Are biological markers useful to predict whether depressive patients will effectively respond to antidepressants?

(1)

1

Thesis

MSc in Brain and Cognitive Science

Track Cognitive Neuroscience

Are biological markers useful to predict whether depressive

patients will effectively respond to antidepressants?

Major Depressive Disorder (MDD) is a common psychiatric disorder. Despite the availability of numerous of treatment options for MDD, research have shown that antidepressant yields only modest rates of response and remission. Clearly, there is an urgent need to develop more effective treatment strategies for MDD patients. One possible approach towards the development of novel pharmacotherapeutic strategies for MDD involves identifying subpopulations of depressed patients who are more likely to experience the benefits of a given treatment versus placebo, or versus a second treatment. Furthermore, there is an urgent need for objective diagnostic tests. In this thesis, we will attempt to summarize the literature focusing on biological markers, which can serve as diagnostic marker or as a predictor of treatment outcome. This includes genomic, proteomic, metabolomics, structural and functional neuroimaging biomarkers. Further research should focus on subsets within the MDD population to find sufficient biomarkers.

Ilke van Loon 5695279

July 2012

University of Amsterdam

Brain & Cognition

Supervisor: Huib van Dis

Co-assessor: Bob Bermond

(2)

2

Part 1. Background

1.1 Major Depressive Disorder

Major Depressive Disorder (MDD) is a common mental disorder which diagnosis is based on patient’s symptoms according to criteria set forth in the Diagnostic and Statistical Manual (DSMIV, 2000). Table 1 gives an overview of depressive symptoms as characterize in the DSM-IV. Symptoms rating scales like Hamilton Depression Rate Scale (HDRS) and Montgomery Asberg Depression Rate Scale (MADRS) are currently being used for MDD diagnosis, and to measure treatment improvement. Establishing a diagnosis with the aid of these tests strongly depends on the subjective interpretation of the evaluator and patients truthfulness in reporting symptoms

(Paez-Pereda & Panhuysen, 2008; Raelson & Belouchi, 2008). This means that currently diagnosis of MDD is not based on objective laboratory results, but rather on a highly variable set of symptoms (Mossner et al., 2007). Moreover, symptoms differ considerably between affected individuals with respect to presence, frequency, severity, and topography (Iris, 2008). Accordingly, MDD should not be considered as a single disease, but as a heterogeneous syndrome (Nestler et al., 2002). This heterogeneity in symptoms has complicated the search for the causes, etiology and effective treatments of MDD (Nester et al, 2002;Iris, 2008). However, there is a strong need for diagnosis of MDD in a reliable, reproducible and objective manner (Paez-Pereda & Panhuysen, 2008).

Globally, MDD is one of the main causes of personal dysfunctioning and affects almost 16% of the world’s population (Kessler et al., 2003). The extreme prevalence causes a financial burden of approximately 1% of the Gross Domestic Product (GDP) in developed counties (Belzeaux et al., 2010). According to the GIPdatabank, antidepressants (ADs) have been used by ± 960.000 people in the Netherlands during 2011, and the associated costs of prescribed drugs were around €96 million (www.gipdatabank.nl). If no clinical progress is achieved, MDD will become the second cause of disability in 2030 (Mathers and Loncar, 2006) and financial consequences will rise. Clinical efficiency is necessary for both improvement of the patient quality of life and decrease of the social-economic burden.

1.2 Research focus is moving from monoamine system toward molecular mechanisms

Medication developed approximately 50 years ago for treatment of other disorders turned out to elevate the mood of psychiatric patients. Further research showed that these drugs increase the extracellular levels of serotonin and norepinephrine by inhibiting the reuptake in nerve endings. These finding were the basis for the monoamine hypothesis. It is currently understood that this hypothesis does not only suggest a deviation or an unbalance of serotonin (5-HT) and norepinephrine (NE), but also of dopamine (DA). Despite 50 years of research, this theory cannot entirely explain the effects of ADs. For example, when an AD is ingested, the blockade of the target transporter occurs within several minutes (Ruhé, 2008). On the other hand, the clinical effects of antidepressants take

Table 1. Diagnostic criteria for Major Depression

Depressed mood Irritability Low self esteem

Feeling of hopelessness, worthlessness, and guilt Decreased ability to concentrate and think Decreased or increased appetite Weight loss or weight gain Insomnia or hypersomnia

Low energy, fatigue, or increased agitation

Decreased interested in pleasurable stimuli (e.g. sex, food, social interactions)

Recurrent thoughts of death and suicide

Table adapted from Nestler et al. (2002). A diagnosis of depression is made when a certain number of the above symptoms are reported for longer than a 2 week period of time, and when the symptoms disrupt normal social and occupational functioning (see DSMIV, 2000).

(4)

4 much longer than this direct pharmacological effect. Therefore, the antidepressant effects cannot only based on increased neurotransmitters by decreased 5-HT reuptake (Ruhé, 2008). In other words, because clinical effects of ADs are delayed, researchers believed that ADs function by long-lasting downstream changes instead of just elevating neurotransmitter levels (Lee et al., 2010). Additional researches that associated alteration of DA and NE system with MDD found inconclusive results. Although it is commonly accepted that monoamine deviations are involved in MDD, the view that these deviations do explain some aspects of MDD, but not all gets more and more support (Toups & Trivedi, 2012). Consequent studies focused on receptors, intracellular signal transduction molecules and genes of the monoamine neurotransmitters. This was the first step towards the chemical hypothesis of MDD which presumes that MDD is caused by a structural or functional changes in specific molecules in the brain, and that ADs function by counteracting these molecular changes (Castren, 2005).

Current research try to identify downstream neurobiological changes (induced by ADs) and thereby reaching for a detailed view to the pathophysiology of MDD (Lee et al., 2010). Research conducted today is not limited to the monoamine system, but is also aimed at identifying other dysfunctional molecules. Moreover, they try to find biomarkers that indicate a biological change associated with MDD. This biomarker should indicate the presence and severity of the condition and predict drug or other treatment response as well as the clinical prognosis (Mossner et al., 2007). Take in mind that MDD is a heterogeneous disease and therefore MDD may be due to multiple factors. This makes the search for biomarkers complex (Lee et al., 2010).

1.3 Treatment

Today’s MDD treatments remain sub-optimal. Although psychiatrists have several medical treatment possibilities for MDD, there is presently no method available for selecting a specific treatment that fits the individual clinical condition of the patient best (Serretti et al., 2008). As a result, while psychiatrists are testing different medical treatments, many of their patients remain unrelieved from their mental condition. Usually patients receive first a selective serotonin reuptake inhibitor (SSRI). Although ±35% of patients achieve remission (disappearance of symptoms) after 6-14 weeks, non-response (<50% decrease of symptoms) is a major clinical problem affecting around 50% of SSRI-treated patients (Rush et al., 2006). Switching to or additional administration of another monoamine medication remain ineffective in approximately 30-40% of patients (Rush et al., 2006). It is currently not understood why certain patients recover and others do not show signs of improvement, even after having used several types of ADs. Further disadvantages of using ADs are: (1) treatment response usually occurs after a couple of weeks, (2) chronic treatment with AD is required for clinical effects and, (3) users suffer from a wide spectrum of undesired side effects (Lee et al., 2010).

