Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

27.1 Introduction

Major depressive disorder (MDD) occurs frequently, affecting 7.1 % people each year and 14.4 % of people over the course of a lifetime [1]. According to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), the diagnosis of MDD requires at least five symptoms, present for at least 2 weeks; among the five symptoms at least one should be depressed mood and/or loss of interest or pleasure [2]. Other symptoms include change in sleep, appetite, fatigue/energy loss, feelings of worthlessness or guilt, diminished concentration, and suicidal thoughts [2, 3]. In addition to suffering from impairment in normal functioning, patients with MDD experience increased rates of other comorbid medical illnesses, including diabetes, arthritis, cardiovascular disease as well as comorbid psychiatric disorders [4–10].

Effective treatments for MDD are greatly needed, since MDD is very common and associated with high costs of health care as well as high morbidity and mortality [8, 9, 11–13]. Unfortunately, partial or suboptimal response to treatment occurs commonly in MDD, and only 30–40 % of patients receiving treatment will fully remit (achieving near-complete resolution of symptoms) [14–17]. The Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study is one of the largest and more comprehensive studies of MDD treatments ever conducted. Responses to up to four successive treatment steps were studied in individuals with MDD. Remission was achieved in less than one third of patients receiving first- and second-line treatments; rates of remission were significantly lower in patients with treatment-resistant depression (TRD, i.e., those who have had failed two or more consecutive treatments) [14, 15]. The cumulative rate of remission was 67 % after all four levels. These sobering results indicate that the current treatment options are not very effective; even when effective treatment effects are very slow. Each level of treatment requires 12 weeks; a full year of treatments may be required in order to achieve significant positive outcomes (remission) in two thirds of the patients. This problem of treatment-resistant depression has led to the continued search for novel, more effective antidepressant treatments [18–25].

Given the current only partially effective antidepressant treatments and the imperfect, trial-and-error methods for treatment selection in MDD, predictive biomarkers could be particularly useful. Ideally, biomarkers predictive of treatment response to specific therapies could eliminate multiple and lengthy treatment steps by selecting the right treatment from the start.

The use of biomarkers to predict or identify a disease or disorder is an essential strategy in many areas of medicine, and is increasingly studied in psychiatric research. The United States Food and Drug Administration (FDA) defines a biomarker as an objective measurement of a normal biological process, a pathological, biological process or an objective measurement that indicates response to a therapeutic [26]. Biomarkers are commonly used for diseases such as cancer [27–29], however, no biomarker is routinely used clinically in psychiatry, despite the assumption that biological changes underlie or contribute to many psychiatric problems [30, 31]. Biomarkers could have several functions in psychiatry, including clarification of the etiology of psychiatric problems, and could be used as predictors of therapeutic response [32].

Many different biological parameters may potentially serve as biomarkers, such as specific genetic or epigenetic abnormalities, metabolites, proteins, or brain activity patterns. For example, specific changes in theta power on prefrontal leads in electroencephalograms (EEG) from baseline to week 1 of treatment have been used to predict antidepressant treatment outcomes at week 8 [33]. Moreover, EEG data has been shown to be specific enough to guide clinical treatment in patients with MDD [34]. Both resting-state EEGs as well as evoked potentials have been explored as biomarkers predictive of antidepressant treatment response [35]. EEGs may provide a useful tool for aiding in antidepressant therapy however access to the technology and imperfect predictive ability have limited its development so far. It is likely that, similar to other areas of medicine, biomarkers useful in psychiatry will involve a composite of clinical and biological factors [35]; proteins found in human biomaterials could play a significant role.

Brain neuroanatomical changes, as measured by MR morphometry, diffusion tensor imaging (DTI), or even functional MRI may also provide depression biomarkers. For example, DTI studies have suggested that white matter abnormalities may be present in treatment-resistant depression [36]. DTI has also been used to indicate that MDD may be associated with abnormal microstructure in brain reward/aversion regions, specifically the ventral tegmental area and dorsolateral prefrontal white matter [37]. While neuroimaging provides the potential to understand the impairment in neurocircuitry underlying TRD, it is unclear whether such results are specific enough to become useful biomarkers for diagnosis or treatment selection. Brain imaging techniques are also not easily accessible to all clinicians and expensive. Such instruments tend to be limited to major medical centers. Therefore, additional biomarkers, more easily deployed in the general population and based on blood or saliva samples that can be easily analyzed in remote labs, warrant exploration.

