Brain Structural Effects of Antidepressant Treatment in Major Depression

Abstract

Depressive disorder is a very frequent and heterogeneous syndrome. Structural imaging techniques offer a useful tool in the comprehension of neurobiological alterations that concern depressive disorder. Altered brain structures in depressive disorder have been particularly located in the prefrontal cortex (medial prefrontal cortex and orbitofrontal cortex, OFC) and medial temporal cortex areas (hippocampus). These brain areas belong to a structural and functional network related to cognitive and emotional processes putatively implicated in depressive symptoms. These volumetric alterations may also represent biological predictors of response to pharmacological treatment. In this context, major findings of magnetic resonance (MR) imaging, in relation to treatment response in depressive disorder, will here be presented and discussed.

INTRODUCTION

Depression is a very common disease, being an important cause of burden worldwide [1]. Patients with depressive disease have different clinical outcome with some of them facing a benign course of illness and others presenting severe, recurrent and remitting episodes [2]. To date, the neuropathological alterations that sustain this wide syndromic entity and the mechanisms behind drug response are still not completely understood [3, 4]. Many hypothesis have been proposed regarding the alterations involved in the neurobiology of depressive disorder: most of them have been focused on brain areas that are implicated in the circuits of serotonin and norepinephrine [5], which are also the main target neurotransmitters of antidepressants [6-9]. Impaired neural circuits that involve these neurotransmitters and encode cognitive and emotional functions have been observed in depressive disorder [10]. Brain areas that are implicated in these networks have been studied with neuroimaging techniques that have revealed specific structural [11-13] and functional alterations [14] features.

Magnetic resonance imaging (MRI) studies on depressive disorder have shown volume reductions in frontal regions, hippocampus, putamen and caudate nucleus [15-18], anterior cingulate [19], amygdala [20], as well as white matter hyperintense lesions [21]. Alterations in limbic and prefrontal areas are suggested to be involved in affective symptoms, such as exaggerated response to negative emotions, guilt, hopelessness and despair, whereas alterations in hypothalamus, locus coeruleus and periacqueductal grey matter may be involved in neurovegetative and neuroendocrine alterations, such as sleep and appetite disturbances, loss of weight, psychomotor retardation or agitation [22].

Imaging methods are required to overcome classical nosological definitions based on syndromic clinical descriptions, by offering reliable models of neuro-morphological alterations that can explain those phenotypes commonly considered as part of the disease [23]. In addition, imaging methods have also been applied to evaluate treatment response to antidepressants among responder and non-responder patients, in order to establish trait markers of depression biological features of refractory illness and predictors of clinical outcome [24, 25]. While phar-macotherapy is often prescribed as an effective intervention for depressive disorder, not all patients who undergo antidepressant treatment get significant amelioration, with almost one third of them failing to achieve remission even after several pharmacological trials [26]. Despite prescriptions of pharmacological therapy for depressive disorder are incresing world wide, treatment response is still uncertain and there is a lack of indicators of those group of patients who can actually benefit from antidepressant therapy or not. In general, neurobiological features are relevant key factors in understanding the course of illness in psychiatric disorders, therefore their identification would help both clinical and research progress. Characterizing homogeneous diagnostic groups based on neural features will facilitate genetic investigations on the etiology of depression and will ameliorate the effectiveness of interventions [27, 28]. In this perspective, evidence from imaging research has shed light on the importance of early diagnosis and intervention on major depression [29-31]. In particular, MRI offers, in vivo, the possibility to explore the structural basis of the response mechanisms to antidepressants, allowing to improve the understanding of how these compounds work on the brain, and ultimately lead to clinical improvement.

We here first review the most robust findings on brain areas involved in depressive disorder and then debate them in relationship to antidepressant treatment response.

MAJOR STRUCTURAL BRAIN ALTERATIONS IN DEPRESSIVE DISORDER

Whole Brain

Total brain volume and whole gray matter volume have not consistently been reported to be altered in adult and geriatric patients with depressive disorder [32-37], as recently confirmed by a meta-analysis [38]. White matter hyperintensities have also been reported, particularly in older patients [38-41].