All currently ADs act in a similar way, increasing the availability of monoamine neurotransmitters in the brain. The development of new treatments for MDD is limited. Unfortunately, in the last decades no improved treatment options have been developed for MDD (Roiser et al., 2012). It is believed that the identification of reliable biomarkers can be used to develop new and adequate treatments (Harmer et al., 2011).

(5)

5 1.4 Prediction of treatment outcome

At present, it is not possible to reliably predict whether an individual patient will respond better to one treatment approach or to the other. Therefore, clinical guidelines still recommend waiting for at least 4-6 weeks before switching to or additional administration of another AD when an adequate treatment response is not achieved (Serretti et al., 2008). Prediction of AD effectiveness for each individual patients will reduce the time a patient must continue to live with the effects of MDD. An useful predictor indicates whether a certain AD is likely to lead to remission.

Current MDD research focus on finding these predictors, however it is still unknown whether any of candidate predictors are sufficient to fulfil diagnostic criteria as diagnostic predictor for major depressive disorder.

In this thesis we gave an overview of biomarkers with high potency to become useful for identification of MDD and of biomarkers with high potency to become a treatment outcome predictor (Chapter 2 and 3, respectively). Furthermore, we discuss the usefulness of these biomarkers and investigated whether the research focus shift from monoamine systems towards these biomarkers is a good or bad alteration (Chapter 4). Finally, we will give further research recommendations (Chapter 5).

(6)

6

2. Biological markers in MDD research

2.1 Biomarkers

In 2001 The Biomarkers Definitions Working Group of the National Institutes of Health Group defined a biomarker as “a characteristic that is objectively measured and evaluated as an indicator of

normal biological processes, pathogenic processes, or pharmacological responses to a therapeutic intervention.” This was the third definition of biomarkers, and seems to be the most suitable

definition for a heterogeneous disease like major depression disorder (Iris, 2008). In other words a biological marker describe a biological change associated with the disease that could be used to indicate the presence and severity of the condition and predict drug or other treatment response as well as the clinical prognosis (Mossner et al., 2007). Additionally, in order to be diagnostically useful, biomarkers needs to provide sufficiently high levels of sensitivity and specificity (>80%) in the detection and correct classification of distinct disorders (Ritsner & Gotteman, 2009). Furthermore, biomarkers should be reproducible, reliable, inexpensive, non-invasive and easily accessible in order to ensure their application in daily clinical practice (Schneider et al., 2011). This broadly definition ensures a huge number of biomarkers. In this chapter we will see that one way to distinguish biomarkers is to look at their molecular characteristics; whether they are genomic, transcriptomes, proteome or metabonome biomarkers. Consider that human genomes accounts for ~20,500 genes, ~100,000 transcriptomes, ~1.000.000 proteomes and ~2.500-3.000 metabonomes. This means that the total number of biomarkers of interest can be estimated to be ~1.133.000 (Ritsner & Gotteman, 2009). For obvious reasons, we cannot provide a list of all these potential biomarkers. In this chapter we will give an overview of biomarkers with high potency to become a useful biomarker for MDD. Although a large number of biomarkers are under investigation, no biomarkers are used nowadays in clinical practice.

2.2 Overview potential biomarkers for MDD

We can distinguish three different ways to categorize depression biomarkers. The first approach looked into the molecular characteristics of biomarkers; for example DNA and proteins. The second approach focussed on supposed biological deficiencies of depressed patients; for example cortisol and neurotransmitters. And finally, neuroimaging methods are used to find differences between depressed patients and healthy controls in general; for example functional MRI shows brain activation differences between both subject groups. In this chapter we will look at these biomarker categories.

2.2.1 First approach: biomarkers based on molecular characteristics

We can distinguish four types of biochemical markers; genomic, transcriptomic, proteomic and metabolomic. Table 2 gives an overview of characteristics of these types. Each kind of biomarker has advantages and disadvantages as shown in this table. In summary, genomic biomarkers such as small nucleotide polymorphisms (SNPs) are readily available, however they have (probably) a low penetrance and do not directly or predictably impact on subject’s phenotype (Schwarz and Bahn, 2008). Transcriptomic biomarkers have a limited availability and therefore not a first choice for

(7)

7 biomarker identification (Paez-Pereda & Panhuysen, 2008). Both proteomic and metabolimic states are closely associated with an organism’s phenotype, however access to proper biofluid is limited. Cerebrospinal fluid (CSF) might be considered as the sources that most closely reflect brain activity, but it is not easily accessible on a routine, risk-free basis that is likely to be acceptable to patients (only obtained by a highly invasive method called lumbar puncture). Other biofluids, although easily collected, is further removed from brain function (Leuchter et al., 2010). Initial studies involved with proteomics and metabolomics have yielded very exciting and promising results that justify further studies over the years to come (Raedler et al., 2009).

Table 2. biomarkers based on molecular characteristics

Type Indicator Access to proper target tissue

Advantages Disadvantages Potential biomarkers for

depression

Genomic Small nucleotide polymorphisms (SNPs)

Excellent -Genome is rather constant

-New technologies make complete genome SNP investigation possible

- Many disease genes have a low penetrance and do not directly or predictably impact on an organism’s phenotype

- Doubts whether genome-wide associations studies will uncover biomarkers with strong predictive values

Polymorphisms in genes which code for a HT receptor; 5-HTT; MAOA.

Transcriptomic Change in gene expression, measured via transcript abundance Very limited; expression in the central nervous system not suitable, however expression in peripheral blood leucocytes could be an alternative

-Reflects only genes which are activated

- Levels of transcriptomes are not directly

proportional to the expression level of the proteins they code for

5HTT mRNA; LIM mRNA Proteomic Change in protein concentration or modification Limited; access of CSF tissue (related to brain processes) hard, however access to other biofluids (e.g. blood, urine, plasma) good.

-The state of a proteome is in close association with an organism’s phenotype -New techniques made analyse of several thousand proteins in one single step possible

- Complicated: a proteome differs from cell to cell and constantly changes - High invasive; expression in the cerebrospinal fluid (CSF) can be obtained only by lumbar puncture

-High number of possible biomarkers

the protein glyoxalase 1 (Glx 1);

BDNF

Metabolomic Enzyme activity, concentration of metabolites or other molecules

Limited; access of CSF tissue (related to brain processes) hard, however access to other biofluids (e.g. blood, urine, plasma) good.

-Some biofluids non-invasively

-relative small number of biomarkers

- Metabolomic changes in biofluids do not specifically reflects maladaptive brain function

Glucose; dopamine, serotonin; IL-1, IL-6, GFAP; NGF; valine, leucine

Adapted from (Jain, 2010; Mossner et al., 2007; Peaz-Pereda & Panhysen, 2008; Ritsner & Gottesman, 2009; Schwarz & Bahn, 2008). 5-HTT: serotonin transporter; BDNF: brain derived neurotrophin factor; CSF: cerebrospinal fluid; GFAP: Glial fibrillary acidic protein; IL: interleukins; LIM: PDLIM5 is a small protein that modulated neural calcium signalling; mRNA: messenger ribonucleic acid; NGF: nerve growth factor.

(8)

8 2.2.2 Second approach: Deficiencies of MDD patients as focus of biomarkers

This approach tries to find clinical biomarkers that link maladaptive processes of MDD patients with biological alterations. Once affected biological processes are identified, single parameters representative of these alterations can be chosen as biomarkers (Paez-Pereda & Panhuysen, 2008). In this chapter we will focus on four supposed affected processes; deficiencies of the neurotransmitter system, deficiencies of the hypothalamus-pituitary-adrenal axis, abnormal neurotrophin levels and inflammatory processes. Take in mind that these topics are only a part of this research field.

