Candidate genes expression profile associated with antidepressants response in the gendep study: differentiating between baseline ‘predictors’ and longitudinal ‘targets’

Neuropsychopharmacology (2012), 1–9& 2012 American College of Neuropsychopharmacology. All rights reserved 0893-133X/12 Candidate Genes Expression Profile Associated withAntidepressants Response in the GENDEP Study: Differentiatingbetween Baseline ‘Predictors' and Longitudinal ‘Targets' Annamaria Cattaneo1, Massimo Gennarelli1,2, Rudolf Uher3, Gerome Breen3, Anne Farmer3, Katherine J Aitchison3,4, Ian W Craig3, Christoph Anacker5, Patricia A Zunsztain5, Peter McGuffin3 and Carmine M Pariante*,5 Department of Biomedical Sciences and Biotechnology, Genetic and Biology Section, University of Brescia, Brescia, Italy; 2Genetic Unit, IRCCS San Giovanni di Dio, Fatebenefratelli Centre, Brescia, Italy; 3Institute of Psychiatry, MRC Social, Genetic and Developmental Psychiatry, King's College London, London, UK; 4Department of Psychiatry, University of Alberta, Edmonton, Canada; 5Department of Psychological Medicine, Institute of Psychiatry, Section of Perinatal Psychiatry and Stress, Psychiatry and Immunology (SPI-lab), King's College London, London, UK To improve the ‘personalized-medicine' approach to the treatment of depression, we need to identify biomarkers that, assessed before starting treatment, predict future response to antidepressants (‘predictors'), as well as biomarkers that are targeted by antidepressants and change longitudinally during the treatment (‘targets'). In this study, we tested the leukocyte mRNA expression levels of genes belonging to glucocorticoid receptor (GR) function (FKBP-4, FKBP-5, and GR), inflammation (interleukin (IL)-1a, IL-1b, IL-4, IL-6, IL-7, IL-8, IL-10, macrophage inhibiting factor (MIF), and tumor necrosis factor (TNF)-a), and neuroplasticity (brain-derived neurotrophic factor (BDNF), p11 and VGF), in healthy controls (n ¼ 34) and depressed patients (n ¼ 74), before and after 8 weeks of treatment with escitalopram or nortriptyline, as part of the Genome-based Therapeutic Drugs for Depression study. Non-responders had higher baseline mRNA levels of IL-1b ( þ 33%), MIF ( þ 48%), and TNF-a ( þ 39%). Antidepressants reduced the levels of IL-1b (  6%) and MIF (  24%), and increased the levels of GR ( þ 5%) and p11 ( þ 8%), but these changes were not associated with treatment response. In contrast, successful antidepressant response was associated with a reduction in the levels of IL-6 (  9%) and of FKBP5 (  11%), and with an increase in the levels of BDNF ( þ 48%) and VGF ( þ 20%)—that is, response was associated with changes in genes that did not predict, at the baseline, the response. Our findings indicate a dissociation between ‘predictors' and ‘targets' of antidepressant responders. Indeed, while higher levels of proinflammatory cytokines predict lack of future response to antidepressants, changes in inflammation associated with antidepressant response are not reflected by all cytokines at the same time. In contrast, modulation of the GR complex and of neuroplasticity is needed to observe a therapeutic antidepressant effect.
Neuropsychopharmacology advance online publication, 19 September 2012; Keywords: antidepressant; depression; glucocorticoid receptor; inflammation; neuroplasticity; treatment response; personalised medicine available, only a third of patients respond adequately totreatment, and up to half of them relapse within 1 year Antidepressants are commonly prescribed drugs, but the Unfortunately, we still cannot predict the treatment protocols are dictated by clinical practice and likelihood of response of an individual patient to a specific personal preferences rather than by a ‘biomarker-based' drug . Therefore, there is a pressing need to personalized medicine approach (These drugs identify biomarkers that, assessed before starting treatment, are used in patients with major depression, one of the most ‘predict' future response, as well as biomarkers that are common psychiatric disorders and a leading cause of ‘targeted' by antidepressants and change longitudinally disability worldwide (). However, during antidepressant treatment. In addition to fostering despite the increasing variety of antidepressants currently personalized medicine, establishing ‘predictors' and ‘target'biomarkers could lead to the identification of novel *Correspondence: Professor CM Pariante, Department of Psycholo- pathophysiological pathways relevant to depression, and gical Medicine, Institute of Psychiatry, Section of Perinatal Psychiatry thus novel mechanisms for designing therapeutic strategies.
and Stress, Psychiatry and Immunology, Kings College London, Room Based on the current conceptualization of depression, we 2-055, The James Black Centre, 125 Coldharbour Lane, London SE5 suggest that hypothesis-driven, blood-based biomarker 9NU, UK, Tel: þ 44 0 20 7848 0807, Fax: þ 44 0 20 7848 0986,E-mail: analysis should focus on the biological systems that Received 30 April 2012; revised 13 July 2012; accepted 30 July 2012 have been more consistently described as abnormal in Gene expression profile in the GENDEP study depression: the glucocorticoid receptor (GR) complex, serum and in the leukocytes mRNA of depressed patients, inflammation, and neuroplasticity .
and that, in turn, pharmacological and non-pharmacologi- One of the most consistent biological findings in depres- cal antidepressant therapies increase BDNF to levels similar sion is a hyperactivity of the hypothalamic–pituitary– to those in healthy controls ; adrenal (HPA) axis as shown by a multitude of studies describing high levels of levels of BDNF- and neuroplasticity-related genes are cortisol, the main HPA axis hormone, in the context of therefore another class of important candidate biomarkers reduced function of the GR, the cortisol receptor primarily in relation with antidepressant response.
