Serum diagnosis of diffuse large B-cell lymphomas and further identification of response to therapy using SELDI-TOF-MS and tree analysis patterning

Vitreous Humor Proteomic Profile in Patients With Vitreoretinal Lymphoma

Purpose

Vitreoretinal lymphoma is a high-grade malignant non-Hodgkin lymphoma with poor prognosis. The objective of this study was to elucidate the proteome profile of the vitreous in patients with vitreoretinal lymphoma (VRL), aiming to advance understanding of the pathophysiology of VRL.

Methods

Comprehensive proteomic analyses of vitreous humor using liquid chromatography with tandem mass spectrometry were performed for 10 patients with VRL, 10 control patients with idiopathic epiretinal membrane or macular hole, and 10 patients with ocular sarcoidosis. Differentially expressed proteins (DEPs) were identified by comparing VRL with controls and sarcoidosis, and functional pathway analysis was performed. Finally, vitreous concentrations of representative DEPs that were significantly upregulated in proteomics study were measured by ELISA using a separate cohort.

Results

In total, 1594 proteins were identified in the vitreous humor of VRL, control, and sarcoidosis samples. Also, 282 DEPs were detected in VRL, 249 upregulated and 33 downregulated, compared with controls. Enrichment pathway analysis showed alterations in proteasome-related pathways. Compared to controls and sarcoidosis, 14 DEPs in VRL showed significant upregulation. In the validation study, ELISA confirmed significantly higher vitreous concentrations of PSAT1, YWHAG, and 20S/26S proteasome complex in VRL compared with controls and sarcoidosis. Among the upregulated DEPs, vitreous PITHD1 and NCSTN concentrations correlated positively with vitreous IL-10 concentrations.

Conclusions

This study highlights aberrations in protein expression pattern in the vitreous of patients with VRL. The DEPs identified in this study may play pivotal roles in VRL pathogenesis, providing insights to enhance understanding of VRL pathophysiology and contribute to the development of VRL biomarkers.

Functional imaging in combination with mutation status aids prediction of response to inhibiting B-cell receptor signaling in lymphoma

Aberrant B-cell receptor (BCR) signaling is known to contribute to malignant transformation. Two small molecule inhibitors targeting BCR pathway signaling include ibrutinib, a Bruton’s tyrosine kinase (BTK) inhibitor, and idelalisib, a specific Phosphatidylinositol-4,5-bisphosphate 3-kinase delta (PI3Kδ) inhibitor, both of which have been approved for use in haematological malignancies. Despite the identification of various diffuse large B-cell lymphoma (DLBCL) subtypes, mutation status alone is not sufficient to predict patient response and therapeutic resistance can arise. Herein we apply early molecular imaging across alternative activated B-cell (ABC) and germinal center B-cell (GCB) DLBCL subtypes to investigate the effects of BCR pathway inhibition. Treatment with both inhibitors adversely affected cell growth and viability. These effects were partially predictable based upon mutation status. Accordingly, very early 2-deoxy-2-[18F]fluoro-D-glucose positron emission tomography (¹⁸F-FDG-PET) and 3’-deoxy-3’[18F]-fluorothymidine positron emission tomography (¹⁸F-FLT-PET) reported tumour regression and reductions in tumour metabolism and proliferation upon treatment. Furthermore, matrix-assisted laser desorption ionization imaging mass spectrometry (MALDI IMS) identified alterations in the proteome of a model of ABC DLBCL upon treatment with ibrutinib or idelalisib. In conclusion we demonstrate that very early molecular imaging adds predictive value in addition to mutational status of DLBCL that may be useful in directing patient therapy.

Determining protein biomarkers for DLBCL using FFPE tissues from HIV negative and HIV positive patients

DLBCL is the most common lymphoma subtype occurring in older populations as well as in younger HIV infected patients. The current treatment options for DLBCL are effective for most patients yet the relapse rate is high. While many biomarkers for DLBCL exist, they are not in clinical use due to low sensitivity and specificity. In addition, these biomarkers have not been studied in the HIV context. Therefore, the identification of new biomarkers for HIV negative and HIV positive DLBCL, may lead to a better understanding of the disease pathology and better therapeutic design. Protein biomarkers for DLBCL were determined using MALDI imaging mass spectrometry (IMS) and characterised using LC-MS. The expression of one of the biomarkers, heat shock protein (Hsp) 70, was confirmed on a separate cohort of samples using immunohistochemistry. The biomarkers identified in the study consisted of four protein clusters including glycolytic enzymes, ribosomal proteins, histones and collagen. These proteins could differentiate between control and tumour tissue, and the DLBCL immunohistochemical subtypes in both cohorts. The majority (41/52) of samples in the confirmation cohort were negative for Hsp70 expression. The HIV positive DLBCL cases had a higher percentage of cases expressing Hsp70 than their HIV negative counterparts. The non-GC subtype also frequently overexpressed Hsp70, confirming MALDI IMS data. The expression of Hsp70 did not correlate with survival in both the HIV negative and HIV positive cohort. This study identified potential biomarkers for HIV negative and HIV positive DLBCL from FFPE tissue sections. These may be used as diagnostic and prognostic markers complementary to current clinical management programmes for DLBCL.

Serum protein profiling in diffuse large B-cell lymphoma

PURPOSE

The aim of this pilot study was to conduct a nontargeted exploratory proteomics profiling analysis on sera obtained from patients diagnosed with diffuse large B-cell lymphoma (DLBCL) with the goal of identifying disease-specific biomarkers.

EXPERIMENTAL DESIGN

Sera from 87 participants (57 chemotherapy-naïve diffuse DLBCL patients, 30 controls frequency-matched by age group and World Health Organization (WHO) BMI categories) that were part of a large San Francisco Bay Area case-control study of non-Hodgkin lymphoma were analyzed by liquid chromatography combined with tandem mass spectrometry.

RESULTS

Thirty-five proteins (p-adjusted <0.05) were identified as differentially abundant between the DLBCL patients at various disease stages as compared to the controls. Of these, five proteins were randomly selected for further confirmation by ELISA: adiponectin (AdipoQ), cluster of differentiation 14 (CD14), heparin sulfate proteoglycan core protein (HSPG2), extracellular matrix 1 (ECM1), and alpha-1-antichymotrypsin (ACT). These proteins were statistically significantly elevated by 68.8, 37.0, 61.6, 68.0, and 32.0%, respectively, in DLBCL patient sera as compared to controls.

CONCLUSION AND CLINICAL RELEVANCE

These preliminary data when combined with other cancer-related data regarding these proteins warrant continued research in clinical and large prospective studies to clarify the role for these biomarkers in DLBCL pathogenesis and/or prognosis.

Machine learning for biomarker identification in cancer research - developments toward its clinical application

The patterns identified from the systematically collected molecular profiles of patient tumor samples, along with clinical metadata, can assist personalized treatments for effective management of cancer patients with similar molecular subtypes. There is an unmet need to develop computational algorithms for cancer diagnosis, prognosis and therapeutics that can identify complex patterns and help in classifications based on plethora of emerging cancer research outcomes in public domain. Machine learning, a branch of artificial intelligence, holds a great potential for pattern recognition in cryptic cancer datasets, as evident from recent literature survey. In this review, we focus on the current status of machine learning applications in cancer research, highlighting trends and analyzing major achievements, roadblocks and challenges toward its implementation in clinics.

Potential role of proteomics in the diagnosis of lymphoma: a meta-analysis

Surface-enhanced laser desorption/ionisation time-of-flight mass spectrometry (SELDI-TOF-MS) has been approved for identifying biomarkers and diagnosing many diseases such as lymphomas. It is arguable whether the SELDI technique has its value of diagnostic accuracy for lymphomas. The purpose of our study is to determine the diagnostic accuracy of SELDI-TOF-MS for diagnosing lymphomas. The Cochrane Central Register of Controlled Trials, MEDLINE, Pub Med, EMBASE, the Chinese Biomedical Literature Database, the China Academic Journals Full-text Database, and the Chinese Scientific Journals Database were searched systematically for potential studies. Reference lists of included studies and review articles were also reviewed. All studies that reported data on patients with a confirmed diagnosis of lymphomas and that compared the measurement of SELDI-TOF-MS with pathology standard were considered for inclusion. Eleven studies were included in the systematic review. The ranges of the diagnostic value of SELDI-TOF-MS for lymphoma were as follows: sensitivity (SEN) was 0.69-0.96; specificity (SPE) was 0.70-1.00; positive likelihood ratio (PLR) was 2.99-96.09; negative likelihood ratio (NLR) was 0.04-0.35; and diagnostic odds ratio (DOR) was 18.13-1250.71, respectively. Further, we analysed serum samples as a subgroup, and the pooled endpoints were as follows: pooled SEN was 0.89 (0.85-0.91); pooled SPE was 0.91 (0.88-0.93); pooled PLR was 12.35 (5.36-28.44); pooled NLR was 0.13 (0.09-0.20); and pooled DOR was 101.04 (39.57-258.04), respectively. SELDI-TOF-MS showed high accuracy in identifying lymphoma and could be a useful screening tool for diagnosing lymphoma patients.

Assessment of pre-diagnosis biomarkers of immune activation and inflammation: insights on the etiology of lymphoma

The DNA-modifying processes that are involved in B lymphocyte activation, somatic hypermutation (SHM), and IgH class switch recombination (CSR) have the potential to lead to genetic errors that lead to the genesis of B cell cancers, such as lymphoma. Given the potential contribution of these immune mechanisms to the development of cancer, assessment of the expression of cytokines, and other immune stimulatory molecules that drive B cell activation, prior to lymphoma diagnosis, may provide insights into the etiology of these cancers. Here, we review studies that have examined prediagnosis protein biomarkers for non-Hodgkin lymphoma (NHL), both AIDS-related NHL, as well as NHL seen in immunocompetent populations. Overall, these studies provide support for the notion that B cell hyper-activation is elevated preceding the appearance of AIDS-NHL, particularly those forms of AIDS-NHL that are not driven by EBV infection and that presumably arise from errors in IgH CSR and SHM. In more limited studies, it appears that dysregulation of cytokine production also precedes the diagnosis of NHL in HIV-negative persons. The availability of prediagnosis serum/plasma from cohort studies provides unique opportunities for proteomic approaches to identify novel prediagnosis etiologic biomarkers for NHL.

Transthyretin, identified by proteomics, is overabundant in pancreatic juice from pancreatic carcinoma and originates from pancreatic islets

Analyses of pancreatic juice by proteomics have identified many proteins that are overabundant in pancreatic cancer (PC) juice. The mechanism by which secretion of these proteins occur remains unclear. Pancreatic juice was collected from patients with three pancreatic diseases: PC, chronic pancreatitis (CP), and simple choledocholithiasis (CDS), and analyzed by 2-DE, MALDI-TOF/MS, and Western blot. Five PC cell lines, 30 PC tissues and their corresponding adjacent pancreatic tissues were used to validate the expression of genes which code for overabundant proteins in PC juice. The mRNA and protein levels were measured by RT-PCR and immunohistochemistry, respectively. Using proteomics, it was demonstrated that the protein transthyretin (TTR) was upregulated more than 2-fold in PC juice compared with CP and CDS, while apolipoprotein A-I, lithostathine, and regenerating islet-derived 1 beta precursor were downregulated more than 2-fold. Western blots confirmed that TTR was overabundant in the PC juice. However, TTR mRNA was not detected in any of the five PC cell lines, and was only detected in islet cells. By microscopy, it was shown that islet architecture was almost completely destroyed, and the islet's maximum diameter appeared larger in PC tissues than in normal. Some overabundant proteins in PC juice, such as TTR expressed only in islets, leak into the pancreatic ductal system due to hyperplasia and architectural damage in PC tissues. The destruction of organ and tissue architecture by tumor growth may result in novel tumor markers even if the markers are not secreted directly by tumor cells.

Protein profiling in pathology: analysis and evaluation of 239 frozen tissue biopsies for diagnosis of B-cell lymphomas

PURPOSE

We determined the potential value of protein profiling of tissue samples by assessing how precise this approach enables discrimination of B-cell lymphoma from reactive lymph nodes, and how well the profiles can be used for lymphoma classification.

EXPERIMENTAL DESIGN

Protein lysates from lymph nodes (n=239) from patients with the diagnosis of reactive hyperplasia (n=44), follicular lymphoma (n=63), diffuse large B-cell lymphoma (n=43), mantle cell lymphoma (n=47), and chronic lymphocytic leukemia/small lymphocytic B-cell lymphoma (n=42) were analysed by SELDI-TOF MS. Data analysis was performed by (i) classification and regression tree-based analysis and (ii) binary and polytomous logistic regression analysis.

RESULTS

After internal validation by the leave-one-out principle, both the classification and regression tree and logistic regression classification correctly identified the majority of the malignant (87 and 96%, respectively) and benign cases (73 and 75%, respectively). Classification was less successful since approximately one-third of the cases of each group were misclassified according to the histological classification. However, an additional mantle cell lymphoma case that was misclassified as chronic lymphocytic leukemia/small lymphocytic B-cell lymphoma initially was identified based on the protein profile.

CONCLUSIONS AND CLINICAL RELEVANCE

SELDI-TOF MS protein profiling allows for reliable identification of the majority of malignant lymphoma cases; however, further validation and testing robustness in a diagnostic setting is needed.