2.2.2.1 Dysfunction neurotransmitters

One of the leading hypotheses assumed that MDD is caused by deficiencies of the monoamine neurotransmitters. Especially, the 5-HT system is investigated. Due to this, studies have focussed on abnormalities of biochemical parameters that are involved with this system (e.g. receptors, transporters, gene polymorphisms and RNAs which are coding for such particles).

A meta-analysis by Ellis & Salmond (1994) investigated the binding of imipramine to blood platelets. Imipramine binds to 5-HT transporter and represent a model system for serotonergic neurons. In addition, imipramine is a tricyclic antidepressant. This meta-analysis showed that imipramine binding to platelets is generally decreased in MDD patients and that imipramine binding could be a biological marker of depression (Ellis and Salmond, 1994). To our knowledge, in the past decade this biomarker is not further investigated.

Multiple studies investigated genetic variants of 5-HT receptors (e.g. polymorphisms of C-1019G promotor and G861C). A trend of decreased 5-HT1A receptor expression appears to be a robust finding in MDD. Whether other alterations of receptor types are associated with the disease is unclear due to contradictory results (Mossner et al., 2007).

Polymorphisms of the 5-HT synthesizing enzyme tryptophan hydroxylase 2 (TPH2) are also investigated. TPH2 is exclusively expressed in neuronal cell types and therefore an interesting candidate. A recently published study showed that while previous studies have suggested that several TPH2 genetic variants could be associated with MDD, they could not confirm these findings. In contrast, they found that none of the investigated polymorphisms were associated with MDD, except for one polymorphism that was linked to the severity of MDD (Serretti et al., 2011). Moreover, functional variant of TPH2 have been shown to be important in the pathogenesis of obsessive-compulsive disorder (Mossner et al., 2006). It seems that even if TPH2 variants are sensitive for MDD, they are also sensitive for other psychiatric disorders (Mossner et al., 2007). Due to this non-specificity, TPH2 is unsuitable as MDD biomarker.

In only a few studies biomarkers of catecholamine system are investigated. Catechol-O-Methyltransferase (COMT) is involved in the catabolic pathways of norepinephrine (NE) and DA. Subjects with val/val polymorphism have a higher COMT activity than subjects with met/met polymorphism and therefore lower extracellular NE and DA levels. It seems that met/met subjects and heterozygous subjects (i.e. val/met) are vulnerable to develop MDD compared to val/val subjects. However, studies are still inconclusive (Aberg et al., 2011;Baekken et al., 2008).

(9)

9

2.2.2.2 Deficiencies of the HPA-axis

MDD is associated with a malfunction of the regulation of emotions and cognition in the response to stress. Therefore, one of the leading hypotheses supposed an alteration of brain areas involved in the control of stress response. This brain circuit is called the hypothalamus-pituitary-adrenal axis. Stress can activate this brain circuit to produce hormones like corticotrophin-releasing hormone (CRH), adrenocorticotrophic hormone (ACTH), and cortisol. Abnormal levels of these hormones have consistently observed in depressed patients (Ising et al., 2005).

Biomarkers which can measure deficiencies of the HPA-axis are the Dexamethasone suppression test (i.e. evaluating the pituitary negative feedback inhibition of the HPA axis by

administration of Dexamethasone and measuring cortisol and ACTH), CRH test (i.e. corticotrophin releasing hormone test in which corticotrophin is administered intravenously and plasma cortisol is measured) and Dex/CRH test (combined; sensitive to impaired glucocorticoid receptor signalling at the pituitary level as well as to the effects of increased secretions of central neuropeptides). Especially

the latter is sensitive to antidepressant treatment (reviewed in Ising & Holsboer, 2008). These biochemical parameters are among the most consistent observations in MDD patients (Ising et al., 2005). However, only a proportion of depressed patients show HPA axis alterations and this limits the value of these biochemical parameters as universal makers of depression (Paez-Pereda & Panhuysen, 2008). In addition, elevated hormonal responses to the combined Dex/CRH test does not uniquely characterize MDD, but can also be observed in other psychiatric disorders (Ising et al., 2005).

2.2.2.3 Abnormal neurotrophin levels

Long-term stress appears to reduce the expression level of brain derived neurotrophic factor (BDNF) (Lee et al., 2010). Neurotrophins are proteins that induce the survival, development and function of neurons; these proteins help to stimulate and control neurogenesis.

Although still controversial, studies have reported a reduction of the hippocampus size in depressed patients, which could be the result of cell death and changes in neuroplasticity in this brain area (Geuze et al., 2005). Research had focus on whether reduced levels of BDNF could explain the reduction in the number of dendrites and synapses in the hippocampus. Some studies showed reduced BDNF levels in MDD patients (Cunha et al., 2006;Nestler et al., 2002), however this reduction is also observed in other neurological diseases (Nestler et al., 2002). A recently appeared study made subsets within a MDD population and distinguish them on severity and it specific symptomatic dimension. They showed no relationship between those subsets and BDNF levels (Jevtovic et al., 2011). The lack of selectively and consistent results precludes the use of BDNF as biomarker for MDD. Worth noting, there is some evidence that ADs increased BDNF levels in hippocampus (Chen et al., 2001). These findings raise the possibility that antidepressant-induced upregulation of BDNF could help repair some stress-induced damage to hippocampal neurons and protect vulnerable neurons from further damage. This could also explain why an antidepressant response is delayed: it would require sufficient time for levels of BDNF to gradually rise and exert their neurotrophic effects (Nestler et al., 2002). Furthermore, this raise the possibility that only a subset of MDD patients (e.g. stressed subjects) have reduced BDNF levels, while not others.

(10)

10

2.2.2.4 Inflammatory markers

Ten years ago, the idea arose that inflammatory processes might contribute to the development of MDD. To investigate this hypothesis researchers especially studied cytokine release in MDD patients. Cytokines are small glycoproteins that function as signalling molecules between immune cells. Robust evidence showed that increased HPA axis activity stimulate cytokine release (Li et al., 2011). Furthermore, several lines of evidence suggesting a bidirectional influence; cytokines inhibit or enhance (depends upon several factors) release of neurotransmitters and neurotransmitters inhibit or enhance release of cytokines (Serretti et al., 2008).

Some studies reported that MDD patients had increased levels of a variety of peripheral inflammatory biomarkers when compared with non-depressed subjects (Mossner et al., 2007;Raison and Miller, 2011). These elevated levels in MDD patients may be related to psychological stress or maladaptive neurotransmitter levels (Mossner et al., 2007). A meta-analysis showed inconsistent results whether cytokine levels are altered in MDD patients compared to controls (Eller et al., 2008). Previous study also showed that ADs exert anti-inflammatory effects and anti-inflammatory agents are adjunctive treatment for MDD (Li et al., 2011). However, inflammation is neither necessary nor sufficient to cause MDD (Raison and Miller, 2011) and this limits the value of these biochemical parameters as universal makers of MDD.

