Search bioRxiv⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/018747

Mass spectrometry-based quantitation of Her2 in gastroesophageal tumor tissue: Comparison to IHC and FISH

Daniel Catenacci·Wei-Li Liao·Lei Zhao·Emma Whitcomb·Les Henderson·peng Xu·Emily O'Day·Sheeno Thyparambil·Kathleen Bengali·Jamar Uzzell·Marlene Darflur·Fabiola Cecchi·Adele Blackler·David Krisman·Yung-Jue Bang·John Hart·Shu-Yuan Xiao·Sang Mee Lee·Jon Burrows·Todd Hembrough
View original source ↗Export metadata ↓

Abstract

BackgroundTrastuzumab showed survival benefit for Her2-positive gastroesophageal cancers (GEC). Immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) currently determine eligibility for trastuzumab-based therapy. However, both assays are low throughput with various limitations.\n\nMethodsWe developed a selected reaction monitoring mass spectrometric (SRM-MS) assay and quantified levels (amol/ug) of Her2-SRM in cell lines (n=27) and GEC tissues (n=139). We compared Her2-SRM expression with IHC/FISH, seeking to determine optimal SRM expression cut-offs to identify HER2 amplification.\n\nResultsAfter demonstrating assay development, precision, and stability, Her2-SRM measurement was observed to be highly concordant with HER2/CEP17 ratio, particularly in a multivariate regression model adjusted for SRM-expression of Met, Egfr, Her3, and HER2-heterogeneity covariates, and their interactions (cell lines r2=0.9842; FFPE r2=0.7643). In GEC tissues, Her2-SRM was detected in 71.2% of cases, and 12.3% were identified as HER2+. ROC curves demonstrated HER2-SRM levels to have high specificity (100%) at an upper-level cut-off of >750 amol/g and sensitivity (75%) at lower-level cut-off of <450 amol/ug. We observed an equivocal-zone between 450-750 amol/ug, analogous to IHC2+, but less frequent (9-16% of cases versus 36-41%).\n\nSignificanceCompared to IHC, SRM-MS provided more objective and quantitative Her2 expression with excellent HER2/CEP17 FISH correlation and fewer equivocal cases. Along with the multiplex capability for other relevant oncoproteins, these results demonstrated a refined HER2 expression assay for clinical application.

Source connections

Explore related subjects

PathologyPreprint

Keep this discovery

BibTeXRIS

Daniel Catenacci, Wei-Li Liao, Lei Zhao, Emma Whitcomb, Les Henderson, peng Xu, Emily O'Day, Sheeno Thyparambil, Kathleen Bengali, Jamar Uzzell, Marlene Darflur, Fabiola Cecchi, Adele Blackler, David Krisman, Yung-Jue Bang, John Hart, Shu-Yuan Xiao, Sang Mee Lee, Jon Burrows, Todd Hembrough. 2015-04-30. Mass spectrometry-based quantitation of Her2 in gastroesophageal tumor tissue: Comparison to IHC and FISH. https://doi.org/10.1101/018747

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2016

Using Machine Learning to Parse Breast Pathology Reports

1PurposeExtracting information from Electronic Medical Record is a time-consuming and expensive process when done manually. Rule-based and machine learning techniques are two approaches to solving this problem. In this study, we trained a machine learning model on pathology reports to extract pertinent tumor characteristics, which enabled us to create a large database of attribute searchable pathology reports. This database can be used to identify cohorts of patients with characteristics of interest.\n\nMethodsWe collected a total of 91,505 breast pathology reports from three Partners hospitals: Massachusetts General Hospital (MGH), Brigham and Womens Hospital (BWH), and Newton-Wellesley Hospital (NWH), covering the period from 1978 to 2016. We trained our system with annotations from two datasets, consisting of 6,295 and 10,841 manually annotated reports. The system extracts 20 separate categories of information, including atypia types and various tumor characteristics such as receptors. We also report a learning curve analysis to show how much annotation our model needs to perform reasonably.\n\nResultsThe model accuracy was tested on 500 reports that did not overlap with the training set. The model achieved accuracy of 90% for correctly parsing all carcinoma and atypia categories for a given patient. The average accuracy for individual categories was 97%. Using this classifier, we created a database of 91,505 parsed pathology reports.\n\nConclusionsOur learning curve analysis shows that the model can achieve reasonable results even when trained on a few annotations. We developed a user-friendly interface to the database that allows physicians to easily identify patients with target characteristics and export the matching cohort. This model has the potential to reduce the effort required for analyzing large amounts of data from medical records, and to minimize the cost and time required to glean scientific insight from this data.