ESIprot: a universal tool for charge state determination and molecular weight calculation of proteins from electrospray ionization mass spectrometry data

Electrospray ionization (ESI) ion trap mass spectrometers with relatively low resolution are frequently used for the analysis of natural products and peptides. Although ESI spectra of multiply charged protein molecules also can be measured on this type of devices, only average spectra are produced for the majority of naturally occurring proteins. Evaluating such ESI protein spectra would provide valuable information about the native state of investigated proteins. However, no suitable and freely available software could be found which allows the charge state determination and molecular weight calculation of single proteins from average ESI-MS data. Therefore, an algorithm based on standard deviation optimization (scatter minimization) was implemented for the analysis of protein ESI-MS data. The resulting software ESIprot was tested with ESI-MS data of six intact reference proteins between 12.4 and 66.7 kDa. In all cases, the correct charge states could be determined. The obtained absolute mass errors were in a range between -0.2 and 1.2 Da, the relative errors below 30 ppm. The possible mass accuracy allows for valid conclusions about the actual condition of proteins. Moreover, the ESIprot algorithm demonstrates an extraordinary robustness and allows spectral interpretation from as little as two peaks, given sufficient quality of the provided m/z data, without the necessity for peak intensity data. ESIprot is independent from the raw data format and the computer platform, making it a versatile tool for mass spectrometrists. The program code was released under the open-source GPLv3 license to support future developments of mass spectrometry software.

Mass spectrometry based proteomics profiling as diagnostic tool in oncology: current status and future perspective

Proteomics analysis has been heralded as a novel tool for identifying new and specific biomarkers that may improve diagnosis and monitoring of various disease states. Recent years have brought a number of proteomics profiling technologies. Although proteomics profiling has resulted in the detection of disease-associated differences and modification of proteins, current proteomics technologies display certain limitations that are hampering the introduction of these new technologies into clinical laboratory diagnostics and routine applications. In this review, we summarize current advances in mass spectrometry based biomarker discovery. The promises and challenges of this new technology are discussed with particular emphasis on diagnostic perspectives of mass-spectrometry based proteomics profiling for malignant diseases.

When one chip is not enough: augmenting the validity of SELDI-TOF proteomic profiles of clinical specimens

To improve recovery, selectivity and reproducibility of SELDI-TOF analyses, we found it necessary to modify manufacturer's recommended protocols on sample and chip preparation. To yield reproducible denaturing conditions we verified concentrations of denaturing, reducing and lipid-solubilizing agents. We improved sorption of molecules of interest and reproducibility of analyses by introducing the preconditioning step and alkaline/acidic elutions for normal phase chips. The ratio that reproducibly decomposed the specimen was urea 9 mol l(-1) + DTT 10 mmol l(-1) + CHAPS 20 g l(-1). For sample denaturation, 100 microl of the fresh mixture was added to 100 microl of the specimen. Our modification of a chip processing increased recovery of the NP20 chip by up to 400% as assessed by total ion current. We obtained the range of mass accuracy of 0.02-0.04% and response precision between 30-40% of m/z+. We observed about 50% peak overlap. To obtain approximately 92% of possible peaks three chip selectivities, IMAC, H50 and normal phase with alkaline wash should be used. The selectivity of the SELDI chips is affected by unspecific interactions of a sample with a chip backbone. The system is compatible with matrix-based biological materials and does not suffer from urea interference and sensitivity to covalently bound alkaline ions. The technique is reasonably suitable for semiquantitative screening in the mammalian low-molecular weight cellular, tissue and plasma proteome.

Examining the significance of fingerprint-based classifiers

BACKGROUND

Experimental examinations of biofluids to measure concentrations of proteins or their fragments or metabolites are being explored as a means of early disease detection, distinguishing diseases with similar symptoms, and drug treatment efficacy. Many studies have produced classifiers with a high sensitivity and specificity, and it has been argued that accurate results necessarily imply some underlying biology-based features in the classifier. The simplest test of this conjecture is to examine datasets designed to contain no information with classifiers used in many published studies.

RESULTS

The classification accuracy of two fingerprint-based classifiers, a decision tree (DT) algorithm and a medoid classification algorithm (MCA), are examined. These methods are used to examine 30 artificial datasets that contain random concentration levels for 300 biomolecules. Each dataset contains between 30 and 300 Cases and Controls, and since the 300 observed concentrations are randomly generated, these datasets are constructed to contain no biological information. A modest search of decision trees containing at most seven decision nodes finds a large number of unique decision trees with an average sensitivity and specificity above 85% for datasets containing 60 Cases and 60 Controls or less, and for datasets with 90 Cases and 90 Controls many DTs have an average sensitivity and specificity above 80%. For even the largest dataset (300 Cases and 300 Controls) the MCA procedure finds several unique classifiers that have an average sensitivity and specificity above 88% using only six or seven features.

CONCLUSION

While it has been argued that accurate classification results must imply some biological basis for the separation of Cases from Controls, our results show that this is not necessarily true. The DT and MCA classifiers are sufficiently flexible and can produce good results from datasets that are specifically constructed to contain no information. This means that a chance fitting to the data is possible. All datasets used in this investigation are available on the web.