involved in HPA axis regulation during stress. This reduced In this study, to capture a comprehensive picture of the GR function, or glucocorticoid resistance, is particularly biological and clinical interaction between these three bio- evident in patients with treatment-resistant depression logical systems in relationship with antidepressant treat- ment, we examined the gene expression (blood mRNA via persistent glucocorticoid resistance during antidepressant PaxGene tubes) of 15 genes belonging to the GR (three genes), treatment is associated with early relapse ; inflammation (nine genes), and neuroplasticity (three Moreover, polymorphisms in the GR genes) pathways. We studied a well-characterized group of (or NR3C1) gene, and in the gene for the GR-associated, depressed patients from the Genome-based Therapeutic FKBP-5 co-chaperone protein, have been shown to regulate Drugs for Depression (GENDEP) study GR function and to predict antidepressant treatment response ), before and after 8 weeks of treatment with one of two pharmacologically different antidepressants: the selective ). Therefore, the expression levels of GR-related genes serotonin reuptake inhibitor, escitalopram, and the tricyclic are important candidate biomarkers in relation with noradrenaline reuptake inhibitor, nortryptline (and in a group of matched controls). We selected all patients (n ¼ 74) A second biological system potentially involved in who were drug-free for at least 2 weeks before enrolling into antidepressant response is inflammation. Pro-inflammatory the trial, and had provided PaxGene tubes at both baseline cytokines, and in particular interleukin (IL)-1b, IL-6, and and follow-up (after 8 weeks of antidepressants). We mea- tumor necrosis factor (TNF)-a, are increased in depressed sured the transcriptional levels of the following genes: patients as compared with controls in for the GR complex, FKBP-4, FKBP-5, and GR; for the turn, antidepressants have been shown to have anti- inflammatory system, IL-1a, IL-1b, IL-4, IL-6, IL-7, IL-8, inflammatory effects (), and anti- IL-10, macrophage inhibiting factor (MIF), and TNF-a; and inflammatory drugs, such as celecoxib and TNF-a antago- for neuroplasticity, BDNF, p11, and VGF (non-acronymic).
nists, have been shown to have antidepressant properties or We wished to answer three questions: first, which genes at to improve antidepressant response ( baseline (ie, before starting antidepressant treatment) Interestingly in this context, higher levels of inflammation differentiate depressed patients vs controls; second, which seem to identify depressed patients who are less likely to genes, again at baseline, predict treatment response to respond to antidepressant treatment ; subsequent antidepressant treatment (‘predictors'); and third, which genes, assessed prospectively (ie, at baseline polymorphisms in immune genes, such as in IL-1b, IL-11, and after 8 weeks of antidepressant treatment) change in and TNF-a, have been associated with reduced responsive- parallel with treatment response (‘targets').
ness to antidepressant therapy ; . Consistent with this notion, theelevated levels of IL-1b, IL-6, and TNF-a tend to diminish in MATERIALS AND METHODS parallel with antidepressant response (; Of note is also the proposed modelthat the above-mentioned glucocorticoid resistance results The GENDEP project is an open-label part-randomized from the direct molecular action of the activated inflam- multicentre pharmacogenetic study with two active phar- matory pathways on the GR complex, and that, in turn, the macological treatment arms glucocorticoid resistance maintains the inflammatory status It was designed to establish clinical and genetic by reducing the inhibitory control of endogenous gluco- determinants of therapeutic response to two antidepres- corticoids on the immune system sants with different primary modes of action: nortriptyline This model suggests a common pathogenic process under- and escitalopram. The overall study has been extensively lying both the HPA axis and the inflammatory abnormal- described before (see Supplementary Material for details).
ities in depression, and thus identifies expression levels of In total, 9 psychiatric centers in 8 European countries inflammatory genes as further important candidate bio- recruited 811 adult outpatients (296 men and 514 women), markers in relation with antidepressant response.
aged between 19 and 72 (mean age 42.5, SD ¼ 11.8) suffering Finally, one of the potential mechanisms by which from unipolar depression of at least moderate severity excessive HPA axis activity and inflammatory responses according to International Classification of Diseases 10/ may contribute to the pathogenesis of depression is through Diagnostic and Statistical Manual of Mental Disorders, inhibition of neurotrophic factors and disturbance of fourth edition, and established by the semi-structured neuroplasticity The neurotrophin, SCAN interview.
brain-derived neurotrophic factor (BDNF), is the most Severity of depressive symptoms and treatment response studied molecule within the neuroplasticity network, and we was assessed by weekly administration of three established and others have shown that BDNF levels are reduced in the measures of depression severity: the clinician-rated 10-item Gene expression profile in the GENDEP studyA Cattaneo et al Montgomery–Asberg Depression Rating Scale (MADRS; ), the 17-item Hamilton Data were analyzed using the Statistical Package for Social Rating Scale for Depression (HRSD–17; ), Sciences, version 17.0 (SPSS). Continuous variables are and the Beck Depression Inventory (BDI; ).
presented as mean±SD or SEM, as indicated. Categorical The average participant in the original sample was in his/ variables were tested by means of w2 and Fisher's tests.
her second episode of depression and scored 28.7 (SD ¼ 6.7) Univariate analysis of variance was used for comparing the on the MADRS, 21.8 (SD ¼ 5.3) on the 17-item HDRS, and mean values of the mRNA levels of the genes of interest, at 28.2 (SD ¼ 9.7) on the BDI.
baseline, in patients vs controls and in responders vs non- A psychometric analysis has found that the MADRS was responders. Changes over time were analyzed using the the most internally consistent and informative of the three repeated-measures General Linear Model with time (T0 and scales , and therefore in the pharmaco- T8) and response (yes/no) as factors. The Greenhouse– genetic analyses already published and Geisser correction was applied to degrees of freedom when in the current gene expression study, we have used the the sphericity assumption was violated. Parametric correla- MADRS score as a primary outcome measure of ‘treatment tion analyses using Pearson's coefficient were used to test response'. Response to antidepressant medication was the association between genes and the improvement in the quantified as percentage reduction in the MADRS score depressive symptoms measured as changes in the MADRS from baseline to week 12, and responders were identified as score. Linear regression analyses were used to test for patients with a reduction in MADRS450%.
predictors of the treatment outcome.
For this study, we selected all patients who had been drug-free for at least 2 weeks before entering into the trial and Biomarkers Differences between Patients at Baseline, provided a baseline and a follow-up PaxGene tube (n ¼ 74).
Their average (SD) age was 38.3±10.9 and, there were 31males and 43 females. The average participant in our study Patients (at baseline) and controls differed in the expression was in his/her second episode of moderately severe of most of the examined genes (Specifically, we depression and scored, at baseline, 28.7 (SD ¼ 4.2) on the found that depressed patients, as compared with controls, MADRS, 20.7 (SD ¼ 4.1) on the HRSD–17, and 27.5 had higher FKBP-5 mRNA levels ( þ 27%, F ¼ 69.4, (SD ¼ 10.2) on the BDI. There were no significant differ- po0.0001) and lower GR mRNA levels (  18%, F ¼ 63.2, ences between patients treated with escitalopram (n ¼ 38) or po0.0001). Moreover, they had higher mRNA levels of nortryptiline (n ¼ 36), in age (38±12.4 vs 36±9.4, p ¼ 0.25), IL-1b, ( þ 48%, F ¼ 117.9, po0.0001), IL-6 ( þ 24%, F ¼ 86.3, gender (F/M was 20/18 vs 23/13, p ¼ 0.2), and in the po0.0001), MIF ( þ 32%, F ¼ 34.8, po0.0001), and TNF-a response rate (responders/non-responders was 26/12 vs 25/ ( þ 58%, F ¼ 87.7, po0.0001), and lower levels of IL-4 11, p ¼ 0.6).
(  9%, F ¼ 5.6, p ¼ 0.02). Finally, depressed patients had Controls were recruited in London (UK), through adver- lower mRNA levels of BDNF (  24%, F ¼ 46.5, po0.0001), tisement in local newspapers, hospitals, and job centers, as p11 (  16%, F ¼ 12.1, p ¼ 0.001), and VGF (  36%, F ¼ 37.3, well as from existing volunteer databases. Controls were screened using the Psychosis Screening Questionnaire), and were excluded if they Baseline Differences in Biomarkers between Responders met criteria for a present or past psychotic disorder, or if and Non-Responders (‘Predictors') taking any kind of hormonal treatment; their average agewas 35.2 (SD ¼ 8), and there were 19 males and 15 females.
As mentioned above, treatment response was defined as a There were no significant differences in age and gender percentage reduction 450% in the MADRS score from between patients and controls (p ¼ 0.14 for age, and p ¼ 0.13 baseline to week 12. In this sample, we had 51 responders for gender distribution).
and 23 non-responders: 26 responders to escitalopram, 12non-responders to escitalopram, 25 responders to nortrypt- Gene Expression Analyses line, and 11 non-responders to nortryptline. There were nodifferences in age (38.3±1.6 vs 38.4±2.2 years, Fo0.1, We measured the leukocytes mRNA levels of the above- p ¼ 0.98), gender distribution (F/M ¼ 31/20 vs 12/11, w2 ¼ 0.5, mentioned candidate genes involved in GR function (FKBP- p ¼ 0.3), or baseline MADRS (26.8±0.6 vs 25.0±0.8, 4, FKBP-5, and GR), inflammatory system (IL1-a, IL-1b, IL- F ¼ 3.0, p ¼ 0.09) between responders and non-responders; 4, IL-6, IL-7, IL-8, IL-10, MIF, and TNF-a), and neuroplas- moreover, study center also did not influence treatment ticity (BDNF, p11, and VGF). Each sample was assayed in response (p ¼ 0.3).
duplicate, and each target gene was normalized to the We compared the baseline mRNA levels of each gene in patients who did not respond to treatment vs patients who 3-phosphate dehydrogenase, beta-actin, and beta-2-micro- did Only three genes were differentially globulin. The expression levels of each target gene were expressed: specifically, non-responders had higher mRNA normalized to the geometric mean of all three reference levels of the three pro-inflammatory cytokines, IL-1b ( þ 33%, genes, and the Pfaffl method was used to determine relative F ¼ 55.9, po0.0001), MIF ( þ 48%, F ¼ 14.6, po0.0001), and target gene expression of each gene in patients as compared TNF-a ( þ 39%, F ¼ 39.4, po0.0001). Moreover, for MIF with controls (see Supplementary Material for details).
levels we observed a significant drug  response interaction Gene expression profile in the GENDEP study Table 1 Expression Levels of Genes Belonging to GR Functionality (FKBP-4, FKBP-5, and GR), Inflammation (IL-1a, IL-1b, IL-4, IL-6, IL-7, IL-8,IL-10, TNF-a, and MIF), and Neuroplasticity (BDNF, p11, and VGF) in Controls and Depressed Patients with Statistics (p and F values andPercentage Changes) Controls (mean±SEM) Patients (mean±SEM) Percentage change (%) Table 2 Expression Levels of Genes Belonging to GR Functionality (FKBP-4, FKBP-5, and GR), Inflammation (IL-1a, IL-1b, IL-4, IL-6, IL-7, IL-8,IL-10, TNF-a, and MIF), and Neuroplasticity (BDNF, p11, and VGF) in Non-Responder and Responder Patients Non-responder (n ¼ 23) Responder (n ¼ 51) Drug  response interaction Statistics (F and p values) are presented for both main effects and for the drug  response interaction (26 responders to escitalopram, 12 non-esponders toescitalopram, 25 responders to nortryptline, and 11 non-responders to nortryptline).
(F ¼ 4.4, p ¼ 0.04): this was due to the difference between as changes in the MADRS score between week 0 and week non-responders and responders being larger for non- 12 (DMADRS), both in the overall sample and separately responders to nortriptyline ( þ 56%, F ¼ 73.2, po0.0001) based on the drug used. As expected, the expression than for those non-responders to escitalopram ( þ 39%, levels of IL-1b, MIF, and TNF-a at baseline were all strongly F ¼ 36.9, po0.0001). There was no drug  response inter- and negatively correlated with the treatment outcome, action for any of the other genes.
both in the entire group (IL-1b, r ¼  0.56; MIF, r ¼  0.62; We further examined the relative contributions of the and TNF-a, r ¼  0.44; all po0.0001), and also separately three cytokines in predicting treatment response measured in the two samples based on the drug used (for Gene expression profile in the GENDEP studyA Cattaneo et al escitalopram: IL-1b, r ¼  0.54, p ¼ 0.001; MIF, r ¼  0.73, (F ¼ 4.4, p ¼ 0.039) interactions. Namely, IL-6 levels po0.0001; and TNF-a, r ¼  0.39, p ¼ 0.016; and for decreased significantly in responders (  9%, F ¼ 20.3, nortriptyline, IL-1b, r ¼  0.65 po0.0001; MIF, r ¼  0.56 po0.0001), and this was present for both responders to po0.0001; and TNF-a, r ¼  0.68, po0.0001). We then run escitalopram (  12%, F ¼ 14.0, p ¼ 0.001) and to nortripty- a linear regression model to identify the relative contribu- line (  6%, 0.2, F ¼ 6.6, p ¼ 0.02). In non-responders there tions of the three cytokines to the prediction of response. As was no overall effect ( þ 1%, F ¼ 0.4, p ¼ 0.5) but, when the shown in the best predictive model was obtained two drugs were analyzed separately, IL-6 did not change in when the three cytokines were all included in the model, the non-responders to escitalopram (  2%, F ¼ 0.5, p ¼ 0.5) both in the overall samples (46% of the variance) and and increased in the non-responders to nortriptyline separately in the escitalopram-treated group (53% of the ( þ 7%, F ¼ 5.8, p ¼ 0.037).
variance) and in nortriptyline-treated group (48% of the Finally, four genes were regulated by antidepressant treatment, irrespective of the antidepressant used or oftreatment response, as shown by a main effect of time in theabsence of any response  time or drug  timeinteractions.
Change in Biomarkers and Relationship with Treatment Specifically, antidepressant treatment significantly reduced Response (‘Targets') the expression levels of IL-1b (  6%, F ¼ 7.9, p ¼ 0.006) and To investigate the effect of 8 weeks of antidepressant treat- MIF (  24%, F ¼ 16.4, po0.0001), and increased GR mRNA ment with escitalopram or nortriptyline on gene expression levels ( þ 5%, F ¼ 7.3, p ¼ 0.009) and p11 levels ( þ 8%, (and its relationship with treatment response), we compared F ¼ 8.4, p ¼ 0.005).
the change in mRNA levels of each gene between baseline(T0) and week 8 (T8). These data are presented in Three genes were regulated by antidepressant treatment but in responders only, and regardless of the antidepressant To provide evidence supporting a personalized-medicine used, as shown by significant response  time interactions approach to the treatment of depression, we have assessed but no drug  time interactions. Specifically, antidepressant the blood mRNA expression of 15 candidate genes across treatment significantly reduced FKBP5 levels only in three biological systems implicated in the pathogenesis of patients who responded to the treatment (  11%, F ¼ 16.4, depression and in the action of antidepressants: GR po0.0001), whereas no effect was observed in non-res- complex, inflammation, and neuroplasticity. We have used ponders (  2%, F ¼ 0.6, p ¼ 0.45; response  time interaction, a well-characterized group of drug-free depressed patients F ¼ 4.4, p ¼ 0.04; drug  timeinteraction, F ¼ 0.05, p ¼ 0.8).
who entered a randomized trial with two different anti- Moreover, antidepressant treatment significantly increased depressants, and we have assessed genes that differentiate VGF expression only in responders ( þ 20%, F ¼ 15.4, patients from controls, predict future antidepressants response, po0.0001) but not in non-responders (  3%, F ¼ 0.002, or change in association with response. The main finding is p ¼ 0.97; response  time interaction, F ¼ 4.4, p ¼ 0.039; a dissociation between genes that predict treatment response drug  timeinteraction, F ¼ 0.03, p ¼ 0.8). Finally, antide- (‘predictors') and genes that change longitudinally in patients pressant treatment increased BDNF expression more in the who respond (‘targets'). Specifically, among the 15 genes, responders ( þ 48%, F ¼ 126.4, po0.0001) than in the non- only higher levels of the three inflammation-related genes, responders ( þ 21%, F ¼ 49.4, po0.0001; response  time IL-1b, MIF, and TNF-a, predict lack of response to interaction, F ¼ 17.8, po0.0001; drug  time interaction, antidepressants, and successful antidepressant response is F ¼ 3.6, p ¼ 0.062).
not associated with a reduction in the levels of these genes.
IL-6 was regulated by antidepressant treatment but in a In contrast, successful antidepressant response is associated drug- and response-specific way, that is, in the presence of with a reduction in the levels of the inflammation-related both response  time (F ¼ 10.0, p ¼ 0.002) and drug  time gene, IL-6, and of the GR-associated gene, FKBP-5; and with Table 3 Adjusted R2 Values and Significance of Linear Regression Model to Assess the Contribution of TNF-a, MIF, and IL-1b, Alone or inCombination, in Predicting the Treatment Response in the Whole Sample, in the Escitalopram-Treated Group and in the Nortriptyline-Treated Group Whole sample (n ¼ 74) Escitalopram-treated group (n ¼ 38) Nortriptyline-treated group (n ¼ 36) Effects of each single cytokine as predictor of treatment response Adjusted R2 ¼ 0.19, po0.0001 Adjusted R2 ¼ 0.13, p ¼ 0.016 Adjusted R2 ¼ 0.45, po0.0001 Adjusted R2 ¼ 0.37, po0.0001 Adjusted R2 ¼ 0.51, po0.0001 Adjusted R2 ¼ 0.29, po0.0001 Adjusted R2 ¼ 0.31, po0.0001 Adjusted R2 ¼ 0.27, p ¼ 0.001 Adjusted R2 ¼ 0.41, po0.0001 Combined effects of cytokines as predictor of treatment response Adjusted R2 ¼ 0.19, po0.0001 Adjusted R2 ¼ 0.13, p ¼ 0.016 Adjusted R2 ¼ 0.45, po0.0001 Adjusted R2 ¼ 0.40, po0.0001 Adjusted R2 ¼ 0.50, po0.0001 Adjusted R2 ¼ 0.47, po0.0001 TNF-a þ MIF þ IL-1b Adjusted R2 ¼ 0.46, po0.0001 Adjusted R2 ¼ 0.53, po0.0001 Adjusted R2 ¼ 0.48, po0.0001 Gene expression profile in the GENDEP study Table 4 Expression Levels of Genes Belonging to GR Functionality (FKBP-4, FKBP-5, and GR), Inflammation (IL-1a, IL-1b, IL-4, IL-6, IL-7, IL-8,IL-10, TNF-a, and MIF), and Neuroplasticity (BDNF, p11, and VGF) in Depressed Patients before (Patients T0) and after 8 Weeks ofAntidepressant Treatment (Patients T8) Time  response interaction Time  drug interaction Statistics (F and p values) are presented for the time effect as well as the time  response and time  drug interactions. The numbers of patients for each group were:26 responders to escitalopram, 12 non-responders to escitalopram, 25 responders to nortryptline, and 11 non responders to nortryptline).
an increase in the neuroplasticity-associated genes, VGF sants have anti-inflammatory properties ; and BDNF—that is, in genes that are not associated with the here we also demonstrate that the changes baseline prediction of treatment response.
in inflammation associated with antidepressant response The baseline levels of TNF-a, IL1-b, and MIF, together are not reflected by all cytokines at the same time. This is predict the treatment outcome for both antidepressants, with perhaps not surprising, because the action of a single clinically significant effect sizes of around 50% of the cytokine is regulated within a complex network, where variance (Moreover, although each single multiple pro-inflammatory cytokines maybe involved in the cytokine strongly correlates with the treatment response, the antidepressant response, and a reduction in inflammation best predictive model is present when we include all three may be signaled by an increase in the expression of negative cytokines in the linear regression, suggesting that each regulators of cytokine action as well as by a reduction in cytokine taps also in non-shared molecular mechanisms. For cytokines level. It is also possible that a simple reduction in example, IL1-b and TNF-a, but not MIF, activate the the levels of some inflammatory biomarkers, such as IL1-b indoleamine 2,3-dioxygenase pathway and MIF, is not sufficient for reversing the biological ), which is responsible for the changes underlying the depressive symptoms, because catabolism of tryptophan to kynurenine, and results in the potentially downstream effectors pathways, such as MAPK production of the neurotoxic and depressogenic metabolites, and nuclear factor kappa B, remain abnormal 3-hydroxykynurenine and quinolinic acid ). At the opposite end, persistently high levels of other Indeed, we have recently shown that this pathway is inflammatory cytokines, such as TNF-a, may be a marker of activated by IL-1b also in human neurons ( specific forms of depression that are less responsive to In contrast, MIF, but not IL-1b and TNF-a, is released antidepressant treatment, eg, depressed patients with a in response to glucocorticoids, and, when secreted, it history of childhood trauma tend to have higher inflam- renders immune cells less sensitive to the anti-inflammatory matory biomarkers and to be less effects of glucocorticoids ( responsive to pharmacological intervention It is also of note that antidepressant treatment in our ; in this case, the increased study reduces the levels of IL1-b and MIF, but this reduction cytokine levels may be an underlying ‘trait feature' conferring is not associated with treatment response. In contrast, levels treatment resistance, and may not be modifiable by of TNF-a, which are elevated in non-responders, are not modified by antidepressant treatment; and levels of IL-6 The expression of four genes changes only in patients decrease following antidepressants, but in responders only.
who respond to treatment, but not in those who do not Of note, this last finding is remarkably consistent with the respond. These genes are across the three biological systems previous paper by , who also found investigated: the GR-associated gene, FKBP-5; the pro- that IL-6 plasma levels after 8 weeks of antidepressant inflammatory cytokine, IL-6 (as mentioned above); and the treatment were reduced in responders only. Although these neuroplasticity associated genes, VGF and BDNF. The data confirm previous evidence showing that antidepres- association between reduction in FKBP-5 and response to Gene expression profile in the GENDEP studyA Cattaneo et al antidepressants has never, to our knowledge, been de- the two drugs, that is, belonging to the serotonergic and the scribed before; but previous studies have demonstrated that noradrenergic pathways, respectively. This is also in line polymorphisms in the gene encoding this co-chaperone are with the results of our previous genetic study in the same associated with antidepressant response The main limitation of this study is that our finding have recently shown that the pattern of cannot yet be used to change clinical practice. Notwith- RNA levels of GR-stimulated genes, including FKBP-5, standing the strengths of the randomized, longitudinal design, discriminate between patients and controls these findings need to be replicated in an independent sample.
Taken together, our data (see also the changes in GR Moreover, the predictive biomarkers that we have identified discussed below) suggest that depression is characterized by are not specific to one or the other of the antidepressants, the coexistence of higher FKBP-5 and lower GR, leading to and as such could not be used to guide the choice of anti- GR resistance, and that successful antidepressant treatment depressants but rather to identify patients that potentially requires normalization of GR function. Finally, treatment may be helped by early access to adjuvant therapies.
response was associated with an increase in BDNF and VGF.
Moreover, it is possible that these gene expression changes The evidence for a role of BDNF in depression is abundant are not causally involved in the treatment response, and It is also of note that the indeed we have not investigated potential biological mechan- recent study from describing patients isms that are changed by antidepressants and may underlie who responded to an open-labeled treatment with venlafax- the gene expression changes that we have described, such as ine, also showed an increase in serum BDNF by week 3 of changes in immune cells composition or in cortisol levels.
treatment. Also of interest is a large cross-sectional study In conclusion, our findings identify for the first time that again showing that depressed the baseline levels of MIF, IL1-b, and TNF-a are ‘predictors' patients had lower levels of serum BDNF, whereas remitted of antidepressant treatment response. Moreover, we show patients (a different group) had normal levels. There are less that an enhancement of GR function and an improvement data on the role of VGF in depression, but previous studies in neuroplasticity are needed to observe a response to have shown that VGF modulates hippocampal function and antidepressant therapies, suggesting that future antidepres- behavior through an effect that is BDNF-dependent and that sant strategies should specifically target these pathways.
it is involved both in the pathogenesis of depression and inthe effects of antidepressants We also find that GR and p11 levels are lower in depressed patients compared with controls, and that their levels The GENDEP project was supported by a European Commis- increase after antidepressants, but not in relationship with sion Framework 6 grant (contract reference: LSHB-CT-2003- treatment response. Both lower levels of GR and lower levels 503428). This specific project has been supported by a grant of p11, which is considered a GR-target gene via two specific from the Commission of European Communities Seventh glucocorticoid response elements in its promoter Framework Programme (Collaborative Project Grant Agree- , have been described before in depressed ment no. 22963, Mood Inflame) and a Clinician Scientist Fellowship from the Medical Research Council, UK (G108/ knowledge, this is the first report describing prospective 603) to CMP; by a NARSAD young investigator award to changes in p11 levels; and only one previous study has PAZ; by a grant from the Psychiatry Research Trust, UK assessed GR expression prospectively using radioactive (McGregor 97) to CMP and AC; by grants from the Italian dexamethasone binding in peripheral blood mononuclear Ministry of Health (Ricerca Corrente) and Regione Lom- cells, and, again, found a reduction of GR density in depressed bardia (ID:17387Sal-13) to MG; and by salary support to patients (compared with controls) and an increase after CMP, GB, and PZ, and a Studentship to CA, from the antidepressant treatment ). Of note, we National Institute for Health Research Mental Health have consistently shown that antidepressants regulate the Biomedical Research Centre at South London and Maudsley function and the expression of the GR NHS Foundation Trust and King's College London.
Moreover, a recent study in rats has shown that escitalopram increases brain p11 levels concomitantly to adecrease in the p11 promoter methylation ( All funding regarding this project is described in the Indeed, our own research in human neuronal stem acknowledgment section. Dr Pariante in the last 3 years has cells has recently shown that antidepressants directly increase received fees as a speaker or as a member of advisory board, the function of the GR and, via a GR-dependent mechanism, as well as research funding, from pharmaceutical companies the expression of p11: an action that is required for the effects that commercialize or are developing antidepressants, such of antidepressants on neurogenesis as Lilly, Servier, and Janssen. All other authors declare no It is interesting to note that, except for IL-6, we have conflict of interest.
found that all the examined genes are modulated in thesame way by both drugs, even if escitalopram and nortry-ptiline have different modes of action. We believe this is due to the fact that we have examined genes that belong to Anacker C, Zunszain PA, Carvalho LA, Pariante CM (2011). The pathways that are in common to both drugs glucocorticoid receptor: pivot of depression and of antidepres- rather than genes that are modulated differently by sant treatment? Psychoneuroendocrinology 36: 415–425.
Gene expression profile in the GENDEP study Anacker C, Zunszain PA, Cattaneo A, Carvalho LA, Garabedian MJ, treatment in people with depression: a meta-analysis. Psychol Thuret S et al (2011). Antidepressants increase human Med 16: 1–12.
hippocampal neurogenesis by activating the glucocorticoid Horstmann S, Lucae S, Menke A, Hennings JM, Ising M, Roeske D receptor. Mol Psychiatry 16: 738–750.
et al (2010). Polymorphisms in GRIK4, HTR2A, and FKBP5 show Bauer ME, Papadopoulos A, Poon L, Perks P, Lightman SL, interactive effects in predicting remission to antidepressant Checkley S et al (2003). Altered glucocorticoid immunoregula- treatment. Neuropsychopharmacology 35: 727–740.
tion in treatment resistant depression. Psychoneuroendocrinol- Ising M, Horstmann S, Kloiber S, Lucae S, Binder EB, Kern N et al ogy 28: 49–65.
(2007). Combined dexamethasone/corticotropin releasing hor- Bebbington P, Nayani T (1995). The psychosis screening mone test predicts treatment response in major depression-a questionnaire. Int J Methods Psychiatr Res 5: 11–19.
potential biomarker? Biol Psychiatry 62: 47–54.
Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J (1961). An Janssen DG, Caniato RN, Verster JC, Baune BT (2010). A inventory for measuring depression. Arch Gen Psychiatry 4: psychoneuroimmunological review on cytokines involved in antidepressant treatment response. Hum Psychopharmacol 25: Benedetti F, Lucca A, Brambilla F, Colombo C, Smeraldi E (2002).
Interleukine-6 serum levels correlate with response to anti- Juruena MF, Cleare AJ, Papadopoulos AS, Poon L, Lightman S, depressant sleep deprivation and sleep phase advance. Prog Pariante CM (2010). The prednisolone suppression test in Neuropsychopharmacol Biol Psychiatry 26: 1167–1170.
depression: dose-response and changes with antidepressant Binder EB (2009). The role of FKBP5, a co-chaperone of the treatment. Psychoneuroendocrinology 35: 1486–1491.
glucocorticoid receptor in the pathogenesis and therapy of Juruena MF, Pariante CM, Papadopoulos AS, Poon L, Lightman S, affective and anxiety disorders. Psychoneuroendocrinology 1: Cleare AJ (2009). Prednisolone suppression test in depression: prospective study of the role of HPA axis dysfunction in Bocchio-Chiavetto L, Bagnardi V, Zanardini R, Molteni R, Nielsen treatment resistance. Br J Psychiatry 194: 342–349.
MG, Placentino A et al (2010). Serum and plasma BDNF levels in Keers R, Uher R, Gupta B, Rietschel M, Schulze TG, Hauser J et al major depression: a replication study and meta-analyses. World J (2010). Stressful life events, cognitive symptoms of depression Biol Psychiatry 11: 763–773.
and response to antidepressants in GENDEP. J Affect Disord 127: Calfa G, Kademian S, Ceschin D, Vega G, Rabinovich GA, Volosin M (2003). Characterization and functional significance of Kitaichi N, Kotake S, Mizue Y, Sasamoto Y, Goda C, Iwabuchi K glucocorticoid receptors in patients with major depression: et al (2000). High-dose corticosteroid administration induces modulation by antidepressant treatment. Psychoneuroendocri- increase of serum macrophage migration inhibitory factor in nology 28: 687–701.
patients with Vogt-Koyanagi-Harada's disease. Microbiol Im-munol 44: 1075–1077.
Capuron L, Neurauter G, Musselman DL, Lawson DH, Nemeroff Lanquillon S, Krieg JC, Bening-Abu-Shach U, Vedder H (2000).
CB, Fuchs D et al (2003). Interferon-alpha-induced changes in Cytokine production and treatment response in major depres- tryptophan metabolism. relationship to depression and parox- sive disorder. Neuropsychopharmacol 22: 370–379.
etine treatment. Biol Psychiatry 54: 906–914.
Lee BH, Kim YK (2010). The roles of BDNF in the pathophysiology Cattaneo A, Bocchio-Chiavetto L, Zanardini R, Milanesi E, of major depression and in antidepressant treatment. Psychiatry Placentino A, Gennarelli M (2010a). Reduced peripheral brain- Investig 7: 231–235.
derived neurotrophic factor mRNA levels are normalized Lekman M, Laje G, Charney D, Rush AJ, Wilson AF, Sorant AJ et al by antidepressant treatment. Int J Neuropsychopharmacol 13: (2008). The FKBP5-gene in depression and treatment response— an association study in the Sequenced Treatment Alternatives Cattaneo A, Sesta A, Calabrese F, Nielsen G, Riva MA, Gennarelli M to Relieve Depression (STAR*D) Cohort. Biol Psychiatry 63: (2010b). The expression of VGF is reduced in leukocytes of depressed patients and it is restored by effective antidepressant Melas PA, Rogdaki M, Lennartsson A, Bjo¨rk K, Qi H, Witasp A treatment. Neuropsychopharmacology 35: 1423–1428.
et al (2011). Antidepressant treatment is associated with Chopra K, Kumar B, Kuhad A (2011). Pathobiological targets of epigenetic alterations in the promoter of P11 in a genetic model depression. Expert Opin Ther Targets 15: 379–400.
of depression. Int J Neuropsychopharmacol 20: 1–11.
Danese A, Moffitt TE, Pariante CM, Ambler A, Poulton R, Caspi A Menke A, Arloth J, Pu¨tz B, Weber P, Klengel T, Mehta D et al (2008). Elevated inflammation levels in depressed adults with (2012). Dexamethasone stimulated gene expression in peripheral a history of childhood maltreatment. Arch Gen Psychiatry 65: blood is a sensitive marker for glucocorticoid receptor resistance in depressed patients. Neuropsychopharmacology 37: 1455–1464.
Dowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK Molendijk ML, Bus BA, Spinhoven P, Penninx BW, Kenis G, et al (2010). A meta-analysis of cytokines in major depression.
Prickaerts J et al (2011). Serum levels of brain-derived neuro- Biol Psychiatry 67: 446–457.
trophic factor in major depressive disorder: state-trait issues, Gustavsson A, Svensson M, Jacobi F, Allgulander C, Alonso J, clinical features and pharmacological treatment. Mol Psychiatry Beghi E et al (2011). Cost of disorders of the brain in Europe. Eur 16: 1088–1095.
Neuropsychopharmacol 21: 718–779.
Montgomery SA, Asberg M (1979). A new depression scale Hamilton M (1967). Development of a rating scale for primary designed to be sensitive to change. Br J Psychiatry 134: 382–389.
depressive illness. Brit J Clin Psychol 6: 278–296.
Nanni V, Uher R, Danese A (2012). Childhood maltreatment Hannestad J, DellaGioia N, Bloch M (2011). The effect of predicts unfavorable course of illness and treatment outcome in antidepressant medication treatment on serum levels of depression: a meta-analysis. Am J Psychiatry 169: 141–151.
inflammatory cytokines: a meta-analysis. Neuropsychopharmacol Nemeroff CB, Heim CM, Thase ME, Klein DN, Rush AJ, Schatzberg 36: 2452–2459.
AF et al (2003). Differential responses to psychotherapy versus Haroon E, Raison CL, Miller AH (2012). Psychoneuroimmunology pharmacotherapy in patients with chronic forms of major meets neuropsychopharmacology: translational implications of depression and childhood trauma. Proc Natl Acad Sci USA 100: the impact of inflammation on behaviour. Neuropsychopharma- cology 37: 137–162.
O'Connor JC, Andre´ C, Wang Y, Lawson MA, Szegedi SS, Lestage J Hiles SA, Baker AL, de Malmanche T, Attia J (2012). Interleukin-6, et al (2009). Interferon-gamma and tumor necrosis factor-alpha C-reactive protein and interleukin-10 after antidepressant mediate the upregulation of indoleamine 2,3-dioxygenase and Gene expression profile in the GENDEP studyA Cattaneo et al the induction of depressive-like behavior in mice in response to take inhibitors (SSRIs) on human T lymphocyte function and bacillus Calmette-Guerin. J Neurosci 29: 4200–4209.
gene expression. Eur Neuropsychopharmacol 17: 774–780.
Pandey GN, Dwivedi Y, Rizavi HS, Ren X, Zhang H, Pavuluri MN Thakker-Varia S, Jean YY, Parikh P, Sizer CF, Jernstedt Ayer J, (2010). Brain-derived neurotrophic factor gene and protein Parikh A et al (2010). The neuropeptide VGF is reduced in human expression in pediatric and adult depressed subjects. Prog bipolar postmortem brain and contributes to some of the behavioral Neuropsychopharmacol Biol Psychiatry 34: 645–651.
and molecular effects of lithium. J Neurosci 30: 9368–9380.
Pariante CM, Alhaj HA, Arulnathan VE, Gallagher P, Hanson A, Thase ME (2006). Preventing relapse and recurrence of depression: Massey E et al (2012). Central glucocorticoid receptor-mediated a brief review of therapeutic options. CNS Spectr 11: 12–21.
effects of the antidepressant, citalopram, in humans: a study using Uher R (2011). Genes, environments and individual differences in EEG and cognitive testing. Psychoneuroendocrinology 37: 618–628.
response to treatment. Harv Rev Psychiatry 19: 109–124.
Pariante CM, Hye A, Williamson R, Makoff A, Lovestone S, Kerwin Uher R, Farmer A, Maier W, Rietschel M, Hauser J, Marusic A et al RW (2003). The antidepressant clomipramine regulates cortisol (2008). Measuring depression: comparison and integration of intracellular concentrations and glucocorticoid receptor expres- three scales in the GENDEP study. Psychol Med 38: 289–300.
sion in fibroblasts and rat primary neurones. Neuropsychophar- Uher R, Huezo-Diaz P, Perroud N, Smith R, Rietschel M, Mors O macol 28: 1553–1561.
et al (2009). Genetic predictors of response to antidepressants in Pariante CM, Lightman SL (2008). The HPA axis in major the GENDEP project. Pharmacogenomics J 9: 225–233.
depression: classical theories and new developments. Trends Uher R, Perroud N, Ng MY, Hauser J, Henigsberg N, Maier W et al Neurosci 31: 464–468.
(2010). Genome-wide pharmacogenetics of antidepressant re- Pariante CM, Miller AH (2001). Glucocorticoid receptors in major sponse in the GENDEP project. Am J Psychiatry 167: 555–564.
depression: relevance to pathophysiology and treatment. Biol Uher R, Tansey KE, Malki K, Perlis RH (2012). Biomarkers Psychiatry 49: 391–404.
predicting treatment outcome in depression: what is clinically Pariante CM, Pearce BD, Pisell TL, Owens MJ, Miller AH (1997).
significant? Pharmacogenomics 13: 233–240.
Wong ML, Dong C, Maestre-Mesa J, Licinio J (2008). Polymorph- receptor by the antidepressant desipramine. Mol Pharmacol 52: isms in inflammation-related genes are associated with suscept- ibility to major depression and antidepressant response. Mol Pariante CM, Thomas SA, Lovestone S, Makoff A, Kerwin RW Psychiatry 13: 800–812.
(2004). Do antidepressants regulate how cortisol affects the Yoshimura R, Hori H, Ikenouchi-Sugita A, Umene-Nakano W, brain? Psychoneuroendocrinol 29: 423–447.
Ueda N, Nakamura J (2009). Higher plasma interleukin-6 (IL-6) Raison CL, Borisov AS, Woolwine BJ, Massung B, Vogt G, Miller level is associated with SSRI- or SNRI-refractory depression.
AH (2010). Interferon-alpha effects on diurnal hypothalamic- Prog Neuropsychopharmacol Biol Psychiatry 33: 722–726.
pituitary-adrenal axis activity: relationship with proinflamma- Yu YW, Chen TJ, Hong CJ, Chen HM, Tsai SJ (2003). Association tory cytokines and behavior. Mol Psychiatry 15: 535–547.
study of the interleukin-1 beta (C-511T) genetic polymorphism Raison CL, Dantzer R, Kelley KW, Lawson MA, Woolwine BJ, Vogt with major depressive disorder, associated symptomatology, and G et al (2010). CSF concentrations of brain tryptophan and antidepressant response. Neuropsychopharmacol 28: 1182–1185.
kynurenines during immune stimulation with IFN-alpha: Zhang L, Li H, Su TP, Barker JL, Maric D, Fullerton CS et al (2008).
relationship to CNS immune responses and depression. Mol Traumatic Stress Brain Study Group, Ursano R.p11 is up- Psychiatry 15: 393–403.
regulated in the forebrain of stressed rats by glucocorticoid Ribeiro SC, Tandon R, Grunhaus L, Greden JF (1993). The DST as a acting via two specific glucocorticoid response elements in the predictor of outcome in depression: a meta-analysis. Am J p11 promoter. Neuroscience 153: 1126–1134.
Psychiatry 150: 1618–1629.
Zunszain PA, Anacker C, Cattaneo A, Carvalho LA, Pariante CM Rojas PS, Fritsch R, Rojas RA, Jara P, Fiedler JL (2011). Serum (2011). Glucocorticoids, cytokines and brain abnormalities brain-derived neurotrophic factor and glucocorticoid receptor in depression. Prog Neuropsychopharmacol Biol Psychiatry 35: levels in lymphocytes as markers of antidepressant response in major depressive patients: a pilot study. Psychiatry Res 189: 239–245.
Zunszain PA, Anacker C, Cattaneo A, Choudhury S, Musaelyan K, Sluzewska A, Sobieska M, Rybakowski JK (1997). Changes in Myint AM et al (2012). Interleukin-1b: a new regulator of the acute-phase proteins during lithium potentiation of antidepres- kynurenine pathway affecting human hippocampal neurogen- sants in refractory depression. Neuropsychobiology 35: 123–127.
esis. Neuropsychopharmacology 37: 939–949.
Snyder JS, Soumier A, Brewer M, Pickel J, Cameron HA (2011).
Adult hippocampal neurogenesis buffers stress responses anddepressive behaviour. Nature 476: 458–461.
This work is licensed under the Creative Spijker AT, van Rossum EF (2012). Glucocorticoid sensitivity in mood disorders. Neuroendocrinology 95: 179–186.
Derivative Works 3.0 Unported License. To view a copy Taler M, Gil-Ad I, Lomnitski L, Korov I, Baharav E, Bar M et al (2007). Immunomodulatory effect of selective serotonin reup- Supplementary Information accompanies the paper on the Neuropsychopharmacology website )

Source: http://people.musc.edu/~carsonds/B2B%20Projects/Hit%20or%20Miss/predictor%20of%20antidepressant.pdf