Genomics has witnessed a major search for biomarkers in psychiatry and candidate genomic biomarkers have been found for several psychiatric conditions, most notably, serotonin transport protein polymorphisms as predictors of depression and anxiety [38–41]. Specific gene mutations may interact with environmental factors to ultimately increase the risk of psychiatric problems [42–45]. However, despite the initial promise of genetic information, a recent mega-analysis of genome-association studies in MDD failed to find consistent genetic associations [46]. The search for gene mutations as psychiatric biomarkers may be flawed, since genomic information does not necessarily reflect active protein levels [47] or tell the researcher about possible important posttranslational modifications (PTMs) such as glycosylation, phosphorylation, or formation/destruction of disulfide bridges to keep/disturb the protein’s three-dimensional structure [48]. Epigenetics, studying the transcription status of genes, is probably a better reflection of temporal changes in gene expression; but the study of epigenetics with respect to biomarkers in MDD is still in its infancy. Additional proteomic information may expand possibilities for a disorder’s identification in clinical tests. Proteins represent the functional molecules in a biological system; therefore, study of proteins may take a researcher closer to identifying the cause of a disorder and could also suggest targets for therapeutics. Protein profiling of candidate biomarkers in psychiatry is therefore an area of/with great potential [49, 50]. Mass spectrometry (MS) is currently the technique of choice for proteomic studies [32, 49–73].

A recent PubMed search of depression and biomarkers reveals the extent to which different biomarker approaches are emphasized in the field of depression research. Using the search terms “depression biomarker” and adding a variable third term, yielded the following results: genomic = 306 articles, EEG = 112 articles, neuroimaging = 103 articles, proteomic = 34 articles, mass spectrometry = 40 articles (Fig. 27.1). This search underscores the need for expansion into the proteomic realm in the field of depression research.

Fig. 27.1
figure 1

PubMed search results for “depression biomarker” plus genomic, electroencephalogram (EEG), neuroimaging, proteomics, or mass spectrometry

Full size image

MS-based proteomics is very useful for protein biomarker identification for several reasons. Using a variety of proteomics approaches, a researcher can detect and determine multiple properties of a specific sample [55, 56, 61, 63, 65–67]. Virtually all information about a protein/peptide such as mass, sequence information, and the charge state can be measured and identified using MS-based proteomic methods [66, 69, 70, 72, 74, 75]. Knowledge-based methods for protein biomarker discovery can also be employed, such as enzyme-linked immunoabsorbant assay (ELISA) [76], Western Blotting [72, 77], or immunohistochemistry [72, 78, 79]. These techniques can be used independently, with the downside being that the “discovery” aspect is negated by the need to identify a target protein. They may also be used to validate and expand upon the more comprehensive screening provided by MS. A general proteomics approach is presented in Fig. 27.2. Here the protein samples from various sources are either fractionated, digested, and then analyzed by MS (liquid chromatography-tandem mass spectrometry), followed by data analysis, validation, and follow up, or analyzed without a priori digestion (i.e., by matrix-assisted laser desorption ionization MS or MALDI-MS) followed by data analysis, verification, and follow up. Many times, when the expertise and instrumentation are available, both methods are used simultaneously and their outcome complements each other.

Fig. 27.2
figure 2

Schematic of a full proteomics experiment. The protein sample is separated/fractionated under non-denaturing (1) and denaturing (2) conditions, digested (3) and analyzed by LC-MS/MS (4) (bottom-up proteomics). Database search and data analysis (5) led to identification of proteins, post-translational modifications (PTMs) in proteins and protein–protein interactions (PPIs), that were verified and then validated by Western blotting (6), and then monitored by single reaction monitoring (SRM), multiple reaction monitoring (MRM), and/or cross-validation. The original protein samples can also be analyzed by MALDI-MS (8) that leads to identification of specific protein/peptide patterns whose data analysis (10) leads to identification of proteins, PTMs, and PPIs. The fractionated samples in (2) can also be investigated without using electrophoretic denaturing conditions (2A) in a process called in solution digestion, or analyzed without any digestion (2B) in a process called top-down proteomics. In a third alternative, the fractionated samples from (2) are analyzed without digestion directly by MALDI-MS (2C). © 2013 Wormwood KL, et al. Reproduced from Wormwood et al., J. Proteomics Bioinform 2013, S5, http://dx.doi.org/10.4172/jpb.S5-001

Full size image

27.1.1 Protein Biomarkers in MDD

Depression has indeed been associated with several protein biomarkers, which have been identified using both directed methods as well as mass spectrometry [49, 80]. For example, serum brain-derived neurotrophic factor (BDNF) appears to be decreased relative to healthy controls, based on immunological measurement techniques [81–85]. A recent study using ELISA confirmed that ketamine treatment indeed results in rapid upregulation of BDNF in those patients who are responders, measured in blood sera. Ketamine has been recently identified as producing rapid antidepressant response [86], which is in contrast to many antidepressants currently used, which take several weeks to produce a response [87]. Greater understanding of the factors producing BDNF upregulation may help in predicting responders to ketamine or possibly other antidepressants. MS and comprehensive proteomics may further expand upon directed approaches to yield more biomarker candidates.

27.1.2 Use of Animal Models and MS

MS and proteomic approaches have been utilized to identify putative biomarkers in depression animal models. An animal study examined the effects of chronic stress on proteins found in the hippocampus of rats. MALDI-TOF-MS was utilized to identify found 27 potential protein markers that were dysregulated relative to non-stressed control animals. These proteins were known to have roles in neurogenesis, oxidative metabolism, transcription, and signal transduction [88]. This is particularly relevant to depression based on the observation that the hippocampus may decrease in volume in individuals who are depressed [89]. A separate study Using iTRAQ labeling, a label-based method used to identify and quantify proteins, coupled with MS, quantified 2,000 proteins from rat hippocampus in a stress model of depression. Seventy-three proteins were found to be differentially expressed. Deficits in vesicular release proteins (SNCA, SYN-1, and AP-3) were found with a relationship to stress susceptibility. Increased expression of a sodium channel protein (SCN9A) also predicted susceptibility to stress [90]. Such susceptibility could provide a model for biomarkers predictive of depression, although naturally these results would need to be replicated in humans.

27.1.3 Protein Biomarkers in MDD: Human Studies

In addition to animal models, MS has indeed also yielded biomarker insights in humans. Alawam et al., used MALDI-MS with a C18 magnetic beads protocol to analyze serum samples from 39 individuals with depression and 30 controls. C18 magnetic silica beads are used for protein and peptide sample concentration, desalting, and fractionation. C18 beads are coated with C18 alkyl groups, which allow them to provide a reversed phase surface chemistry and, through fractionation of proteins and peptides, reduce sample complexity. No protein signals distinguished participants with depression from controls, however, analysis of individual peptide signals identified three signals that were distinct in the depression group. The specific identities of these peptides were however, not reported [91]. Protein identification is the next step to establish biomarkers, although comprehensive protein fingerprints via spectral differences could also serve as a potential diagnostic.

Proteomic analysis of brain tissue is an approach that can be used to identify MDD biomarkers, although naturally the approach is limited to post-mortem tissue. A proteomic study using difference gel electrophoresis (DIGE) identified 59 potential biomarkers in cerebral cortex and 11 in amygdala in post-mortem brain tissue from suicide victims. Proteins identified were those that control biological functions and structures such as metabolism, the redox system, the cytoskeleton, synaptic function, and proteolysis [92]. Another group used shotgun proteomics to analyze post-mortem dorsolateral prefrontal cortex brain tissue from 24 MDD patients and 12 matched controls [93]. Shotgun proteomics refers to the use of bottom-up proteomics techniques in identifying proteins in complex mixtures. This method usually takes advantage of combining HPLC with MS. Distinct protein fingerprints resulted from significantly, differentially expressed proteins between subgroups of MDD patients (with and without psychosis) as well as between MDD subjects and healthy controls. Differentially expressed proteins between MDD patients and healthy controls were those involved in metabolism, transport, cell communication and signalling, cell growth and maintenance, protein metabolism, and regulation of nucleic acid metabolism [93].

