61 to 2.21 CTCs/250 µL, p = 0.0207) while no significant change was observed in the vehicle group (23.26 to 11.89 CTCs/250 µL, p = 0.8081). This reduction in CTC burden in the treatment group correlated with tumor growth inhibition indicating CTC burden is a promising biomarker of response to treatment in preclinical models. Mutant enriched sequencing of isolated CTCs confirmed that they harbored KRAS G12V mutations, identical to the matched tumors. In the long-term, PDX mice are a useful preclinical model for furthering our understanding of CTCs. Clinically, mutational analysis of CTCs and serial monitoring of CTC burden may be used as a minimally invasive approach to predict and monitor treatment response to guide therapeutic regimens.

Feasibility analysis of high resolution tissue image registration using 3-D synthetic data

BACKGROUND

Registration of high-resolution tissue images is a critical step in the 3D analysis of protein expression. Because the distance between images (~4-5μm thickness of a tissue section) is nearly the size of the objects of interest (~10-20μm cancer cell nucleus), a given object is often not present in both of two adjacent images. Without consistent correspondence of objects between images, registration becomes a difficult task. This work assesses the feasibility of current registration techniques for such images.

METHODS

We generated high resolution synthetic 3-D image data sets emulating the constraints in real data. We applied multiple registration methods to the synthetic image data sets and assessed the registration performance of three techniques (i.e., mutual information (MI), kernel density estimate (KDE) method [1], and principal component analysis (PCA)) at various slice thicknesses (with increments of 1μm) in order to quantify the limitations of each method.

RESULTS

Our analysis shows that PCA, when combined with the KDE method based on nuclei centers, aligns images corresponding to 5μm thick sections with acceptable accuracy. We also note that registration error increases rapidly with increasing distance between images, and that the choice of feature points which are conserved between slices improves performance.

CONCLUSIONS

We used simulation to help select appropriate features and methods for image registration by estimating best-case-scenario errors for given data constraints in histological images. The results of this study suggest that much of the difficulty of stained tissue registration can be reduced to the problem of accurately identifying feature points, such as the center of nuclei.

Feasibility of multiplex quantum dot stain using primary antibodies from four distinct host animals

We discuss the feasibility of multiplex QD stain for four biomarkers and our progress in finding four suitable biomarkers from four different hosts. There is a demand for using more than three fluorescent probes on a single tissue sample for disease detection to offer a more reliable prediction of disease progression. We developed a protocol for targeting four biomarkers using four primary antibodies from four different animal hosts. We performed primary-secondary antibody binding assays on nitrocellulose paper and stained breast cancer microarray slides with known expression of ER, PR, and HER2. We identified the lack of a standard protocol and the limited supply of primary antibodies from hosts other than rabbit and mouse in the market as key challenges. The results show variable success in both assays, but indicate future potential for this protocol with more development.

caCORRECT2: Improving the accuracy and reliability of microarray data in the presence of artifacts

BACKGROUND

In previous work, we reported the development of caCORRECT, a novel microarray quality control system built to identify and correct spatial artifacts commonly found on Affymetrix arrays. We have made recent improvements to caCORRECT, including the development of a model-based data-replacement strategy and integration with typical microarray workflows via caCORRECT's web portal and caBIG grid services. In this report, we demonstrate that caCORRECT improves the reproducibility and reliability of experimental results across several common Affymetrix microarray platforms. caCORRECT represents an advance over state-of-art quality control methods such as Harshlighting, and acts to improve gene expression calculation techniques such as PLIER, RMA and MAS5.0, because it incorporates spatial information into outlier detection as well as outlier information into probe normalization. The ability of caCORRECT to recover accurate gene expressions from low quality probe intensity data is assessed using a combination of real and synthetic artifacts with PCR follow-up confirmation and the affycomp spike in data. The caCORRECT tool can be accessed at the website: http://cacorrect.bme.gatech.edu.

RESULTS

We demonstrate that (1) caCORRECT's artifact-aware normalization avoids the undesirable global data warping that happens when any damaged chips are processed without caCORRECT; (2) When used upstream of RMA, PLIER, or MAS5.0, the data imputation of caCORRECT generally improves the accuracy of microarray gene expression in the presence of artifacts more than using Harshlighting or not using any quality control; (3) Biomarkers selected from artifactual microarray data which have undergone the quality control procedures of caCORRECT are more likely to be reliable, as shown by both spike in and PCR validation experiments. Finally, we present a case study of the use of caCORRECT to reliably identify biomarkers for renal cell carcinoma, yielding two diagnostic biomarkers with potential clinical utility, PRKAB1 and NNMT.

CONCLUSIONS

caCORRECT is shown to improve the accuracy of gene expression, and the reproducibility of experimental results in clinical application. This study suggests that caCORRECT will be useful to clean up possible artifacts in new as well as archived microarray data.

Adaptive control model reveals systematic feedback and key molecules in metabolic pathway regulation

Robust behavior in metabolic pathways resembles stabilized performance in systems under autonomous control. This suggests we can apply control theory to study existing regulation in these cellular networks. Here, we use model-reference adaptive control (MRAC) to investigate the dynamics of de novo sphingolipid synthesis regulation in a combined theoretical and experimental case study. The effects of serine palmitoyltransferase over-expression on this pathway are studied in vitro using human embryonic kidney cells. We report two key results from comparing numerical simulations with observed data. First, MRAC simulations of pathway dynamics are comparable to simulations from a standard model using mass action kinetics. The root-sum-square (RSS) between data and simulations in both cases differ by less than 5%. Second, MRAC simulations suggest systematic pathway regulation in terms of adaptive feedback from individual molecules. In response to increased metabolite levels available for de novo sphingolipid synthesis, feedback from molecules along the main artery of the pathway is regulated more frequently and with greater amplitude than from other molecules along the branches. These biological insights are consistent with current knowledge while being new that they may guide future research in sphingolipid biology. In summary, we report a novel approach to study regulation in cellular networks by applying control theory in the context of robust metabolic pathways. We do this to uncover potential insight into the dynamics of regulation and the reverse engineering of cellular networks for systems biology. This new modeling approach and the implementation routines designed for this case study may be extended to other systems. Supplementary Material is available at www.liebertonline.com/cmb .

WebPK, a web-based tool for custom pharmacokinetic simulation

Drug bioavailability is a major failing point of new pharmaceuticals i.e. drugs fail to reach their target or fail to stay there long enough for therapeutic effect. Compounding this issue, significant variability exists between patients and how they metabolize and distribute a drug. We present WebPK, a web-based tool for simulation of custom pharmacokinetic models. Model parameters can be entered manually or uploaded as a file. Simulation computations are performed on the server side, and thus require minimal client resources, which makes WebPK suitable for mobile devices. Time series biodistribution data are returned to the user in graphical and numerical form for quick interpretation or archiving. Results generated from WebPK are consistent with previously published pharmacokinetic models. This work is expected to provide physicians with access to easy simulation of patient pharmacokinetic profiles, which will allow for the prescription of more efficient and personalized drug regimens. URL: http://webpk.bme.gatech.edu.