2.2.3 Third approach: Neuroimaging biomarkers

Neuroimaging technology has afforded the ability to investigate neurophysiological, neuroanatomical and neurochemical correlates of mood disorders in vivo. Neuroimaging techniques such as structural magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), diffusion tensor imaging (DTI), electroencephalography (EEG) and positron emission tomography (PET) showed evidence for underlying biological factors of MDD.

Changes in specific parameters measured by these methods probably represent biological alterations that are mechanistically related to both the cause of MDD and the working of ADs. These changes could be, therefore, considered intermediate phenotypes between the genetic and biochemical cause of MDD and the complex psychological and behavioural manifestation of the symptoms of MDD (Paez-Pereda & Panhuysen, 2008).

Candidate MRI biomarkers measure brain volume differences in MDD patients compared to healthy controls. Volume reduction was found in the anterior cingulate cortex, orbitofrontal cortex, the hippocampus, the putamen and the caudate nucleus in the patient group compared to controls in a meta-analysis (Koolschijn et al., 2009). Inconsistent results of structural alterations of the amygdala were found (reviewed in Schneider et al., 2011). Also a fMRI study found that impaired cognitive emotional regulation was associated with impaired downregulation of amygdala activity in MDD patients compared with controls (Erk et al., 2010). According to Drevets and colleagues (2001) PET studies showed abnormal cerebral blood flow and glucose metabolism in brain regions that play important roles in depression; such as prefrontal/ cingulate cortex, hippocampus, striatum, amygdala, and thalamus. The abnormalities in many of these regions are, to some extent, mood-state-dependent, implicating areas where neurophysiological activity may increase or decrease to mediate or respond to the emotional and cognitive manifestations of the depressive syndrome. During antidepressant drug treatment, some of these state-dependent changes reverse in those patients who respond to treatment, however they do not entirely normalize during symptoms

(11)

11 remission in many of these regions (Drevets, 2001). In contrast, a review of DTI studies did not reveal significant differences in diffusivity (Sexton et al., 2009).

There are several important limitations in this field of research. Even when similar tasks were used, brain activation measured with fMRI was frequently discrepant among studies. Most imaging studies examined small samples limiting statistical power (Cerullo et al., 2009). Additionally, most studies were confounding by patients taking psychotropic medications. Furthermore, task performance may affect brain activation and mood likely affect brain activation; it may not be valid to compare brain activation from groups in different affective states (Schneider et al., 2011).

(12)

12

3. Current findings potential biomarkers which predict treatment

outcome

3.1 Treatment outcome predictors

During the past several years, we have achieved a deeper understanding of the etiology and pathophysiology of MDD. However, this improved understanding has not translated to improved treatment outcome (Leuchter et al., 2010). There are three major “paths” towards improvement of treatment outcome; (1) developing new antidepressants; (2) combining pharmacological agents and (3) identifying subpopulations of depressed patients who are more likely to experience the benefits of a given (existing) treatment versus placebo, or versus a second treatment (Papakostas and Fava, 2008). In this chapter we will focus on the third approach; biomarkers which may be used to predict treatment success.

The ideal biomarker is measurable at diagnosis to assist in selection of the first treatment. However, thus far no pre-treatment predictors were found which are able to predict which specific treatment is likely to benefit a particular patient (Leuchter et al., 2010). Much research now is aimed to identifying biomarkers that emerge early in the course of treatment and may indicate whether the medication that the patient is receiving is likely to lead to remission (Leuchter et al., 2010). Furthermore, some studies investigate the biological differences between responders and non-responders after long-term AD treatment.

When ineffective treatment attempts can be reduced to a minimum, MDD patients will suffer less and the financial burden will decrease. The identification of disease-related biomarkers is urgently needed to advance the diagnosis, treatment and management of complex psychiatric disorders such as MDD (Schwarz and Bahn, 2008). At the present time, no biomarkers have sufficiently proven utility to be ready for clinical application (Leuchter et al., 2010).

3.2 Overview potential treatment outcome predictors

To date, numerous studies have explored several potential biomarker of outcome. The majority of these studies are involved with genetics, proteomics, metabolomics and brain imaging. In the context of this thesis it is impossible to discuss all appeared studies which are involved with this topic, however we will discuss biomarkers which show promising results for predicting clinical response. Two recent appeared reviews gave an overview of the most promising biomarkers in this field and we will summarize their findings (Leuchter et al., 2010;Papakostas and Fava, 2008). Additionally, we will summarize results of others.

3.2.1 Genetic markers

The majority of studies involved with pharmacogenomics (i.e. they measure influence of genetic variation on drug response in patients) focus on genes coding for proteins directly involved in the monoaminergic system, including tryptophan hydrolase (TPH), serotonin transporter (5-HTT), the serotonin receptors, the monoamine oxidase enzyme (MAO), and the catechol-O-methyltransferase enzyme (COMT). While some studies suggest that patients with a specific polymorphism in one of these genes may respond poorly to SSRIs (or other antidepressants) than those without such a polymorphisms, other studies did not confirm their findings. Relatively fewer studies have focused on

(13)

13 genes coding for proteins not directly related to the monoaminergic system. Single studies found associations between the presents of a polymorphisms and clinical response to antidepressants, however further research is needed to confirm their findings. To date, no polymorphism is localized to predict treatment response (for details see (Papakostas and Fava, 2008). Despite some clinics use pharmacogenomics results (reviewed in (Kirchheiner et al., 2003) already in practice (Hoop et al., 2010).

3.2.2 MDD deficiencies markers

In both reviews limited data is reported about markers that are involved with neurotransmitter functioning, neurotropic levels, HPA-axis functioning and inflammation processes, and therefore additional studies are discussed.

3.2.2.1 neurotransmitters functioning

A couple of studies investigated whether neurotransmitter functioning could be used to predict treatment outcome. Some studies showed differences in monoamine functioning between responders and non-responders. One study showed lower plasma 5-HT concentration in responders compared to non-responders after 1 day of treatment (Alvarez et al., 1999). Also a significant association between treatment response and tryptophan/ large neutral amino acid ratio was found (Lucini et al., 1996). Furthermore, Ueda and colleagues (2002) measured both plasma 3-Methoxy-4-hydroxyphenylglycol (MPHG; a metabolite of norepinephrine degradation) and plasma homovanillic acid (pHVA; a metabolite of catecholamine) levels. They showed that responders to sulpiride had significantly lower pHVA levels before administration of sulpiride than did non-responders or controls. In contrast responders to fluvoxamine had significantly higher plasma pMHPG levels before administration of fluvoxamine than did non-responders or controls (Ueda et al., 2002).

In contrast, other studies found no differences in monoamine functioning. No differences of both 5-HT and DA transporter availability was found between both conditions (Cavanagh et al., 2006). Furthermore, no differences in urinary catecholamine levels (measured norepinephrine, epinephrine, metanephrine and normetanephrine levels) in responders and non-responders to cognitive behavioural therapy was found (Oei et al., 2010). No biomarkers based on monoaminergic alterations can predict treatment outcome yet.