In addition to the analysis of brain tissue, MS and proteomic approaches have been used to analyze bodily fluids. The advantage of this approach is naturally the possibility of identifying biomarkers in live subjects, with the hope that the proteins identified reflect brain processes or have the potential for diagnosis and/or prediction of response. Cerebrospinal fluid (CSF) is naturally the closest fluid to brain tissue, and therefore may be most suitable for reflecting brain content. The downside of CSF use is the invasive nature of collection and need for a spinal tap. CNS has been analyzed using MALDI-TOF-MS to compare individuals with MDD versus controls. This study reported 11 proteins and 144 peptide features that were significantly different in the CSF from depressed patients versus controls. The investigators also found differences in the phosphorylation pattern of several CSF proteins, underscoring the additional utility of MS for identifying PTMs as well as protein levels. Identified dysregulated proteins in this study were those involved with neuroprotection, neuronal development, sleep regulation, and amyloid plaque deposition in aging brain [94], reflecting possible disruptions of these systems by depression, involvement of these systems in the etiology of depression, or both.

Blood plasma or serum is more accessible than CSF, and may also be used for proteomic identification in MDD. Depression has been associated with disturbances in cholesterol transport and metabolism, which is mediated by apolipoproteins [32]. Indeed, a study utilizing multidimensional liquid chromatography-tandem mass spectrometry with validation by immunoblotting or ELISA to analyze blood plasma, several protein biomarkers that are altered in depressed individuals. The primary function of the identified proteins was in lipid metabolism and immunoregulation [95]. This is consistent with prior observations of increased disease incidence with depression [4–10] as well as lipid perturbations that have been observed in depression [32].

Urine analysis could be most convenient and amenable to an MDD diagnostic or predictive test. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was used to corroborate levels of urinary serotonin in individuals with depression taking antidepressants. A strong correlation was measured between urinary serotonin levels measured by ELISA. Using ELISA, serotonin levels detected in depressed patients were significantly lower than in control subjects and 5-hydroxy-tryptophane and/or selective serotonin re-uptake inhibitors significantly increased urinary serotonin levels [96]. Biological monitoring of depression and response to antidepressants may therefore be possible via urinary analysis.

Saliva is an accessible biofluid that has not been fully exploited in depression research. Saliva collection is noninvasive and convenient. In all, 2,290 proteins have been found in saliva compared with 2,698 proteins found in plasma. Nearly 40 % of proteins believed to be candidate markers for diseases such as cancer, cardiovascular disease, and stroke can be found in whole saliva, Therefore, saliva presents a convenient, noninvasive, accessible bodily fluid for analysis [97]. Potential salivary biomarkers have been identified for other disorders, specifically Alzheimer’s disease [98] and autism spectrum disorder [99]. Although we have not been able to identify analyses of MDD salivary proteome, salivary cortisol in MDD has been analyzed using directed methods (competitive radioimmunoassay for cortisol) rather than the comprehensive screening provided by MS. High salivary cortisol was not found to be a risk factor for MDD, however, low mean salivary cortisol concentration and a small difference between morning and evening cortisol concentrations were identified as possible risk factors [100]. Saliva may be an unexploited biofluid for MS analysis in MDD that warrants further exploration.

27.1.4 Metabolomics for MS-Based Biomarker Discovery in MDD

Based on prevalent hypotheses that small molecules, such as monoamine neurotransmitters, are dysregulated in MDD [2], metabolomics is an approach that may yield biomarkers in MDD. Metabolomics is the investigation of the metabolites, or small molecules that exist on a cell or a bodily fluid at a particular timepoint, and can be performed using specialized MS approaches. In a study comprising 26 subjects with MDD versus controls, blood sera content was measured using a 3200 QTRAP MS system. The investigators found depletions in tryptophan, lysine, and gamma amino butyric acid (GABA) in the subjects with MDD [101]. Proton nuclear magnetic resonance ((1)H NMR) spectrometry has also been used to distinguish metabolomic differences in MDD plasma relative to control plasma. Investigators have reported use of this method to accurately diagnose MDD in 26 subjects with sensitivity and specificity of 92.8 % and 83.3 %, respectively [102]. The prevalent monoamine theory of depression as well as the known mechanism of action of many antidepressant medications on neurotransmitters (small molecules) suggests that metabolomics may be particularly useful for MS-based exploration of biomarkers in MDD. The ability to measure metabolites in blood, urine, or saliva could provide a useful diagnostic and predictive test for MDD.

27.1.5 Protein Preparation Methods

Preparing samples for MS analysis can be a crucial component of biomarker discovery. In fact, it may be considered the most important component of the MS analysis. Depending on how the sample is prepared, one can get either useless results or significant results with high clinical relevance. This is well reflected in the NCI Clinical Proteomic Tumor Analysis Consortium (CPTAC) study, in which different proteomics labs from the United States, with different instrumentation and instrument settings obtained different results [103]. For example, the methods by which samples are collected are crucial, including how they are processed, frozen (and how many times they are thawed). Instrument settings need to be identical, for high reproducibility. Furthermore, It also matters whether the samples to be analyzed (i.e., sera) are depleted or not and if so, whether the depletion is performed under identical conditions in different labs. At its best, due to the secretome and/or functional degradome (truncated proteins with physiological significance), as well as posttranslational proteomics (labile or transient/reversible phosphorylations, acetylations, methylations, glycosylations, etc.), the samples have to be kept as intact as possible and then processed. For example, one study identified transferrin and fibrinogen as potential biomarkers, based on 6 M HCL hydrolyzation of serum proteins followed by MALDI-TOF MS [104]. However, HCl incubation caused rapid protein decomposition. To distinguish between samples from individuals with MDD versus control samples, the same investigators reported that protein hydrolysis using 20 % TFA and sinapinic acid were the optimal reagents for protein hydrolysis and found that the same markers, fibrinogen and transferrin, could be distinguished as potential biomarkers for MDD, with higher peaks identified for fibrinogen relative to controls and lower peaks for transferrin (relative intensity) [105]. However, analysis of the forced, uncontrolled fragmentation of the proteins from the samples may be dangerous, as it may be difficult to reproduce and translate into clinical settings. Therefore, sample preparation is and will always be the most crucial part of the proteomics that will allow one to obtain significant results with clinical relevance.

27.1.6 Effects of Antidepressant Treatment on Protein Biomarkers

In addition to understanding depression, naturally MS-based proteomics may ultimately elucidate markers that correspond with outcomes of antidepressant treatment. Initial proteomic analysis in animal models indicates possible protein changes associated with the use of antidepressant medications. A proteomic study analyzed monoamine reuptake inhibitors (MAOIs) and protein changes that occurred in rat hippocampal cytosolic extract. Either venlafaxine (primarily noradrenergic) or fluoxetine (primarily serotonergic) was systemically administered to adult rats for 2 weeks, after which 2D gel electrophoresis identified 31 proteins that were upregulated and two that were down-regulated. MALDI TOF MS was used to identify these proteins. Treatment with both antidepressants led to the upregulation of several factors associated with central nervous system function. These include those associated with neurogenesis (IGF-1, glia maturation factor) neuronal process outgrowth, and maintenance (hippocampal cholinergic neurostimulating peptide [HCNP]-precursor; PCTAIRE-3; serine protease inhibitor 2.1), and anti-apoptotic activity (dimethylargininase-1 l-N,N-dimethylarginine dimethylaminohydrolase-1 [DDAH1], antioxidant protein-2 [AOP-2], and pyruvate dehydrogenase-E1 [PDH-E1]) [106]. Whether similar markers can be measured in human bodily fluids remains to be seen.

Using directed protein measurements (immunoblot and electrophoretic mobility shift assays) of total cyclic AMP response element binding protein (tCREB) in T-lymphocytes, it has been found that subjects with MDD and low baseline tCREB were significantly more likely to respond (78 %) to selective serotonin reuptake inhibitor (SSRI) treatment than individuals with high baseline tCREB (36 %) [107]. This study illustrates that biomarkers that are predictive of antidepressant response could be very useful to directing and selecting treatments. The question remains as to whether tCREB is a specific marker for SSRI response, or anti-depressant response in general. MS and proteomic have the potential to reveal further protein biomarkers and expand upon this therapeutic area, which will become increasingly important as more treatment options with different mechanisms of action become available.

27.2 Concluding Remarks

Although various biomarkers for MDD have been identified using an assortment of methods, the true test of a biomarker will be its validity, reproducibility, and prediction of therapeutic response. In addition, biomarkers for use in clinical tests will need to be identified in an inexpensive, quick, and convenient manner. For this reason, protein tests may be particularly useful. Proteins can be derived from easily obtainable bodily fluids, such as blood and even saliva. MS has the potential to identify new protein or even metabolite targets that may ultimately be used in tests. Such tests may be conducted via MS, or using directed protein-identification methods that are quick, cheap, and easy. The hope is to develop a form of personalized medicine for MDD that can direct the choice of treatment, particularly in light of the high rate of depression relapse and the high rate of treatment-resistance.