Hippocampal and Amygdalar Volume Changes in Major Depressive Disorder: A Targeted Review and Focus on Stress

Abstract

Medial temporal lobe structures have long been implicated in the pathogenesis of major depressive disorder. Although findings of smaller hippocampal and amygdalar volumes are common, inconsistencies remain in the literature. In this targeted review, we examine recent and significant neuroimaging papers examining the volumes of these structures in major depressive disorder. A targeted PubMed/Google Scholar search was undertaken focusing on volumetric neuroimaging studies of the hippocampus and amygdala in major depressive disorder. Where possible, mean volumes and accompanying standard deviations were extracted allowing computation of Cohen’s d_s effect sizes. Although not a meta-analysis, this allows a broad comparison of volume changes across studies. Thirty-nine studies in total were assessed. Hippocampal substructures and amygdale substructures were investigated in 11 and 2 studies, respectively. The hippocampus was more consistently smaller than the amygdala across studies, which is reflected in the larger cumulative difference in volume found with the Cohen’s d_s calculations. The left and right hippocampi were, respectively, 92% and 91.3% of the volume found in controls, and the left and right amygdalae were, respectively, 94.8% and 92.6% of the volume of controls across all included studies. The role of stress in temporal lobe structure volume reduction in major depressive disorder is discussed.

Keywords

amygdala chronic stress cornu ammonis hippocampus major depressive disorder medial temporal lobe magnetic resonance imaging neuroimaging

Introduction

Major depressive disorder (MDD) is an increasingly prevalent condition resulting in significant morbidity and mortality.¹ Consistent with the variation in illness presentation, diagnostic criteria for MDD encapsulate a wide range of signs and symptoms.² Core symptoms of MDD comprise pervasively low mood, increased fatiguability and anhedonia.³ However, deficits in neurocognitive function frequently accompany these affective phenomena and include impairment of memory, visuospatial processing and attention.⁴,⁵ This heterogeneity of illness presentation is unsurprising given the multifactorial aetiology of the condition.⁶ Such heterogeneity implicates several brain regions in the neuropathology of MDD.⁷

Characterisation of the links between the neuropathology of depression and a patient’s symptoms is of fundamental importance to academic psychiatrists worldwide. This is evident in the proliferation of neuroimaging studies over the last few years. Significant advances have been made towards understanding the neurobiology of MDD, with imaging studies identifying multiple loci exhibiting volumetric differences in depressed patients versus healthy controls. In particular, medial temporal lobe (MTL) structures such as the amygdala and hippocampus, along with frontal regions, have been implicated as potential biomarkers for depression, with most studies demonstrating volume reduction.⁸,⁹ Many studies have emerged within the literature focusing on hippocampal and amygdalar volumes, but these have yielded some contradictory findings. Understanding these findings may aid our understanding of MDD neurobiology and focus targeted research into these structures.

Background

Ambiguity surrounds the anatomical terminology of these temporal lobe structures, and the definitions of these terms vary in the literature. The hippocampus proper consists of the allocortical cornu ammonis subfields 1-4 (CA1-4), and the hippocampal formation consists of the dentate gyrus (DG) medially along its transverse axis, and the subiculum inferiorly.⁹ Recent research has also illustrated long-axis specialisation of the hippocampus, with both anatomical and functional segregation noted along the anterior and posterior segments.¹⁰ The CA1-4 subfields are components of the canonical trisynaptic circuit (Figure 1). This comprises the perforant path from the entorhinal cortex to granule cells of the DG, which in turn project mossy fibres to pyramidal cells in CA3 from which Schaffer collaterals project to CA1.⁷ The hippocampus is the cardinal neural structure in memory, emotional and cognitive processing, transmitting information across the cortex and to the hypothalamus via the classic Papez circuit.¹¹ The amygdala is arguably even more critical to emotional processing, with reciprocal connections to almost every part of the cortex (Figure 2). There are numerous anatomical and functional regions of the amygdala, which can be further divided into nuclei and nuclear groups. Although multiple classification systems exist, amygdala nuclei are commonly categorised into three groups: the deep laterobasal amygdala containing the lateral (LA) and basal nuclei; the superficial cortical-like nuclei; and centromedial amygdala containing the central (CE) and medial nuclei.¹² Morphometric deviations in both hippocampal and amygdalar substructures have been also studied in MDD more recently with the advent of higher resolution magnetic resonance imaging (MRI) and advanced cortical segmentation.

Figure 1.

Classic trisynaptic circuit. The classic trisynaptic circuit is demonstrated with the green arrows. Information flows into the hippocampus from the adjacent entorhinal cortex through the perforant pathway. It is then processed through the dentate, CA3 and CA1 substructures via the mossy fibre and Schaffer collateral pathways. Information from the entorhinal cortex also bypasses the trisynpatic circuit and enters the CA regions directly by collateral perforant pathways (blue arrows). This facilitates parallel processing of information through the hippocampus. CA: cornu ammonis.

Figure 2.

Hippocampal and amygdalar functional connectivity. The major efferent and afferent connections of the hippocampus and amygdala are shown. Internal and environmental information enters the hippocampus via the entorhinal cortex and outflows via the fornix, cingulum and entorhinal cortex. The amygdala has reciprocal connections with almost every cortical area, including the hippocampus. Amygdalar information also outflows via the stria terminalis and amygdalofugal pathways. ACC: anterior cingulate cortex; EC: entorhinal cortex; Hypo: hypothalamus; PCC: posterior cingulate cortex; SGC: subgenual cortex.

The role of stressful life events in triggering and perpetuating major depressive episodes (MDEs) has received considerable attention in the literature.^13–15 The hippocampus and the amygdala are key players in mediating behavioural and neuroendocrine responses to stress¹⁶ as well as being vulnerable to the neurotoxic effects of stress themselves.¹⁷ Indeed, the role of stress in the development of MDEs has been increasingly investigated in recent years, and a number of mechanisms have been proposed to explain this link including hypothalamic–pituitary–adrenal (HPA) axis dysfunction, neuroinflammation and neurotransmitter perturbations.^18–20

Significant heterogeneity exists in studies of these two key MTL substructures, most notably in those of the amygdala.^21–24 Difficulties in the analysis of multiple papers on this topic arise for several reasons. Firstly, there is a significant disparity in anatomical and imaging definitions of the hippocampus and the amygdale used in research. Automated tissue-segmentation protocols using atlas probabilistic based segmentation, such as the Desikan–Killiany atlas, are an emerging method of examination of neural substructures and allow for greater standardisation and reduced inter-study variability in the future. Secondly, findings must be interpreted with caution due to inter-study variability in terms of study design, including variation in clinical phenotype (including first-episode, remitted, recurrent and treatment-resistant depression), depression severity, patient age, the role of stressful life events and the effects of medication. For example, it has been suggested that multiple MDEs and stress may result in neurotoxic effects, exacerbating global neuronal loss across multiple neural substructures including the hippocampus.¹⁵,²⁵,²⁶ Furthermore, antidepressant medication has been shown to ameliorate such changes in these neural structures, and therefore, patient medication status is an important factor in the interpretation of study findings.²⁷ These factors likely contribute to the heterogeneity of study findings. It is, therefore, difficult to extrapolate these findings to all patients who fall under the broad singular clinical definition of MDD.

The aims of our study were threefold: (1) to review studies examining global and substructural volumes in the hippocampus and amygdala in depressed patients compared to non-depressed individuals, (2) to generate an approximate volume difference for each structure (limitation of the methods not withstanding) and (3) explore the role of stress and neurotoxic processes in these studies. It is hoped that this review will contribute to a deeper understanding of the role of these MTL structures in MDD.

Methods

A targeted search of the literature was undertaken to review recent papers on this subject. We searched MEDLINE, Embase and Cochrane Library databases, using the keywords ‘depression’, ‘amygdala’, ‘hippocampus’, ‘MRI’ and ‘volume’, along with the correct medical subject heading codes and abbreviations. These were paired with the appropriate Boolean operators, and the search was conducted for papers published from 1 January 2009 to present 25 June 2019.

Inclusion criteria were (1) adult patients with a primary diagnosis of MDD by DSM-IV/DSM-V or ICD-10 by a psychiatrist; (2) comparison to healthy controls who were screened for neurological, psychiatric and other medical disorders that may affect brain structure; (3) assessment of volumetric differences using MRI; (4) patients reporting depressive symptoms without a formal diagnosis of MDD; and (5) systematic reviews and meta-analyses of studies fitting the above specifications. Studies were commonly excluded for examining a paediatric cohort. Studies which solely analysed volume following treatment, and articles which assessed volumetric differences post-mortem were also excluded. The references of the included papers were inspected for relevant articles, and other key articles known to the reviewers were also included in our review.

A Cohen’s d_s effect size was calculated by extraction of means and standard deviations of hippocampal and amygdala volumes reported in each study. Only studies that reported means and standard deviation were included in this calculation. Cohen’s d_s is a modified calculation of the Cohen’s d score allowing pooling of means from samples that are not equal in size to be compared. This metric provides a standardised difference between means, i.e. allowing a comparison of the effect sizes between different studies. The calculation was made using the following equation:

\begin{array}{l} C ohen' s d_{s} = \frac{M_{2} - M_{1}}{Pooled S D} \\ P ooled S D = \sqrt{\frac{(n_{1} - 1) \times S D_{1}^{2} + (n_{2} - 1) \times S D_{2}^{2}}{n_{1} + n_{2}}} \\ M_{1} = The mean o f group 1 \\ M_{2} = The mean o f group 2 \\ S D_{1} = The s \tan dard deviation o f group 1 \\ S D_{2} = The s \tan dard deviation o f group 2 \\ n_{1} = The size o f group 1 \\ n_{2} = The size o f group 2 \end{array}

A negative Cohen’s d_s value from our data denotes that the structure volume is smaller in MDD, while a positive value means it is larger in depression. The further away the value is from 0, the larger the difference between controls and depressed in each study. The use of this metric allows for a standardised comparison of the effect size between the means of the groups being discussed across different studies. A percentage decrease in the size of volumes was determined by calculation of the mean value of the average volume sizes in different studies and dividing the mean from depressed participants by the mean from non-depressed participants.

Results

Thirty-nine studies were identified with 21 measuring hippocampus volumes, 9 measuring amygdalar volumes and 9 measuring both hippocampal and amygdalar volumes. A total of 7270 MDD patients and 12,996 controls were investigated across all studies. In addition, six meta-analyses of hippocampal volumes in MDD and four meta-analyses of amygdalar volumes in MDD were also returned. Moreover, 24 studies found smaller hippocampal volumes in MDD (80% of hippocampal studies) and 12 studies found smaller amygdala volumes in MDD (67%); 13 hippocampal studies used automated segmentation, 9 used manual segmentation and 6 used a combination of both. Eight amygdalar studies used automated segmentation, 6 used manual segmentation and 4 used a combination of both; 11 hippocampal studies examined the structure at a substructural level compared to only 2 amygdalar studies. Sixteen hippocampal studies investigated MDD patients with medication compared to 8 without medication, with 4 not reporting medication status. Moreover, 12 amygdalar studies investigated MDD patients with medication compared to 4 without medication, with 1 not reporting medication status. Only two studies overall appeared to control for medication effects.²⁶,²⁸

Using the Cohen’s d_s calculation as the standardized difference between the means, smaller bilateral hippocampi across all studies in MDD was revealed (mean effect size −0.341, range −1.211 to 0.673). The left and right hippocampi were respectively 92% and 91.3% the volume found in controls. Similarly, Cohen’s d_s calculation revealed smaller bilateral amygdalae (mean effect size −0.701, range: −2.927, −0.0293) with the left and right amygdalae showing 94.8% and 92.6% of the volume of controls across all studies.

Discussion

This targeted review explored neuroimaging studies measuring hippocampal (Table 1) and amygdala (Table 2) volumes in MDD. Of the 39 studies returned, the majority showed smaller hippocampi and amygdalae in the disorder. Although not a meta-analysis, by pooling the means and standard deviations from studies that supplied the data we were able to calculate an estimate of the overall differences in hippocampal and amygdala volumes in MDD (Table 3). Across the studies, both structures were smaller in MDD, with the hippocampus showing an approximate 8% volume difference bilaterally and the amygdala showing an approximate 7% difference on the right and 5% difference on the left.

Global Hippocampal Changes in MDD

Within the neuroimaging literature, smaller hippocampal volumes are amongst the most widely replicated findings in MDD. We identified 24 studies showing smaller hippocampi compared to controls. Our Cohen’s d_s calculation strongly supports this wider consensus with an 8% hippocampal volume reduction bilaterally. This is consistent with meta-analyses investigating the hippocampus as a single structure^29–31 or as part of a greater limbic system analysis²¹,³²,³³ reporting volume reductions of 4%–10%. Meta-analyses of voxel-based morphometry data also provides evidence of hippocampal grey matter loss in MDD.^34–36 Most available evidence indicates that volume reduction is associated with longer durations of depressive illness (i.e. MDD chronicity), a greater number of depressive episodes and earlier age of onset of MDD.⁹,³⁷ From these findings emerged the concept of the so-called cumulative depressive load, a combination of duration of episodes and number of episodes over a person’s lifetime.³⁸ A cumulative depressive load of longer than 2 years is associated with smaller hippocampi in MDD.³¹,³⁸ This suggests that a ‘depressive dose’ effect depending on the length of time the individual suffers from active depression may influence hippocampal size. However, hippocampal changes have also been found in first episode patients,⁹,³⁹ suggesting that changes in the structure either occur early in the illness or smaller hippocampal size may represent a vulnerability for the development of MDD.

Some studies show that hippocampal volumes are smaller during depressive episodes and resolve upon illness resolution.⁴⁰ Hippocampal size may represent a state marker of the illness, i.e. present during episodes and absent in recovery. The neuroplastic capacity of the hippocampus is illustrated by normalization/recovery of volumes following pharmacological treatment for MDD,²⁸,^40–43 further giving weight to the notion that hippocampal volume may serve as a biomarker of disease state. One study found that hippocampal volumes increase greater than controls following long-term (three years) antidepressant treatment.⁴⁴ Neurogenesis in the DG of the hippocampus has been proposed as a potential mechanism for the action of antidepressants.⁴⁵

A minority of studies found hippocampal volumes were not smaller in MDD. In particular, two large studies with a combined sample size of 1767 patients failed to demonstrate volumetric differences in the hippocampus between depressed and non-depressed individuals.¹,⁴⁶ There are some difficulties in interpreting these results. Shen et al. used a composite of self-reported symptoms and hospital admission data to estimate probable MDD, rather than an operational MDD diagnosis based on strict structured clinical interviews.⁴⁶ Ancelin et al. investigated a community-based sample and most of the MDD patient group only had one episode at some point in their lifetime, potentially accounting for the lack of hippocampal findings.¹

Although care is needed when interpreting these findings in their totality, the majority of available evidence points towards a disease process at work in the hippocampus in MDD. Hippocampal atrophy may already be present at first episode, with greater volume reduction occurring with further depressive episodes as the disease progresses.⁹,²⁵,²⁹ However, the hippocampus is not a unitary structure and has many individual yet interdependent parts. Changes may occur discretely in subfields rather than across the whole structure. Recently, increased efforts have been made to localise differences found in MDD to individual substructures.

Dentate Gyrus

Various structural neuroimaging studies have found dentate volume reductions along with other substructure changes in depression.⁹,⁴²,⁴⁷,⁴⁸ A study comparing 52 medication-free patients to 51 healthy controls found that the number of prior depressive episodes was correlated with reduced dentate volumes using FreeSurfer 5.3 automated segmentation.⁴⁸ Using the same methods, 20 medication naïve females with MDD were found to have smaller dentate regions compared to 21 controls.⁴⁷ A high-field (4.7 T) manual segmentation study showed smaller dentate volume in nine unmedicated patients compared to eleven medicated patients and twenty seven controls as well as smaller CA (1–3) in the depressed patients.⁴² The dentate was also found to be smaller in a study of 83 depressed individuals compared to 80 controls, using FreeSurfer 6.0 with the difference more pronounced on the left and in those with recurrent depression.⁹

The dentate represents the input circuitry of the hippocampus (Figure 1), funnelling information from the entorhinal cortex into the hippocampus.¹¹ It is one of only two locations where adult neurogenesis occurs.⁴⁹ Decreased dentate neurogenesis may play a role in MDD,⁷ and increased dentate neurogenesis may be a mechanism for the action of antidepressants.⁴⁵ The granule cells of the dentate express high concentrations of glucocorticoid and mineralocorticoid receptors⁵⁰ and appear to be exceptionally vulnerable to circulating cortisol levels compared to other brain regions.⁷ This may be particularly important in MDD which is associated with HPA axis and cortisol perturbations.

Subiculum

The subiculum has been shown to be smaller in some MRI studies in depression (see Table 1).⁹,⁴⁷,⁵¹ A high-field (7 T) manual tracing study of 13 patients with one episode of MDD and 5 patients with multiple episodes, showed smaller subicular volumes with multiple episodes, but these findings are limited in power by small sample size.⁵¹ Moreover, 20 medication naïve females with MDD were found to have smaller subiculum regions compared to 21 controls in a FreeSurfer 5.3 study.⁴⁷ Using the same automated technique, the subiculum was found to be smaller in a study of thirty older patients with MDD.⁵² A FreeSurfer 6.0 study found the subiculum to be smaller in 83 depressed individuals compared to 80 controls, with the difference more pronounced on the left and in those recurrent depression.⁹ The left subiculum was also smaller in a similar study of 102 patients with MDD and 135 controls.⁵³ Hippocampal atrophy has also been found in individuals who reported depressive symptoms,⁵⁴ and this atrophy is correlated to an increasing course of these symptoms.⁵⁵

Table 1.

Hippocampal studies in MDD.

Study	Segmentation	MDD/HC	Drug Status	Illness Duration	Male/Female	Key finding
Studies showing smaller hippocampus in MDD
Han et al.⁵⁴	Automated (FreeSurfer 6.0)	102/135	80 medicated	44.76 months	99/138	Bilateral atrophy in CA1, CA4, granule layer, molecular layer, whole hippocampus in cMDD [Cohen’s d = −0.4976(left), −0.4802(right)]
						Left CA2/3 and right presubiculum and subiculum atrophy in cMDD.
						Degree of atrophy not correlated with illness duration.
Roddy et al.⁹	Automated (FreeSurfer 6.0)	80/83	64 medicated	29 months	57/106	Bilateral atrophy in CA1-4, DG, subiculum in cMDD, more pronounced on left and in rMDD.
						Left CA1 atrophy most marked change overall.
						CA2-4 only structures showing atrophy in first MDE.
Yao et al.⁷⁰	Automated (FIRST)	25/28	19 medicated	-	30/23	Bilateral CA1, subiculum and hippocampal atrophy in cMDD. [Cohen’s d = −0.2917 (left), −0.5819 (right)]

Zaremba et al.,⁹⁰	Automated(SPM8)	213/213	202 medicated	30.58 months	198/228	Atrophy correlated to number of inpatient hospitalisations.
				(Cumulative lifetime duration)
Han et al.⁴⁷	Automated (FreeSurfer 5.3)	20/21	Medication free	114.22 months	0/41	Left subiculum, CA2–4, DG and whole hippocampus atrophy in cMDD [Cohen’s d = −0.5819].
						Right subiculum atrophy in cMDD.
Treadway et al.⁴⁸	Automated (MAGeT Brain)	51/52	Medication free	–	56/47	DG atrophy correlated to number of MDEs.

Brown et al.⁵⁴	Automated (FIRST)	1946 persons	480 medicated	-	804/1132	Severity of depressive symptoms correlated to global hippocampal atrophy, but not left/right individually,
						Correlation strongest in older persons.
Stratmann et al.²⁶	Automated (SPM8)	132/132	126 medicated	14.66 months	114/150	Bilateral hippocampus and parahippocampus atrophy in cMDD.
						Right hippocampus atrophy correlated to number of MDEs.
Arnone et al.⁴⁰	Automated (SPM8)	64/66	Medication free	–	37/93	Bilateral hippocampus atrophy in cMDD and remitted untreated MDD. State marker of illness.
						Hippocampus enlargement in medicated MDD.
						Included in Zhao³⁵ meta-analysis
Huang et al.⁴²	Manual	20/27	11 medicated	–	18/29	CA1-3 and DG atrophy in unmedicated cMDD.
Buddeke et al.⁵⁵	Manual	636 persons	∼50 medicated	–	515/121	Atrophy correlated with severity of depressive symptoms.
Maller et al.,⁹⁴	Manual	182/76	Unclear	–	135/123	Global hippocampal atrophy most marked in tails in TRD. [Cohen’s d = −0.8302 (left), −0.6805(right)]
Schuhmacher et al.²⁷	Manual	86/87	86 medicated	1-year fMDD, 15.51 years rMDD	77/96	Bilateral hippocampus atrophy in rMDD. [Cohen’s d = −0.7373(left), −0.7777 (right)]
Cole et al.⁹²	Manual	37/37	Medication free	–	18/56	Right hippocampus atrophy in first MDE.
						Bilateral hippocampus atrophy in rMDD.
						Subiculum and CA1 atrophy most marked, predominantly in tail. CA2-3 also displayed atrophy.
						Included in Santos 2018, Cole 2011 meta-analysis
Malykhin et al.²⁸	Manual	39/34	16 medicated	-	17/56	Right hippocampus head, and right total hippocampus atrophy in cMDD. [Cohen’s d = −0.5417]
						Bilateral hippocampus tail atrophy in cMDD.
						Hippocampus body enlargement in medicated MDD patients vs. HCs and unmedicated cMDD.
Bearden et al.⁴¹	Manual	31/31	Medication free	11.42 years	14/48	Left hippocampus atrophy, particularly CA1 and subiculum, in severe MDD.
						Left hippocampal enlargement with previous medication.
						Included in Santos 2018 meta-analysis
Kronmuller et al.,⁹⁵	Manual	57/30	57 medicated	5 years	35/52	Left hippocampus atrophy in male patients with first MDE. [Cohen’s d = −1.2106]
				34.90 weeks current episode		No hippocampus atrophy in female patients with first MDE.
						No difference between hippocampus volume in rMDD vs. HCs.
						Included in Cole 2011 meta-analysis.
Santos et al.,⁹⁶	Mixed	1327/1004	Medication free	Range: 7–57 months	965/1366	Bilateral hippocampus atrophy in MDD
Schmaal et al.²⁵	Mixed	1728/7199	623 Medicated			Bilateral hippocampus atrophy in rMDD, but not in first MDE.
						Earlier age of MDD onset correlated with hippocampus atrophy.
						Proportion of medication in studies did not moderate alterations in hippocampus volumes.
Zhao et al.³⁵	Mixed	400/424	Medication free	Range: 0.22–7.5 years first onset	142/258	Bilateral hippocampus and parahippocampus atrophy in MDD.
						Bilateral hippocampus atrophy correlated to illness duration.
						Right hippocampus atrophy correlated to percentage of female patients in each study.
Cole et al.²⁹	Mixed	191/282	93 medicated	months average = 14.4 (range is from 4.7–25.2)	169/304	Bilateral hippocampus atrophy in first MDE.
Videbech and Ravnkilde³⁰	Mixed	351/279	N/A	–	–	Bilateral hippocampus atrophy in MDD.
						Right hippocampus atrophy correlated to number of MDEs.
Belleau et al.¹⁵	N/A	N/A	N/A	–	–	HPA-axis dysfunction, inflammation, oxidative stress, neurotransmitter abnormalities are neurotoxic and facilitate atrophy in MDD.
						Stress-sensitisation: with each successive MDE, it takes less stress to trigger these neurotoxic pathways.
Malykhin and Coupland⁹¹	N/A	N/A	N/A	–	–	Hippocampus atrophy in MDD specific to the DG and CA subfields.
						Antidepressant treatment causes DG enlargement by stimulating neurogenesis, and reverses hippocampus atrophy in MDD.
Studies not showing smaller hippocampus in MDD
Ancelin et al.¹	Automated (FreeSurfer 5.3)	162/448	31 medicated	–	290/230	No hippocampus atrophy in lifetime MDD vs. HCs. [Cohen’s d = 0.6739]

Brown et al.⁷²	Automated (FreeSurfer 6.0)	24/20	Medication free	80.7 months	44/14	No hippocampal subfield atrophy in depressed patients and no correlation to illness duration or severity

Shen et al.⁴⁶	Automated (FAST)	354/803	N/A	–	529/628	No hippocampus atrophy in MDD vs. HCs.

Schermuly et al.⁴³	Automated (SPM2)	15/13	15 medicated	Duration illness: 35.7 m	9/19	No hippocampus atrophy in MDD vs. healthy controls. [Cohen’s d = 0.3461]
				Duration episode: 7.0 m		Posterior hippocampus enlargement during treatment with medication.
Wisse et al.⁵¹	Manual	13/34	4 medicated	–	18/29	No hippocampus subfield atrophy in patients who had MDE in previous 7 years vs. HCs. [Cohen’s d = 0]
						Subiculum atrophy correlated to number of MDEs.
Bora et al.⁶⁹	Mixed	986/937	Mixed	–	878/1045	No hippocampus atrophy in MDD vs. HCs.

The table is divided into studies showing smaller hippocampus in MDD and those showing no smaller hippocampus in MDD. Each section is stratified into automated, manual and mixed segmentation techniques. CA, cornu ammonis; cMDD, current major depressive disorder; DG, dentate gyrus; FAST, FMRIBs automated segmentation tool; FIRST, FMRIBs integrated and registration segmentation tool; HC, healthy controls; N/A, not available; MAGet, multiple automatically generated templates; MDD, major depressive disorder; MDE, major depressive episode; rMDD, recurrent major depression; SPM, statistical parametric mapping.

Table 2.

Amygdalar studies in MDD.

Study	Segmentation	MDD/HC	Drug Status	Illness Duration	Male/Female	Key finding
Studies showing smaller amygdala in MDD
Ancelin et al.¹	Automated (FreeSurfer 5.3)	162/448	31 medicated	–	290/230	Lower global amygdala volume in men associated with lifetime diagnosis of MDD. [Cohen’s d = −2.9271]
Brown et al.⁷²	Automated (FreeSurfer 6.0)	24/20	Medication free	80.7 months	44/14	Left cortical nucleus, left accessory basal nucleus, bilateral corticoamygdaloid transition area – significant negative associations with MDD severity. No relationship to duration.
Daftary et al.⁷⁵	Automated (FreeSurfer 4.4)	1797 persons	320 medicated	–	726/1071	Right amygdala volume negatively correlated to depressive symptom severity in 18–39 years age group.
Yao et al.⁷⁰	Automated (FSL)	25/28	19 medicated	–	30/23	Atrophy of bilateral LA, BLVM; and right LA, BLVM, CE, ASTR, ACO and AAA.
						Right amygdala atrophy [Cohen’s d = −0.5581]
Stratmann et al.²⁶	Automated (SPM8)	132/132	126 medicated	14.66 months	114/150	Right amygdala volume negatively correlated to number of MDEs.
Zavorotnyy et al.,⁹⁷	Manual	23/30	18 medicated	4.4	12/41	Larger left amygdala in late-onset MDD [Cohen’s d = −0.02933]
						Amygdala volume negatively correlated to illness duration.
Schuhmacher et al.²⁷	Manual	86/87	86 medicated	1-year fMDD, 15.51 years rMDD	77/96	Bilateral amygdala atrophy in first MDE. [Cohen’s d = −0.5452(left), −0.7988 (right)]
						Bilateral volume normalisation with medication in recurrent MDD.
Kronenberg et al.⁷⁴	Manual	24/14	Medication free	–	15/23	Bilateral amygdala atrophy in cMDD. [Cohen’s d = −0.8294 (left), −0.8519 (right)]
						Amygdala volume negatively correlated to number of MDEs.
						Volume not related to severity of MDE.
Sheline et al.⁷¹	Manual	20/20	14 medicated	–	0/40	Bilateral laterobasal nuclei atrophy, no global atrophy. [Cohen’s d = −0.94984 (left core nuclei), −0.96168 (right core nuclei)]
Zhao et al.³⁵	Mixed	400/424	Medication free	Range: 0.22–7.5 years first onset	142/258	Right amygdala atrophy in first-episode medication-naïve MDD.
Bora et al.⁶⁹	Mixed	986/937	Mixed	–	878/1045	Amygdala atrophy in medication-free first MDE vs. HCs and rMDD.
Hamilton et al.⁶⁸	Mixed	299/268	183 medicated	–	136/431	Amygdala atrophy in unmedicated MDD vs. HCs.
						Amygdala volume increased in medicated MDD vs. HCs.
Studies not showing smaller amygdala in MDD
Yuksel et al.,⁹⁸	Automated (SPM8)	37/54	Medicated	–	50/41	No differences in amygdala volume in MDD vs. HCs initially.
						Significant grey matter volume increase in bilateral amygdala in MDD during follow up compared to HCs.
						Milder courses of MDD exhibited increased grey matter volume increases in left amygdala compared to severe course of MDD on follow up.
						Volumetric changes in left amygdala may be reversible and dependent on clinical phenotype.
Sandu et al.⁷⁷	Automated (SPM8)	81/44	74 medicated	Non TRD: 18.01 Y	60/65	No difference in amygdala volume between HCs and group consisting of TRD and remitted MDD.
				TRD: 18.9 Y		Amygdala enlargement in TRD vs. remitted MDD.
						Larger amygdala may indicate vulnerability to develop medication resistance.
Shen et al.⁴⁶	Automated (FAST)	354/803	N/A	–	529/628	No difference in amygdala volumes in MDD vs. HCs.
van Eijndhoven et al.⁷⁶	Manual	40/20	Medication free	current first episode: 7.1 months	20/40	Amygdala enlargement in cMDD vs. HCs (Cohen’s d = 0.4974 (left), 0.5786 (right))
				recovered first episode: 21.6 months		No difference in amygdala volume between remitted MDD and HCs. Included in Cole 2011 meta-analysis
Lorenzetti et al.⁷³	Manual	56/21	33 medicated in previous 6 months	cMDD: 11.45	26/61	No amygdala volume difference in cMDD vs. HCs.
				rMDD: 9.04		Left amygdala enlargement in remitted MDD vs. HCs.
Schmaal et al.²⁵	Mixed	1728/7199	623 Medicated	–	–	Smaller amygdala with earlier age of onset; however, this did not survive correction for multiple comparisons.

Note: The table is divided into studies showing smaller amygdala in MDD and those showing no smaller amygdala in MDD. Each section is stratified into automated, manual and mixed segmentation techniques. AAA: anterior amygdaloid area; ACO: anterior cortical; ASTR: amygdalostriatal transition area; BLVM: basolateral ventrolateral; CA: cornu ammonis; CE: Central; cMDD: current major depressive disorder; DG: dentate gyrus; FAST: FMRIBs automated segmentation tool; HC: healthy controls; HPA: hypothalamic-pituitary-adrenal; LA: lateral; MDD: major depressive disorder; MDE: major depressive episode; N/A: not available; rMDD: recurrent major depression; SPM: statistical parametric mapping.

Table 3.

Cohen’s ds for hippocampal and volumetric studies in MDD.

		No depression			Depression			Cohen’s d_s
Study	Area	Mean	SD	N	Mean	SD	N	Cohen’s d_s
Amygdala studies
Ancelin et al.¹	Global amygdala	2634.1	14.88	448	2580.7	25.34	162	−2.9271056
Yao et al.⁷⁰	Left amygdala	1405.3	349.9	28	1247.7	540.7	25	−0.3503127
Yao et al.⁷⁰	Right amygdala	1510.7	511.3	28	1205.6	583.9	25	−0.5581094
Zavorotnyy et al.,⁹⁷	Left amygdala	2.101	0.292	30	2.093	0.245	23	−0.029334
Zavorotnyy et al.,⁹⁷	Right amygdala	1.977	0.242	30	1.915	0.176	23	−0.2870146
Schuhmacher et al.²⁷	Left amygdala	0.136	0.021	87	0.124	0.0246	32	−0.5451756
Schuhmacher et al.²⁷	Right amygdala	0.136	0.0188	87	0.12	0.0231	32	−0.7988252
Kronenberg et al.⁷⁴	Left amygdala	1.97	0.26	14	1.71	0.34	24	−0.8294118
Kronenberg et al.⁷⁴	Right amygdala	1.97	0.27	14	1.74	0.27	24	−0.8518519
Sheline et al.⁷¹	Left amygdala	1782	296	20	1650	310	20	−0.43553
Sheline et al.⁷¹	Right amygdala	1752	295	20	1724	304	20	−0.09348
Hippocampus studies
Han et al.⁵³	Left hippocampus	3471.5	376.32	135	3290.4	346.66	102	−0.4976274
	Right hippocampus	3589.7	386.91	135	3414.5	333.22	102	−0.4802339
	Left CA1	669.69	88.449	135	631.78	83.848	102	−0.4382812
	Right CA1	704.07	85.319	135	660.78	76.182	102	−0.5309772
	Left CA4	269.87	33.777	135	253.74	32.272	102	−0.4869254
	Right CA4	276.57	33.936	135	263.89	29.735	102	−0.3938791
	Left CA2-3	216.62	33.64	135	202.5	35.014	102	−0.4123575
	Right subiculum	429.86	51.484	135	411.72	41.205	102	−0.3832026
Yao et al.⁷⁰	Left hippocampus	4701.4	991.6	28	4303.5	1687.2	25	−0.2917425
Yao et al.⁷⁰	Right hippocampus	5161.7	926.6	28	4494.7	1351.2	25	−0.5819352
Han et al.⁴⁷	Left subiculum	640.79	46.89	21	586.9	77.4	20	−0.8472077
	Left CA2-3	927.31	104.38	21	862.45	108.92	20	−0.6083518
	Left CA4-DG	490.94	62.51	21	527.62	59.99	20	0.598415
	Left hippocampus	2940.7	256.66	21	2737.2	341.09	20	−0.6767011
	Right hippocampus	3040.1	310.46	21	2849	391.43	20	−0.5423863
	Right subiculum	655.14	52.48	21	603.3	76.97	20	−0.7906793
Huang et al.⁴²	CA1-3	1618	282	27	1646	277	20	0.1000358
Huang et al.⁴²	DG	1281	179	27	1261	244	20	−0.0957278
Maller et al.,⁹⁴	Left hippocampus	2773.4	409.85	76	2446.5	386.93	182	−0.830178
Maller et al.,⁹⁴	Right hippocampus	2629.3	403.73	76	2344.4	424.76	182	−0.6805215
Schuhmacher et al.²⁷	Left hippocampus	0.23	0.0285	87	0.207	0.0337	86	−0.737337
Schuhmacher et al.²⁷	Right hippocampus	0.233	0.0299	87	0.209	0.0318	86	−0.7777292
Malykhin et al.²⁸	Left hippocampus	2798	257	34	2747	297	39	−0.1827156
Malykhin et al.²⁸	Right hippocampus	2982	244	34	2829	312	39	−0.5417117
Kronmuller et al.,⁹⁵	Left hippocampus	3.19	0.25	30	2.81	0.36	26	−1.2105645
Kronmuller et al.,⁹⁵	Right hippocampus	3.3	0.29	30	3	0.57	26	−0.6783556
van Eijndhoven et al.⁷⁶	Left amygdala	2222	412	20	2389	236	20	0.4974117
van Eijndhoven et al.⁷⁶	Right amygdala	2153	396	20	2358	307	20	0.5785965
Ancelin et al.¹	Global hippocampus	7033	33.88	448	7061	57.73	162	0.6738832
Ota et al.,⁹⁹	Global hippocampus	5583.7	489.9	35	5578.4	521.8	19	−0.0105752
Wisse et al.⁵¹	Global hippocampus	3.1	0.5	34	3.1	0.4	13	0
Schermuly et al.⁴³	Global hippocampus	2.67	0.29	13	2.83	0.57	15	0.346063

Note: A Cohen’s d_s metric is provided for a standardized comparison of the effect size between the depressed and non-depressed group means. Meta-analyses and studies providing insufficient information (e.g. no mean volumes or SDs) for calculation of Cohen’s d_s have been excluded from the table. Cohen’s d_s for specific subfields have also been excluded in instances where the study does not report the metrics needed for Cohen’s d_s calculation. CA: cornu ammonis; DG: dentate gyrus; N: number; SD: standard deviation.

The subiculum is part of the exit circuitry of the hippocampus (Figure 1) and, along with CA1, outputs information back to the adjacent entorhinal cortex and to the wider cortex and hypothalamus via the fornix.¹¹ Subicular output inhibits HPA activity resulting in reduced circulating cortisol. Smaller subiculums found in MDD may disrupt the negative feedback mechanisms of the HPA axis causing escalating cortisol levels, known to be detrimental to mood and cognition⁵⁶,⁵⁷ as well as directly damaging brain areas including the DG.⁵⁰,⁵⁸

Ca1

The CA1 has been found to be bilaterally smaller in a study of 102 patients with chronic MDD and 135 controls.⁵³ The left CA1 region was also smaller in 83 depressed individuals compared to 80 controls and was reported as a predictor of depressive illness duration.⁹ Previous studies that have identified changes in various other hippocampal substructures, however, have failed to demonstrate CA1 changes.⁵⁹ Failure to do so may be accounted by the well-documented methodological issues with previous versions of automated segmentation that have underestimated the size of CA1.⁵¹

The CA1 region has a particularly high expression of multiple 5-HT receptor subtypes⁶⁰ and is the most vulnerable hippocampal substructure to excitotoxicity.⁶¹ Long-term corticosteroid exposure, a model of MDD and chronic stress in rats, has been shown to lead to attenuated serotonin responses in the CA1 region.⁶² Evidence showing extensive post-mortem CA1 neural apoptosis⁶³ and reduced CA1 thickness⁶⁴ in MDD also strongly suggests CA1 involvement in the disorder.

Ca2/3

Hippocampal segmentation often combines these two subfields into a single CA2/3 region due to the inability to distinguish the CA2 from the adjacent CA3 region using MR imaging, the diminutive size of CA2 and the similarity between the pyramidal layers of both subfields. Left CA2/3 volume in MDD was reduced in a study of 83 depressed patients compared to 80 controls, particularly in patients with recurrent depression,⁹ and also of 102 patients with MDD and 135 controls.⁵³ CA3 was also found to be reduced in Parkinson’s patients with comorbid depression. These volumes normalized following L-DOPA treatment.⁵⁹ Similar reductions were found in CA2/3 subfields in an unmedicated MDD female only cohort.⁴⁷ Smaller CA2/3 subfields have been found in later life patients⁵² and associated with cerebrovascular events.⁶⁵

Although small, the CA2 subfield receives prominent hypothalamic input, including directly from the paraventricular nucleus, suggesting a role in HPA feedback.⁶⁶ The CA3 displays rich connectivity within the hippocampus and has roles as a hippocampal ‘pacemaker’ critical for encoding and decoding of information.⁵⁹,⁶⁷ Smaller CA2 and CA3 in MDD could result in the aberrant stress/HPA responses found in the disorder as well as the neurocognitive difficulties found in some individuals with MDD.

Global Amygdala Changes in MDD

While a relative abundance of literature exists detailing structural alterations of the hippocampus in MDD, there is a relative dearth of research detailing amygdala volumes in the illness. We identified 12 studies showing smaller amygdala compared to controls. Our Cohen’s d_s calculation showed a 5.2% reduction in amygdala volume on the left and a 7.4% reduction on the right in MDD patients. This largely corresponds with previous meta-analyses focusing on the amygdala as a single structure⁶⁸ or part of a broader brain analyses.²⁵,³⁵,⁶⁹ Structural neuroimaging studies of the amygdala in the literature are hindered by the same inter-study variability as those of hippocampal studies. Due to the particular difficulties in categorization and identification of amygdalar nuclei, there has been less focus in measuring amygdalar subregions. However, effort has focused on identifying global changes and also the lateralisation of differences. Debate has also centred around whether amygdala volume alterations represent a state marker for depression.

Amygdala Substructures

Unfortunately, due to difficulties in substructure definitions and visualization on MRI, only a handful of studies have investigated the amygdala at a deeper level in MDD. A recent automated segmentation study of 25 patients and 28 controls reported specific subregional volumetric loss not apparent when examining global amygdala volume.⁷⁰ Specifically, the lateral and basolateral ventromedial nuclei were bilaterally smaller in MDD, along with right central and anterior cortical nucleus volume reductions. Smaller laterobasal nuclei were found bilaterally in a manual tracing study of twenty MDD patients.⁷¹ A recent 7 T automated segmentation of study of 24 patients and 20 controls found no reduction in whole volumes or nuclei volumes in MDD, but did find the right lateral and left cortical and accessory basal nuclei correlated negatively with depressive symptoms.⁷²

Amygdala as a State Marker of Illness

It is clear that there is significantly greater heterogeneity in the findings from studies of the amygdala, particularly when comparing studies with differing disease stages. Many have established the case that volumetric alterations may represent a trait of neurobiological vulnerability for depression.⁷³ A trend towards increasing atrophy with increasing number of depressive episodes has also been detected.⁷⁴ A previous meta-analysis has found evidence of reduction of volume in the right amygdala in first episode depression.⁶⁹ Younger patients aged 18–39 years reporting depressive symptoms but not formally diagnosed with MDD have also been shown to display atrophy in the right amygdala.⁷⁵ No relationship between depression severity and smaller amygdala volume has been found.⁷⁴ This is in contrast to conflicting studies showing enlarged amygdalae in first-episode depression is correlated to episode severity.⁷⁶

These volume changes have been shown to resolve in patients treated with antidepressant medication. A meta-analysis shows that smaller amygdala volumes found in MDD increase in size with medications, even beyond that seen in controls.⁶⁸ Conversely, normalisation of volumes of initial enlarged amygdala after treatment has also been described.⁷⁶ This may be underpinned by protection from excess glucocorticoid stimulation. In one study there appeared to be a correlation between bilateral normalization of amygdalar volumes following medication administration (in the context of pre-treatment bilateral atrophy) with reductions in HPA-axis activity.²⁷ Suggesting that the effects of antidepressants on circulating cortisol may reverse amygdala volume loss from MDD. Additionally, others have concluded that amygdala volume may be a predictor of treatment response, finding higher amygdala volumes in treatment-resistant patients versus non-treatment resistant patients.⁷⁷ However, it must be noted that comorbid anxiety may be present in these cohorts, and this could have an effect on amygdala volumes.⁷⁸

Although amygdalar volume has been suggested to be a marker of treatment response, one study did not demonstrate volumetric differences between currently depressed individuals and controls; however, there was volumetric enlargement in patients in remission not related to the use of medication.⁷³ This was suggested to represent a vulnerability factor for depression relapse and may constitute a reaction to the stress induced by previous depressive episodes enhancing synaptogenesis in the amygdala. This is in contradiction to other findings whereby stress in the previous six months has been associated with atrophy of the left amygdala.⁷⁹

Stress, Neurotoxicity and Temporal Lobe Structures in MDD

HPA axis dysfunction, neuroinflammation and neurotransmitter perturbations are implicated in the link between MDD and stress. Belleau et al. proposed a unifying model of these mechanisms.¹⁵ In this model, chronic life stress is proposed to trigger dysregulation of the HPA axis, ultimately resulting in altered levels of basal cortisol in those with MDD.¹⁹ This finding has been widely replicated with a majority of studies indicating HPA-axis overactivity, perhaps underpinning the association between MDD and chronic metabolic and neurodegenerative diseases.¹⁸,⁸⁰,⁸¹ HPA activity depends on balanced integration of amygdalar activation and hippocampal inhibition of the periventricular nucleus of the hypothalamus.⁸²,⁸³ MDD may cause a shift from hippocampal inhibition to amygdala-mediated release of corticotrophin releasing hormone from the hypothalamus.²⁷ These increased cortisol levels have been associated with damage to both amygdala and hippocampal neurons.

This hyperactive endocrine state cultures a neuroinflammatory milieu, coupled with direct immune activation by stress itself.¹⁵ Chronic stress, therefore, promotes the expression of pro-inflammatory cytokines such as interleukins 6 and 1β and tumour necrosis factor alpha. This is supported by reports of elevations in inflammatory marker levels in the hippocampus in animal models of chronic unpredictable stress.⁸⁴ The effects of peripheral pro-inflammatory cytokines on central nervous system microglia may result in reduced hippocampal neurogenesis.²⁰ Additionally, pro-inflammatory cytokines have been shown to activate indoleamine 2, 3-dioxygenase, an enzyme that shunts tryptophan into the kynurenine pathway, yielding neurotoxic end-products such as 3-hydroxykynurenine and quinolinic acid. Kynurenine pathway activation and associated elevated immune response has been shown to be associated with smaller hippocampi in MDD.⁸⁵ Abnormal membrane turnover in the hippocampus in MDD is greater in patients with highly recurrent illness and is further evidenced by magnetic resonance spectroscopy (MRS) studies.⁸⁶

Following from the shift to a pro-inflammatory state and the production of neurotoxic end-products, as well as a direct effect from chronic stress, perturbations in neurotransmission also occur in the pathogenesis of MDD. In particular, there are changes in glutamatergic signalling, as seen in MRS studies.⁸⁷,⁸⁸ Glutamate, the primary excitatory neurotransmitter in the central nervous system exerts an excitotoxic effect on neurons. Disruptions in glutamatergic signalling may result from excessive release and reduced clearance from the synaptic cleft, thereby facilitating direct neurotoxic effects.

Stress Sensitisation and Recurrent Depressive Episodes

The ‘stress sensitisation’ or ‘kindling’ model of affective illness¹³ proposes that while an initial MDE often follows a period of intense life stress, the strength of the relationship between life stressors and subsequent MDEs declines as a function of the number of MDEs. This leads to the unfortunate situation where progressively lower levels of stress have the ability to trigger recurrent MDEs as the disease progresses. An extension of the kindling model is the ‘stress autonomy’ model, which proposes that recurrent MDEs can generate independent of life stressors.¹⁵ Assuming that the above formulation of pathogenesis correctly encapsulates the neurobiological underpinnings of MDD, it could be postulated that lesser levels of stress may trigger these increasing neuroendocrine and inflammatory disturbances with each successive MDE. One strong line of evidence for the kindling model is the clear association between early life stressors and MDD.¹⁴ Opel et al. have recently reported a significant association between childhood maltreatment and hippocampal volume in both healthy controls and individuals with MDD, and to the development of MDD.¹⁴,¹⁷ Childhood maltreatment has also been linked to an earlier onset of MDD, perhaps explained by the induction of HPA-axis and immune dysregulation.⁸⁹ In contrast to this, Lenze et al. did not find an association between childhood maltreatment and hippocampal volume.

Ultimately, a cumulative effect of chronic stress, and of the neurotoxic implications of such stress in MDEs, may be reflected in morphometric disturbances. Naturally, numerous investigators have proposed the hippocampus as a marker of illness progression, with increasing atrophy with each successive depressive episode.¹⁵,²⁵,²⁶ Zaremba et al. proposed that atrophy is not related to total illness duration, but to the number of hospitalisations, suggesting that increasing atrophy may be linked to the severity of the depressive episodes. Attempts have also been made to localise these volume reductions. CA1-4 and DG have consistently been shown to display the most marked changes.⁹,⁴¹,⁴²,⁷⁰,⁹⁰,^91,92

Limitations

The main limitation of any review of hippocampal and amygdalar volumes is the difficulty in comparing results across the variety of different techniques used. Of the 39 studies examined, 30 utilised automated tissue segmentation protocols in some fashion and the remainder relied solely on manual tracing. Although automated tissue segmentation and increasingly capable analysis software allow for more reliable and replicable identification of hippocampal and amygdalar subfields, there is some debate over the validity of these techniques compared to the ‘gold standard’ manual tracing. Automated techniques allow much faster assessment of greater numbers of subjects and will probably become more common in the future as dataset numbers increase. Voxel-based methods have been shown to exhibit relative insensitivity in segmenting tissue into grey and white matter in regions such as the hippocampus wherein grey and white matter sheets are convoluted into each other, perhaps resulting in non-significant findings in some cases.⁶⁹ Similarly, studies use different definitions of amygdalar and hippocampal volumes. A review of MRI volumetrics has suggested that approximately 60 different anatomical guidelines exist for hippocampal volumes.⁹³ To view ‘volume reduction’ in the hippocampus and amygdala as the sole characteristic of MDD-related changes in brain structure may also be overly simplistic. Rather, both differences in structure and connectivity/function may serve as more accurate illness biomarkers. Clearly, these subtle changes in MDD have yet to be fully elucidated. The Cohen’s d_s metric used in this study facilitates a standardised comparison of the effect sizes across different studies, yielding an easily interpretable if somewhat blunt measure of overall effect size for each structure. Unfortunately, a full meta-analysis identifying a common effect size, controlling for the multiple confounders and causes of heterogeneity across all these studies is outside the scope of this review. This is currently in progress.

Future Directions

Considerable work is needed to further elucidate the precise link between neuropathology of MTL structures and the clinical presentation of depression. Larger studies are needed with standardised patient groups (age, sex, demographics, etc.), disease characteristics (numbers of episodes, medication status, etc.), MRI protocols and anatomical definitions. As MRI technology and computer processing advances, greater focus on individual substructures, particularly with respect to the understudied substructures of the amygdala, may provide further insight. As noted above, homogenisation of the definitions of these MTL structures and their substructures will permit greater standardisation across studies and inter-study comparison. No study to date has focused on automated segmentation of the hippocampal subfields into their anterior and posterior components. It is known that the hippocampus has anatomical and functional differences along its anterior–posterior axis¹⁰ (e.g. a lower proportion of dentate is found in the anterior division of the hippocampus than in the posterior hippocampus).²⁸ Newer automated techniques are now able to divide the hippocampus and substructures longitudinally along this axis. Not only will future use of these fine grain techniques improve standardisation, but the greater anatomical specificity may form the basis for future hypotheses based on precise substructure localisation. This may yield a more specific and sensitive biomarker – a holy grail of academic psychiatry.

Conclusion

On balance, the overwhelming consensus across all the studies is that both the hippocampus and amygdala are smaller in MDD. The hippocampus most commonly exhibits atrophy proportional to disease chronicity and the amygdala potentially representing a state marker of illness. Several factors contribute to the heterogeneity of study findings, including the impact of stress, number of depressive episodes, impact of medication and the age of depression onset. It is therefore necessary to be cognisant of the impact of these factors in the interpretation and extrapolation of study findings. Recent developments in automated tissue segmentation software have allowed a more nuanced examination of the volumetric changes in the hippocampal and amygdalar substructures. This presents an exciting prospect for future research in this field. Future understanding of the relationship between stress, cognitive and affective processing in depression, and these key temporal lobe structures may help elucidate the pathophysiology of MDD.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Anurag Nasa

Darren Roddy

References

Ancelin

M-L

Carrière

Artero

, et al. Lifetime major depression and grey-matter volume. J Psychiatry Neurosci. 2019; 44(1): 45–53.

American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 5th ed. Washington, DC: American Psychiatric Association Publishing; 2013.

American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (DSM-5®). Washington, DC: American Psychiatric Association Publishing; 2013.

Sweeney

Kmiec

Kupfer

DJ.

Neuropsychologic impairments in bipolar and unipolar mood disorders on the CANTAB neurocognitive battery. Biol Psychiatry. 2000; 48(7): 674–684.

Cornwell

Salvadore

Colon-Rosario

, et al. Abnormal hippocampal functioning and impaired spatial navigation in depressed individuals: evidence from whole-head magnetoencephalography. Am J Psychiatry. 2010; 167(7): 836–844.

Kaltenboeck

Harmer

The neuroscience of depressive disorders: a brief review of the past and some considerations about the future. Brain Neurosci Adv. 2018; 2: 2398212818799269.

Adam Samuels

Leonardo

Hen

Hippocampal subfields and major depressive disorder. Biol Psychiatry. 2015; 77(3): 210–211.

Sheline

Liston

McEwen

BS.

Parsing the hippocampus in depression: chronic stress, hippocampal volume, and major depressive disorder. Biol Psychiatry. 2019; 85(6): 436–438.

Roddy

Farrell

Doolin

, et al. The hippocampus in depression: more than the sum of its parts? Advanced hippocampal substructure segmentation in depression. Biol Psychiatry. 2019; 85(6): 487–497.

10.

Poppenk

Evensmoen

Moscovitch

Nadel L. Long-axis specialization of the human hippocampus. Trends Cogn Sci (Regul Ed). 2013; 17(5): 230–240.

11.

Weininger

Roman

Tierney

, et al. Papez’s forgotten tract: 80 years of unreconciled findings concerning the thalamocingulate tract. Front Neuroanat. 2019; 13: 14.

12.

Yang

Wang

JZ.

From structure to behavior in basolateral amygdala-hippocampus circuits. Front Neural Circuits. 2017; 11: 86.

13.

Post

RM.

Transduction of psychosocial stress into the neurobiology of recurrent affective disorder. Am J Psychiatry. 1992; 149(8): 999–1010.

14.

Opel

Redlich

Zwanzger

, et al. Hippocampal atrophy in major depression: a function of childhood maltreatment rather than diagnosis? Neuropsychopharmacology. 2014; 39(12): 2723–2731.

15.

Belleau

Treadway

Pizzagalli

DA.

The impact of stress and major depressive disorder on hippocampal and medial prefrontal cortex morphology. Biol Psychiatry. 2019; 85(6): 443–453.

16.

Herman

Cullinan

WE.

Neurocircuitry of stress: central control of the hypothalamo-pituitary-adrenocortical axis. Trends Neurosci. 1997; 20(2): 78–84.

17.

Frodl

O'Keane

How does the brain deal with cumulative stress? A review with focus on developmental stress, HPA axis function and hippocampal structure in humans. Neurobiol Dis. 2013; 52: 24–37.

18.

Stetler

Miller

GE.

Depression and hypothalamic-pituitary-adrenal activation: a quantitative summary of four decades of research. Psychosom Med. 2011; 73(2): 114–126.

19.

Slavich

Irwin

MR.

From stress to inflammation and major depressive disorder: a social signal transduction theory of depression. Psychol Bull. 2014; 140(3): 774–815.

20.

Moylan

Maes

Wray

Berk M. The neuroprogressive nature of major depressive disorder: pathways to disease evolution and resistance, and therapeutic implications. Mol Psychiatry. 2013; 18(5): 595–606.

21.

Kempton

Salvador

Munafo

, et al. Structural neuroimaging studies in major depressive disorder. Meta-analysis and comparison with bipolar disorder. Arch Gen Psychiatry. 2011; 68(7): 675–690.

22.

Greenberg

Payne

MacFall

Steffens DC, Krishnan RR. Hippocampal volumes and depression subtypes. Psychiatry Res. 2008; 163(2): 126–132.

23.

Phillips

Batten

Tremblay

Aldosary F, Blier P. A prospective, longitudinal study of the effect of remission on cortical thickness and hippocampal volume in patients with treatment-resistant depression. Int J Neuropsychopharmacol. 2015; 18(8). doi: 10.1093/ijnp/pyv037

24.

Rusch

Abercrombie

Oakes

Schaefer SM, Davidson RJ. Hippocampal morphometry in depressed patients and control subjects: relations to anxiety symptoms. Biol Psychiatry. 2001; 50(12): 960–964.

25.

Schmaal

Veltman

van Erp

, et al. Subcortical brain alterations in major depressive disorder: findings from the ENIGMA major depressive disorder working group. Mol Psychiatry. 2016; 21(6): 806–812.

26.

Stratmann

Konrad

Kugel

, et al. Insular and hippocampal gray matter volume reductions in patients with major depressive disorder. PLoS One. 2014; 9(7): e102692.

27.

Schuhmacher

Mossner

Jessen

, et al. Association of amygdala volumes with cortisol secretion in unipolar depressed patients. Psychiatry Res. 2012; 202(2): 96–103.

28.

Malykhin

Carter

Seres

Coupland NJ. Structural changes in the hippocampus in major depressive disorder: contributions of disease and treatment. J Psychiatry Neurosci. 2010; 35(5): 337–343.

29.

Cole

Costafreda

McGuffin

Fu CH. Hippocampal atrophy in first episode depression: a meta-analysis of magnetic resonance imaging studies. J Affect Disord. 2011; 134(1-3): 483–487.

30.

Videbech

Ravnkilde

Hippocampal volume and depression: a meta-analysis of MRI studies. Am J Psychiatry. 2004; 161(11): 1957–1966.

31.

McKinnon

Yucel

Nazarov

MacQueen GM. A meta-analysis examining clinical predictors of hippocampal volume in patients with major depressive disorder. J Psychiatry Neurosci. 2009; 34(1): 41–54.

32.

Koolschijn

van Haren

Lensvelt-Mulders

, et al. Brain volume abnormalities in major depressive disorder: a meta-analysis of magnetic resonance imaging studies. Hum Brain Mapp. 2009; 30(11): 3719–3735.

33.

Arnone

McIntosh

Ebmeier

, et al. Magnetic resonance imaging studies in unipolar depression: systematic review and meta-regression analyses. Eur Neuropsychopharmacol. 2012; 22(1): 1–16.

34.

Wise

Radua

Via

, et al. Common and distinct patterns of grey-matter volume alteration in major depression and bipolar disorder: evidence from voxel-based meta-analysis. Mol Psychiatry. 2017; 22(10): 1455–1463.

35.

Zhao

Huang

, et al. Brain grey matter abnormalities in medication-free patients with major depressive disorder: a meta-analysis. Psychol Med. 2014; 44(14): 2927–2937.

36.

Yue

, et al. Voxelwise meta-analysis of gray matter reduction in major depressive disorder. Prog Neuropsychopharmacol Biol Psychiatry. 2012; 36(1): 11–16.

37.

Roddy

O’Keane

Cornu ammonis changes are at the core of hippocampal pathology in depression. Chronic Stress (Thousand Oaks). 2019; 3: 2470547019849376.

38.

Sheline

YI.

Depression and the hippocampus: cause or effect?

Biol Psychiatry. 2011; 70(4): 308–309.

39.

Frodl

Meisenzahl

Zetzsche

, et al. Hippocampal changes in patients with a first episode of major depression. Am J Psychiatry. 2002; 159(7): 1112–1118.

40.

Arnone

McKie

Elliott

, et al. State-dependent changes in hippocampal grey matter in depression. Mol Psychiatry. 2013; 18(12): 1265–1272.

41.

Bearden

Thompson

Avedissian

, et al. Altered hippocampal morphology in unmedicated patients with major depressive illness. ASN Neuro. 2009; 1(4):AN20090026.

42.

Huang

Coupland

Lebel

, et al. Structural changes in hippocampal subfields in major depressive disorder: a high-field magnetic resonance imaging study. Biol Psychiatry. 2013; 74(1): 62–68.

43.

Schermuly

Wolf

Lieb

Stoeter P, Fellgiebel A. State dependent posterior hippocampal volume increases in patients with major depressive disorder. J Affect Disord. 2011; 135(1–3): 405–409.

44.

Frodl

Jäger

Smajstrlova

, et al. Effect of hippocampal and amygdala volumes on clinical outcomes in major depression: a 3-year prospective magnetic resonance imaging study. J Psychiatry Neurosci. 2008; 33(5): 423–430.

45.

Santarelli

Saxe

Gross

, et al. Requirement of hippocampal neurogenesis for the behavioral effects of antidepressants. Science. 2003; 301(5634): 805–809.

46.

Shen

Reus

Cox

, et al. Subcortical volume and white matter integrity abnormalities in major depressive disorder: findings from UK Biobank imaging data. Sci Rep. 2017; 7(1): 5547.

47.

Han

Won

Sim

Tae WS. Hippocampal subfield analysis in medication-naive female patients with major depressive disorder. J Affect Disord. 2016; 194: 21–29.

48.

Treadway

Waskom

Dillon

, et al. Illness progression, recent stress, and morphometry of hippocampal subfields and medial prefrontal cortex in major depression. Biol Psychiatry. 2015; 77(3): 285–294.

49.

Samuels

Hen

Neurogenesis and affective disorders. Eur J Neurosci. 2011; 33(6): 1152–1159.

50.

Joëls

Role of corticosteroid hormones in the dentate gyrus. Prog Brain Res. 2007; 163: 355–370.

51.

Wisse

Biessels

Stegenga

, et al. Major depressive episodes over the course of 7 years and hippocampal subfield volumes at 7 tesla MRI: the PREDICT-MR study. J Affect Disord. 2015; 175: 1–7.

52.

Lim

Hong

Jung

, et al. Automated hippocampal subfields segmentation in late life depression. J Affect Disord. 2012; 143(1–3): 253–256.

53.

Han

K-M

Kim

Kang

, et al. Hippocampal subfield volumes in major depressive disorder and bipolar disorder. Eur Psychiatry. 2019; 57: 70–77.

54.

Brown

Hughes

McColl

Peshock R, King KS, Rush AJ. Association of depressive symptoms with hippocampal volume in 1936 adults. Neuropsychopharmacol. 2014; 39(3): 770–779.

55.

Buddeke

Kooistra

Zuithoff

, SMART Study Groupet al. Hippocampal volume and the course of depressive symptoms over eight years of follow-up. Acta Psychiatr Scand. 2017; 135(1): 78–86.

56.

Holsboer

The corticosteroid receptor hypothesis of depression. Neuropsychopharmacology. 2000; 23(5): 477–501.

57.

Qin

Rizak

Feng

, et al. Prolonged secretion of cortisol as a possible mechanism underlying stress and depressive behaviour. Sci Rep. 2016; 6: 30187.

58.

Joels

Karst

Alfarez

, et al. Effects of chronic stress on structure and cell function in rat hippocampus and hypothalamus. Stress. 2004; 7(4): 221–231.

59.

Györfi

Nagy

Bokor

, et al. Reduced CA2–CA3 hippocampal subfield volume is related to depression and normalized by L-DOPa in newly diagnosed Parkinson’s disease. Front Neurol. 2017; 8: 84.

60.

Dale

Pehrson

Jeyarajah

, et al. Effects of serotonin in the hippocampus: how SSRIs and multimodal antidepressants might regulate pyramidal cell function. CNS Spectr. 2016; 21(2): 143–161.

61.

Coultrap

Nixon

Alvestad

Valenzuela CF, Browning MD. Differential expression of NMDA receptor subunits and splice variants among the CA1, CA3 and dentate gyrus of the adult rat. Brain Res Mol Brain Res. 2005; 135(1–2): 104–111.

62.

Karten

Nair

van Essen

Joëls M. Long-term exposure to high corticosterone levels attenuates serotonin responses in rat hippocampal CA1 neurons. Proc Natl Acad Sci USA. 1999; 96(23): 13456–13461.

63.

Lucassen

Muller

Holsboer

, et al. Hippocampal apoptosis in major depression is a minor event and absent from subareas at risk for glucocorticoid overexposure. Am J Pathol. 2001; 158(2): 453–468.

64.

Stockmeier

Mahajan

Konick

, et al. Cellular changes in the postmortem hippocampus in major depression. Biol Psychiatry. 2004; 56(9): 640–650.

65.

Choi

Jung

Lee CU, Park YH, Lim HK. Cerebral vascular burden on hippocampal subfields in first-onset drug-naïve subjects with late-onset depression. J Affect Disord. 2017; 208: 47–53.

66.

Zhang

Hernandez

Synaptic innervation to rat hippocampus by vasopressin-immuno-positive fibres from the hypothalamic supraoptic and paraventricular nuclei. Neuroscience. 2013; 228: 139–162.

67.

Buzsáki

Theta oscillations in the hippocampus. Neuron. 2002; 33(3): 325–340.

68.

Hamilton

Siemer

Gotlib

IH.

Amygdala volume in major depressive disorder: a meta-analysis of magnetic resonance imaging studies. Mol Psychiatry. 2008; 13(11): 993–1000.

69.

Bora

Fornito

Pantelis

Yücel M. Gray matter abnormalities in major depressive disorder: a meta-analysis of voxel based morphometry studies. J Affect Disord. 2012; 138(1-2): 9–18.

70.

Yao

, et al. Morphological changes in subregions of hippocampus and amygdala in major depressive disorder patients. Brain Imaging Behav 2020; 14(3): 653-667.

71.

Sheline

Gado

Price

JL.

Amygdala core nuclei volumes are decreased in recurrent major depression. Neuroreport. 1998; 9(9): 2023–2028.

72.

Brown

Rutland

Verma

, et al. Structural MRI at 7T reveals amygdala nuclei and hippocampal subfield volumetric association with major depressive disorder symptom severity. Sci Rep. 2019; 9(1): 1–10.

73.

Lorenzetti

Allen

Whittle

Yücel M. Amygdala volumes in a sample of current depressed and remitted depressed patients and healthy controls. J Affect Disord. 2010; 120(1–3): 112–119.

74.

Kronenberg

Tebartz van Elst

Regen

Deuschle M, Heuser I, Colla M. Reduced amygdala volume in newly admitted psychiatric in-patients with unipolar major depression. J Psychiatr Res. 2009; 43(13): 1112–1117.

75.

Daftary

Van Enkevort

Kulikova

Legacy M, Brown ES. Relationship between depressive symptom severity and amygdala volume in a large community-based sample. Psychiatry Res Neuroimag. 2019; 283: 77–82.

76.

van Eijndhoven

van Wingen

van Oijen

, et al. Amygdala volume marks the acute state in the early course of depression. Biol Psychiatry. 2009; 65(9): 812–818.

77.

Sandu

Artiges

Galinowski

, et al. Amygdala and regional volumes in treatment-resistant versus nontreatment-resistant depression patients. Depress Anxiety. 2017; 34(11): 1065–1071.

78.

Moore

Bertels

Phan KL, Dolcos F, Dolcos S. Smaller amygdala volume and increased neuroticism predict anxiety symptoms in healthy subjects: a volumetric approach using manual tracing. Neuropsychologia. Advance online publication. doi: 10.1016/j.neuropsychologia.2017.11.008.

79.

Sublette

Galfalvy

Oquendo

, et al. Relationship of recent stress to amygdala volume in depressed and healthy adults. J Affect Disord. 2016; 203: 136–142.

80.

Farrell

Doolin

O' Leary

, et al. DNA methylation differences at the glucocorticoid receptor gene in depression are related to functional alterations in hypothalamic-pituitary-adrenal axis activity and to early life emotional abuse. Psychiatry Res. 2018; 265: 341–348.

81.

Nickle

Stanley

Middlemas

DS.

Corticosterone induces depressive-like behavior in female peri-pubescent rats, but not in pre-pubescent rats. Chronic Stress (Thousand Oaks). 2020; 4: 2470547020923711.

82.

Jacobson

Sapolsky

The role of the hippocampus in feedback regulation of the hypothalamic-pituitary-adrenocortical axis. Endocr Rev. 1991; 12(2): 118–134.

83.

LeDoux

Iwata

Cicchetti

Reis DJ. Different projections of the central amygdaloid nucleus mediate autonomic and behavioral correlates of conditioned fear. J Neurosci. 1988; 8(7): 2517–2529.

84.

Iwata

Ota

, et al. Psychological stress activates the inflammasome via release of adenosine triphosphate and stimulation of the purinergic type 2X7 receptor. Biol Psychiatry. 2016; 80(1): 12–22.

85.

Doolin

Allers

Pleiner

, et al. Altered tryptophan catabolite concentrations in major depressive disorder and associated changes in hippocampal subfield volumes. Psychoneuroendocrinology. 2018; 95: 8–17.

86.

Milne

MacQueen

Yucel

Soreni N, Hall GB. Hippocampal metabolic abnormalities at first onset and with recurrent episodes of a major depressive disorder: a proton magnetic resonance spectroscopy study. Neuroimage. 2009; 47(1): 36–41.

87.

Djavadian

RL.

Serotonin and neurogenesis in the hippocampal dentate gyrus of adult mammals. Acta Neurobiol Exp (Wars). 2004; 64(2): 189–200.

88.

Tavares

Tasca

Santos

, et al. Quinolinic acid stimulates synaptosomal glutamate release and inhibits glutamate uptake into astrocytes. Neurochem Int. 2002; 40(7): 621–627.

89.

Lenze

Xiong

Sheline

YI.

Childhood adversity predicts earlier onset of major depression but not reduced hippocampal volume. Psychiatry Res. 2008; 162(1): 39–49.

90.

Zaremba D, Enneking V, Meinert S, et al. Effects of cumulative illness severity on hippocampal gray matter volume in major depression: a voxel-based morphometry study. Psychol Med. 2018; 48(14): 2391-2398.

91.

Malykhin

Coupland

NJ.

Hippocampal neuroplasticity in major depressive disorder. Neuroscience. 2015; 309: 200–213.

92.

Cole

Toga

Hojatkashani

, et al. Subregional hippocampal deformations in major depressive disorder. J Affect Disord. 2010; 126(1–2): 272–277.

93.

Geuze

Vermetten

Bremner

JD.

MR-based in vivo hippocampal volumetrics: 1. Review of methodologies currently employed. Mol Psychiatry. 2005; 10(2): 147–159.

94.

Maller JJ, Daskalakis ZJ, Thomson RH, Daigle M, Barr MS, Fitzgerald PB. Hippocampal volumetrics in treatment-resistant depression and schizophrenia: the devil's in de-tail. Hippocampus. 2012; 22(1): 9-16.

95.

Kronmüller KT, Schröder J, Köhler S, et al. Hippocampal volume in first episode and recurrent depression. Psychiatry Res. 2009; 174(1): 62-66.

96.

Santos MAO, Bezerra LS, Carvalho ARMR, Brainer-Lima AM. Global hippocampal atrophy in major depressive disorder: a meta-analysis of magnetic resonance imaging studies. Trends Psychiatry Psychother. 2018; 40(4): 369-378.

97.

Zavorotnyy M, Zöllner R, Schulte-Güstenberg LR, et al. Low left amygdala volume is associated with a longer duration of unipolar depression. J Neural Transm (Vienna). 2018; 125(2): 229-238.

98.

Yüksel D, Engelen J, Schuster V, et al. Longitudinal brain volume changes in major depressive disorder. J Neural Transm (Vienna). 2018; 125(10): 1433-1447.

99.

Ota M, Sato N, Hidese S, et al. Structural differences in hippocampal subfields among schizophrenia patients, major depressive disorder patients, and healthy subjects. Psychiatry Res Neuroimaging. 2017; 259: 54-59.