Neuroimaging and Depression with Inadequate Treatment Response

Frank Andrew Kozel, MD, MS, and Mark S. George, MD

Dr. Kozel is assistant professor in the Department of Psychiatry and Behavioral Neurosciences at the Brain Stimulation Lab & Center for Advanced Imaging Research at the Medical University of South Carolina (MUSC), and psychiatry/neuroscience fellow at the Ralph H. Johnson Veterans Administration (VA) Medical Center, both in Charleston, SC.

Dr. George is distinguished professor of psychiatry, radiology and neurology, director of the Center for Advanced Imaging Research, and director of the Brain Stimulation Laboratory at MUSC.

Disclosure: Dr. Kozel has received research support from AstraZeneca, Cephos, Cyberonics, GlaxoSmithKline, the National Institute of Health, MUSC Intramural Funding, and the VA. Dr. George is a consultant to Aventis and Neotonus; is on the speaker’s bureaus of Cyberonics, Eli Lilly, GlaxoSmithKline, Janssen, Mediphysics/Amersham, Neotonus, Parke Davis, and Philips; has received grant support from Cortex, Cyberonics, Dupont, Eli Lilly, GlaxoSmithKline, Janssen, Mediphysics/Amersham, Parke Davis, and Solvay/Duphar; and has a formal research collaboration with Medtronic and Philips.

Funding/support: This work was supported by a VA neuroscience/psychiatric research fellowship awarded to Dr. Kozel.

Please direct all correspondence to: Frank Andrew Kozel, MD, MUSC Psychiatry Department, 67 President St, PO Box 250861, Charleston, SC 29425; Tel: 843-876-5142; Fax: 843-792-5702; E-mail: [email protected].

Focus Points

Abstract

Depression with an inadequate treatment response presents a significant public health problem that is poorly understood. Neuroimaging has been used as a tool to study the neurobiologic nature of this disorder. The results of these studies, however, have been inconsistent. A lack of clear definitions regarding inadequate treatment response combined with differing imaging methodologies has made interpreting the limited data challenging. Currently, there is insufficient neuroimaging data to conclusively determine whether treatment-resistant depression is a unique biological entity or merely a spectrum of other depressions. Although neuroimaging cannot presently help clinicians in making individual treatment decisions, it does hold future promise for providing a means to investigate the nature of this illness and possibly direct treatment in the future.

Introduction

Depression that does not respond adequately to treatment is a significant global public health problem.¹ Approximately 50% to 60% of patients treated with antidepressants do not achieve remission.^2,3 Attempts have been made to understand the nature of this inadequate treatment response with the hope of being able to make more effective treatment decisions. One method of scientific inquiry involves the use of neuroimaging to study brain correlates of inadequate treatment response. The most commonly employed modalities are structural imaging, such as computed tomography and magnetic resonance imaging (MRI), and functional imaging, such as single photon emission tomography (SPECT), positron emission tomography (PET), blood oxygen level dependent (BOLD) functional MRI (fMRI), and magnetic resonance spectroscopy (MRS). Each modality measures different aspects of the brain (ie, structure, function, chemical composition, etc.) with varying strengths and weaknesses.

Neuroimaging measures differences in the brains of patients with resistant depression versus a comparison group. Most often these analyses involve comparisons between a group of patients having some degree of resistant depression with groups of patients with depression that has responded to treatment and/or a control group without depression. Structural imaging has largely been used to measure volumetric differences (whole brain and regionally specific areas), differences in tissue density, and location and number of lesions. Functional imaging measures a broad array of brain activity depending on the technique. Examples of functional imaging include the measurement of blood flow (^99m–technetium hexamethylpropleneamineoxime [^99mTc-HAMPAO] SPECT, 0-15 PET, BOLD fMRI) that is thought to be tightly coupled with regional brain activity, metabolism (F-18-fluorodeoxyglycose PET [FDG PET]), receptor function (iomazenil SPECT, [11C]raclopride PET), and chemical composition (MRS). In addition to the multiple modalities of measurement, the methods of analysis have also been quite varied.

The lack of a consensus regarding a definition for depression that has not adequately responded to treatment^4,5 makes the comparability of neuroimaging results of this condition difficult. Even the most commonly used term, “treatment-resistant depression” (TRD), has been questioned. Another label proposed is “difficult–to-treat depression.”⁶ For this review, we have chosen to use TRD because it is the term most often used in the literature. The definition of TRD, however, is variable across the literature. A number of classification schemes have been proposed in order to define and quantify the degree of treatment resistance.^4,7-10 Although there is no consistent definition for TRD, a typical minimum criteria is inadequate response to at least one antidepressant trial of adequate dose and duration.⁴

There have been numerous studies of depression using neuroimaging.^11-13 Many, but not all, studies revealed differences from controls in brain structure and function. Unfortunately, the differences were not consistent across studies. One possible explanation for the lack of consistency is the disparity in the populations studied. If one does not study the same population (ie, different diseases or disease states), one would not expect to get the same results (ie, different neuroimaging findings). One source of disparity could be differences in neuropathological etiology. Depression is presently a syndrome (cluster of signs and symptoms)-based diagnosis¹⁴ versus an etiological one. Although the subjects across the studies were defined as suffering from depression by currently used criteria based on symptoms, they may actually have had different neuropathological diseases that are presently undefined. This is very difficult to test due to the lack of a “gold-standard” neuropathological diagnosis. A related possible difference in populations studied could be differences in level of severity and/or duration of illness. MacQueen and colleagues¹⁵ found differences in the hippocampal size as measured by MRI between subjects who were having their first episode of depression and subjects who had suffered multiple episodes of depression. Frodl and colleagues¹⁶ found differences in amygdala volume between subjects with first-episode depression, subjects with recurrent episodes of depression, and control subjects. Because TRD has shown a particular treatment response (not responded) and severity (occurred for extended period of time), one may be able to reduce the differences in populations studied and acquire more consistent results. The research investigating neuroimaging correlates of depression that has not adequately responded to treatment will be the focus of this review.

Neuroimaging Correlates of Treatment Resistance

In an attempt to delineate neuroimaging correlates of treatment resistance, a number of strategies have been employed. One strategy is to use a cross-sectional design to determine neuroimaging differences of subjects with depression who have responded to treatment, subjects with depression who have not responded to treatment, and subjects with no history of depression.

Shah and colleagues¹⁷ used voxel-based analysis of structural MRIs to compare 20 subjects with TRD, 20 subjects with depression who responded to treatment, and 20 subjects who had no psychiatric disease. Reduced grey-matter density was found in the left temporal cortex and there was a trend for reduction in the right hippocampus in subjects with TRD versus controls and subjects with recovered depression. No significant differences were found between subjects with recovered depression and control subjects in grey-matter density.

In a follow-up study, Shah and colleagues¹⁸ used a similar approach of voxel-based analysis with the addition of volumetric analysis to compare the TRD group with subjects who had recovered from depression and with control subjects. Using voxel-based analysis, differences in tissue density were found in the right putamen, bilateral hippocampal/parahippocampal areas, and right medial and superior frontal gyri. The TRD group had less right frontal cortex and right caudate tissue than controls.¹⁸

Although there was some consistency across the two studies, there were also differences. This highlights an important point of whether different groups defined as treatment resistant by history and symptoms may be neurobiologically different. When performing a group analysis, an assumption is made that the sample of subjects in each group studied is in some way a uniform group and is representative of a particular population. Unfortunately, there are very little data to support TRD as a particular subtype of depression.³ This makes generalizations from the group sampled to the population of patients with TRD very tenuous.

