Construction and Validation of a Protein Prognostic Model for Lung Squamous Cell Carcinoma.

Lung squamous cell carcinoma (LUSCC), as the major type of lung cancer, has high morbidity and mortality rates. The prognostic markers for LUSCC are much fewer than lung adenocarcinoma. Besides, protein biomarkers have advantages of economy, accuracy and stability. The aim of this study was to construct a protein prognostic model for LUSCC. The protein expression data of LUSCC were downloaded from The Cancer Protein Atlas (TCPA) database. Clinical data of LUSCC patients were downloaded from The Cancer Genome Atlas (TCGA) database. A total of 237 proteins were identified from 325 cases of LUSCC patients based on the TCPA and TCGA database. According to Kaplan-Meier survival analysis, univariate and multivariate Cox analysis, a prognostic prediction model was established which was consisted of 6 proteins (CHK1_pS345, CHK2, IRS1, PAXILLIN, BRCA2 and BRAF_pS445). After calculating the risk values of each patient according to the coefficient of each protein in the risk model, the LUSCC patients were divided into high risk group and low risk group. The survival analysis demonstrated that there was significant difference between these two groups (p= 4.877e-05). The area under the curve (AUC) value of the receiver operating characteristic (ROC) curve was 0.699, which suggesting that the prognostic risk model could effectively predict the survival of LUSCC patients. Univariate and multivariate analysis indicated that this prognostic model could be used as independent prognosis factors for LUSCC patients. Proteins co-expression analysis showed that there were 21 proteins co-expressed with the proteins in the risk model. In conclusion, our study constructed a protein prognostic model, which could effectively predict the prognosis of LUSCC patients.

Multiclass cancer classification in fresh frozen and formalin-fixed paraffin-embedded tissue by DigiWest multiplex protein analysis.

Histomorphology and immunohistochemistry are the most common ways of cancer classification in routine cancer diagnostics, but often reach their limits in determining the organ origin in metastasis. These cancers of unknown primary, which are mostly adenocarcinomas or squamous cell carcinomas, therefore require more sophisticated methodologies of classification. Here, we report a multiplex protein profiling-based approach for the classification of fresh frozen and formalin-fixed paraffin-embedded (FFPE) cancer tissue samples using the digital western blot technique DigiWest. A DigiWest-compatible FFPE extraction protocol was developed, and a total of 634 antibodies were tested in an initial set of 16 FFPE samples covering tumors from different origins. Of the 303 detected antibodies, 102 yielded significant correlation of signals in 25 pairs of fresh frozen and FFPE primary tumor samples, including head and neck squamous cell carcinomas (HNSC), lung squamous cell carcinomas (LUSC), lung adenocarcinomas (LUAD), colorectal adenocarcinomas (COAD), and pancreatic adenocarcinomas (PAAD). For this signature of 102 analytes (covering 88 total proteins and 14 phosphoproteins), a support vector machine (SVM) algorithm was developed. This allowed for the classification of the tissue of origin for all five tumor types studied here with high overall accuracies in both fresh frozen (90.4%) and FFPE (77.6%) samples. In addition, the SVM classifier reached an overall accuracy of 88% in an independent validation cohort of 25 FFPE tumor samples. Our results indicate that DigiWest-based protein profiling represents a valuable method for cancer classification, yielding conclusive and decisive data not only from fresh frozen specimens but also FFPE samples, thus making this approach attractive for routine clinical applications.

Digital quantitative assessment of PD-L1 using digital spatial profiling.

The assessment of programmed death 1 ligand 1 (PD-L1) expression by Immunohistochemistry (IHC) is the US Food and Drug Administration (FDA)-approved predictive marker to select responders to checkpoint blockade anti-PD-1/PD-L1 axis immunotherapies. Different PD-L1 immunohistochemistry (IHC) assays use different antibodies and different scoring methods in tumor cells and immune cells. Multiple studies have compared the performance of these assays with variable results. Here, we investigate an alternative method for assessment of PD-L1 using a new technology known as digital spatial profiling. We use a previously described standardization tissue microarray (TMA) to assess the accuracy of the method and compare digital spatial profiler (DSP) to each FDA-approved PD-L1 assays, one LDT assay and three quantitative fluorescence assays. The standardized cell line Index tissue microarray contains 10 isogenic cells lines in triplicates expressing various ranges of PD-L1. The dynamic range of PD-L1 digital counts was measured in the ten cell lines on the Index TMA using the GeoMx DSP assay and read on the nCounter platform. The digital method shows very high correlation with immunohistochemistry scored with quantitative software and with quantitative fluorescence. High correlation of PD-L1 digital DSP counts were seen between rows on the same Index TMA. Finally, experiments from two Index TMAs showed reproducibility of DSP counts were independent of variable slide storage time over a three-week period after antibody labeling but before collection of cleaved tags. In summary, DSP appears to have quantitative potential comparable to quantitative immunohistochemistry. It is possible that this technology could be used as a PD-L1 protein measurement system for companion diagnostic testing for immune therapy.

Bayesian data integration and variable selection for pan-cancer survival prediction using protein expression data.

Accurate prognostic prediction using molecular information is a challenging area of research, which is essential to develop precision medicine. In this paper, we develop translational models to identify major actionable proteins that are associated with clinical outcomes, like the survival time of patients. There are considerable statistical and computational challenges due to the large dimension of the problems. Furthermore, data are available for different tumor types; hence data integration for various tumors is desirable. Having censored survival outcomes escalates one more level of complexity in the inferential procedure. We develop Bayesian hierarchical survival models, which accommodate all the challenges mentioned here. We use the hierarchical Bayesian accelerated failure time model for survival regression. Furthermore, we assume sparse horseshoe prior distribution for the regression coefficients to identify the major proteomic drivers. We borrow strength across tumor groups by introducing a correlation structure among the prior distributions. The proposed methods have been used to analyze data from the recently curated "The Cancer Proteome Atlas" (TCPA), which contains reverse-phase protein arrays-based high-quality protein expression data as well as detailed clinical annotation, including survival times. Our simulation and the TCPA data analysis illustrate the efficacy of the proposed integrative model, which links different tumors with the correlated prior structures.

Protein chip and bioinformatic analyses of differentially expressed proteins involved in the effect of hydrogen-rich water on myocardial ischemia-reperfusion injury.

Background: The differentially expressed proteins (DEPs) involved in the effect of hydrogen-rich water on myocardial ischemia reperfusion injury (MIRI) and their biological processes and signaling pathway were analyzed. Methods: 20 Wistar rats were randomly and equally divided into a control and a hydrogen-rich group. Hearts were removed and fixed in a Langendorff device. The control group was perfused with K-R solution, and the hydrogen-rich water group was perfused with K-R solution + hydrogen-rich water. Protein was extracted from the ventricular tissues, and GSR-CAA-67 was used to identify the DEPs between two groups. DEPs were analyzed through bioinformatic methods. Results: Compared with the control group, in the treatment group, the expression of 25 proteins was obviously decreased (P<0.05). For the DEPs, 359 biological processes, including the regulation of signaling pathways, immune reaction and formation of cardiovascular endothelial cells, were selected by GO enrichment analysis. Five signaling pathways were selected by KEGG pathway enrichment analysis. Conclusions: 25 proteins that are involved in hydrogen-water reducing MIRI were selected by high-throughput GSR-CAA-67. The biological processes and metabolic pathways involved in the DEPs were summarized, providing theoretical evidence for the clinical application of hydrogen-rich water.

RPPAware: A software suite to preprocess, analyze and visualize reverse phase protein array data.