FRONTAL LOBES AND CINGULATE

Patients with depressive disorder have been shown lower volume in prefrontal/orbitofrontal cortex (OFC) [32, 42-44] and anterior cingulate [33, 37, 38, 42, 45, 46]. These findings were not always replicated in other research studies [38, 47]. Interestingly, shrinkage of medial prefrontal cortex in drug naïve patients [48], along with anterior cingulate volume reduction, seem to progress over time at a faster rate than healthy controls [49]. In regards of cortical folding, lower gyrification index has been reported in orbitofrontal and cingulate cortices as well as in the insula and temporal operculum [12].

TEMPORAL LOBES

Evidence of reduced temporal gray matter has been found by some studies [42, 50], though not confirmed by others [33, 51, 52]. Furthermore, an inverse relationship between superior temporal gyrus volume and length of illness has been shown in unmedicated patients [53]. On the other hand, greater cortical thickness was reported on some paralimbic areas located in the temporal lobes, such as temporal poles in first episode [54] and child-adolescence depression [55].

BASAL GANGLIA AND THALAMUS

Volume reduction in putamen and caudate nucleus have been reported in depression [36-38, 42, 56, 57]. Hyperintensities in putamen and globus pallidus have also been shown, mostly in elderly patients [58, 59], being possible predictors of lack of response in geriatric population [60-63].

Reduced thalamus have been observed in patients with depression by some reports [34, 42, 48], whereas others found no difference [64, 65].

HIPPOCAMPUS AND AMYGDALA

Smaller hippocampal volumes have consistently been observed in chronic, acute and non remitting patients with depression [18, 49, 66-69], as confirmed by a comprehensive meta-analysis [15], with progressive reduction over time [70]. Amygdala has also been reported as reduced by some studies [71-75], particularly in unmedicated [53, 76, 77] and recurrent patients [51, 53, 78]. However, preserved [74, 79-84] or even enlarged amygdala volumes [49, 84-88] have been observed. Inconsistency across findings has been attributed to differences in illness stage, age and gender composition of samples and effect of treatment [15, 89, 90].

CEREBELLUM AND BRAINSTEM

Depressed patients have been reported to have smaller cerebellar [42, 91, 92] and brain stem [92] volumes compared to healthy controls. However, the literature is limited by the paucity of the studies.

VENTRICLES AND CEREBROSPINAL FLUID (CSF)

A vast majority of studies has shown larger volume of third [93-96], lateral ventricles [95, 97-101] and CSF in patients with depressive disorder, whereas some others found preserved size [33, 92, 102-106]. A recent meta-analysis reported a trend of enlargement of CSF among patients with depression, but remarked the heterogeneity across studies [38].

CORPUS CALLOSUM

Enlarged corpus callosum has been observed by one study [107], though not confirmed by others [36, 57, 108].

PITUITARY

A reduced volume in pituitary gland has been reported by one study [109], whereas others did not find any difference [110], even longitudinally after remission [111]. Morover, in a recent meta-analysis that included psychotic depression [112], pituitary gland was larger in depressed patients, relative to healthy controls [38].

EFFECTS OF ANTIDEPRESSANT TREATMENT ON BRAIN MORPHOLOGY IN DEPRESSION

Several imaging studies have assessed the effects of antidepressant treatment - in terms of duration, efficacy and compounds - on brain volumes in patients with depressive disorder. The purpose of this kind of investigation is twofold: on one hand to investigate how treatment affects brain anatomy; on the other hand, to understand the mechanism of action of antidepressants. Furthermore, clinical outcome of patients with depressive disorder can be fairly variable and prediction of treatment response is a challenge for development of reliable therapies [113].

As previously mentioned, there is a convergent research line focusing on hippocampus as a possible biomarker of depressive disorder, even in relation to clinical outcome and treatment effect. Indeed, a smaller hippocampal volume has been associated with severity of depression [114, 115], early onset [116-118], refractory illness [70, 114, 119, 120], longer duration of untreated depression [121], comorbidity with childhood abuse [122] and high levels of disease burden [72, 123-125] or anxiety [126, 127].

In this context, increased right hippocampal volumes have been found in female responders compared to non-responders after eight weeks of fluoxetine treatment [114]; female responders also had larger caudate nucleus compared to male responders and to female non responders [128, 129]. This may indicate a modulatory response effect influenced by gender [67, 130]. In another (one-year) longitudinal study non remitting patients to various antidepressants (including fluxovamine, paroxetine, sertraline, citalopram, venlafaxine, mirtazapine, amitriptyline, doxepine, trimipramine and reboxetine) had lower bilateral hippocampal volumes both at baseline and at follow-up, compared to remitted patients [70], whereas patients who remitted at 3 years follow-up had lower shrinkage of hippocampus [49]. In agreement with these observations, larger hippocampal volumes at baseline predicted remission after antidepressant treatment [131, 132], whereas lower volumes predicted relapse or lack of remission [119, 131]. However, it has to be noted that no effects on hippocampal size after remission mostly with SSRIs have been found in patients with major depression by one study [68].

Morover, imaging studies have produced interesting findings on the structural effects of antidepressant treatment in prefrontal areas. Larger frontal cortical thickness [131] and medial frontal gyrus, dorsolateral prefrontal cortex (DLPFC) and cingulate cortex volumes [24, 133, 134] predicted remission after antidepressant treatment [84, 128]. Furthermore, effective treatment with fluoxetine and sertraline determined enlargement in middle frontal gyrus, DLPFC and OFC [48, 135]. Accordingly, it has been shown that remission correlates to more preserved volumes of anterior cingulate, dorsomedial prefrontal cortex and DLPFC over time [49, 136]. Finally, geriatric patients previously exposed to antidepressant treatment had larger OFC volumes compared to drug-naïve patients [32].

However, even though there is convergent evidence from MRI studies of antidepressant effect on an altered prefrontal-limbic network, some controversial results deserve consideration. Drug-naïve, first-episode patients with depression and comorbid panic disorder treated with duloxetine for 6 weeks had only subtle enlargement of infero-frontal areas, although they underwent clinical amelioration [137]. Furthermore, in two other longitudinal studies, no volumetric changes were observed to SSRIs or nortriptiline, though in presence of clinical response [68, 138], and no predictive property of hippocampal volume was observed [66].

DISCUSSION

Based on the above reviewed literature, it looks that depressive disorder is characterized by an altered structural network that encompasses reduced volumes of OFC, anterior cingulate, hippocampus, and striatum with enlarged ventricles. Also, in regards to the structural effects of antidepressants, taken together the literature’s results underlie the implication of hippocampus, DLPFC and cingulate cortex in their neurobiological mechanisms. In this regards, other imaging techniques that specifically assess white matter, such as diffusion tensor imaging (DTI), corroborated these data showing an impaired fiber integrity connecting cingulate, DLPFC, and hippocampus in non remitter patients [136, 139]. Therefore, such brain areas may represent the biological markers of treatment response and outcome in depressive disorder.

Effective antidepressant treatment might have a neurobiological impact on depressive disorder by reducing structural shrinkage processes in hippocampus and prefrontal cortex, based on a putative neuroprotective or neuro-modulatory effect [140, 141]. In this perspective, antidepressants SSRIs have caused an increase of volume in cingulate subdivisions and precuneus in healthy controls under short administration, confirming a structural remodeling, independent of depressive illness, by serotonergic neurotransmission [142]. Serotonin and norepinephrine have indeed been observed in some reports to enhance neuro-trophic factors, such as BDNF, that increases neurogenesis on grey matter [143]. Furthermore, according to fMRI studies, antidepressants affect this network by reversing hyperactivation of limbic areas to emotional stimuli and by enhancing frontal cortex and cingulate top-down modulatory influence on subcortical structures [144-146].

However, larger studies, focused on specific compounds administered longitudinally to drug-naive patients are needed to finally clarify the impact of antidepressants on brain morphology in major depression. Indeed, a major limitation of the studies presented here is the low sample size, with most studies below 50 subjects and only one over 100 [62, 131]. Moreover, it is impossible to draw conclusions on the effect of single specific compounds because many studies include multiple antidepressants [49, 67, 68, 84, 119, 132, 133, 147] and very few studies focused on the effect of “non SSRIs” antidepressants [138, 148, 149]. Although these designs are closer to real world interventions which apply multiple drugs [150] they limit the possibility to investigate the contribute of different compounds to the brain altered circuits, in order to plan more effective and targeted interventions.

CONFLICT OF INTEREST

The authors confirm that this article content has no conflict of interest.