Machine learning pipeline to analyze clinical and proteomics data: experiences on a prostate cancer case.

Proteomic-based analysis is used to identify biomarkers in blood samples and tissues. Data produced by devices such as mass spectrometry requires platforms to identify and quantify proteins (or peptides). Clinical information can be related to mass spectrometry data to identify diseases at an early stage. Machine learning techniques can be used to support physicians and biologists in studying and classifying pathologies. We present the application of machine learning techniques to define a pipeline aimed at studying and classifying proteomics data enriched using clinical information. The pipeline allows users to relate established blood biomarkers with clinical parameters and proteomics data. The proposed pipeline entails three main phases: (i) feature selection, (ii) models training, and (iii) models ensembling. We report the experience of applying such a pipeline to prostate-related diseases. Models have been trained on several biological datasets. We report experimental results about two datasets that result from the integration of clinical and mass spectrometry-based data in the contexts of serum and urine analysis. The pipeline receives input data from blood analytes, tissue samples, proteomic analysis, and urine biomarkers. It then trains different models for feature selection, classification and voting. The presented pipeline has been applied on two datasets obtained in a 2 years research project which aimed to extract hidden information from mass spectrometry, serum, and urine samples from hundreds of patients. We report results on analyzing prostate datasets serum with 143 samples, including 79 PCa and 84 BPH patients, and an urine dataset with 121 samples, including 67 PCa and 54 BPH patients. As results pipeline allowed to identify interesting peptides in the two datasets, 6 for the first one and 2 for the second one. The best model for both serum (AUC=0.87, Accuracy=0.83, F1=0.81, Sensitivity=0.84, Specificity=0.81) and urine (AUC=0.88, Accuracy=0.83, F1=0.83, Sensitivity=0.85, Specificity=0.80) datasets showed good predictive performances. We made the pipeline code available on GitHub and we are confident that it will be successfully adopted in similar clinical setups.

Development of a predictive model to distinguish prostate cancer from benign prostatic hyperplasia by integrating serum glycoproteomics and clinical variables.

BACKGROUND: Prostate Cancer (PCa) represents the second leading cause of cancer-related death in men. Prostate-specific antigen (PSA) serum testing, currently used for PCa screening, lacks the necessary sensitivity and specificity. New non-invasive diagnostic tools able to discriminate tumoral from benign conditions and aggressive (AG-PCa) from indolent forms of PCa (NAG-PCa) are required to avoid unnecessary biopsies. METHODS: In this work, 32 formerly N-glycosylated peptides were quantified by PRM (parallel reaction monitoring) in 163 serum samples (79 from PCa patients and 84 from individuals affected by benign prostatic hyperplasia (BPH)) in two technical replicates. These potential biomarker candidates were prioritized through a multi-stage biomarker discovery pipeline articulated in: discovery, LC-PRM assay development and verification phases. Because of the well-established involvement of glycoproteins in cancer development and progression, the proteomic analysis was focused on glycoproteins enriched by TiO2 (titanium dioxide) strategy. RESULTS: Machine learning algorithms have been applied to the combined matrix comprising proteomic and clinical variables, resulting in a predictive model based on six proteomic variables (RNASE1, LAMP2, LUM, MASP1, NCAM1, GPLD1) and five clinical variables (prostate dimension, proPSA, free-PSA, total-PSA, free/total-PSA) able to distinguish PCa from BPH with an area under the Receiver Operating Characteristic (ROC) curve of 0.93. This model outperformed PSA alone which, on the same sample set, was able to discriminate PCa from BPH with an AUC of 0.79. To improve the clinical managing of PCa patients, an explorative small-scale analysis (79 samples) aimed at distinguishing AG-PCa from NAG-PCa was conducted. A predictor of PCa aggressiveness based on the combination of 7 proteomic variables (FCN3, LGALS3BP, AZU1, C6, LAMB1, CHL1, POSTN) and proPSA was developed (AUC of 0.69). CONCLUSIONS: To address the impelling need of more sensitive and specific serum diagnostic tests, a predictive model combining proteomic and clinical variables was developed. A preliminary evaluation to build a new tool able to discriminate aggressive presentations of PCa from tumors with benign behavior was exploited. This predictor displayed moderate performances, but no conclusions can be drawn due to the limited number of the sample cohort. Data are available via ProteomeXchange with identifier PXD035935.

Data-Independent Acquisition Mass Spectrometry of EPS-Urine Coupled to Machine Learning: A Predictive Model for Prostate Cancer.

Prostate cancer (PCa) is annually the most frequently diagnosed cancer in the male population. To date, the diagnostic path for PCa detection includes the dosage of serum prostate-specific antigen (PSA) and the digital rectal exam (DRE). However, PSA-based screening has insufficient specificity and sensitivity; besides, it cannot discriminate between the aggressive and indolent types of PCa. For this reason, the improvement of new clinical approaches and the discovery of new biomarkers are necessary. In this work, expressed prostatic secretion (EPS)-urine samples from PCa patients and benign prostatic hyperplasia (BPH) patients were analyzed with the aim of detecting differentially expressed proteins between the two analyzed groups. To map the urinary proteome, EPS-urine samples were analyzed by data-independent acquisition (DIA), a high-sensitivity method particularly suitable for detecting proteins at low abundance. Overall, in our analysis, 2615 proteins were identified in 133 EPS-urine specimens obtaining the highest proteomic coverage for this type of sample; of these 2615 proteins, 1670 were consistently identified across the entire data set. The matrix containing the quantified proteins in each patient was integrated with clinical parameters such as the PSA level and gland size, and the complete matrix was analyzed by machine learning algorithms (by exploiting 90% of samples for training/testing using a 10-fold cross-validation approach, and 10% of samples for validation). The best predictive model was based on the following components: semaphorin-7A (sema7A), secreted protein acidic and rich in cysteine (SPARC), FT ratio, and prostate gland size. The classifier could predict disease conditions (BPH, PCa) correctly in 83% of samples in the validation set. Data are available via ProteomeXchange with the identifier PXD035942.

Bioengineering and medical informatics education in MD programs: perspectives from three Italian experiences.

BACKGROUND: Given the impact of bioengineering and medical informatics technologies in health care, the design and implementation of education programs able to combine medical curricula with a proper teaching on engineering and informatics is now of paramount importance. In Italy, this goal has to fit in with the existing higher education system, which is structured into Bachelor programs and Master programs. Medicine and Surgery programs, instead, are designed as a six-year single-cycle Degree Program in Medicine and Surgery which comprises both class attendance and hospital internship and training. This program allows students to become Medical Doctors (MD). The different organization of this University program makes it not easy to introduce further contents, namely hard science courses, in the educational program. Notwithstanding this, we present here some recent innovative programs aimed at widening MD curriculum by including biomedical engineering and informatics subjects. In particular, we will introduce three of them. Two are joint-degree programs, the first between Humanitas University and Politecnico di Milano (MEDTEC School), and the second between University of Calabria and University Magna Graecia of Catanzaro (Medicina e Chirurgia TD). The Third one is a Professional Master coupled with an MD degree, based on a joint program among Pavia University, Pisa University, the Institute of Advanced studies in Pavia and the Scuola Superiore S. Anna in Pisa (MEET). CONTRIBUTION: The paper provides a description of the fundamental design principles of the three above mentioned programs, and explores some aspects of the teaching modules, highlighting their positive aspects. In particular, we show how the three different programs allow students to enrich their knowledge by studying engineering subjects and innovative methods and technologies, as well as their applications to patient care. CONCLUSIONS: The MEDTEC program is the first degree program at Italian and international scale which integrates medical and engineering subjects. In the following years, other programs were issued in Italy, defining similar education programs to couple a degree in medicine education with bioengineering and medical informatics, among which Medicina e Chirurgia TD and MEET. We believe the experiences described here in this paper represent the possibility of bridging the gap between medical and technological competencies.

Impaired insulin-stimulated myocardial glucose metabolic rate is associated with reduced estimated myocardial energetic efficiency in subjects with different degrees of glucose tolerance.

BACKGROUND: Alterations in myocardial mechano-energetic efficiency (MEEi), which represents the capability of the left ventricles to convert the chemical energy obtained by oxidative metabolism into mechanical work, have been associated with cardiovascular disease. Although whole-body insulin resistance has been related to impaired myocardial MEEi, it is unknown the relationship between cardiac insulin resistance and MEEi. Aim of this study was to evaluate the relationship between insulin-stimulated myocardial glucose metabolic rate (MrGlu) and myocardial MEEi in subjects having different degrees of glucose tolerance. METHODS: We evaluated insulin-stimulated myocardial MrGlu using cardiac dynamic positron emission tomography (PET) with 18F-Fluorodeoxyglucose (18F-FDG) combined with euglycemic-hyperinsulinemic clamp, and myocardial MEEi in 57 individuals without history of coronary heart disease having different degrees of glucose tolerance. The subjects were stratified into tertiles according to their myocardial MrGlu values. RESULTS: After adjusting for age, gender and BMI, subjects in I tertile showed a decrease in myocardial MEEi (0.31 ± 0.05 vs 0.42 ± 0.14 ml/s*g, P = 0.02), and an increase in myocardial oxygen consumption (MVO2) (10,153 ± 1375 vs 7816 ± 1229 mmHg*bpm, P < 0.0001) as compared with subjects in III tertile. Univariate correlations showed that insulin-stimulated myocardial MrGlu was positively correlated with MEEi and whole-body glucose disposal, and negatively correlated with waist circumference, fasting plasma glucose, HbA1c and MVO2. In a multivariate regression analysis running a model including several CV risk factors, the only variable that remained significantly associated with MEEi was myocardial MrGlu (ß 0.346; P = 0.01). CONCLUSIONS: These data suggest that an impairment in insulin-stimulated myocardial glucose metabolism is an independent contributor of depressed myocardial MEEi in subjects without history of CHD.

Detection of pan-cancer surface protein biomarkers via a network-based approach on transcriptomics data.

Cell surface proteins have been used as diagnostic and prognostic markers in cancer research and as targets for the development of anticancer agents. Many of these proteins lie at the top of signaling cascades regulating cell responses and gene expression, therefore acting as 'signaling hubs'. It has been previously demonstrated that the integrated network analysis on transcriptomic data is able to infer cell surface protein activity in breast cancer. Such an approach has been implemented in a publicly available method called 'SURFACER'. SURFACER implements a network-based analysis of transcriptomic data focusing on the overall activity of curated surface proteins, with the final aim to identify those proteins driving major phenotypic changes at a network level, named surface signaling hubs. Here, we show the ability of SURFACER to discover relevant knowledge within and across cancer datasets. We also show how different cancers can be stratified in surface-activity-specific groups. Our strategy may identify cancer-wide markers to design targeted therapies and biomarker-based diagnostic approaches.

Atrial myxomas arise from multipotent cardiac stem cells.

AIMS: Cardiac myxomas usually develop in the atria and consist of an acid-mucopolysaccharide-rich myxoid matrix with polygonal stromal cells scattered throughout. These human benign tumours are a valuable research model because of the rarity of cardiac tumours, their clinical presentation and uncertain origin. Here, we assessed whether multipotent cardiac stem/progenitor cells (CSCs) give rise to atrial myxoma tissue. METHODS AND RESULTS: Twenty-three myxomas were collected and analysed for the presence of multipotent CSCs. We detected myxoma cells positive for c-kit (c-kitpos) but very rare Isl-1 positive cells. Most of the c-kitpos cells were blood lineage-committed CD45pos/CD31pos cells. However, c-kitpos/CD45neg/CD31neg cardiac myxoma cells expressed stemness and cardiac progenitor cell transcription factors. Approximately ≤10% of the c-kitpos/CD45neg/CD31neg myxoma cells also expressed calretinin, a characteristic of myxoma stromal cells. In vitro, the c-kitpos/CD45neg/CD31neg myxoma cells secrete chondroitin-6-sulfate and hyaluronic acid, which are the main components of gelatinous myxoma matrix in vivo. In vitro, c-kitpos/CD45neg/CD31neg myxoma cells have stem cell properties being clonogenic, self-renewing, and sphere forming while exhibiting an abortive cardiac differentiation potential. Myxoma-derived CSCs possess a mRNA and microRNA transcriptome overall similar to normal myocardium-derived c-kitpos/CD45neg/CD31negCSCs , yet showing a relatively small and relevant fraction of dysregulated mRNA/miRNAs (miR-126-3p and miR-335-5p, in particular). Importantly, myxoma-derived CSCs but not normal myocardium-derived CSCs, seed human myxoma tumours in xenograft's in immunodeficient NOD/SCID mice. CONCLUSION: Myxoma-derived c-kitpos/CD45neg/CD31neg CSCs fulfill the criteria expected of atrial myxoma-initiating stem cells. The transcriptome of these cells indicates that they belong to or are derived from the same lineage as the atrial multipotent c-kitpos/CD45neg/CD31neg CSCs. Taken together the data presented here suggest that human myxomas could be the first-described CSC-related human heart disease.

Melanoma Detection by Means of Multiple Instance Learning.

We present an application to melanoma detection of a multiple instance learning (MIL) approach, whose objective, in the binary case, is to discriminate between positive and negative sets of items. In the MIL terminology these sets are called bags and the items inside the bags are called instances. Under the hypothesis that a bag is positive if at least one of its instances is positive and it is negative if all its instances are negative, the MIL paradigm fits very well with images classification, since an image (bag) is in general classified on the basis of some its subregions (instances). In this work we have applied a MIL algorithm on some clinical data constituted by color dermoscopic images, with the aim to discriminate between melanomas (positive images) and common nevi (negative images). In comparison with standard classification approaches, such as the well known support vector machine, our method performs very well in terms both of accuracy and sensitivity. In particular, using a leave-one-out validation on a data set constituted by 80 melanomas and 80 common nevi, we have obtained the following results: accuracy = 92.50%, sensitivity = 97.50% and specificity = 87.50%. Since the results appear promising, we conclude that a MIL technique could be at the basis of more sophisticated tools useful to physicians in melanoma detection.

H-Ferritin Affects Cisplatin-Induced Cytotoxicity in Ovarian Cancer Cells through the Modulation of ROS.

Reactive oxygen species (ROS) mediates cisplatin-induced cytotoxicity in tumor cells. However, when cisplatin-induced ROS do not reach cytotoxic levels, cancer cells may develop chemoresistance. This phenomenon can be attributed to the inherited high expression of antioxidant protein network. H-Ferritin is an important member of the antioxidant system due to its ability to store iron in a nontoxic form. Altered expression of H-Ferritin has been described in ovarian cancers; however, its functional role in cisplatin-based chemoresistance of this cancer type has never been explored. Here, we investigated whether the modulation of H-Ferritin might affect cisplatin-induced cytotoxicity in ovarian cancer cells. First, we characterized OVCAR3 and OVCAR8 cells for their relative ROS and H-Ferritin baseline amounts. OVCAR3 exhibited lower ROS levels compared to OVCAR8 and greater expression of H-Ferritin. In addition, OVCAR3 showed pronounced growth potential and survival accompanied by the strong activation of pERK/pAKT and overexpression of c-Myc and cyclin E1. When exposed to different concentrations of cisplatin, OVCAR3 were less sensitive than OVCAR8. At the lowest concentration of cisplatin (6 µM), OVCAR8 underwent a consistent apoptosis along with a downregulation of H-Ferritin and a consistent increase of ROS levels; on the other hand, OVCAR3 cells were totally unresponsive, H-Ferritin was almost unaffected, and ROS amounts met a slight increase. Thus, we assessed whether the modulation of H-Ferritin levels was able to affect the cisplatin-mediated cytotoxicity in both the cell lines. H-Ferritin knockdown strengthened cisplatin-mediated ROS increase and significantly restored sensitivity to 6 µM cisplatin in resistant OVCAR3 cells. Conversely, forced overexpression of H-Ferritin significantly suppressed the cisplatin-mediated elevation of intracellular ROS subsequently leading to a reduced responsiveness in OVCAR8 cells. Overall, our findings suggest that H-Ferritin might be a key protein in cisplatin-based chemoresistance and that its inhibition may represent a potential approach for enhancing cisplatin sensitivity of resistant ovarian cancer cells.

On the reliability of measurements for a stent positioning simulation system.

BACKGROUND AND OBJECTIVE: Computer aided simulations are useful to support the physician in many steps of the surgical activity, but also in pre-surgical patient classification and in post-surgical diagnosis and treatment decisions. At a broader level, computerized technologies and infrastructures permeate every aspect of the medical activity, from patient management to surgery and patients' follow up with outcomes analyses. Radiography assisted surgery is often used in hemodynamic surgery to study and support cardio-circulatory stents positioning with the use of radioscopy coupled with contrast liquid injected into the vessels. Computer based surgery instruments (both software and hardware) are used to support clinicians during interventions, e.g., to reduce radioscopy time exposure, to minimize errors and to estimate tissues and organs dimension. In this paper we present the use of a newly developed system which supports physicians during transcatheter percutaneous coronary interventions. METHODS: This paper presents a Java-based tool which acquires images from angiographic equipment during surgery procedures. An high performance image acquisition module has been used and a stent simulation environment module is available to simulate stent positioning and to measure vessels. Operators may acquire images, perform measurements and simulations on DICOM images. We performed tests off-line on images to validate the reliability of the tool. Real cases and on line tests have been performed by operators showing the robustness of the system to be used in surgery room. The system has been integrated in the surgery room control panel and allows (i) vascular images acquisition, (ii) vessels and coronary measurement and (iii) stent positioning simulations. The tool is an aid for the physician for both measuring tissues or lesions and for defining the stent's geometry and position before its deployment in the patient's vessels. RESULTS: Experiments have been performed on lesions and vessels by different operators using the system and an available commercial system, on both real patient cases and synthetic images designed with a CAD. It has been tested on 76 images extracted from real angiography cases and on 11 synthetic images created by using CAD. Five different operators performed 2128 measurements for the real cases images (for both Cartesio and CAAS tools) and 112 for the synthetic dataset. Results show the efficacy of the system compared with the commercial one by means of several statistical tests. CONCLUSIONS: The proposed system is a reliable tool for hemodynamic surgery and can be used both for decision support in stent positioning procedures and for didactic training of new physicians.

An Innovative Framework for Bioimage Annotation and Studies.

The collection and analysis of clinical data are needed to investigate diseases and to define medical protocols and treatments. Bioimages, medical annotations and patient history are clinical data acquired and studied to perform a correct diagnosis and to propose an appropriate therapy. Currently, hospital departments manage these data using legacy systems which do not often allow data integration among different departments or health structures. Thus, in many cases clinical information sharing and exchange are difficult to implement. This is also the case for biomedical images for which data integration or data overlapping is usually not available. Image annotations and comparison can be crucial for physicians in many case studies. In this paper, a general purpose framework for bioimage management and annotations is proposed. Moreover, a simple-to-use information system has been developed to integrate clinical and diagnosis codes. The framework allows physicians (1) to integrate DICOM images from different platforms and (2) to report notes and highlights directly on images, thus offering, among the others, to query and compare similar clinical cases. This contribution is the result of a framework aimed to support oncologists in managing DICOM images and clinical data from different departments. Data integration is performed using a here-proposed XML-based module also utilized to trace temporal changes in image annotations.

Adult cardiac stem cells are multipotent and robustly myogenic: c-kit expression is necessary but not sufficient for their identification.

Multipotent adult resident cardiac stem cells (CSCs) were first identified by the expression of c-kit, the stem cell factor receptor. However, in the adult myocardium c-kit alone cannot distinguish CSCs from other c-kit-expressing (c-kitpos) cells. The adult heart indeed contains a heterogeneous mixture of c-kitpos cells, mainly composed of mast and endothelial/progenitor cells. This heterogeneity of cardiac c-kitpos cells has generated confusion and controversy about the existence and role of CSCs in the adult heart. Here, to unravel CSC identity within the heterogeneous c-kit-expressing cardiac cell population, c-kitpos cardiac cells were separated through CD45-positive or -negative sorting followed by c-kitpos sorting. The blood/endothelial lineage-committed (Lineagepos) CD45posc-kitpos cardiac cells were compared to CD45neg(Lineageneg/Linneg) c-kitpos cardiac cells for stemness and myogenic properties in vitro and in vivo. The majority (~90%) of the resident c-kitpos cardiac cells are blood/endothelial lineage-committed CD45posCD31posc-kitpos cells. In contrast, the LinnegCD45negc-kitpos cardiac cell cohort, which represents ⩽10% of the total c-kitpos cells, contain all the cardiac cells with the properties of adult multipotent CSCs. These characteristics are absent from the c-kitneg and the blood/endothelial lineage-committed c-kitpos cardiac cells. Single Linnegc-kitpos cell-derived clones, which represent only 1-2% of total c-kitpos myocardial cells, when stimulated with TGF-ß/Wnt molecules, acquire full transcriptome and protein expression, sarcomere organisation, spontaneous contraction and electrophysiological properties of differentiated cardiomyocytes (CMs). Genetically tagged cloned progeny of one Linnegc-kitpos cell when injected into the infarcted myocardium, results in significant regeneration of new CMs, arterioles and capillaries, derived from the injected cells. The CSC's myogenic regenerative capacity is dependent on commitment to the CM lineage through activation of the SMAD2 pathway. Such regeneration was not apparent when blood/endothelial lineage-committed c-kitpos cardiac cells were injected. Thus, among the cardiac c-kitpos cell cohort only a very small fraction has the phenotype and the differentiation/regenerative potential characteristics of true multipotent CSCs.

Microfluidic device for continuous single cells analysis via Raman spectroscopy enhanced by integrated plasmonic nanodimers.

In this work a Raman flow cytometer is presented. It consists of a microfluidic device that takes advantages of the basic principles of Raman spectroscopy and flow cytometry. The microfluidic device integrates calibrated microfluidic channels- where the cells can flow one-by-one -, allowing single cell Raman analysis. The microfluidic channel integrates plasmonic nanodimers in a fluidic trapping region. In this way it is possible to perform Enhanced Raman Spectroscopy on single cell. These allow a label-free analysis, providing information about the biochemical content of membrane and cytoplasm of the each cell. Experiments are performed on red blood cells (RBCs), peripheral blood lymphocytes (PBLs) and myelogenous leukemia tumor cells (K562).

A proteomics approach to identify changes in protein profiles in serum of Familial Adenomatous Polyposis patients.

Familial adenomatous polyposis (FAP) is one of the most important clinical hereditary forms of inherited susceptibility to colorectal cancer and is characterized by a high degree of phenotypic heterogeneity. We used a mass spectrometry driven-proteomic strategy to identify serum molecules differently expressed in FAP patients. The data obtained were subsequently processed by bioinformatic analysis and confirmed by Western blotting. Significant differences were highlighted in the expression of serum proteins of FAP patients. In particular, two proteins (alpha-2-HS-glycoprotein and apoliprotein D) were down-regulated (about 0.5- and 0.7-fold, respectively) in carpeting versus diffuse FAP patients and healthy donors, while alpha-2-antiplasmin was up-regulated (about 1.4-fold). Moreover, mass spectrometry approach enabled us to identify serum biomarkers specific for two distinct clinical form of FAP, i.e. carpeting and diffuse FAP. In particular, vitronectin was up-regulated (more than 1.4-fold) in diffuse FAP patients versus carpeting FAP and versus healthy donors, and two additional proteins (Haptoglobin and alpha-1-acid glycoprotein 1) were up-regulated in 2 out of 3 carpeting FAP patients. Our study suggests that mass spectrometry combined to a strong bioinformatics analysis is a valuable tool for the identification of quali/quantitative differences in the serum proteome of otherwise indistinguishable FAP phenotypes. Moreover, the definition of a proteomic profile, supported by the supervised classification, is a powerful and highly sensitive approach for the identification molecular signatures that are able to outperform the traditional disease markers and can therefore be efficiently applied for the diagnosis and clinical management of FAP patients.

The EIPeptiDi tool: enhancing peptide discovery in ICAT-based LC MS/MS experiments.

BACKGROUND: Isotope-coded affinity tags (ICAT) is a method for quantitative proteomics based on differential isotopic labeling, sample digestion and mass spectrometry (MS). The method allows the identification and relative quantification of proteins present in two samples and consists of the following phases. First, cysteine residues are either labeled using the ICAT Light or ICAT Heavy reagent (having identical chemical properties but different masses). Then, after whole sample digestion, the labeled peptides are captured selectively using the biotin tag contained in both ICAT reagents. Finally, the simplified peptide mixture is analyzed by nanoscale liquid chromatography-tandem mass spectrometry (LC-MS/MS). Nevertheless, the ICAT LC-MS/MS method still suffers from insufficient sample-to-sample reproducibility on peptide identification. In particular, the number and the type of peptides identified in different experiments can vary considerably and, thus, the statistical (comparative) analysis of sample sets is very challenging. Low information overlap at the peptide and, consequently, at the protein level, is very detrimental in situations where the number of samples to be analyzed is high. RESULTS: We designed a method for improving the data processing and peptide identification in sample sets subjected to ICAT labeling and LC-MS/MS analysis, based on cross validating MS/MS results. Such a method has been implemented in a tool, called EIPeptiDi, which boosts the ICAT data analysis software improving peptide identification throughout the input data set. Heavy/Light (H/L) pairs quantified but not identified by the MS/MS routine, are assigned to peptide sequences identified in other samples, by using similarity criteria based on chromatographic retention time and Heavy/Light mass attributes. EIPeptiDi significantly improves the number of identified peptides per sample, proving that the proposed method has a considerable impact on the protein identification process and, consequently, on the amount of potentially critical information in clinical studies. The EIPeptiDi tool is available at http://bioingegneria.unicz.it/~veltri/projects/eipeptidi/ with a demo data set. CONCLUSION: EIPeptiDi significantly increases the number of peptides identified and quantified in analyzed samples, thus reducing the number of unassigned H/L pairs and allowing a better comparative analysis of sample data sets.