Reverse Phase Protein Arrays (RPPA) is a high-throughput technology used to profile levels of protein expression. Handling the large datasets generated by RPPA can be facilitated by appropriate software tools. Here, we describe RPPAware, a free and intuitive software suite that was developed specifically for analysis and visualization of RPPA data. RPPAware is a portable tool that requires no installation and was built using Java. Many modules of the tool invoke R to utilize the statistical features. To demonstrate the utility of RPPAware, data generated from screening brain regions of a mouse model of Down syndrome with 62 antibodies were used as a case study. The ease of use and efficiency of RPPAware can accelerate data analysis to facilitate biological discovery. RPPAware 1.0 is freely available under GNU General Public License from the project website at http://downsyndrome.ucdenver.edu/iddrc/rppaware/home.htm along with a full documentation of the tool.

Integration of pan-cancer transcriptomics with RPPA proteomics reveals mechanisms of epithelial-mesenchymal transition.

Integrating data from multiple regulatory layers across cancer types could elucidate additional mechanisms of oncogenesis. Using antibody-based protein profiling of 736 cancer cell lines, along with matching transcriptomic data, we show that pan-cancer bimodality in the amounts of mRNA, protein, and protein phosphorylation reveals mechanisms related to the epithelial-mesenchymal transition (EMT). Based on the bimodal expression of E-cadherin, we define an EMT signature consisting of 239 genes, many of which were not previously associated with EMT. By querying gene expression signatures collected from cancer cell lines after small-molecule perturbations, we identify enrichment for histone deacetylase (HDAC) inhibitors as inducers of EMT, and kinase inhibitors as mesenchymal-to-epithelial transition (MET) promoters. Causal modeling of protein-based signaling identifies putative drivers of EMT. In conclusion, integrative analysis of pan-cancer proteomic and transcriptomic data reveals key regulatory mechanisms of oncogenic transformation.

Identification of differentially expressed peptides in high-throughput proteomics data.

With the advent of high-throughput proteomics, the type and amount of data pose a significant challenge to statistical approaches used to validate current quantitative analysis. Whereas many studies focus on the analysis at the protein level, the analysis of peptide-level data provides insight into changes at the sub-protein level, including splice variants, isoforms and a range of post-translational modifications. Statistical evaluation of liquid chromatography-mass spectrometry/mass spectrometry peptide-based label-free differential data is most commonly performed using a t-test or analysis of variance, often after the application of data imputation to reduce the number of missing values. In high-throughput proteomics, statistical analysis methods and imputation techniques are difficult to evaluate, given the lack of gold standard data sets. Here, we use experimental and resampled data to evaluate the performance of four statistical analysis methods and the added value of imputation, for different numbers of biological replicates. We find that three or four replicates are the minimum requirement for high-throughput data analysis and confident assignment of significant changes. Data imputation does increase sensitivity in some cases, but leads to a much higher actual false discovery rate. Additionally, we find that empirical Bayes method (limma) achieves the highest sensitivity, and we thus recommend its use for performing differential expression analysis at the peptide level.

Statistical Models for the Analysis of Isobaric Tags Multiplexed Quantitative Proteomics.

Mass spectrometry is being used to identify protein biomarkers that can facilitate development of drug treatment. Mass spectrometry-based labeling proteomic experiments result in complex proteomic data that is hierarchical in nature often with small sample size studies. The generalized linear model (GLM) is the most popular approach in proteomics to compare protein abundances between groups. However, GLM does not address all the complexities of proteomics data such as repeated measures and variance heterogeneity. Linear models for microarray data (LIMMA) and mixed models are two approaches that can address some of these data complexities to provide better statistical estimates. We compared these three statistical models (GLM, LIMMA, and mixed models) under two different normalization approaches (quantile normalization and median sweeping) to demonstrate when each approach is the best for tagged proteins. We evaluated these methods using a spiked-in data set of known protein abundances, a systemic lupus erythematosus (SLE) data set, and simulated data from multiplexed labeling experiments that use tandem mass tags (TMT). Data are available via ProteomeXchange with identifier PXD005486. We found median sweeping to be a preferred approach of data normalization, and with this normalization approach there was overlap with findings across all methods with GLM being a subset of mixed models. The conclusion is that the mixed model had the best type I error with median sweeping, whereas LIMMA had the better overall statistical properties regardless of normalization approaches.

Protein network construction using reverse phase protein array data.

In this paper, we introduce a novel computational method for constructing protein networks based on reverse phase protein array (RPPA) data to identify complex patterns in protein signaling. The method is applied to phosphoproteomic profiles of basal expression and activation/phosphorylation of 76 key signaling proteins in three breast cancer cell lines (MCF7, LCC1, and LCC9). Temporal RPPA data are acquired at 48h, 96h, and 144h after knocking down four genes in separate experiments. These genes are selected from a previous study as important determinants for breast cancer survival. Interaction networks are constructed by analyzing the expression levels of protein pairs using a multivariate analysis of variance model. A new scoring criterion is introduced to determine relevant protein pairs. Through a network topology based analysis, we search for wiring patterns to identify key proteins that are associated with significant changes in expression levels across various experimental conditions.

Technical Advances of the Recombinant Antibody Microarray Technology Platform for Clinical Immunoproteomics.

In the quest for deciphering disease-associated biomarkers, high-performing tools for multiplexed protein expression profiling of crude clinical samples will be crucial. Affinity proteomics, mainly represented by antibody-based microarrays, have during recent years been established as a proteomic tool providing unique opportunities for parallelized protein expression profiling. But despite the progress, several main technical features and assay procedures remains to be (fully) resolved. Among these issues, the handling of protein microarray data, i.e. the biostatistics parts, is one of the key features to solve. In this study, we have therefore further optimized, validated, and standardized our in-house designed recombinant antibody microarray technology platform. To this end, we addressed the main remaining technical issues (e.g. antibody quality, array production, sample labelling, and selected assay conditions) and most importantly key biostatistics subjects (e.g. array data pre-processing and biomarker panel condensation). This represents one of the first antibody array studies in which these key biostatistics subjects have been studied in detail. Here, we thus present the next generation of the recombinant antibody microarray technology platform designed for clinical immunoproteomics.

Quantitative Proteomic Atlas of Ubiquitination and Acetylation in the DNA Damage Response.

Execution of the DNA damage response (DDR) relies upon a dynamic array of protein modifications. Using quantitative proteomics, we have globally profiled ubiquitination, acetylation, and phosphorylation in response to UV and ionizing radiation. To improve acetylation site profiling, we developed the strategy FACET-IP. Our datasets of 33,500 ubiquitination and 16,740 acetylation sites provide valuable insight into DDR remodeling of the proteome. We find that K6- and K33-linked polyubiquitination undergo bulk increases in response to DNA damage, raising the possibility that these linkages are largely dedicated to DDR function. We also show that Cullin-RING ligases mediate 10% of DNA damage-induced ubiquitination events and that EXO1 is an SCF-Cyclin F substrate in the response to UV radiation. Our extensive datasets uncover additional regulated sites on known DDR players such as PCNA and identify previously unknown DDR targets such as CENPs, underscoring the broad impact of the DDR on cellular physiology.

Statistical and bioanalytical considerations for establishing a depletion criterion for specificity testing during immunogenicity assessment of a biotherapeutic.

Immunogenicity assessment of fully human monoclonal antibody-based biotherapeutics requires sensitive and specific ligand binding assays. One of the components of specificity is the depletion of signal by a relevant biotherapeutic that is commonly based on an arbitrary depletion criterion of inhibition of the original response or reduction of the signal below the screening assay cut point (ACP). Hence, there is a need to develop a statistically derived physiologically relevant specificity criterion. We illustrate an optimization approach to determine the concentration of biotherapeutic required for the specificity evaluation. Naïve donor sample sets with and without circulating drug and antitherapeutic/drug antibody (ADA) were prepared. Next, a depletion cut point (DCP) using naïve and ADA-containing donor sets with the optimized biotherapeutic concentration was evaluated. A statistically derived design of experiment was used to establish a validated DCP. A reliable DCP requires naïve (no ADA) donors treated only with an optimized concentration of biotherapeutic. The additional DCPs generated using two distinct concentrations of ADA-spiked sample sets led to a physiologically irrelevant criterion that was not necessarily representative of real-time samples. This increased the risk of false positives or negatives. In this study, well-defined bioanalytical and statistical methods were employed to validate a DCP to confirm the presence of biotherapeutic specific ADA in human serum samples. A physiologically relevant and effective strategy to confirm specificity in immune reactive samples, especially those that are close to the ACP, is proposed through this study.

A computational framework for the analysis of peptide microarray antibody binding data with application to HIV vaccine profiling.

We present an integrated analytical method for analyzing peptide microarray antibody binding data, from normalization through subject-specific positivity calls and data integration and visualization. Current techniques for the normalization of such data sets do not account for non-specific binding activity. A novel normalization technique based on peptide sequence information quickly and effectively reduced systematic biases. We also employed a sliding mean window technique that borrows strength from peptides sharing similar sequences, resulting in reduced signal variability. A smoothed signal aided in the detection of weak antibody binding hotspots. A new principled FDR method of setting positivity thresholds struck a balance between sensitivity and specificity. In addition, we demonstrate the utility and importance of using baseline control measurements when making subject-specific positivity calls. Data sets from two human clinical trials of candidate HIV-1 vaccines were used to validate the effectiveness of our overall computational framework.

Isolated IgE reactivity to native walnut vicilin-like protein (nJug r 2) on ISAC™ microarray is due to cross-reactive carbohydrate epitopes.

BACKGROUND: The last version of the microarray-based testing ImmunoCAP ISAC 112™ includes the native walnut (Junglans regia) molecules 2S albumin (nJug r 1), vicilin (nJug r 2) and lipid transfer protein (nJug r 3). In view of the many unexpected cases of isolated positivity to nJug r 2 occurring in daily practice, we evaluated the association of these reactivities with clinical symptoms, as well as the relationship between sIgE and nJug r 2 and cross-reactive carbohydrate determinants (CCDs). METHODS: Sera from 320 consecutive allergic outpatients tested by ImmuoCAP ISAC™ 112 were considered. The medical records of all nJug r 2 positive patients were reviewed to assess clinical symptoms related to walnut allergy. A linear regression analysis was performed to evaluate the correlation between nJug r 2 and CCDs (nMUXF3) sIgE values, and a CAP inhibition assay was carried out to confirm the possible cross-reactivity between CCDs and nJug r 2. RESULTS: Thirty-seven out of 320 sera tested (11.6%) were positive to nJug r 2. Among them three (8.1%) and eight (21.6%) scored positive for nJug r 1 and nJug r 3 as well, respectively. Twenty-seven (73%) sera showed isolated nJug r 2 positivity. Only nJug r 1 reactors had symptoms referred to walnut allergy. Twenty-five/37 nJug r 2-positive sera (67.6%) showed a simultaneous positivity to nMUXF3 and a significant correlation (p<0.0001) between the IgE levels to nJug r 2 and nMUXF3 (r²=0.787). After incubation with nMUXF3 a complete inhibition of sIgE reactivity to both nMUXF3 and nJug r 2 was shown. CONCLUSIONS: The unexpected isolated sIgE reactivity to nJug r 2 found by ImmunoCAP ISAC™ 112 is frequently related to reactivity to cross-reactive carbohydrate epitopes and it is lacking clinical significance.

Modeling skewness in human transcriptomes.

Gene expression data are influenced by multiple biological and technological factors leading to a wide range of dispersion scenarios, although skewed patterns are not commonly addressed in microarray analyses. In this study, the distribution pattern of several human transcriptomes has been studied on free-access microarray gene expression data. Our results showed that, even in previously normalized gene expression data, probe and differential expression within probe effects suffer from substantial departures from the commonly assumed symmetric gaussian distribution. We developed a flexible mixed model for non-competitive microarray data analysis that accounted for asymmetric and heavy-tailed (Student's t distribution) dispersion processes. Random effects for gene expression data were modeled under asymmetric Student's t distributions where the asymmetry parameter (λ) took values from perfect symmetry (λ = 0) to right- (λ>0) or left-side (λ>0) over-expression patterns. This approach was applied to four free-access human data sets and revealed clearly better model performance when comparing with standard approaches accounting for traditional symmetric gaussian distribution patterns. Our analyses on human gene expression data revealed a substantial degree of right-hand asymmetry for probe effects, whereas differential gene expression addressed both symmetric and left-hand asymmetric patterns. Although these results cannot be extrapolated to all microarray experiments, they highlighted the incidence of skew dispersion patterns in human transcriptome; moreover, we provided a new analytical approach to appropriately address this biological phenomenon. The source code of the program accommodating these analytical developments and additional information about practical aspects on running the program are freely available by request to the corresponding author of this article.

Development of multigene expression signature maps at the protein level from digitized immunohistochemistry slides.

Molecular classification of diseases based on multigene expression signatures is increasingly used for diagnosis, prognosis, and prediction of response to therapy. Immunohistochemistry (IHC) is an optimal method for validating expression signatures obtained using high-throughput genomics techniques since IHC allows a pathologist to examine gene expression at the protein level within the context of histologically interpretable tissue sections. Additionally, validated IHC assays may be readily implemented as clinical tests since IHC is performed on routinely processed clinical tissue samples. However, methods have not been available for automated n-gene expression profiling at the protein level using IHC data. We have developed methods to compute expression level maps (signature maps) of multiple genes from IHC data digitized on a commercial whole slide imaging system. Areas of cancer for these expression level maps are defined by a pathologist on adjacent, co-registered H&E slides, allowing assessment of IHC statistics and heterogeneity within the diseased tissue. This novel way of representing multiple IHC assays as signature maps will allow the development of n-gene expression profiling databases in three dimensions throughout virtual whole organ reconstructions.

A time-series DDP for functional proteomics profiles.

Using a new type of array technology, the reverse phase protein array (RPPA), we measure time-course protein expression for a set of selected markers that are known to coregulate biological functions in a pathway structure. To accommodate the complex dependent nature of the data, including temporal correlation and pathway dependence for the protein markers, we propose a mixed effects model with temporal and protein-specific components. We develop a sequence of random probability measures (RPM) to account for the dependence in time of the protein expression measurements. Marginally, for each RPM we assume a Dirichlet process model. The dependence is introduced by defining multivariate beta distributions for the unnormalized weights of the stick-breaking representation. We also acknowledge the pathway dependence among proteins via a conditionally autoregressive model. Applying our model to the RPPA data, we reveal a pathway-dependent functional profile for the set of proteins as well as marginal expression profiles over time for individual markers.

A Bayesian hierarchical model for identifying epitopes in peptide microarray data.

Peptide Microarray Immunoassay (PMI for brevity) is a novel technology that enables researchers to map a large number of proteomic measurements at a peptide level, providing information regarding the relationship between antibody response and clinical sensitivity. PMI studies aim at recognizing antigen-specific antibodies from serum samples and at detecting epitope regions of the protein antigen. PMI data present new challenges for statistical analysis mainly due to the structural dependence among peptides. A PMI is made of a complete library of consecutive peptides. They are synthesized by systematically shifting a window of a fixed number of amino acids through the finite sequence of amino acids of the antigen protein as ordered in the primary structure of the protein. This implies that consecutive peptides have a certain number of amino acids in common and hence are structurally dependent. We propose a new flexible Bayesian hierarchical model framework, which allows one to detect recognized peptides and bound epitope regions in a single framework, taking into account the structural dependence between peptides through a suitable latent Markov structure. The proposed model is illustrated using PMI data from a recent study about egg allergy. A simulation study shows that the proposed model is more powerful and robust in terms of epitope detection than simpler models overlooking some of the dependence structure.