Using a similar cross-sectional study design, Hornig and colleagues¹⁹ used SPECT scanning to assess differences in regional cerebral blood flow in unmedicated subjects with TRD. Eight presently depressed subjects with a history of treatment resistance (the TRD group), 13 presently depressed subjects without a history of treatment resistance (the non-TRD group), and 16 controls were compared to determine differences between the groups. A significant increase for TRD versus non-TRD and controls was found in the ratio of activity in the hippocampus-amygdala area relative to cortex. These findings are in contrast to those of Mayberg and colleagues²⁰ who compared 13 medicated TRD subjects with 11 non-depressed controls. They found a reduction in relative blood flow in the bilateral frontal and anterior temporal cortex, anterior cingulate gyrus, and caudate using the cerebellum as a control region. Although the results of these studies are quite different, comparing them is difficult due to the different criteria used to define TRD, the differences in analysis methods, and the differences in the medication status of the subjects with depression.

Kimbrell and colleagues²¹ investigated the brain correlates of depression using FDG PET. Medication-free subjects (n=75 [38 with depression, 37 controls]) with a history of unipolar depression who were referred to the National Institute of Mental Health for clinical research and medication trials were compared to a control group in a cross-sectional manner. Although the degree of treatment resistance was not formally assessed, the group studied consisted of patients who had not responded to multiple treatment interventions (M.S. George, MD, verbal communication, 2004). No difference was found in global regional cerebral glucose metabolism between the two groups. The subjects with depression were then placed into two groups: one group (n=11) was symptomatic at the time of imaging (Hamilton Rating Scale for Depression [HAM-D] score >22); and one group (n=9) was asymptomatic at the time of imaging (HAM-D score <10). These two groups were compared to 11 separate matched controls. Compared to matched controls, the symptomatic subjects had significantly decreased regional cerebral glucose metabolism in the right dorsolateral and bilateral medial prefrontal cortex, bilateral anterior paralimbic areas, the temporal pole, and the insula on the absolute regional analysis. Conversely, the asymptomatic subjects did not show any differences from the nine matched controls. These results highlight the importance of “state” in interpreting and comparing neuroimaging results in depression.

Using a cross-sectional design, Kumari and colleagues²² compared differences in brain activation using BOLD fMRI in six subjects with TRD and six healthy controls. With a task of generating affect using picture-caption pairs, differences in relative activation were found in a number of regions. The TRD group had a relative decrease in activation for various contrasts in the anterior cingulate, left medial frontal gyrus, and left hippocampus. An increased blood flow response for subjects with TRD using various contrasts was found in the temporal lobe, inferior frontal gyrus, subgenual cingulate, striatum, and brain stem. Because a group with depression that was not treatment resistant was not included, no statements about which brain correlates were associated with treatment resistance versus depression could be made.

Taken as a group, the literature concerning the neuropathological correlates of TRD is too variable to make any conclusions. What is unknown is whether the variability in results is due to fundamental neurobiological differences in the “depressions” that have been grouped as TRD, or or due to the use of various measurement and analysis techniques. Further work on larger samples of subjects will be required to answer this very important question.

Neuroimaging Predictors and Correlates of Response to Treatment

Another method to investigate the neural correlates of treatment resistance is to test for predictors of treatment response and correlates of treatment response using various imaging modalities. This method has been employed for various treatment options including pharmacotherapy, cognitive-behavioral therapy (CBT), sleep deprivation, and somatic treatments, such as electroconvulsive therapy (ECT), transcranial magnetic stimulation (TMS), and stereotactic anterior cingulotomy surgery. If predictors of treatment response can be identified, then this information can be used to better understand the neurobiology of response and nonresponse and to select more optimal treatments. Similarly, knowing the correlates of treatment response provides the benefit of a greater understanding of what is neurobiologically associated with response and potentially offers a way to test more quickly whether a treatment will work instead of having to wait weeks to months.