3.2.2.2 HPA-axis markers

Several studies investigated whether maladaptive functioning of the HPA-axis predict treatment outcome. Most studies indicated that administration of a DEX suppression test or a CRH test alone, did not predict treatment outcome (Isogawa et al., 2005;Papakostas et al., 2010). Furthermore, cortisol concentrations cannot serve as a biomarker (Muck-Seler et al., 2002). However, combined administration of the DEX suppression test and the CRH test (DEX/CRH) is a potential biomarker that may predict clinical outcome. Results of a couple of studies suggest that early change (within 1 or 2 weeks) in HPA system regulation assessed with repeated DEX/CRH tests is associated with treatment response (Ising et al., 2007;Ising et al., 2005;Isogawa et al., 2005;Paslakis et al., 2010;Schule et al., 2009). However, early improvement of HPA system dysfunction is not a sufficient condition for a favourable response (Schule et al., 2009).

(14)

14

3.2.2.3 Neurotropic markers

As shown in chapter 2, BDNF is clearly implicated in the pathogenesis and treatment of MDD. Three studies investigated whether serum BDNF levels could be used to predict treatment outcome (Rojas et al., 2011;Tadic et al., 2011;Yoshimura et al., 2007). Tadic and colleagues (2011) showed that the absence of an early increase of sBDNF in conjunction with early non-improvement might be a highly specific peripheral marker predictive for treatment failure in patients with MDD. Rojas and colleagues (2011) concluded that responders showed an early improvement in parallel with a rise in sBDNF levels during the ﬁrst two weeks of treatment with venlafaxine. Both studies showed that sBDNF levels might represent biomarkers that could be used to predict responses. Yoshimura and colleagues showed that responders to antidepressant had increased sBDNF levels at 8 weeks, but not 4 weeks. In contrast, sBDNF levels were not changed in non-responders. This study showed that sBDNF levels did not changed in the first weeks and therefore cannot be used as predictor.

3.2.2.4 Inflammatory markers

Less studies measured cytokine levels in responders and non-responders to ADs. The difference between responders and non-responders has been assessed in MDD subjects after a six-week treatment period with an antidepressant. They showed that tumor necrosis factor- alpha levels (TNF-α; a cytokine) normalized only in responders. Furthermore, pretreatment production of interleukin-6 (IL-6; a cytokine) was significantly decreased in responders, while increased in non-responders (Lanquillon et al., 2000). Another study showed that based on the currently available data, it is still unknown whether pretreatment levels of immune system markers can be used to predict responses to ADs (Eller et al., 2009).

3.2.3 Neuroimaging markers

Some studies investigated structural differences between responders and non-responders. The most robust evidence showed that larger hippocampal volumes predict better response after 8 weeks of pharmacotherapy in two separate samples (MacQueen et al., 2008;Vakili et al., 2000). Furthermore, in a prospective study, smaller hippocampal volumes were predictive of clinical outcome 3 years later (Frodl et al., 2008). In another prospective study, larger hippocampal volume was associated with a lower probability of relapse in men at a 2-year follow up (Kronmuller et al., 2008). The predictive utility of structural data is not limited to the hippocampus, as grey matter density in the ACC and posterior cingulate cortex was also predictive of clinical remission following 8 weeks of fluoxetine treatment (Costafreda et al., 2009). Limited data is available whether abnormalities in white matter pathways are associated with treatment outcome. However, preliminary data showed that application of diffusion tensor imaging has predictive potential for delineating treatment responders (for details see Leuchter et al., 2010).

Two types of functional MRI data have demonstrated the most promise as biomarkers of treatment outcome: (1) connectivity analysis in resting state. Anand and colleagues showed that corticolimbic connectivity increased as scores on the Hamilton Depression Rating Scale decreased during treatment, suggesting that assessment of resting state corticolimbic connectivity could be useful for predicting antidepressant treatment response (Anand et al., 2005a). And (2) task-related activations, specifically the viewing of negative emotional facial expressions. When viewing negative

(15)

15 faces MDD patients show different brain activation than healthy subjects. In a treatment study, 8 weeks of fluoxetine administration ameliorated the deficient connectivity between the amygdala and anterior cingulate (Anand et al., 2005b). These changes in task-related reactivity are complementary to differences between treatment responders and non-responders in resting state connectivity within corticolimbic circuits (Anand et al., 2007). These studies are reviewed in Leuchter et al., 2010. The best-documented brain functional biomarker for predicting antidepressant treatment response is quantitative electroencephalography (QEEG). QEEG signals are generated by assemblages of neurons in the cortex and deeper structures and as such provide a global measure of brain function. Both reviews described that responders to ADs differ from non-responders in QEEG power, either in the resting state or during simple tasks. The results of several studies suggest a decrease in theta concordance for prefrontal EEG leads during the first week of treatment with either an SSRI, an SNRI, or a variety of depressants, to predict greater symptom improvement following 4 to 10 weeks of treatment. Furthermore, a decrease in prefrontal EEG theta cordance was found during the week immediately predicting the initiation of treatment of MDD with antidepressants (fluoxetine, venlafaxine) or placebo is related to the likelihood of responding to antidepressants but not placebo following 9 weeks of treatment. In summary, a number of studies showed that QEEG measurements can be a predictor of treatment outcome in MDD patients (for details see Leuchter et al., 2010; Papakostas & Fava, 2008).

(16)

16

Part 4. Discussion

4.1 Potential biomarkers and challenges

Even though some progress has been made in the identification of potential biomarkers, biomarker research is still in its infancy. Productive efforts have been made in the field of research towards biomarkers which involve stress-induced alterations in the hormone system (i.e. HPA-axis) and proteins involved in stimulation and controlling neurogenesis (i.e. BDNF). Although numerous studies investigated these biomarkers, their usefulness has not yet been proved.

No diagnostic tests or wide clinical use of biomarkers has yet emerged from biomarker research. We argued that in particular the heterogeneity and the complexity of MDD slowed down the identification of a suitable biomarker. One of the problems in biomarker identification is that a biomarker needs to provide sufficiently high levels of sensitivity and specificity in the detection and correct classification of distinct disorders (Ritsner & Gotteman, 2009), but we wonder whether that is possible in such a heterogeneous disorder like MDD. Other challenges in biomarker identification are limited availability of proper tissue that can be accessed non-invasively, complexity of the brain, poor clinical diagnostic tools and lack of models for validation.

In our view, MDD is most likely the result of biological alterations, but the environment also plays a role. For example, socio-economic status and traumatic experiences may modify the onset and severity of MDD. This means that biomarkers can only tell clinicians limited information about patient likelihood for developing a psychiatric disorder. As a result, it is unwise to make generalization or predictions based solely on biomarkers (Raelson & Belouchi, 2008).

4.2 Diagnostic dilemma

Another major concern about current biomarker research is the use of current diagnostic tools to distinguish between depressed and non-depressed conditions. These diagnostic tools are unreliable and subjective according to some researchers (reviewed in (Raelson & Belouchi, 2008) and a meta-analysis confirmed this view. This study showed that evaluators correctly identified depression in less than half of more than 50000 reported cases (Mitchell et al., 2009). This means that current validation of potential biomarkers depends on the interpretation of evaluators, while this research field needs objective tools to measure sensitivity and specificity of potential biomarkers. In other words, researchers are trying to find objective biological markers to improve current diagnostic tools, while using subjective diagnostic tools during this process. Suppose that researchers want to validate a ‘true’ biomarker for MDD. By using the current diagnostic tools the depressed and non-depressed conditions may not be pure (e.g. non-depressed patients included in depressed condition) which could provoke that no significant differences will be found between both conditions with regard to this marker. This means that even if researchers investigated a true biomarker, current research method may not recognize it.

This diagnostic dilemma is complicated even more by the fact that diagnosis of MDD is based upon DSM-IV criteria, while it is unknown whether these criteria reflect specific brain dysfunction. Furthermore, clinical symptoms are frequently not specific for only one neuropsychiatric disorder but show significant overlap between different disorders (e.g. mood, sleep). In addition, many neuropsychiatric disorders often occur co-morbid with other neuropsychiatric disorders (e.g. MDD with anxiety disorders and/ or substance use disorders) (Raedler et al., 2009). This makes the search

(17)

17

for ‘true’ biomarkers (i.e. associated with only one psychiatric disorders) complicated. Furthermore, symptom-rating scales like the Hamilton Depression Rate Scale (HDRS) and the Montgomery Asberg Depression Rate Scale (MADRS) are mostly used to measure treatment progress. However, these tasks do not measure only the core depressive symptoms (i.e. depressed mood and anhedonia), but also the sub-symptoms (e.g. sleep distraction and change in appetite). This means that changed symptom-rating scores do not reflect only improvement or deterioration of core MDD symptoms and researchers should be more aware of this when trying to identify ‘true’ biomarkers. Furthermore, these researchers should remember that DSM-IV is just a diagnostic handbook in which disorders are characterized by symptoms. Moreover, many psychiatric diagnoses can be seen as labelling deviant behaviour than as ‘real’ diseases. In our opinion, the dominated thinking that we have developed to characterize MDD and other neuropsychiatric disorders on basis of the symptomatic criteria described by DSM-IV has become a barrier to understand the biological alterations linked to the disorder. Moreover, we think that no biomarkers are found so far due to heterogeneous of MDD and due to similar symptoms in other neuropsychiatric disorders.

To reduce heterogeneity of MDD patients, further research should create subsets within the patient population. One example is to split up the MDD population into subgroups that manifest a certain symptom (e.g. hypersomnia or insomnia). In that case, biomarker differences between research conditions can be correlated with this specific symptom of MDD. Stratification of patients is not limited to symptoms, but patients can be stratified also by their age, treatment response, genetic information, cognitive task performance or availability of specific monoamine receptors etcetera. By creating such subsets both heterogeneity of subject groups and influence of DSM-IV criteria will decrease and due to this researchers can better validate potential biomarkers for specific subsets.

4.3 Are biological markers useful to predict whether MDD patients will

effectively respond to antidepressants, or not?

This research field aims to identify biomarkers that emerge early in the course of treatment and indicate whether the received medication is likely to lead to remission (Leuchter et al., 2010). Therefore, these biomarkers will help to determine which AD(s) are effective in each individual. Results of studies involved with biomarkers that predict outcome of ADs are promising. The most robust biochemical test is the combined Dex/ CRH test that showed that the hormonal response measured with this test correlates with treatment response (Paez-Pereda & Panhuysen, 2008). However according to Schule and colleagues (2009), early hormonal response is not a sufficient condition for a favourable response. So, unfortunately, this test may recommend discontinuing with an effective antidepressant. Other promising results include associations of treatment outcome with monoamine markers, sBDNF levels or with QEEG signals. Further research is needed to investigate the effectiveness of these biomarkers.

Although no polymorphism is localized to predict treatment response, pharmacogenomics is already used in clinics, although to a limited extent. Hoop and colleagues (2010) investigated the ordering of genotyping by clinicians who treated depressed patients and they found that these tests were ordered approximately 20 times a year. Most clinicians only did so for patients who were not responding to medical therapy. In our opinion clinicians may better not use genotypes to determine their medicine prescription, because treatment prediction of polymorphisms are still inconclusive. So even though the idea of personalized medicine is attractive to many, it will not be feasible until further research confirms its effectiveness (Raelson & Belouchi, 2008).

(18)

18

4.4 Is the shift of focus moving from monoamine systems toward molecular

mechanisms: a good or bad alteration?

Numerous studies investigated whether deficiencies of the monoamine system could explain the cause of MDD or explain treatment effects of ADs in patients. Unfortunately it could not. While it is widely embraced that monoamine deficiencies are involved with MDD, it has been accepted that this may explain some but not all aspects of MDD. Nowadays it seems more obvious that MDD is caused by structural or functional changes in particular molecules in the brain, and that ADs function by counteracting these molecular changes. In the past decade, more and more studies have examined potential biomarkers for MDD diagnosis and prediction of treatment outcome. Although no clinical use has yet emerged from biomarker research some results are promising.

But should we be more sceptical in regard to biomarkers? Could we really have the pretention to model neurological deficiencies of MDD? Furthermore, could we make such a model by using small changes of molecules? Somewhat paradoxical, at least in our view, since it becomes more and more clear that MDD is caused by complicated downstream neurobiological changes and that different brain areas are involved with the pathophysiology of MDD (macro level), while researchers started to investigate changes of the smallest particles of the brain (micro level). However, the idea behind this is that small alterations can have major consequences in the brain. This means that further research has to integrate massive amounts of information about small biochemical alterations to understand excessive complicated processes and interactions. This will at least increase knowledge about brain function in general, and probably it will increase knowledge of deficiencies in MDD patients. Therefore, we think that the shift of focus from monoamine systems toward molecular mechanisms was a good alteration. Furthermore, we think that biomarker use will lead to more effective and eventually more accessible diagnoses. Finally, we have high expectations of treatment outcome predictors.

4.5 Ethical issues

Lakhan & Vieira (2011) summarized some ethical issues in regard to biomarkers. They indicated that the main question is whether MDD patients benefit from an objective and technology-based diagnosis. Or whether such diagnoses contribute to discrimination or stigmatization of having a psychiatric disorder. Although an objective diagnosis may help to avoid an unmerited stigmatization, it is unlikely that a biochemical test is 100% reliable, and some subjects will still get an erroneous diagnosis. So, some subjects will get negative test results, while these subjects should need serious care. In contrast, others will get positive test results while MDD symptoms would never have come to express and these subjects were better off without medication.

To overcome these ethical issues, before adapting any biomarker-based technology into medical practices, clinicians have to evaluate the potential risk and benefits of patients. Furthermore, clinicians have to remember that the experiences of patients are the most important, not their DNA, protein profile or underlying neurological abnormalities. The highest priority of clinicians is still to improve quality of life of their patients, even if this means that they have to ignore test results. Imaging that you are a psychiatrist seeing a patient complaining of severe depression. You measure a certain biomarker. A few days later, the patient returns, and you hold the laboratory results in your hands. There is no evidence for altered biochemical processes. Would you decide that because the patient’s measures are normal, the patient cannot be depressed?

(19)

19

Part 5. Recommendations for further research

MDD is a leading cause of disability worldwide and it causes a big financial burden. To date, too many MDD patients suffer too long from the effects of this disorder because treatment is ineffective. Therefore, we have to continue with depression research and the search for effective treatments. One way to do this is to identify biological differences between depressed and non-depressed subjects. Previous studies reported promising results and even a small chance that we can determine the cause of MDD should be grabbed. In this chapter we give some recommendations on how to improve current research methods.

Like we already mentioned researchers should be more aware of the fact that a DSM-IV diagnosis is just a label instead of a ‘real disease’. Instead of continuing with this syndrome dominated thinking, researchers have to create certain subsets within the MDD population. These patients can be stratified on certain behaviourally, symptomatically and/ or biological aspects. In this way subject groups will become more homogeneous and biomarker alterations may therefore be better determined.

One way to understand MDD is to view this clinical syndrome as a final behavioural endpoint reached by many biological pathways (Toups & Travidi, 2012). For example in one patient MDD is caused by alterations of the HPA-axis, while in another it is caused by alterations of sBDNF levels. With this view, there is no single marker that captures enough information to identify all paths. This theory could declare the conflicted results found in previous research and why some patients respond to ADs, while others do not. In order to further investigate this, we recommend researchers to use a panel of biomarkers instead of investigating only one biomarker at a time. By doing so, they can identify whether some patients show biological alterations of one marker compared to healthy subjects, while others show other alterations. This information can be combined with specific subsets within the MDD population. Note that researchers should be aware of the fact that it is unlikely that these pathways are totally independent of one another. This system-wide approach may reflect better deficiencies of individual MDD patients.

Finally, we give some recommendations in regard to biomarker selection. Should further research focus on genetics, transcriptomics, proteomics, metabolomics or neuroimaging? To our view approaches based on genetics are not likely to provide further interesting candidate biomarkers in the near future. The initial goal of identifying a limited number of genes serving as markers or predictors had to be abandoned (Lakhan & Vieira, 2011). The interpretation of the complex relationship between genes and behaviour has shown very limited levels of reproducibility (Iris, 2008). Too many association studies characterized a myriad of different candidate genes, while other studies did not confirm their findings. Genomic biomarkers have (probably) a low penetrance and do not directly or predictably have an impact on a subject’s phenotype (Schwarz and Bahn, 2008). In addition, many of these candidate genes are not specific for one neuropsychiatric disorder but show frequent overlap between different disorders (Lakhan & Vieira, 2011).

Approaches based on inflammatory processes may provide further interesting candidate biomarkers in the near future. However, we are sceptical due to several reasons: (1) elevated inflammatory levels observed in MDD patients are far more modest than increases typically observed in autoimmune or infectious diseases; (2) individuals with high inflammatory activity are often able to retain a good mood and hopeful stance towards their lives; (3) the values for any given inflammatory marker always overlap between groups of depressed and not-depressed subjects; (4) Inflammation is neither necessary nor sufficient to cause MDD (Raison and Miller, 2011).

(20)

20

Approaches based on proteomics and metabolomics are more likely to provide further interesting candidate biomarkers in the near future (Paez-Pereda & Panhuysen, 2008). Recent technical developments have made it possible to screen a high number of proteins in a very short period of time. Thanks to these advances it is now possible to analyse several thousands of proteins in one single step (Lakhan & Vieira, 2011). The disadvantages of this approach are the complicated use of CSF tissue in clinical practice, the high number (several thousand) of proteins that are expressed in brain tissue and the current high cost of analysis (Lakhan & Vieira, 2011). Furthermore, approaches based on neuroimaging are likely to provide further interesting treatment outcome predictors. One disadvantage of this approach is that even though brain activity reflects underlying alterations, these underlying alterations will be still unknown. To improve pharmacological treatment, researchers are interested in biochemical cause(s) of MDD. Therefore, researchers have to investigate biomarkers based on proteomics and metabolomics as well as neuroimaging for MDD patients.

(21)

21

References

Aberg, E., A. Fandino-Losada, L. K. Sjoholm, Y. Forsell, and C. Lavebratt, 2011, The functional

Val158Met polymorphism in catechol-O-methyltransferase (COMT) is associated with depression and motivation in men from a Swedish population-based study: J.Affect.Disord., v. 129, no. 1-3, p. 158-166.

Alvarez, J. C., N. Gluck, A. Fallet, A. Gregoire, J. F. Chevalier, C. Advenier, and O. Spreux-Varoquaux, 1999, Plasma serotonin level after 1 day of fluoxetine treatment: a biological predictor for

antidepressant response?: Psychopharmacology (Berl)., v. 143, no. 1, p. 97-101.

Anand, A., Y. Li, Y. Wang, K. Gardner, and M. J. Lowe, 2007, Reciprocal effects of antidepressant treatment on activity and connectivity of the mood regulating circuit: an FMRI study:

J.Neuropsychiatry Clin.Neurosci., v. 19, no. 3, p. 274-282.

Anand, A., Y. Li, Y. Wang, J. Wu, S. Gao, L. Bukhari, V. P. Mathews, A. Kalnin, and M. J. Lowe, 2005a, Activity and connectivity of brain mood regulating circuit in depression: a functional magnetic resonance study: Biol.Psychiatry, v. 57, no. 10, p. 1079-1088.

Anand, A., Y. Li, Y. Wang, J. Wu, S. Gao, L. Bukhari, V. P. Mathews, A. Kalnin, and M. J. Lowe, 2005b, Antidepressant effect on connectivity of the mood-regulating circuit: an FMRI study:

Neuropsychopharmacology, v. 30, no. 7, p. 1334-1344.

Baekken, P. M., F. Skorpen, E. Stordal, J. A. Zwart, and K. Hagen, 2008, Depression and anxiety in relation to catechol-O-methyltransferase Val158Met genotype in the general population: the Nord-Trondelag Health Study (HUNT): BMC.Psychiatry, v. 8, p. 48-55.

Belzeaux, R., C. Formisano-Treziny, A. Loundou, L. Boyer, J. Gabert, J. C. Samuelian, F. Feron, J. Naudin, and E. C. Ibrahim, 2010, Clinical variations modulate patterns of gene expression and define blood biomarkers in major depression: J.Psychiatr.Res., v. 44, no. 16, p. 1205-1213.

Castren, E., 2005, Is mood chemistry?: Nat.Rev.Neurosci., v. 6, no. 3, p. 241-246.

Cavanagh, J., J. Patterson, S. Pimlott, D. Dewar, J. Eersels, M. F. Dempsey, and D. Wyper, 2006, Serotonin transporter residual availability during long-term antidepressant therapy does not

differentiate responder and nonresponder unipolar patients: Biol.Psychiatry, v. 59, no. 4, p. 301-308. Cerullo, M. A., C. M. Adler, M. P. Delbello, and S. M. Strakowski, 2009, The functional neuroanatomy of bipolar disorder: Int.Rev.Psychiatry, v. 21, no. 4, p. 314-322.

Chen, B., D. Dowlatshahi, G. M. MacQueen, J. F. Wang, and L. T. Young, 2001, Increased hippocampal BDNF immunoreactivity in subjects treated with antidepressant medication: Biol.Psychiatry, v. 50, no. 4, p. 260-265.

Costafreda, S. G., C. Chu, J. Ashburner, and C. H. Fu, 2009, Prognostic and diagnostic potential of the structural neuroanatomy of depression: PLoS.One., v. 4, no. 7, p. e6353.

Cunha, A. B., B. N. Frey, A. C. Andreazza, J. D. Goi, A. R. Rosa, C. A. Goncalves, A. Santin, and F. Kapczinski, 2006, Serum brain-derived neurotrophic factor is decreased in bipolar disorder during depressive and manic episodes: Neurosci.Lett., v. 398, no. 3, p. 215-219.

Drevets, W. C., 2001, Neuroimaging and neuropathological studies of depression: implications for the cognitive-emotional features of mood disorders: Curr.Opin.Neurobiol., v. 11, no. 2, p. 240-249.

(22)

22 Eller, T., V. Vasar, J. Shlik, and E. Maron, 2009, The role of IL-2 and soluble IL-2R in depression and antidepressant response: Curr.Opin.Investig.Drugs, v. 10, no. 7, p. 638-643.

Ellis, P. M., and C. Salmond, 1994, Is platelet imipramine binding reduced in depression? A meta-analysis: Biol.Psychiatry., v. 36, no. 5, p. 292-299.

Erk, S., A. Mikschl, S. Stier, A. Ciaramidaro, V. Gapp, B. Weber, and H. Walter, 2010, Acute and sustained effects of cognitive emotion regulation in major depression: J.Neurosci., v. 30, no. 47, p. 15726-15734.

Frodl, T., M. Jager, I. Smajstrlova, C. Born, R. Bottlender, T. Palladino, M. Reiser, H. J. Moller, and E. M. Meisenzahl, 2008, Effect of hippocampal and amygdala volumes on clinical outcomes in major depression: a 3-year prospective magnetic resonance imaging study: J.Psychiatry Neurosci., v. 33, no. 5, p. 423-430.

Geuze, E., E. Vermetten, and J. D. Bremner, 2005, MR-based in vivo hippocampal volumetrics: 2. Findings in neuropsychiatric disorders: Mol.Psychiatry, v. 10, no. 2, p. 160-184.

Harmer, C. J., P. J. Cowen, and G. M. Goodwin, 2011, Efficacy markers in depression: J.Psychopharmacol., v. 25, no. 9, p. 1148-1158.

Hoop, J. G., M. I. Lapid, R. M. Paulson, and L. W. Roberts, 2010, Clinical and ethical considerations in pharmacogenetic testing: views of physicians in 3 "early adopting" departments of psychiatry: J.Clin.Psychiatry, v. 71, no. 6, p. 745-753.

Iris, F., 2008, Biological modelling in the discovery and validation of cognitive dysfunctions

biomarkers. In: Biomarkers for Depressive Disorders, Springer, New York (Chapter 19), pp. 473-522 Ising, M., and F. Holsboer, 2008, Biomarkers for the development of antidepressant and anxiolytic drugs. In: Biomarkers for Depressive Disorders, Springer, New York (Chapter 17), pp. 427-443 Ising, M. et al., 2007, Combined dexamethasone/corticotropin releasing hormone test predicts treatment response in major depression - a potential biomarker?: Biol.Psychiatry., v. 62, no. 1, p. 47-54.

Ising, M., H. E. Kunzel, E. B. Binder, T. Nickel, S. Modell, and F. Holsboer, 2005, The combined dexamethasone/CRH test as a potential surrogate marker in depression:

Prog.Neuropsychopharmacol.Biol.Psychiatry, v. 29, no. 6, p. 1085-1093.

Isogawa, K., H. Nagayama, T. Tsutsumi, A. Kiyota, J. Akiyoshi, and K. Hieda, 2005, Simultaneous use of thyrotropin-releasing hormone test and combined dexamethasone/corticotropine-releasing

hormone test for severity evaluation and outcome prediction in patients with major depressive disorder: J.Psychiatr.Res., v. 39, no. 5, p. 467-473.

Jain, K.K., 2010. Biomarkers of disorders of the nervous system. In: The handbook of Biomarkers, Springer, New York (Chapter 7), pp. 327-396.

Jevtovic, S., D. Karlovic, A. Mihaljevic-Peles, V. Seric, N. Vrkic, and N. Jaksic, 2011, Serum Brain-derived neurotrophic factor (BDNF): the severity and symptomatic dimensions of depression: Psychiatr.Danub., v. 23, no. 4, p. 363-369.

(23)

23 Kessler, R. C., P. Berglund, O. Demler, R. Jin, D. Koretz, K. R. Merikangas, A. J. Rush, E. E. Walters, and P. S. Wang, 2003, The epidemiology of major depressive disorder: results from the National

Comorbidity Survey Replication (NCS-R): JAMA, v. 289, no. 23, p. 3095-3105.

Kirchheiner, J., L. Bertilsson, H. Bruus, A. Wolff, I. Roots, and M. Bauer, 2003, Individualized medicine - implementation of pharmacogenetic diagnostics in antidepressant drug treatment of major

depressive disorders: Pharmacopsychiatry, v. 36 Suppl 3, p. S235-S243.

Koolschijn, P. C., N. E. van Haren, G. J. Lensvelt-Mulders, H. E. Hulshoff Pol, and R. S. Kahn, 2009, Brain volume abnormalities in major depressive disorder: a meta-analysis of magnetic resonance imaging studies: Hum.Brain Mapp., v. 30, no. 11, p. 3719-3735.

Kronmuller, K. T., J. Pantel, S. Kohler, D. Victor, F. Giesel, V. A. Magnotta, C. Mundt, M. Essig, and J. Schroder, 2008, Hippocampal volume and 2-year outcome in depression: Br.J.Psychiatry, v. 192, no. 6, p. 472-473.

Lakhan, S.E., and K.F. Vieira, 2011,The etical ramifications of biomarker use for mood disorders. In: Handbook of Schizophrenia Spectrum Disorders. Springer, New York (Chapter 18), pp. 421-437 Lanquillon, S., J. C. Krieg, U. Bening-Abu-Shach, and H. Vedder, 2000, Cytokine production and treatment response in major depressive disorder: Neuropsychopharmacology, v. 22, no. 4, p. 370-379.

Lee, S., J. Jeong, Y. Kwak, and S. K. Park, 2010, Depression research: where are we now?: Mol.Brain, v. 3, p. 8-18.

Leuchter, A. F. et al., 2010, Biomarkers to predict antidepressant response: Curr.Psychiatry Rep., v. 12, no. 6, p. 553-562.

Li, M., J. K. Soczynska, and S. H. Kennedy, 2011, Inflammatory biomarkers in depression: an opportunity for novel therapeutic interventions: Curr.Psychiatry Rep., v. 13, no. 5, p. 316-320. Lucini, V., A. Lucca, M. Catalano, and E. Smeraldi, 1996, Predicting value of tryptophan/large neutral amino acids ratio on selective serotonin reuptake inhibitors: J Affect Disord. v. 22, no. 36, p. 129-133. MacQueen, G. M., K. Yucel, V. H. Taylor, K. Macdonald, and R. Joffe, 2008, Posterior hippocampal volumes are associated with remission rates in patients with major depressive disorder:

Biol.Psychiatry, v. 64, no. 10, p. 880-883.

Mathers, C. D., and D. Loncar, 2006, Projections of global mortality and burden of disease from 2002 to 2030: PLoS.Med., v. 3, no. 11, p. e442.

Mitchell, A. J., A. Vaze, and S. Rao, 2009, Clinical diagnosis of depression in primary care: a meta-analysis: Lancet, v. 374, no. 9690, p. 609-619.

Mossner, R., O. Mikova, E. Koutsilieri, M. Saoud, A. C. Ehlis, N. Muller, A. J. Fallgatter, and P. Riederer, 2007, Consensus paper of the WFSBP Task Force on Biological Markers: biological markers in

depression: World J.Biol.Psychiatry, v. 8, no. 3, p. 141-174.

Mossner, R. et al., 2006, Transmission disequilibrium of polymorphic variants in the tryptophan hydroxylase-2 gene in children and adolescents with obsessive-compulsive disorder: