Addressing the challenge of defining valid proteomic biomarkers and classifiers

Peptides as "better biomarkers"? Value, challenges, and potential solutions to facilitate implementation

Peptides carry important functions in normal physiological and pathophysiological processes and can serve as clinically useful biomarkers. Given the ability to diffuse passively across endothelial barriers, endogenous peptides can be examined in several body fluids, including among others urine, blood, and cerebrospinal fluid. This review article provides an update on the recently published literature that reports on investigating native peptides in body fluids using mass spectrometry-based platforms, specifically those studies that focus on the application of peptides as biomarkers to improve clinical management. We emphasize on the critical evaluation of their clinical value, how close they are to implementation, and the associated challenges and potential solutions to facilitate clinical implementation. During the last 5 years, numerous studies have been published, demonstrating the increased interest in mass spectrometry for the assessment of endogenous peptides as potential biomarkers. Importantly, the presence of few successful examples of implementation in patients' management and/or in the context of clinical trials indicates that the peptide biomarker field is evolving. Nevertheless, most studies still report evidence based on small sample size, while validation phases are frequently missing. Therefore, a gap between discovery and implementation still exists.

Assessment and Risk Prediction of Chronic Kidney Disease and Kidney Fibrosis Using Non-Invasive Biomarkers

Effective management of chronic kidney disease (CKD), a major health problem worldwide, requires accurate and timely diagnosis, prognosis of progression, assessment of therapeutic efficacy, and, ideally, prediction of drug response. Multiple biomarkers and algorithms for evaluating specific aspects of CKD have been proposed in the literature, many of which are based on a small number of samples. Based on the evidence presented in relevant studies, a comprehensive overview of the different biomarkers applicable for clinical implementation is lacking. This review aims to compile information on the non-invasive diagnostic, prognostic, and predictive biomarkers currently available for the management of CKD and provide guidance on the application of these biomarkers. We specifically focus on biomarkers that have demonstrated added value in prospective studies or those based on prospectively collected samples including at least 100 subjects. Published data demonstrate that several valid non-invasive biomarkers of potential value in the management of CKD are currently available.

Multiple urinary peptides display distinct sex-specific distribution

Previous studies have established the association of sex with gene and protein expression. This study investigated the association of sex with the abundance of endogenous urinary peptides, using capillary electrophoresis-coupled to mass spectrometry (CE-MS) datasets from 2008 healthy individuals and patients with type II diabetes, divided in one discovery and two validation cohorts. Statistical analysis using the Mann-Whitney test, adjusted for multiple testing, revealed 143 sex-associated peptides in the discovery cohort. Of these, 90 peptides were associated with sex in at least one of the validation cohorts and showed agreement in their regulation trends across all cohorts. The 90 sex-associated peptides were fragments of 29 parental proteins. Comparison with previously published transcriptomics data demonstrated that the genes encoding 16 of these parental proteins had sex-biased expression. The 143 sex-associated peptides were combined into a support vector machine-based classifier that could discriminate males from females in two independent sets of healthy individuals and patients with type II diabetes, with an AUC of 89% and 81%, respectively. Collectively, the urinary peptidome contains multiple sex-associated differences, which may enable a better understanding of sex-biased molecular mechanisms and the development of more accurate diagnostic, prognostic, or predictive classifiers for each individual sex.

Urinary peptidomic liquid biopsy for non-invasive differential diagnosis of chronic kidney disease

BACKGROUND AND HYPOTHESIS

Specific urinary peptides hold information on disease pathophysiology, which, in combination with artificial intelligence, could enable non-invasive assessment of chronic kidney disease (CKD) aetiology. Existing approaches are generally specific for the diagnosis of single aetiologies. We present the development of models able to simultaneously distinguish and spatially visualize multiple CKD aetiologies.

METHODS

The urinary peptide data of 1850 healthy control (HC) and CKD [diabetic kidney disease (DKD), immunoglobulin A nephropathy (IgAN) and vasculitis] participants were extracted from the Human Urinary Proteome Database. Uniform manifold approximation and projection (UMAP) coupled to a support vector machine algorithm was used to generate multi-peptide models to perform binary (DKD, HC) and multiclass (DKD, HC, IgAN, vasculitis) classifications. This pipeline was compared with the current state-of-the-art single-aetiology CKD urinary peptide models.

RESULTS

In an independent test set, the developed models achieved 90.35% and 70.13% overall predictive accuracies, respectively, for the binary and the multiclass classifications. Omitting the UMAP step led to improved predictive accuracies (96.14% and 85.06%, respectively). As expected, the HC class was distinguished with the highest accuracy. The different classes displayed a tendency to form distinct clusters in the 3D space based on their disease state.

CONCLUSION

Urinary peptide data present an effective basis for CKD aetiology differentiation using machine learning models. Although adding the UMAP step to the models did not improve prediction accuracy, it may provide a unique visualization advantage. Additional studies are warranted to further validate the pipeline's clinical potential as well as to expand it to other CKD aetiologies and also other diseases.

Recent Advances in Urinary Peptide and Proteomic Biomarkers in Chronic Kidney Disease: A Systematic Review

Biomarker development, improvement, and clinical implementation in the context of kidney disease have been a central focus of biomedical research for decades. To this point, only serum creatinine and urinary albumin excretion are well-accepted biomarkers in kidney disease. With their known blind spot in the early stages of kidney impairment and their diagnostic limitations, there is a need for better and more specific biomarkers. With the rise in large-scale analyses of the thousands of peptides in serum or urine samples using mass spectrometry techniques, hopes for biomarker development are high. Advances in proteomic research have led to the discovery of an increasing amount of potential proteomic biomarkers and the identification of candidate biomarkers for clinical implementation in the context of kidney disease management. In this review that strictly follows the PRISMA guidelines, we focus on urinary peptide and especially peptidomic biomarkers emerging from recent research and underline the role of those with the highest potential for clinical implementation. The Web of Science database (all databases) was searched on 17 October 2022, using the search terms "marker *" OR biomarker * AND "renal disease" OR "kidney disease" AND "proteome *" OR "peptid *" AND "urin *". English, full-text, original articles on humans published within the last 5 years were included, which had been cited at least five times per year. Studies based on animal models, renal transplant studies, metabolite studies, studies on miRNA, and studies on exosomal vesicles were excluded, focusing on urinary peptide biomarkers. The described search led to the identification of 3668 articles and the application of inclusion and exclusion criteria, as well as abstract and consecutive full-text analyses of three independent authors to reach a final number of 62 studies for this manuscript. The 62 manuscripts encompassed eight established single peptide biomarkers and several proteomic classifiers, including CKD273 and IgAN237. This review provides a summary of the recent evidence on single peptide urinary biomarkers in CKD, while emphasizing the increasing role of proteomic biomarker research with new research on established and new proteomic biomarkers. Lessons learned from the last 5 years in this review might encourage future studies, hopefully resulting in the routine clinical applicability of new biomarkers.

Mass Spectrometry-Based Biomarkers to Detect Prostate Cancer: A Multicentric Study Based on Non-Invasive Urine Collection without Prior Digital Rectal Examination

(1) Background: Prostate cancer (PCa) is the most frequently diagnosed cancer in men. Wide application of prostate specific antigen test has historically led to over-treatment, starting from excessive biopsies. Risk calculators based on molecular and clinical variables can be of value to determine the risk of PCa and as such, reduce unnecessary and invasive biopsies. Urinary molecular studies have been mostly focusing on sampling after initial intervention (digital rectal examination and/or prostate massage). (2) Methods: Building on previous proteomics studies, in this manuscript, we aimed at developing a biomarker model for PCa detection based on urine sampling without prior intervention. Capillary electrophoresis coupled to mass spectrometry was applied to acquire proteomics profiles from 970 patients from two different clinical centers. (3) Results: A case-control comparison was performed in a training set of 413 patients and 181 significant peptides were subsequently combined by a support vector machine algorithm. Independent validation was initially performed in 272 negative for PCa and 138 biopsy-confirmed PCa, resulting in an AUC of 0.81, outperforming current standards, while a second validation phase included 147 PCa patients. (4) Conclusions: This multi-dimensional biomarker model holds promise to improve the current diagnosis of PCa, by guiding invasive biopsies.

Advances, obstacles, and opportunities for machine learning in proteomics

The fields of proteomics and machine learning are both large disciplines, each producing well over 5,000 publications per year. However, studies combining both fields are still relatively rare, with only about 2% of recent proteomics papers including machine learning. This review, which focuses on the intersection of the fields, is intended to inspire proteomics researchers to develop skills and knowledge in the application of machine learning. A brief tutorial introduction to machine learning is provided, and research advances that rely on both fields, particularly as they relate to proteomics tools development and biomarker discovery, are highlighted. Key knowledge gaps and opportunities for scientific advancement are also enumerated.

How (Not) to Generate a Highly Predictive Biomarker Panel Using Machine Learning

This review "teaches" researchers how to make their lackluster proteomics data look really impressive, by applying an inappropriate but pervasive strategy that selects features in a biased manner. The strategy is demonstrated and used to build a classification model with an accuracy of 92% and AUC of 0.98, while relying completely on random numbers for the data set. This "lesson" in data processing is not to be practiced by anyone; on the contrary, it is meant to be a cautionary tale showing that very unreliable results are obtained when a biomarker panel is generated first, using all the available data, and then tested by cross-validation. Data scientists describe the error committed in this scenario as having test data leak into the feature selection step, and it is currently a common mistake in proteomics biomarker studies that rely on machine learning. After the demonstration, advice is provided about how machine learning methods can be applied to proteomics data sets without generating artificially inflated accuracies.

The Implant Proteome—The Right Surgical Glue to Fix Titanium Implants In Situ

Titanium implants are frequently applied to the bone in orthopedic and trauma surgery. Although these biomaterials are characterized by excellent implant survivorship and clinical outcomes, there are almost no data available on the initial protein layer binding to the implant surface in situ. This study aims to investigate the composition of the initial protein layer on endoprosthetic surfaces as a key initiating step in osseointegration. In patients qualified for total hip arthroplasty, the implants are inserted into the femoral canal, fixed and subsequently explanted after 2 and 5 min. The proteins adsorbed to the surface (the implant proteome) are analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS). A statistical analysis of the proteins’ alteration with longer incubation times reveals a slight change in their abundance according to the Vroman effect. The pathways involved in the extracellular matrix organization of bone, sterile inflammation and the beginning of an immunogenic response governed by neutrophils are significantly enriched based on the analysis of the implant proteome. Those are generally not changed with longer incubation times. In summary, proteins relevant for osseointegration are already adsorbed within 2 min in situ. A deeper understanding of the in situ protein–implant interactions in patients may contribute to optimizing implant surfaces in orthopedic and trauma surgery.

A Model to Detect Significant Prostate Cancer Integrating Urinary Peptide and Extracellular Vesicle RNA Data

There is a clinical need to improve assessment of biopsy-naïve patients for the presence of clinically significant prostate cancer (PCa). In this study, we investigated whether the robust integration of expression data from urinary extracellular vesicle RNA (EV-RNA) with urine proteomic metabolites can accurately predict PCa biopsy outcome. Urine samples collected within the Movember GAP1 Urine Biomarker study (n = 192) were analysed by both mass spectrometry-based urine-proteomics and NanoString gene-expression analysis (167 gene-probes). Cross-validated LASSO penalised regression and Random Forests identified a combination of clinical and urinary biomarkers for predictive modelling of significant disease (Gleason Score (Gs) ≥ 3 + 4). Four predictive models were developed: ‘MassSpec’ (CE-MS proteomics), ‘EV-RNA’, and ‘SoC’ (standard of care) clinical data models, alongside a fully integrated omics-model, deemed ‘ExoSpec’. ExoSpec (incorporating four gene transcripts, six peptides, and two clinical variables) is the best model for predicting Gs ≥ 3 + 4 at initial biopsy (AUC = 0.83, 95% CI: 0.77−0.88) and is superior to a standard of care (SoC) model utilising clinical data alone (AUC = 0.71, p < 0.001, 1000 resamples). As the ExoSpec Risk Score increases, the likelihood of higher-grade PCa on biopsy is significantly greater (OR = 2.8, 95% CI: 2.1−3.7). The decision curve analyses reveals that ExoSpec provides a net benefit over SoC and could reduce unnecessary biopsies by 30%.

A Review on Differential Abundance Analysis Methods for Mass Spectrometry-Based Metabolomic Data

This review presents an overview of the statistical methods on differential abundance (DA) analysis for mass spectrometry (MS)-based metabolomic data. MS has been widely used for metabolomic abundance profiling in biological samples. The high-throughput data produced by MS often contain a large fraction of zero values caused by the absence of certain metabolites and the technical detection limits of MS. Various statistical methods have been developed to characterize the zero-inflated metabolomic data and perform DA analysis, ranging from simple tests to more complex models including parametric, semi-parametric, and non-parametric approaches. In this article, we discuss and compare DA analysis methods regarding their assumptions and statistical modeling techniques.

Reproducibility Evaluation of Urinary Peptide Detection Using CE-MS

In recent years, capillary electrophoresis coupled to mass spectrometry (CE-MS) has been increasingly applied in clinical research especially in the context of chronic and age-associated diseases, such as chronic kidney disease, heart failure and cancer. Biomarkers identified using this technique are already used for diagnosis, prognosis and monitoring of these complex diseases, as well as patient stratification in clinical trials. CE-MS allows for a comprehensive assessment of small molecular weight proteins and peptides (<20 kDa) through the combination of the high resolution and reproducibility of CE and the distinct sensitivity of MS, in a high-throughput system. In this study we assessed CE-MS analytical performance with regards to its inter- and intra-day reproducibility, variability and efficiency in peptide detection, along with a characterization of the urinary peptidome content. To this end, CE-MS performance was evaluated based on 72 measurements of a standard urine sample (60 for inter- and 12 for intra-day assessment) analyzed during the second quarter of 2021. Analysis was performed per run, per peptide, as well as at the level of biomarker panels. The obtained datasets showed high correlation between the different runs, low variation of the ten highest average individual log2 signal intensities (coefficient of variation, CV < 10%) and very low variation of biomarker panels applied (CV close to 1%). The findings of the study support the analytical performance of CE-MS, underlining its value for clinical application.

Increased Collagen Turnover Is a Feature of Fibromuscular Dysplasia and Associated With Hypertrophic Radial Remodeling: A Pilot, Urine Proteomic Study

Fibromuscular dysplasia (FMD), a nonatherosclerotic, noninflammatory disease of medium-sized arteries, is an underdiagnosed disease. We investigated the urinary proteome and developed a classifier for discrimination of FMD from healthy controls and other diseases. We further hypothesized that urinary proteomics biomarkers may be associated with alterations in medium-sized, but not large artery geometry and mechanics. The study included 33 patients with mostly multifocal, renal FMD who underwent in depth arterial exploration using ultra-high frequency ultrasound. The cohort was separated in a training set of 23 patients with FMD from Belgium and an independent test set of 10 patients with FMD from Italy. For each set, controls matched 2:1 were selected from the Human Urinary Proteome Database. The specificity of the classifier was tested in 700 additional controls from general population studies, patients with chronic kidney disease (n=66) and coronary artery disease (n=31). Three hundred thirty-five urinary peptides, mostly related to collagen turnover, were identified in the training cohort and combined into a classifier. When applying in the test cohort, the area under the receiver operating characteristic curve was 1.00, 100% specificity at 100% sensitivity. The classifier maintained a high specificity in additional controls (98.3%), patients with chronic kidney (90.9%) and coronary artery (96.8%) diseases. Furthermore, in patients with FMD, the proteomic score was positively associated with radial wall thickness and wall cross-sectional area. In conclusion, a proteomic score has the potential to discriminate between patients with FMD and controls. If confirmed in a wider and more diverse cohort, these findings may pave the way for a noninvasive diagnostic test of FMD.

CD99 and polymeric immunoglobulin receptor peptides deregulation in critical COVID-19: A potential link to molecular pathophysiology?

Identification of significant changes in urinary peptides may enable improved understanding of molecular disease mechanisms. We aimed towards identifying urinary peptides associated with critical course of COVID-19 to yield hypotheses on molecular pathophysiological mechanisms in disease development. In this multicentre prospective study urine samples of PCR-confirmed COVID-19 patients were collected in different centres across Europe. The urinary peptidome of 53 patients at WHO stages 6-8 and 66 at WHO stages 1-3 COVID-19 disease was analysed using capillary electrophoresis coupled to mass spectrometry. 593 peptides were identified significantly affected by disease severity. These peptides were compared with changes associated with kidney disease or heart failure. Similarities with kidney disease were observed, indicating comparable molecular mechanisms. In contrast, convincing similarity to heart failure could not be detected. The data for the first time showed deregulation of CD99 and polymeric immunoglobulin receptor peptides and of known peptides associated with kidney disease, including collagen and alpha-1-antitrypsin. Peptidomic findings were in line with the pathophysiology of COVID-19. The clinical corollary is that COVID-19 induces specific inflammation of numerous tissues including endothelial lining. Restoring these changes, especially in CD99, PIGR and alpha-1-antitripsin, may represent a valid and effective therapeutic approach in COVID-19, targeting improvement of endothelial integrity. This article is protected by copyright. All rights reserved.

A Novel Urinary Proteomics Classifier for Non-Invasive Evaluation of Interstitial Fibrosis and Tubular Atrophy in Chronic Kidney Disease

Non-invasive urinary peptide biomarkers are able to detect and predict chronic kidney disease (CKD). Moreover, specific urinary peptides enable discrimination of different CKD etiologies and offer an interesting alternative to invasive kidney biopsy, which cannot always be performed. The aim of this study was to define a urinary peptide classifier using mass spectrometry technology to predict the degree of renal interstitial fibrosis and tubular atrophy (IFTA) in CKD patients. The urinary peptide profiles of 435 patients enrolled in this study were analyzed using capillary electrophoresis coupled with mass spectrometry (CE-MS). Urine samples were collected on the day of the diagnostic kidney biopsy. The proteomics data were divided into a training (n = 200) and a test (n = 235) cohort. The fibrosis group was defined as IFTA ≥ 15% and no fibrosis as IFTA < 10%. Statistical comparison of the mass spectrometry data enabled identification of 29 urinary peptides with differential occurrence in samples with and without fibrosis. Several collagen fragments and peptide fragments of fetuin-A and others were combined into a peptidomic classifier. The classifier separated fibrosis from non-fibrosis patients in an independent test set (n = 186) with area under the curve (AUC) of 0.84 (95% CI: 0.779 to 0.889). A significant correlation of IFTA and FPP_BH29 scores could be observed Rho = 0.5, p < 0.0001. We identified a peptidomic classifier for renal fibrosis containing 29 peptide fragments corresponding to 13 different proteins. Urinary proteomics analysis can serve as a non-invasive tool to evaluate the degree of renal fibrosis, in contrast to kidney biopsy, which allows repeated measurements during the disease course.

Improved Discrimination of Disease States Using Proteomics Data with the Updated Aristotle Classifier

Mass spectrometry data sets from omics studies are an optimal information source for discriminating patients with disease and identifying biomarkers. Thousands of proteins or endogenous metabolites can be queried in each analysis, spanning several orders of magnitude in abundance. Machine learning tools that effectively leverage these data to accurately identify disease states are in high demand. While mass spectrometry data sets are rich with potentially useful information, using the data effectively can be challenging because of missing entries in the data sets and because the number of samples is typically much smaller than the number of features, two challenges that make machine learning difficult. To address this problem, we have modified a new supervised classification tool, the Aristotle Classifier, so that omics data sets can be better leveraged for identifying disease states. The optimized classifier, AC.2021, is benchmarked on multiple data sets against its predecessor and two leading supervised classification tools, Support Vector Machine (SVM) and XGBoost. The new classifier, AC.2021, outperformed existing tools on multiple tests using proteomics data. The underlying code for the classifier, provided herein, would be useful for researchers who desire improved classification accuracy when using their omics data sets to identify disease states.

Urinary peptides in heart failure: a link to molecular pathophysiology

Aims

Heart failure (HF) is a major public health concern worldwide. The diversity of HF makes it challenging to decipher the underlying complex pathological processes using single biomarkers. We examined the association between urinary peptides and HF with reduced (HFrEF), mid‐range (HFmrEF) and preserved (HFpEF) ejection fraction, defined based on the European Society of Cardiology guidelines, and the links between these peptide biomarkers and molecular pathophysiology.

Methods and results

Analysable data from 5608 participants were available in the Human Urinary Proteome database. The urinary peptide profiles from participants diagnosed with HFrEF, HFmrEF, HFpEF and controls matched for sex, age, estimated glomerular filtration rate, systolic and diastolic blood pressure, diabetes and hypertension were compared applying the Mann–Whitney test, followed by correction for multiple testing. Unsupervised learning algorithms were applied to investigate groups of similar urinary profiles. A total of 577 urinary peptides significantly associated with HF were sequenced, 447 of which (77%) were collagen fragments. In silico analysis suggested that urinary biomarker abnormalities in HF principally reflect changes in collagen turnover and immune response, both associated with fibrosis. Unsupervised clustering separated study participants into two clusters, with 83% of non‐HF controls allocated to cluster 1, while 65% of patients with HF were allocated to cluster 2 (P < 0.0001). No separation based on HF subtype was detectable.

Conclusions

Heart failure, irrespective of ejection fraction subtype, was associated with differences in abundance of urinary peptides reflecting collagen turnover and inflammation. These peptides should be studied as tools in early detection, prognostication, and prediction of therapeutic response.

Urinary Proteomics Associates with COVID-19 Severity: Pilot Proof-of-Principle Data and Design of a Multicentric Diagnostic Study

SARS-CoV-2 infection results in a mild-to-moderate disease course in most patients, allowing outpatient self-care and quarantine. However, in approx. 10% of cases a two- or three-phasic critical disease course with starting from day 7 to 10 is observed. To facilitate and plan outpatient care, biomarkers prognosing such worsening at an early stage appear of outmost importance. In this accelerated article, we report on the identification of urinary peptides significantly associated with SARS-CoV-2 infection, and the development of a multi-marker urinary peptide based test, COVID20, that may enable prognosis of critical and fatal outcomes in COVID-19 patients. COVID20 is composed of 20 endogenous peptides mainly derived from various collagen chains that enable differentiating moderate or severe disease from critical state or death with 83% sensitivity at 100% specificity. Based on the performance in this pilot study, testing in a prospective study on 1000 patients has been initiated. This article is protected by copyright. All rights reserved.

Mortality Risk Profiling of Staphylococcus aureus Bacteremia by Multi-omic Serum Analysis Reveals Early Predictive and Pathogenic Signatures

Staphylococcus aureus bacteremia (SaB) causes significant disease in humans, carrying mortality rates of ∼25%. The ability to rapidly predict SaB patient responses and guide personalized treatment regimens could reduce mortality. Here, we present a resource of SaB prognostic biomarkers. Integrating proteomic and metabolomic techniques enabled the identification of >10,000 features from >200 serum samples collected upon clinical presentation. We interrogated the complexity of serum using multiple computational strategies, which provided a comprehensive view of the early host response to infection. Our biomarkers exceed the predictive capabilities of those previously reported, particularly when used in combination. Last, we validated the biological contribution of mortality-associated pathways using a murine model of SaB. Our findings represent a starting point for the development of a prognostic test for identifying high-risk patients at a time early enough to trigger intensive monitoring and interventions.

Proteomics-Based Machine Learning Approach as an Alternative to Conventional Biomarkers for Differential Diagnosis of Chronic Kidney Diseases

Diabetic nephropathy, hypertension, and glomerulonephritis are the most common causes of chronic kidney diseases (CKD). Since CKD of various origins may not become apparent until kidney function is significantly impaired, a differential diagnosis and an appropriate treatment are needed at the very early stages. Conventional biomarkers may not have sufficient separation capabilities, while a full-proteomic approach may be used for these purposes. In the current study, several machine learning algorithms were examined for the differential diagnosis of CKD of three origins. The tested dataset was based on whole proteomic data obtained after the mass spectrometric analysis of plasma and urine samples of 34 CKD patients and the use of label-free quantification approach. The k-nearest-neighbors algorithm showed the possibility of separation of a healthy group from renal patients in general by proteomics data of plasma with high confidence (97.8%). This algorithm has also be proven to be the best of the three tested for distinguishing the groups of patients with diabetic nephropathy and glomerulonephritis according to proteomics data of plasma (96.3% of correct decisions). The group of hypertensive nephropathy could not be reliably separated according to plasma data, whereas analysis of entire proteomics data of urine did not allow differentiating the three diseases. Nevertheless, the group of hypertensive nephropathy was reliably separated from all other renal patients using the k-nearest-neighbors classifier “one against all” with 100% of accuracy by urine proteome data. The tested algorithms show good abilities to differentiate the various groups across proteomic data sets, which may help to avoid invasive intervention for the verification of the glomerulonephritis subtypes, as well as to differentiate hypertensive and diabetic nephropathy in the early stages based not on individual biomarkers, but on the whole proteomic composition of urine and blood.

State of the art in selection of variables and functional forms in multivariable analysis-outstanding issues

BACKGROUND

How to select variables and identify functional forms for continuous variables is a key concern when creating a multivariable model. Ad hoc 'traditional' approaches to variable selection have been in use for at least 50 years. Similarly, methods for determining functional forms for continuous variables were first suggested many years ago. More recently, many alternative approaches to address these two challenges have been proposed, but knowledge of their properties and meaningful comparisons between them are scarce. To define a state of the art and to provide evidence-supported guidance to researchers who have only a basic level of statistical knowledge, many outstanding issues in multivariable modelling remain. Our main aims are to identify and illustrate such gaps in the literature and present them at a moderate technical level to the wide community of practitioners, researchers and students of statistics.

METHODS

We briefly discuss general issues in building descriptive regression models, strategies for variable selection, different ways of choosing functional forms for continuous variables and methods for combining the selection of variables and functions. We discuss two examples, taken from the medical literature, to illustrate problems in the practice of modelling.

RESULTS

Our overview revealed that there is not yet enough evidence on which to base recommendations for the selection of variables and functional forms in multivariable analysis. Such evidence may come from comparisons between alternative methods. In particular, we highlight seven important topics that require further investigation and make suggestions for the direction of further research.

CONCLUSIONS

Selection of variables and of functional forms are important topics in multivariable analysis. To define a state of the art and to provide evidence-supported guidance to researchers who have only a basic level of statistical knowledge, further comparative research is required.

Differential Abundance Analysis with Bayes Shrinkage Estimation of Variance (DASEV) for Zero-Inflated Proteomic and Metabolomic Data

Mass spectrometry (MS) is frequently used for proteomic and metabolomic profiling of biological samples. Data obtained by MS are often zero-inflated. Those zero values are called point mass values (PMVs). Zero values can be further grouped into biological PMVs and technical PMVs. The former type is caused by true absence of a compound and the later type is caused by a technical detection limit. Methods based on a mixture model have been developed to separate the two types of zeros and to perform differential abundance analysis comparing proteomic/metabolomic profiles between different groups of subjects. However, we notice that those methods may give unstable estimate of the model variance, and thus lead to false positive and false negative results when the number of non-zero values is small. In this paper, we propose a new differential abundance analysis method, DASEV, which uses an empirical Bayes shrinkage method to more robustly estimate the variance and enhance the accuracy of differential abundance analysis. Simulation studies and real data analysis show that DASEV substantially improves parameter estimation of the mixture model and outperforms current methods in identifying differentially abundant features.

Extending the information content of the MALDI analysis of biological fluids via multi-million shot analysis

INTRODUCTION

Reliable measurements of the protein content of biological fluids like serum or plasma can provide valuable input for the development of personalized medicine tests. Standard MALDI analysis typically only shows high abundance proteins, which limits its utility for test development. It also exhibits reproducibility issues with respect to quantitative measurements. In this paper we show how the sensitivity of MALDI profiling of intact proteins in unfractionated human serum can be substantially increased by exposing a sample to many more laser shots than are commonly used. Analytical reproducibility is also improved.

METHODS

To assess what is theoretically achievable we utilized spectra from the same samples obtained over many years and combined them to generate MALDI spectral averages of up to 100,000,000 shots for a single sample, and up to 8,000,000 shots for a set of 40 different serum samples. Spectral attributes, such as number of peaks and spectral noise of such averaged spectra were investigated together with analytical reproducibility as a function of the number of shots. We confirmed that results were similar on MALDI instruments from different manufacturers.

RESULTS

We observed an expected decrease of noise, roughly proportional to the square root of the number of shots, over the whole investigated range of the number of shots (5 orders of magnitude), resulting in an increase in the number of reliably detected peaks. The reproducibility of the amplitude of these peaks, measured by CV and concordance analysis also improves with very similar dependence on shot number, reaching median CVs below 2% for shot numbers > 4 million. Measures of analytical information content and association with biological processes increase with increasing number of shots.

CONCLUSIONS

We demonstrate that substantially increasing the number of laser shots in a MALDI-TOF analysis leads to more informative and reliable data on the protein content of unfractionated serum. This approach has already been used in the development of clinical tests in oncology.

Predicting albuminuria response to spironolactone treatment with urinary proteomics in patients with type 2 diabetes and hypertension

Background

The mineralocorticoid receptor antagonist spironolactone significantly reduces albuminuria in patients with diabetes. Prior studies have shown large between-patient variability in albuminuria treatment response. We previously developed and validated a urinary proteomic classifier that predicts onset and progression of chronic kidney disease. Here, we tested whether the proteomic classifier based on 273 urinary peptides (CKD273) predicts albuminuria response to spironolactone treatment.

Methods

We performed a post hoc analysis in a double-blind randomized clinical trial with allocation to either spironolactone 12.5-50 mg/day (n = 57) or placebo (n = 54) for 16 weeks. Patients were diagnosed with type 2 diabetes and resistant hypertension. Treatment was an adjunct to renin-angiotensin system inhibition. Primary endpoint was the percentage change in urine albumin to creatinine ratio (UACR). Capillary electrophoresis mass spectrometry was used to quantify urinary peptides at baseline. The previously validated combination of 273 known urinary peptides was used as proteomic classifier.

Results

Spironolactone reduced UACR relative to placebo by 50%, although with a large between-patient variability in UACR response (5th to 95th percentile, 7 to 312%). An interaction was detected between CKD273 and treatment assignment (β = -1.09, P = 0.026). Higher values of CKD273 at baseline were associated with a larger reduction in UACR in the spironolactone group (β = -0.70, P = 0.049), but not in the placebo group (β = 0.39, P = 0.25). Stratified in tertiles of baseline CKD273, reduction in UACR was greater in the highest tertile, 63% (95% confidence interval: 35-79%), as compared with the two other tertiles combined, 16% (-17 to 40%) (P = 0.011).

Conclusions

A urinary proteomics classifier can be used to identify individuals with type 2 diabetes who are more likely to show an albuminuria-lowering response to spironolactone treatment. These results suggest that urinary proteomics may be a valuable tool to tailor therapy, but confirmation in a larger clinical trial is required.

Clinical Proteomics in Kidney Disease: From Discovery to Clinical Application

Proteome analysis has been applied in multiple studies in the context of chronic kidney disease, aiming at improving our knowledge on the molecular pathophysiology of the disease. The approach is generally based on the hypothesis that proteins are key in maintaining kidney function, and disease is a clinical consequence of a significant change of the protein level. Knowledge on critical proteins and their alteration in disease should in turn enable identification of ideal biomarkers that could guide patient management. In addition, all drugs currently employed target proteins. Hence, proteome analysis also promises to enable identifying the best suited therapeutic target, and, in combination with biomarkers, could be used as the rationale basis for personalized intervention. To assess the current status of proteome analysis in the context of CKD, we present the results of a systematic review, of up-to-date scientific research, and give an outlook on the developments that can be expected in near future. Based on the current literature, proteome analysis has already seen implementation in the management of CKD patients, and it is expected that this approach, also supported by the positive results generated to date, will see advanced high-throughput application.

Urinary peptidome analyses for the diagnosis of chronic kidney disease in dogs

Chronic kidney disease (CKD) is clinically important in canine medicine. Current diagnostic tools lack sensitivity for detection of subclinical CKD. The aim of the present study was to evaluate urinary peptidome analysis for diagnosis of CKD in dogs. Capillary electrophoresis coupled to mass spectrometry analysis demonstrated presence of approximately 5400 peptides in dog urine. Comparison of urinary peptide abundance of dogs with and without CKD led to the identification of 133 differentially excreted peptides (adjusted P for each peptide <0.05). Sequence information was obtained for 35 of these peptides. This 35 peptide subset and the total group of 133 peptides were used to construct two predictive models of CKD which were subsequently validated by researchers masked to results in an independent cohort of 20 dogs. Both models diagnosed CKD with an area under the receiver operating characteristic (ROC) curve of 0.88 (95% confidence intervals [CI], 0.72-1.0). Most differentially excreted peptides represented fragments of collagen I, indicating possible association with fibrotic processes in CKD (similar to the equivalent human urinary peptide CKD model, CKD273). This first study of the urinary peptidome in dogs identified peptides that were associated with presence of CKD. Future studies are needed to validate the utility of this model for diagnosis and prediction of progression of canine CKD in a clinical setting.

CE-MS-based urinary biomarkers to distinguish non-significant from significant prostate cancer

Background

Prostate cancer progresses slowly when present in low risk forms but can be lethal when it progresses to metastatic disease. A non-invasive test that can detect significant prostate cancer is needed to guide patient management.

Methods

Capillary electrophoresis/mass spectrometry has been employed to identify urinary peptides that may accurately detect significant prostate cancer. Urine samples from 823 patients with PSA (<15 ng/ml) were collected prior to biopsy. A case–control comparison was performed in a training set of 543 patients (n_Sig = 98; n_non-Sig = 445) and a validation set of 280 patients (n_Sig = 48, n_non-Sig = 232). Totally, 19 significant peptides were subsequently combined by a support vector machine algorithm.

Results

Independent validation of the 19-biomarker model in 280 patients resulted in a 90% sensitivity and 59% specificity, with an AUC of 0.81, outperforming PSA (AUC = 0.58) and the ERSPC-3/4 risk calculator (AUC = 0.69) in the validation set.

Conclusions

This multi-parametric model holds promise to improve the current diagnosis of significant prostate cancer. This test as a guide to biopsy could help to decrease the number of biopsies and guide intervention. Nevertheless, further prospective validation in an external clinical cohort is required to assess the exact performance characteristics.

Urinary Proteomics as a Tool to Identify Kidney Responders to Dipeptidyl Peptidase-4 Inhibition: A Hypothesis-Generating Analysis from the MARLINA-T2D Trial

PURPOSE

Chronic kidney disease (CKD) is a serious complication of hyperglycemia and treatment options to slow its progression are scarce. Dipeptidyl peptidase-4 (DPP-4) inhibitors are common glucose-lowering drugs in type 2 diabetes (T2D). Among these, linagliptin has been suggested to exert kidney protective effects. It is investigated whether an effect of linagliptin on kidney function could be unmasked by characterizing the urinary proteome profile (UPP) in albuminuric T2D individuals.

EXPERIMENTAL DESIGN

Participants of the MARLINA-T2D trial (NCT01792518) are randomized 1:1 to receive either linagliptin 5 mg or placebo for 24 weeks. A previously developed proteome-based classifier, CKD273, is assessed.

RESULTS

Results confirm a significant correlation between CKD273 and clinical kidney parameters as well as with eGFR decline. Patient stratification using CKD273 at baseline, show a trend toward attenuation of renal function loss in high CKD-risk patients treated with linagliptin. Moreover, characterized are linagliptin affected peptides of which the majority contained a DPP-4 target sequence.

CONCLUSIONS AND CLINICAL RELEVANCE

CKD273 is a promising tool for identifying patients at high risk for CKD progression and may unmask a potential of linagliptin to slow progressive kidney function loss in high CKD-risk patients. UPP characterization reveals a significant impact of linagliptin on urinary peptides.

Urinary proteomic biomarkers in oncology: ready for implementation?

Introduction: Biomarkers are expected to improve the management of cancer patients by enabling early detection and prediction of therapeutic response. Proteins reflect a molecular phenotype, have high potential as biomarkers, and also are key targets for intervention. Given the ease of collection and proximity to certain tumors, the urinary proteome is a rich source of biomarkers and several proteins have been already implemented. Areas covered: We examined the literature on urine proteins and proteome analysis in oncology from reports published during the last 5 years to generate an overview on the status of urine protein and peptide biomarkers, with emphasis on their actual clinical value. Expert commentary: A few studies report on biomarkers that are ready to be implemented in patient management, among others in bladder cancer and cholangiocarcinoma. These reports are based on multi-marker approaches. A high number of biomarkers, though, has been described in studies with low statistical power. In fact, several of them have been consistently reported across different studies. The latter should be the focus of attention and be tested in properly designed confirmatory and ultimately, prospective investigations. It is expected that multi-marker classifiers for a specific context-of-use, will be the preferred path toward clinical implementation.

Nanowell-mediated multidimensional separations combining nanoLC with SLIM IM-MS for rapid, high-peak-capacity proteomic analyses

Mass spectrometry (MS)-based analysis of complex biological samples is essential for biomedical research and clinical diagnostics. The separation prior to MS plays a key role in the overall analysis, with separations having larger peak capacities often leading to more identified species and improved confidence in those identifications. High-resolution ion mobility (IM) separations enabled by Structures for Lossless Ion Manipulation (SLIM) can provide extremely rapid, high-resolution separations and are well suited as a second dimension of separation following nanoscale liquid chromatography (nanoLC). However, existing sample handling approaches for offline coupling of separation modes require microliter-fraction volumes and are thus not well suited for analysis of trace biological samples. We have developed a novel nanowell-mediated fractionation system that enables nanoLC-separated samples to be efficiently preconcentrated and directly infused at nanoelectrospray flow rates for downstream analysis. When coupled with SLIM IM-MS, the platform enables rapid and high-peak-capacity multidimensional separations of small biological samples. In this study, peptides eluting from a 100 nL/min nanoLC separation were fractionated into ~ 60 nanowells on a microfluidic glass chip using an in-house-developed robotic system. The dried samples on the chip were individually reconstituted and ionized by nanoelectrospray for SLIM IM-MS analysis. Using model peptides for characterization of the nanowell platform, we found that at least 80% of the peptide components of the fractionated samples were recovered from the nanowells, providing up to ~tenfold preconcentration for SLIM IM-MS analysis. The combined LC-SLIM IM separation peak capacities exceeded 3600 with a measurement throughput that is similar to current one-dimensional (1D) LC-MS proteomic analyses. Graphical abstract A nanowell-mediated multidimensional separation platform that combines nanoLC with SLIM IM-MS enables rapid, high-peak-capacity proteomic analyses.

Novel Urinary Biomarkers For Improved Prediction Of Progressive Egfr Loss In Early Chronic Kidney Disease Stages And In High Risk Individuals Without Chronic Kidney Disease

Chronic kidney disease is associated with increased risk of CKD progression and death. Therapeutic approaches to limit progression are limited. Developing tools for the early identification of those individuals most likely to progress will allow enriching clinical trials in high risk early CKD patients. The CKD273 classifier is a panel of 273 urinary peptides that enables early detection of CKD and prognosis of progression. We have generated urine capillary electrophoresis-mass spectrometry-based peptidomics CKD273 subclassifiers specific for CKD stages to allow the early identification of patients at high risk of CKD progression. In the validation cohort, the CKD273 subclassifiers outperformed albuminuria and CKD273 classifier for predicting rapid loss of eGFR in individuals with baseline eGFR > 60 ml/min/1.73 m². In individuals with eGFR > 60 ml/min/1.73 m² and albuminuria <30 mg/day, the CKD273 subclassifiers predicted rapid eGFR loss with AUC ranging from 0.797 (0.743-0.844) to 0.736 (0.689-0.780). The association between CKD273 subclassifiers and rapid progression remained significant after adjustment for age, sex, albuminuria, DM, baseline eGFR, and systolic blood pressure. Urinary peptidomics CKD273 subclassifiers outperformed albuminuria and CKD273 classifier for predicting the risk of rapid CKD progression in individuals with eGFR > 60 ml/min/1.73 m². These CKD273 subclassifiers represented the earliest evidence of rapidly progressive CKD in non-albuminuric individuals with preserved renal function.

Personalized Approach and Precision Medicine in Supportive and End-of-Life Care for Patients With Advanced and End-Stage Kidney Disease

Kidney supportive care requires a highly personalized approach to care. Precision medicine holds promise for a deeper understanding of the pathophysiology of symptoms and related syndromes and more precise individualization of prognosis and treatment estimates, therefore providing valuable opportunities for greater personalization of supportive care. However, the major drivers of quality of life are psychosocial, economic, lifestyle, and preference-based, and consideration of these factors and skilled communication are integral to the provision of excellent and personalized kidney supportive care. This article discusses the concepts of personalized and precision medicine in the context of kidney supportive care and highlights some opportunities and limitations within these fields.

Clinical Proteomics: Closing the Gap from Discovery to Implementation

Clinical proteomics, the application of proteome analysis to serve a clinical purpose, represents a major field in the area of proteome research. Over 1000 manuscripts on this topic are published each year, with numbers continuously increasing. However, the anticipated outcome, the transformation of the reported findings into improvements in patient management, is not immediately evident. In this article, the value and validity of selected clinical proteomics findings are investigated, and it is assessed how far implementation has progressed. A main conclusion from this assessment is that to achieve implementation, well-powered clinical studies are required in the appropriate population, addressing a specific clinical need and with a clear context-of-use. Efforts toward implementation, to be feasible, must be supported by the key players in science: publishers and funders. The authors propose a change on objectives, from additional discovery studies toward studies aiming at validation of the plethora of potential biomarkers that have been described, to demonstrate practical value of clinical proteomics. All elements required, potential biomarkers, technologies, and bio-banked samples are available (based on today's literature), hence a change in focus from discovery toward validation and application is not only urgently necessary, but also possible based on resources available today.

Noninvasive diagnosis of chronic kidney diseases using urinary proteome analysis

Background

In spite of its invasive nature and risks, kidney biopsy is currently required for precise diagnosis of many chronic kidney diseases (CKDs). Here, we explored the hypothesis that analysis of the urinary proteome can discriminate different types of CKD irrespective of the underlying mechanism of disease.

Methods

We used data from the proteome analyses of 1180 urine samples from patients with different types of CKD, generated by capillary electrophoresis coupled to mass spectrometry. A set of 706 samples served as the discovery cohort, and 474 samples were used for independent validation. For each CKD type, peptide biomarkers were defined using statistical analysis adjusted for multiple testing. Potential biomarkers of statistical significance were combined in support vector machine (SVM)-based classifiers.

Results

For seven different types of CKD, several potential urinary biomarker peptides (ranging from 116 to 619 peptides) were defined and combined into SVM-based classifiers specific for each CKD. These classifiers were validated in an independent cohort and showed good to excellent accuracy for discrimination of one CKD type from the others (area under the receiver operating characteristic curve ranged from 0.77 to 0.95). Sequence analysis of the biomarkers provided further information that may clarify the underlying pathophysiology.

Conclusions

Our data indicate that urinary proteome analysis has the potential to identify various types of CKD defined by pathological assessment of renal biopsies and current clinical practice in general. Moreover, these approaches may provide information to model molecular changes per CKD.

Urinary proteomics predict onset of microalbuminuria in normoalbuminuric type 2 diabetic patients, a sub-study of the DIRECT-Protect 2 study

Background

Early prevention of diabetic nephropathy is not successful as early interventions have shown conflicting results, partly because of a lack of early and precise indicators of disease development. Urinary proteomics has shown promise in this regard and could identify those at high risk who might benefit from treatment. In this study we investigate its utility in a large type 2 diabetic cohort with normoalbuminuria.

Methods

We performed a post hoc analysis in the Diabetic Retinopathy Candesartan Trials (DIRECT-Protect 2 study), a multi centric randomized clinical controlled trial. Patients were allocated to candesartan or placebo, with the aim of slowing the progression of retinopathy. The secondary endpoint was development of persistent microalbuminuria (three of four samples). We used a previously defined chronic kidney disease risk score based on proteomic measurement of 273 urinary peptides (CKD273-classifier). A Cox regression model for the progression of albuminuria was developed and evaluated with integrated discrimination improvement (IDI), continuous net reclassification index (cNRI) and receiver operating characteristic curve statistics.

Results

Seven hundred and thirty-seven patients were analysed and 89 developed persistent microalbuminuria (12%) with a mean follow-up of 4.1 years. At baseline the CKD273-classifier predicted development of microalbuminuria during follow-up, independent of treatment (candesartan/placebo), age, gender, systolic blood pressure, urine albumin excretion rate, estimated glomerular filtration rate, HbA1c and diabetes duration, with hazard ratio 2.5 [95% confidence interval (CI) 1.4-4.3; P = 0.002] and area under the curve 0.79 (95% CI 0.75-0.84; P < 0.0001). The CKD273-classifier improved the risk prediction (relative IDI 14%, P = 0.002; cNRI 0.10, P = 0.043).

Conclusions

In this cohort of patients with type 2 diabetes and normoalbuminuria from a large intervention study, the CKD273-classifier was an independent predictor of microalbuminuria. This may help identify high-risk normoalbuminuric patients for preventive strategies for diabetic nephropathy.

New developments in transplant proteomics

PURPOSE OF REVIEW

Despite modern immunosuppression, renal allograft rejection remains a major contributor to graft loss. Novel biomarkers may help improve posttransplant outcomes through the early detection and treatment of rejection. Our objective is to provide an overview of proteomics, review recent discovery-based rejection studies, and explore innovative approaches in biomarker development.

RECENT FINDINGS

Urine MMP7 was identified as a biomarker of subclinical and clinical rejection using two-dimensional liquid chromatography tandem-mass spectrometry (LC-MS/MS) and improved the overall diagnostic discrimination of urine CXCL10 : Cr alone for renal allograft inflammation. A novel peptide signature to classify stable allografts from acute rejection, chronic allograft injury, and polyoma virus (BKV) nephropathy was identified using isobaric tag for relative and absolute quantitation (TRAQ) and label-free MS, with independent validation by selected reaction monitoring mass spectrometry (SRM-MS). Finally, an in-depth exploration of peripheral blood mononuclear cells identified differential proteoform expression in healthy transplants versus rejection.

SUMMARY

There is still much in the human proteome that remains to be explored, and further integration of renal, urinary, and exosomal data may offer deeper insight into the pathophysiology of rejection. Functional proteomics may be more biologically relevant than protein/peptide quantity alone, such as assessment of proteoforms or activity-based protein profiling. Discovery-based studies have identified potential biomarker candidates, but external validation studies are required.

Urinary CE-MS peptide marker pattern for detection of solid tumors

Urinary profiling datasets, previously acquired by capillary electrophoresis coupled to mass-spectrometry were investigated to identify a general urinary marker pattern for detection of solid tumors by targeting common systemic events associated with tumor-related inflammation. A total of 2,055 urinary profiles were analyzed, derived from a) a cancer group of patients (n = 969) with bladder, prostate, and pancreatic cancers, renal cell carcinoma, and cholangiocarcinoma and b) a control group of patients with benign diseases (n = 556), inflammatory diseases (n = 199) and healthy individuals (n = 331). Statistical analysis was conducted in a discovery set of 676 cancer cases and 744 controls. 193 peptides differing at statistically significant levels between cases and controls were selected and combined to a multi-dimensional marker pattern using support vector machine algorithms. Independent validation in a set of 635 patients (293 cancer cases and 342 controls) showed an AUC of 0.82. Inclusion of age as independent variable, significantly increased the AUC value to 0.85. Among the identified peptides were mucins, fibrinogen and collagen fragments. Further studies are planned to assess the pattern value to monitor patients for tumor recurrence. In this proof-of-concept study, a general tumor marker pattern was developed to detect cancer based on shared biomarkers, likely indicative of cancer-related features.

The effect of sodium/glucose cotransporter 2 (SGLT2) inhibition on the urinary proteome

Treatment with empagliflozin, an inhibitor of the sodium/glucose cotransporter 2 (SGLT2), is associated with slower progression of diabetic kidney disease. In this analysis, we explored the hypothesis that empagliflozin may have an impact on urinary peptides associated with chronic kidney disease (CKD). In this post-hoc, exploratory analysis, we investigated urine samples obtained from 40 patients with uncomplicated type 1 diabetes (T1D) before and after treatment with empagliflozin for 8 weeks to for significant post-therapy changes in urinary peptides. We further assessed the association of these changes with CKD in an independent cohort, and with a previously established urinary proteomic panel, termed CKD273. 107 individual peptides significantly changed after treatment. The majority of the empagliflozin-induced changes were in the direction of “CKD absent” when compare to patients with CKD and controls. A classifier consisting of these 107 peptides scored significantly different in controls, in comparison to CKD patients. However, empagliflozin did not impact the CKD273 classifier. Our data indicate that empagliflozin induces multiple significant changes in the urinary proteomic markers such as mucin and clusterin. The relationship between empagliflozin-induced proteomic changes and clinical outcomes merits further investigation.

Proteomics approaches in cervical cancer: focus on the discovery of biomarkers for diagnosis and drug treatment monitoring

INTRODUCTION

The HPV virus accounts for the majority of cervical cancer cases. Although a diagnostic tool (Pap Test) is widely available, cervical cancer incidence still remains high worldwide, and especially in developing countries, attributed to a large extent to suboptimal sensitivities of the Pap test and unavailability of the test in developing countries.

AREAS COVERED

Proteomics approaches have been used in order to understand the HPV virus correlation to cervical cancer pathology, as well as to discover putative biomarkers for early cervical cancer diagnosis and drug mode of action. Expert commentary: The present review summarizes the latest in vitro and in vivo proteomic studies for the discovery of putative cervical cancer biomarkers and the evaluation of available drugs and treatments.

A multivariate predictive modeling approach reveals a novel CSF peptide signature for both Alzheimer's Disease state classification and for predicting future disease progression

To determine if a multi-analyte cerebrospinal fluid (CSF) peptide signature can be used to differentiate Alzheimer’s Disease (AD) and normal aged controls (NL), and to determine if this signature can also predict progression from mild cognitive impairment (MCI) to AD, analysis of CSF samples was done on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset. The profiles of 320 peptides from baseline CSF samples of 287 subjects over a 3–6 year period were analyzed. As expected, the peptide most able to differentiate between AD vs. NL was found to be Apolipoprotein E. Other peptides, some of which are not classically associated with AD, such as heart fatty acid binding protein, and the neuronal pentraxin receptor, also differentiated disease states. A sixteen-analyte signature was identified which differentiated AD vs. NL with an area under the receiver operating characteristic curve of 0.89, which was better than any combination of amyloid beta (1–42), tau, and phospho-181 tau. This same signature, when applied to a new and independent data set, also strongly predicted both probability and rate of future progression of MCI subjects to AD, better than traditional markers. These data suggest that multivariate peptide signatures from CSF predict MCI to AD progression, and point to potentially new roles for certain proteins not typically associated with AD.

Urinary peptide-based classifier CKD273: towards clinical application in chronic kidney disease

Capillary electrophoresis coupled with mass spectrometry (CE-MS) has been used as a platform for discovery and validation of urinary peptides associated with chronic kidney disease (CKD). CKD affects ∼ 10% of the population, with high associated costs for treatments. A urinary proteome-based classifier (CKD273) has been discovered and validated in cross-sectional and longitudinal studies to assess and predict the progression of CKD. It has been implemented in studies employing cohorts of > 1000 patients. CKD273 is commercially available as an in vitro diagnostic test for early detection of CKD and is currently being used for patient stratification in a multicentre randomized clinical trial (PRIORITY). The validity of the CKD273 classifier has recently been evaluated applying the Oxford Evidence-Based Medicine and Southampton Oxford Retrieval Team guidelines and a letter of support for CKD273 was issued by the US Food and Drug Administration. In this article we review the current evidence published on CKD273 and the challenges associated with implementation. Definition of a possible surrogate early endpoint combined with CKD273 as a biomarker for patient stratification currently appears as the most promising strategy to enable the development of effective drugs to be used at an early time point when intervention can still be effective.

Metabolomics variable selection and classification in the presence of observations below the detection limit using an extension of ERp

Background

ERp is a variable selection and classification method for metabolomics data. ERp uses minimized classification error rates, based on data from a control and experimental group, to test the null hypothesis of no difference between the distributions of variables over the two groups. If the associated p-values are significant they indicate discriminatory variables (i.e. informative metabolites). The p-values are calculated assuming a common continuous strictly increasing cumulative distribution under the null hypothesis. This assumption is violated when zero-valued observations can occur with positive probability, a characteristic of GC-MS metabolomics data, disqualifying ERp in this context. This paper extends ERp to address two sources of zero-valued observations: (i) zeros reflecting the complete absence of a metabolite from a sample (true zeros); and (ii) zeros reflecting a measurement below the detection limit. This is achieved by allowing the null cumulative distribution function to take the form of a mixture between a jump at zero and a continuous strictly increasing function. The extended ERp approach is referred to as XERp.

Results

XERp is no longer non-parametric, but its null distributions depend only on one parameter, the true proportion of zeros. Under the null hypothesis this parameter can be estimated by the proportion of zeros in the available data. XERp is shown to perform well with regard to bias and power. To demonstrate the utility of XERp, it is applied to GC-MS data from a metabolomics study on tuberculosis meningitis in infants and children. We find that XERp is able to provide an informative shortlist of discriminatory variables, while attaining satisfactory classification accuracy for new subjects in a leave-one-out cross-validation context.

Conclusion

XERp takes into account the distributional structure of data with a probability mass at zero without requiring any knowledge of the detection limit of the metabolomics platform. XERp is able to identify variables that discriminate between two groups by simultaneously extracting information from the difference in the proportion of zeros and shifts in the distributions of the non-zero observations. XERp uses simple rules to classify new subjects and a weight pair to adjust for unequal sample sizes or sensitivity and specificity requirements.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-017-1480-8) contains supplementary material, which is available to authorized users.

Urinary proteomics can define distinct diagnostic inflammatory arthritis subgroups

Current diagnostic tests applied to inflammatory arthritis lack the necessary specificity to appropriately categorise patients. There is a need for novel approaches to classify patients with these conditions. Herein we explored whether urinary proteomic biomarkers specific for different forms of arthritis (rheumatoid arthritis (RA), psoriatic arthritis (PsA), osteoarthritis (OA)) or chronic inflammatory conditions (inflammatory bowel disease (IBD)) can be identified. Fifty subjects per group with RA, PsA, OA or IBD and 50 healthy controls were included in the study. Two-thirds of these populations were randomly selected to serve as a training set, while the remaining one-third was reserved for validation. Sequential comparison of one group to the other four enabled identification of multiple urinary peptides significantly associated with discrete pathological conditions. Classifiers for the five groups were developed and subsequently tested blind in the validation test set. Upon unblinding, the classifiers demonstrated excellent performance, with an area under the curve between 0.90 and 0.97 per group. Identification of the peptide markers pointed to dysregulation of collagen synthesis and inflammation, but also novel inflammatory markers. We conclude that urinary peptide signatures can reliably differentiate between chronic arthropathies and inflammatory conditions with discrete pathogenesis.

Comparison of different statistical approaches for urinary peptide biomarker detection in the context of coronary artery disease

Background

When combined with a clinical outcome variable, the size, complexity and nature of mass-spectrometry proteomics data impose great statistical challenges in the discovery of potential disease-associated biomarkers. The purpose of this study was thus to evaluate the effectiveness of different statistical methods applied for urinary proteomic biomarker discovery and different methods of classifier modelling in respect of the diagnosis of coronary artery disease in 197 study subjects and the prognostication of acute coronary syndromes in 368 study subjects.

Results

Computing the discovery sub-cohorts comprising \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {\scriptscriptstyle \raisebox{1ex}{$2$}\!\left/ \!\raisebox{-1ex}{$3$}\right.} $$\end{document}23 of the study subjects based on the Wilcoxon rank sum test, t-score, cat-score, binary discriminant analysis and random forests provided largely different numbers (ranging from 2 to 398) of potential peptide biomarkers. Moreover, these biomarker patterns showed very little overlap limited to fragments of type I and III collagens as the common denominator. However, these differences in biomarker patterns did mostly not translate into significant differently performing diagnostic or prognostic classifiers modelled by support vector machine, diagonal discriminant analysis, linear discriminant analysis, binary discriminant analysis and random forest. This was even true when different biomarker patterns were combined into master-patterns.

Conclusion

In conclusion, our study revealed a very considerable dependence of peptide biomarker discovery on statistical computing of urinary peptide profiles while the observed diagnostic and/or prognostic reliability of classifiers was widely independent of the modelling approach. This may however be due to the limited statistical power in classifier testing. Nonetheless, our study showed that urinary proteome analysis has the potential to provide valuable biomarkers for coronary artery disease mirroring especially alterations in the extracellular matrix. It further showed that for a comprehensive discovery of biomarkers and thus of pathological information, the results of different statistical methods may best be combined into a master pattern that then can be used for classifier modelling.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-016-1390-1) contains supplementary material, which is available to authorized users.

A capillary electrophoresis coupled to mass spectrometry pipeline for long term comparable assessment of the urinary metabolome

Although capillary electrophoresis coupled to mass spectrometry (CE-MS) has potential application in the field of metabolite profiling, very few studies actually used CE-MS to identify clinically useful body fluid metabolites. Here we present an optimized CE-MS setup and analysis pipeline to reproducibly explore the metabolite content of urine. We show that the use of a beveled tip capillary improves the sensitivity of detection over a flat tip. We also present a novel normalization procedure based on the use of endogenous stable urinary metabolites identified in the combined metabolome of 75 different urine samples from healthy and diseased individuals. This method allows a highly reproducible comparison of the same sample analyzed nearly 130 times over a range of 4 years. To demonstrate the use of this pipeline in clinical research we compared the urinary metabolome of 34 newborns with ureteropelvic junction (UPJ) obstruction and 15 healthy newborns. We identified 32 features with differential urinary abundance. Combination of the 32 compounds in a SVM classifier predicted with 76% sensitivity and 86% specificity UPJ obstruction in a separate validation cohort of 24 individuals. Thus, this study demonstrates the feasibility to use CE-MS as a tool for the identification of clinically relevant urinary metabolites.

A Bayesian algorithm for detecting differentially expressed proteins and its application in breast cancer research

Presence of considerable noise and missing data points make analysis of mass-spectrometry (MS) based proteomic data a challenging task. The missing values in MS data are caused by the inability of MS machines to reliably detect proteins whose abundances fall below the detection limit. We developed a Bayesian algorithm that exploits this knowledge and uses missing data points as a complementary source of information to the observed protein intensities in order to find differentially expressed proteins by analysing MS based proteomic data. We compared its accuracy with many other methods using several simulated datasets. It consistently outperformed other methods. We then used it to analyse proteomic screens of a breast cancer (BC) patient cohort. It revealed large differences between the proteomic landscapes of triple negative and Luminal A, which are the most and least aggressive types of BC. Unexpectedly, majority of these differences could be attributed to the direct transcriptional activity of only seven transcription factors some of which are known to be inactive in triple negative BC. We also identified two new proteins which significantly correlated with the survival of BC patients, and therefore may have potential diagnostic/prognostic values.