Pharmacotherapy and Cognitive-Behavioral Therapy

A number of studies have been performed with various medications and imaging modalities. Passero and colleagues²³ used the Xenon 133 inhalation method to measure cerebral blood flow (CBF) prior to treatment and after 6 months of treatment in order to study correlates of successful treatment. Successful treatment with amitriptyline (n=16) and amineptine (n=10) was associated with increases in left frontal CBF. Initial scans also revealed that baseline resting CBF tended to be lower in depressed subjects compared to 28 age-matched, normal subjects. In comparison, Brody and colleagues²⁴ used FDG PET in 16 subjects taking paroxetine for major depressive disorder (MDD) to investigate both the correlates of response (HAM-D scores >50% and Clinical Global Impression rating of “much” or “very much” improved) and predictors of treatment response. Responders (n=9) had a greater decrease in normalized ventrolateral prefrontal cortex and orbitofrontal cortex metabolism than nonresponders (n=7). Additionally, lower metabolism in the left ventral anterior cingulate gyrus on pre-treatment scans was associated with a better treatment response.

Mayberg and colleagues²⁵ compared four responders (≥50% decrease in HAM-D scores) and four nonresponders (<20% decrease in HAM-D score) to fluoxetine treatment of depression with FDG PET. Clinical response was associated with limbic and striatal decreases (subgenual cingulate, hippocampus, insula and pallidum) and brain stem and dorsal cortical increases (prefrontal, parietal, anterior cingulate, and posterior cingulate). A failed response was associated with an absence of either subgenual cingulate or prefrontal changes.

These studies highlight the difficulty of assimilating these studies into a coherent picture. Because different medications were used, different measurement and analyses performed, and the numbers in each sample were quite small, it is difficult to ascertain which factor was responsible for the differences and similarities in findings.

There is some indication that correlates of treatment response in depression may be different for medications versus CBT. Goldapple and colleagues²⁶ used FDG PET to investigate the neural correlates of treatment response in 14 unmedicated subjects receiving CBT in order to treat MDD (17 subjects started the trial but three dropped out within the first 2 weeks). Of the 14 completers, 9 had a full response (>50% reduction in HAM-D scores) and the other 5 had a partial response (≥35% reduction in HAM-D scores). The researchers²⁶ included all subjects in the comparison of brain metabolism changes from pre to post-treatment. Considering all responders, response was correlated with increased metabolism in hippocampus and dorsal cingulate and decreases in dorsal, ventral, and medial frontal cortex.

Overall, studies of medications and CBT support the concept that changes are required in the brain circuitry for successful treatment of depression. What those regions are, which direction of change is required, and how specific the changes are to treatment modality remain unclear. For a subgroup of depression, late-life onset depression,^27-30 there may be unique predictors of treatment response.

Navarro and colleagues³¹ used SPECT in subjects being treated with nortriptyline for late-life onset depression (n=47). For the 34 remitters, lower left anterior frontocerebellar perfusion ratio predicted response. Unfortunately, comparing these imaging results with other age groups suffering from depression is difficult due to the confounds of age-related differences in neurophysiology as well as the prior mentioned factors. A predictor of response that is unique to the subgroup of geriatric depression, however, is the presence of subcortical hyperintensities.

Fujikawa and colleagues³² (n=41) used structural MRIs in subjects >50 years of age to determine that a severe degree of silent cerebral infarction was associated with requiring a longer hospitalization and more drug-related adverse events for patients who were admitted for pharmacotherapy of unipolar depression. Simpson and colleagues³³ also found that subcortical hyperintensities were significantly increased in subjects who did not respond as well to treatment with antidepressants (n=75).

Using diffusion tensor imaging, a presumed measure of white matter tract structural integrity, Alexopoulos and colleagues³⁴ found that lower fractional anisotropy (less structural integrity) of left and right frontal (15 mm above anterior commissure–posterior commissure plane) white matter was associated with lower remission rates in elderly subjects. In comparison, Baldwin and colleagues³⁵ investigated geriatric subjects who were depressed (n=50; 29 responders, 21 non-responders) and 35 controls. No difference in atrophy or white-matter lesions that would predict treatment response was found. However, a difference was found for nonresponders having a higher periventricular hyperintensity score than controls.

These studies highlight the variability inherent in the literature concerning neuroimaging and TRD. Thus, predictors of response may be unique not only to the modality of treatment, but the age and/or subgroup (eg, late-life onset versus early life onset) of patients suffering from depression.

Sleep Deprivation

Sleep deprivation is an intervention that can temporarily treat depressive symptoms in a subset of patients.³⁶ Functional neuroimaging studies of the treatment predictors and correlates of response in sleep deprivation have produced more consistent results than are present with other treatments. Performing FDG PET, Wu and colleagues³⁷ imaged 36 subjects with depression and 26 controls prior to and after sleep deprivation. Twelve of the subjects with depression had a >40% improvement in symptoms. Compared to healthy controls and nonresponders, responders were found to have significantly higher relative metabolic rates in the ventral anterior cingulate, medial prefrontal cortex, and posterior subcallosal cortex prior to sleep deprivation. Conversely, nonresponders had lower metabolic rates in the right anterior cingulate and medial prefrontal cortex than normal controls. Improvement in depressive symptoms was associated with decreases in the medial prefrontal cortex and frontal pole. Both major findings were results that by and large replicated an earlier study.³⁸

Similarly, Volk and colleagues³⁹ found higher perfusion in the right orbitofrontal and basal cingulate using SPECT to be predictive of treatment response in 15 patients with major depression. This general pattern of response to sleep deprivation being associated with reduction in metabolism as measured by FDG PET in the right anterior cingulate and medial prefrontal cortex has been found in geriatric patients as well.⁴⁰ The consistency of results may be the result of more homogenous patient populations, more uniformity of imaging technique and analysis, or a robust and unique response associated with sleep deprivation. Further work is needed to clarify this very intriguing phenomenon.

Somatic Treatments

Transcranial Magnetic Stimulation

While investigating the neuroimaging correlates of treatment predictors and treatment response in rTMS, being aware of treatment parameters is an important variable. Speer and colleagues⁴¹ found that treatment of depression for 2 weeks (10 sessions) in 10 subjects at different frequencies had differing effects on the resting CBF (rCBF) (1 Hz decreased rCBF and 20 Hz increased rCBF) as measured by 0-15 PET. Loo and colleagues⁴² also found that different frequencies had differing effects on brain changes using SPECT. The blood flow changes, however, were quite different from the changes identified by Speer and colleagues.⁴¹ This could be related to differences in imaging modality, analysis, or patient population studied.

In addition to also showing that TMS frequency can have an impact on neuroimaging, Nahas and colleagues⁴³ have provided functional imaging support for the importance of the distance from coil to cortex (especially in geriatric subjects) on TMS administration that has been found on structural imaging.^44-46 Other TMS parameters identified as having an impact on neuroimaging results include the number of trains delivered.⁴⁷ Clearly, the TMS parameters must be taken into account when interpreting the results of neuroimaging studies of TMS. On a positive note, these variations in imaging results due to parameter variations may eventually help identify better TMS parameter choices (eg, left frontal high frequency TMS) for patients with certain pretreatment imaging characteristics (eg, left frontal hypofrontality). Although this is presently not feasible, the methodology offers great potential. As an example of a study using neuroimaging to investigate correlates and predictors of response for TMS, Nadeau and colleagues⁴⁸ imaged depressed subjects using SPECT prior to and post 10 treatments of 20 Hz rTMS. Less regional blood flow in left amygdala was predictive of treatment response in the six of eight responders. Responders had decreases in orbitofrontal (n=3) and/or anterior cingulate (n=2) and/or right insula (n=3) and right amygdala (n=1). This small study cannot guide a treatment decision but it can be a building block from which further studies can be performed.

Electroconvulsive Therapy

Several studies have investigated the neural correlates of treatment with ECT. Mervaala and colleagues⁴⁹ found that response to ECT for treatment-resistant patients was correlated with significant SPECT ethylcysteinate dimer (ECD) uptake ratios in the right temporal and bilateral parietal cortices as well as increase in uptake of iomazenil uptake (indicates an increase in uptake by benzodiazepine receptors) in all regions except the right temporal lobe. These results suggest increased perfusion and possibly changes in the g-aminobutyric acid system in regions of the brain. Using MRS of the left dorsolateral prefrontal cortex, Michael and colleagues⁵⁰ found that glutamate/glutamine levels were reduced in 12 severely depressed subjects prior to right unilateral ECT. The severity of depression was correlated negatively with glutamate/glutamine levels. After successful treatment, glutamate/glutamine levels increased and were not distinguishable from controls. These studies begin to provide some indication of the mechanism of action of ECT.

Stereotactic Anterior Cingulotomy

Because brain surgery is such an invasive procedure, having a means to predict which patients would be most likely to respond would be quite valuable. In an attempt to look for treatment predictors, Dougherty and colleagues⁵¹ imaged subjects (n=13) using FDG PET prior to undergoing stereotactic anterior cingulotomy for severe TRD. A higher preoperative metabolism in left subgenual prefrontal cortex and left thalamus was significantly correlated with improvement in depressive symptoms. Determining whether these regions will be important predictors of response requires replication, but the method offers a means in which better candidates might be identified preoperatively. These studies also provide a tantalizing hint of the role that neuroimaging may play in the future management of TRD.

Conclusion

The neuroimaging literature concerning TRD presently provides little direction to guide treatment choices. The diversity of results makes any conclusions regarding correlates of treatment resistance, treatment response, or predictors of response very tenuous at best. The potential reasons for these differences are numerous. Besides the discussed methodological variations between studies, the differences may be the result of TRD not comprising a neurobiologically related and distict entity. TRD may be the extreme end of the spectrum of a number of etiologically distinct diseases that have been clumped together and called depression. Despite neuroimaging’s present limitations, it holds great promise as a tool to investigate the nature of TRD. Further work with larger sample sizes, advancing technology, and consistent analysis methods offer the very real possibility that neuroimaging could become a critical tool in the management of TRD in the future. PP

References

1. Greden JF. The burden of recurrent depression: causes, consequences, and future prospects. J Clin Psychiatry. 2001;62(Suppl 22):5-9.

2. O’Reardon JP, Amsterdam JD. Treatment-resistant depression: Progress and limitations. Psychiatr Ann. 1998;28(11):633-640.

3. Fagiolini A, Kupfer DJ. Is treatment-resistant depression a unique subtype of depression? Biol Psychiatry. 2003;53(8):640-648.

4. Fava M. Diagnosis and definition of treatment-resistant depression. Biol Psychiatry. 2003;53(8):649-659.

5. Nierenberg AA, DeCecco LM. Definitions of antidepressant treatment response, remission, nonresponse, partial response, and other relevant outcomes: a focus on treatment-resistant depression. J Clin Psychiatry. 2001;62(Suppl 16):5-9.

6. Kupfer DJ, Charney DS. Difficult-to-treat depression. Biol Psychiatry. 2003;53(8):633-634.

7. Thase ME, Rush AJ. When at first you don’t succeed: sequential strategies for antidepressant nonresponders. J Clin Psychiatry. 1997;58(Suppl 13):23-29.

8. Souery D, Amsterdam J, de Montigny C, et al. Treatment resistant depression: methodological overview and operational criteria. Eur Neuropsychopharmacol. 1999;9(1-2):83-91.

9. Burrows GD, Norman TR, Judd FK. Definition and differential diagnosis of treatment-resistant depression. Int Clin Psychopharmacol. 1994;9(Suppl 2):5-10.

10. Sackeim HA. The definition and meaning of treatment-resistant depression. J Clin Psychiatry. 2001;62(Suppl 16):10-17.

11. George MS, Ketter TA, Post RM. SPECT and PET imaging in mood disorders. J Clin Psychiatry. 1993;54:6-13.

12. Drevets WC. Neuroimaging studies of mood disorders. Biol Psychiatry. 2000;48(8):813-829.

13. Mayberg HS. Positron emission tomography imaging in depression: a neural systems perspective. Neuroimaging Clin N Am. 2003;13(4):805-815.

14. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.

15. MacQueen GM, Campbell S, McEwen BS, et al. Course of illness, hippocampal function, and hippocampal volume in major depression. Proc Natl Acad Sci U S A. 2003;100(3):1387-1392.

16. Frodl T, Meisenzahl EM, Zetzsche T, et al. Larger amygdala volumes in first depressive episode as compared to recurrent major depression and healthy control subjects. Biol Psychiatry. 2003;53(4):338-344.

17. Shah PJ, Ebmeier KP, Glabus MF, Goodwin GM. Cortical grey matter reductions associated with treatment-resistant chronic unipolar depression. Controlled magnetic resonance imaging study. Br J Psychiatry. 1998;172:527-532.

18. Shah PJ, Glabus MF, Goodwin GM, Ebmeier KP. Chronic, treatment-resistant depression and right fronto-striatal atrophy. Br J Psychiatry. 2002;180:434-440.

19. Hornig M, Mozley PD, Amsterdam JD. HMPAO SPECT brain imaging in treatment-resistant depression. Prog Neuropsychopharmacol Biol Psychiatry. 1997;21(7):1097-1114.

20. Mayberg HS, Lewis PJ, Regenold W, Wagner HN Jr. Paralimbic hypoperfusion in unipolar depression. J Nucl Med. 1994;35(6):929-934.

21. Kimbrell TA, Ketter TA, George MS, et al. Regional cerebral glucose utilization in patients with a range of severities of unipolar depression. Biol Psychiatry. 2002;51(3):237-252.

22. Kumari V, Mitterschiffthaler MT, Teasdale JD, et al. Neural abnormalities during cognitive generation of affect in treatment-resistant depression. Biol Psychiatry. 2003;54(8):777-791.

23. Passero S, Nardini M, Battistini N. Regional cerebral blood flow changes following chronic administration of antidepressant drugs. Prog Neuropsychopharmacol Biol Psychiatry. 1995;19(4):627-636.

24. Brody AL, Saxena S, Silverman DH, et al. Brain metabolic changes in major depressive disorder from pre- to post-treatment with paroxetine. Psychiatry Res. 1999;91(3):127-139.

25. Mayberg HS, Brannan SK, Tekell JL, et al. Regional metabolic effects of fluoxetine in major depression: serial changes and relationship to clinical response. Biol Psychiatry. 2000;48(8):830-843.

26. Goldapple K, Segal Z, Garson C, et al. Modulation of cortical-limbic pathways in major depression: treatment-specific effects of cognitive behavior therapy. Arch Gen Psychiatry. 2004;61(1):34-41.

27. Salloway S, Malloy P, Kohn R, et al. MRI and neuropsychological differences in early- and late-life-onset geriatric depression. Neurology. 1996;46(6):1567-1574.

28. Coffey CE, Figiel GS, Djang WT, Saunders WB, Weiner RD. White matter hyperintensity on magnetic resonance imaging clinical and neuroanatomic correlates in the depressed elderly. J Neuropsychiatry. 1989;1(2):135-144.

29. Krishnan KR. Neuroanatomic substrates of depression in the elderly. J Geriatr Psychiatry Neurol. 1993;6(1):39-58.

30. Alexopoulos GS, Meyers BS, