Elsevier

Biological Psychiatry

Volume 51, Issue 5, 1 March 2002, Pages 387-399
Biological Psychiatry

Original article
Principal components of the beck depression inventory and regional cerebral metabolism in unipolar and bipolar depression

https://doi.org/10.1016/S0006-3223(01)01244-6Get rights and content

Abstract

Background: We determined clustering of depressive symptoms in a combined group of unipolar and patients with bipolar disorder using Principle Components Analysis of the Beck Depression Inventory. Then, comparing unipolars and bipolars, these symptom clusters were examined for interrelationships, and for relationships to regional cerebral metabolism for glucose measured by positron emission tomography.

Methods: [18F]-fluoro-deoxyglucose positron emission tomography scans and Beck Depression Inventory administered to 31 unipolars and 27 bipolars, all medication-free, mildly-to-severely depressed. BDI component and total scores were correlated with global cerebral metabolism for glucose, and voxel-by-voxel with cerebral metabolism for glucose corrected for multiple comparisons.

Results: In both unipolars and bipolars, the psychomotor-anhedonia symptom cluster correlated with lower absolute metabolism in right insula, claustrum, anteroventral caudate/putamen, and temporal cortex, and with higher normalized metabolism in anterior cingulate. In unipolars, the negative cognitions cluster correlated with lower absolute metabolism bilaterally in frontal poles, and in right dorsolateral frontal cortex and supracallosal cingulate.

Conclusions: Psychomotor-anhedonia symptoms in unipolar and bipolar depression appear to have common, largely right-sided neural substrates, and these may be fundamental to the depressive syndrome in bipolars. In unipolars, but not bipolars, negative cognitions are associated with decreased frontal metabolism. Thus, different depressive symptom clusters may have different neural substrates in unipolars, but clusters and their substrates are convergent in bipolars.

Introduction

Phenomenological and neurobiological heterogeneity in depression have important implications for understanding and treating affective disorders. Differences in symptomology and neurobiology have been reported between unipolar (UP) and bipolar (BP) depression and for melancholic, endogenomorphic (Klein 1974; Young et al 1986), and atypical depressive subtypes (Nierenberg et al 1998). Identifying depressive subtype differences has become increasingly important as their treatments diverge (Klein 1974; Fawcett et al 1983a; Liebowitz et al 1984a; Liebowitz et al 1984b; Clark et al 1984), especially for UP versus BP depression (Post et al 1987; Post 1995; Sachs et al 2000).

In this study, we determined clustering of depressive symptoms in a combined group of medication-free UP and BP patients using Principle Components Analysis (PCA) of the Beck Depression Inventory (BDI). We then compared these symptom clusters and their interrelationships between UPs and BPs, and examined the relationships of the clusters to regional cerebral metabolism for glucose (rCMRglu) measured by positron emission tomography (PET).

Principal components and other factor analyses of the BDI have been reported for many different psychiatric and nonpsychiatric populations (Beck and Beamesderfer 1974; Beck et al 1988; Steer et al 1986; Startup et al 1992). The vast majority of studies have been in populations with mixed psychiatric diagnoses (Pichot and Lemperiere 1964; Cropley and Weckowicz 1966; Weckowicz et al 1967; Beck and Lester 1973; Lester and Beck 1977; Welch and Ellis 1991; Tanaka and Huba 1984) and in substance abuse patients (Steer et al 1977; Steer et al 1982; Shaw et al 1979; Reynolds and Gould 1981; Clark et al 1985; Louks et al 1989; Haviland et al 1991). There have been only five reports of BDI factor analysis in populations solely composed of primary affective disorder patients, with three reports in UP depression (Brown et al 1995; Startup et al 1992; Steer et al 1987), and two reports in mixed affective populations (Faravelli et al 1986; Steer et al 1989). Patients were antidepressant-free in only one study (Brown et al 1995). All five reports used PCA with orthogonal rotation, a technique we will also use here. No reports specifically examined BPs or compared BPs to UPs.

Abnormal regional CMRglu and cerebral blood flow (CBF) in frontal cortex and paralimbic regions have been found in affective disorder patients compared with healthy controls Dougherty and Rauch 1997, Drevets 1998. Additionally, severity of overall depressive symptoms has been correlated with rCMRglu/CBF in certain brain regions, most consistently in frontal or dorsolateral frontal cortex Buchsbaum et al 1986, Schlegel et al 1989, Baxter et al 1989, Upadhyaya et al 1990, Austin et al 1992, Drevets et al 1992, Yazici et al 1992, Bonne et al 1996, Iidaka et al 1997, Ketter et al 1996.

Specific symptoms or symptom clusters have also been correlated with rCMRglu/CBF in depressed patients, reflecting anergic and anxious depression (Ebmeier et al 1997), psychomotor slowing Mayberg et al 1994, Austin et al 1992, poverty of speech (Dolan et al 1993), anxiety and somatic symptoms (Philpot et al 1993), cognitive impairment Austin et al 1992, Philpot et al 1993, Bench et al 1992, Bench et al 1993, Dolan et al 1992, Dolan et al 1994, and psychotic symptoms Buchsbaum et al 1986, Philpot et al 1993. These studies included all or almost all unipolar patients only. The only study that correlated depressive symptoms with absolute (vs. normalized) regional CBF or CMRglu was a two-dimensional scintigram study (Schlegel et al 1989). One study correlated PCA clustered symptoms with normalized regional CBF determined by SPECT (Bench et al 1993); however, they derived components by PCA of 11 “variables” that had been constructed by a priori grouping of items from the Schedule for Schizophrenia and Affective Disorders and Mini-Mental Status Examination.

In the current study, we derived BDI principal components in a combined group of medication-free UP and BP patients, each with primary affective disorder. We separately compared relationships between BDI components and between individual BDI items for UPs and BPs. We then obtained correlative topographies for BDI components and cerebral metabolism, to uncover neural substrates of these depressive symptom clusters. No a priori grouping of BDI items was used, so that depressive phenomenology determined the depressive symptom clusters. We obtained correlative topographies for absolute, as well as normalized, rCMRglu.

Global normalization lowers variance and may improve detection of processes or associations in more discrete brain regions; however, normalization can compromise detection of processes or associations in regions of brain large enough to affect global metabolism. Absolute metabolism may permit better detection of involvement of large regions of brain. Discrepancies between absolute versus normalized data remain an unresolved problem in the field. We will highlight both the areas of convergent and divergent data, so that readers and future studies can address this issue.

Based on our prior report (Dunn et al, unpublished data, 1997) and the literature Sieden and Dykstra 1977, Bermanzohn and Siris 1992, Graybiel et al 1994, Heinz et al 1994, Salamone et al 1997, Koob and Nestler 1997, Volkow et al 2001, Wexler et al 2001, we hypothesized anhedonia symptoms would correlate with rCMRglu in dopamine-enriched brain regions associated with reward, that is in basal ganglia, medial prefrontal cortex and anterior cingulate. We also hypothesized psychomotor symptoms and their neural substrates would play a more prominent role in BP than UP depression phenomenology and neurobiology APA 1994, Benazzi 1997, Sobin and Sackeim 1997.

Section snippets

Subjects

Sixty primary affective disorder patients enrolled in treatment studies at the National Institute of Mental Health (NIMH) provided BDI results for the PCA, and valid PET scans were obtained from 58 of these patients. Most patients had been referred to NIMH by their psychiatrist because of treatment-resistant illness; the remainder was self-referred. Patients were medication-free ≥ 2 weeks for BDI administration and PET scanning. All studies were NIMH IRB approved, and each patient provided oral

BDI PCA

A four-component model was derived that explained 63.8% of the total variance of all BDI items (Table 1). When only BDI items loading ≥ .6 were included (i.e., above the horizontal line in Table 1), each component had unique items, yet 15 of 21 BDI items were included. These highly loaded BDI components were conceptualized as: negative cognitions, psychomotor-anhedonia, vegetative symptoms, and somatic symptoms, and explained 21.4, 21.2, 12.9 and 8.3% of the variance, respectively. The fourth

BDI principal components analyses

Four depressive symptom clusters were derived from the BDI and conceptualized as negative cognitions, psychomotor-anhedonia, vegetative, and somatic symptoms. Despite the compromised status of the somatic symptom component, the statistical validity of this PCA is strongly supported by the high proportion of total variance (64%) explained by these components when compared with other studies in primary affective disorders, where total variance explained ranged from 44% to 57% (Brown et al 1995;

Acknowledgements

This work was supported in part by Theodore and Vada Stanley. Previously presented in abstract form at the 151st Annual Meeting of the American Psychiatric Association (1988), Toronto, Canada.

References (75)

  • T Iidaka et al.

    Quantitative regional cerebral flow measured by Tc-99 M HMPAO SPECT in mood disorder

    Psychiatry Res

    (1997)
  • T.A Ketter et al.

    Baseline cerebral hypermetabolism associated with carbamazepine response, and hypometabolism with nimodipine response in mood disorders

    Biol Psychiatry

    (1999)
  • L.E Mello et al.

    Neuroanatomy of the basal ganglia

    Psychiatr Clin North Am

    (1997)
  • M.P Philpot et al.

    99 mTc-HMPAO single photon emission tomography in late life depressionA pilot study of regional cerebral blood flow at rest and during a verbal fluency task

    J Affect Disord

    (1993)
  • J.D Salamone et al.

    Behavioral functions of nucleus accumbens dopamineEmpirical and conceptual problems with the anhedonia hypothesis

    Neurosci Biobehav Rev

    (1997)
  • S Schlegel et al.

    Regional cerebral blood flow in depressionAssociations with psychopathology

    J Affect Disord

    (1989)
  • K Zilles

    Cortex

  • G.E Alexander et al.

    Basal ganglia-thalamocortical circuitsParallel substrates for motor, oculomotor, “prefrontal” and “limbic” functions

    Prog Brain Res

    (1990)
  • G.E Alexander et al.

    Parallel organization of functionally segregated circuits linking basal ganglia and cortex

    Annu Rev Neurosci

    (1986)
  • Diagnostic and Statistical Manual of Mental Disorders

    (1994)
  • L.R Baxter et al.

    Cerebral metabolic rates for glucose in mood disorders. Studies with positron emission tomography and fluorodeoxyglucose F 18

    Arch Gen Psychiatry

    (1985)
  • L.R Baxter et al.

    Reduction of prefrontal cortex glucose metabolism common to three types of depression

    Arch Gen Psychiatry

    (1989)
  • C.J Bench et al.

    Regional cerebral blood flow in depression measured by positron emission tomographyThe relationship with clinical dimensions

    Psychol Med

    (1993)
  • C.J Bench et al.

    The anatomy of melancholia - focal abnormalities of cerebral blood flow in major depression

    Psychol Med

    (1992)
  • M.B Carpenter et al.

    Human Neuroanatomy

    (1983)
  • R.M Cohen et al.

    Evidence for common alterations in cerebral glucose metabolism in major affective disorders and schizophrenia

    Neuropsychopharmacology

    (1989)
  • J.L Cummings

    Anatomic and behavioral aspects of frontal-subcortical circuits

    Ann NY Acad Sci

    (1995)
  • J.L Cummings

    Clinical Neuropsychiatry

    (1985)
  • H D’Haenen et al.

    Psychobiological substrate of hedonic capacity

    Clin Neuropharmocol

    (1992)
  • O Devinsky et al.

    Contributions of anterior cingulate cortex to behaviour

    Brain

    (1995)
  • R.J Dolan et al.

    Neuropsychological dysfunction in depressionThe relationship to regional cerebral blood flow

    Psychol Med

    (1994)
  • R.J Dolan et al.

    Regional cerebral blood flow abnormalities in depressed patients with cognitive impairment

    J Neurol Neurosurg Psychiatry

    (1992)
  • R.J Dolan et al.

    Dorsolateral prefrontal cortex dysfunction in the major psychoses; symptom or disease specificity?

    J Neurol Neurosurg

    (1993)
  • D Dougherty et al.

    Neuroimaging and neurobiological models of depression

    Harv Rev Psychiatry

    (1997)
  • W.C Drevets

    Functional neuroimaging studies of depressionThe anatomy of melancholia

    Annu Rev Med

    (1998)
  • W.C Drevets et al.

    Neuroimaging abnormalities in the subgenual prefrontal cortexImplications for the pathophysiology of familial mood disorders

    Mol Psychiatry

    (1998)
  • W.C Drevets et al.

    Subgenual prefrontal cortex abnormalities in mood disorders

    Nature

    (1997)
  • Cited by (158)

    • Neuroanatomical subtypes of schizophrenia and relationship with illness duration and deficit status

      2022, Schizophrenia Research
      Citation Excerpt :

      Deficit status can be debilitating and are associated with motivational deficits which are related to underlying basal ganglia changes (Goldsmith and Rapaport, 2020). Apart from motor functioning, the basal ganglia is thought to play a role in emotional processing, and has been implicated in affective flattening (Bragulat et al., 2007) and anhedonia (Dunn et al., 2002; Plailly et al., 2006) in clinical populations. Ballmaier and colleagues (2008) found an inverse relationship between volume changes in the anterior pole of the putamen and severity of blunted affect in unmedicated patients with SCZ.

    • Cerebral oxidative metabolism mapping in four genetic mouse models of anxiety and mood disorders

      2019, Behavioural Brain Research
      Citation Excerpt :

      In the current study, claustrum was a particularly important variable in CB1 and GRi mouse models. Previously it has been found to be hypoactive in human patients suffering from depression [54,55], and in socially defeated rats [13]. Brain-derived neurotropic factor (BDNF) may serve as an important mediator between stress-driven activation of HPA axis and subsequent post-synaptic remodelling that may lead to mood disorders [56].

    View all citing articles on Scopus
    View full text