Pathology↗
bioRxiv2016

Disrupted cholinergic modulation can underlie abnormal gamma rhythms in the schizophrenia and and auditory hallucination

The pathophysiology of auditory hallucination, a common symptom of schizophrenia, has yet been understood, but during auditory hallucination, primary auditory cortex (A1) shows paradoxical responses. When auditory stimuli are absent, A1 becomes hyperactive, while A1 responses to auditory stimuli are reduced. Such activation pattern of A1 responses during auditory hallucination is consistent with aberrant gamma rhythms in schizophrenia observed during auditory tasks, raising the possibility that the pathology underlying abnormal gamma rhythms can account for auditory hallucination. Moreover, A1 receives top-down signals in the gamma frequency band from an adjacent association area (Par2), and cholinergic modulation regulates interactions between A1 and Par2. In this study, we utilized a computational model of A1 to ask if disrupted cholinergic modulation could underlie abnormal gamma rhythms in schizophrenia. Furthermore, based on our simulation results, we propose potential pathology by which A1 can directly contribute to auditory hallucination.

Pathology↗
bioRxiv2016

H&amp;E-stained Whole Slide Deep Learning Predicts SPOP Mutation State in Prostate Cancer

A quantitative model to genetically interpret the histology in whole microscopy slide images is desirable to guide downstream immuno-histochemistry, genomics, and precision medicine. We constructed a statistical model that predicts whether or not SPOP is mutated in prostate cancer, given only the digital whole slide after standard hematoxylin and eosin [H&E] staining. Using a TCGA cohort of 177 prostate cancer patients where 20 had mutant SPOP, we trained multiple ensembles of residual networks, accurately distinguishing SPOP mutant from SPOP non-mutant patients (test AUROC=0.74, p=0.0007 Fishers Exact Test). We further validated our full metaensemble classifier on an independent test cohort from MSK-IMPACT of 152 patients where 19 had mutant SPOP. Mutants and non-mutants were accurately distinguished despite TCGA slides being frozen sections and MSK-IMPACT slides being formalin-fixed paraffin-embedded sections (AUROC=0.86, p=0.0038). Moreover, we scanned an additional 36 MSK-IMPACT patients having mutant SPOP, trained on this expanded MSK-IMPACT cohort (test AUROC=0.75, p=0.0002), tested on the TCGA cohort (AUROC=0.64, p=0.0306), and again accurately distinguished mutants from non-mutants using the same pipeline. Importantly, our method demonstrates tractable deep learning in this \"small data\" setting of 20-55 positive examples and quantifies each predictions uncertainty with confidence intervals. To our knowledge, this is the first statistical model to predict a genetic mutation in cancer directly from the patients digitized H&E-stained whole microscopy slide. Moreover, this is the first time quantitative features learned from patient genetics and histology have been used for content-based image retrieval, finding similar patients for a given patient where the histology appears to share the same genetic driver of disease i.e. SPOP mutation (p=0.0241 Kosts Method), and finding similar patients for a given patient that does not have have that driver mutation (p=0.0170 Kosts Method).\n\nSignificance StatementThis is the first pipeline predicting gene mutation probability in cancer from digitized H&E-stained microscopy slides. To predict whether or not the speckle-type POZ protein [SPOP] gene is mutated in prostate cancer, the pipeline (i) identifies diagnostically salient slide regions, (ii) identifies the salient region having the dominant tumor, and (iii) trains ensembles of binary classifiers that together predict a confidence interval of mutation probability. Through deep learning on small datasets, this enables automated histologic diagnoses based on probabilities of underlying molecular aberrations and finds histologically similar patients by learned genetic-histologic relationships.\n\nConception, Writing: AJS, TJF. Algorithms, Learning, CBIR: AJS. Analysis: AJS, MAR, TJF. Supervision: MAR, TJF.

Pathology↗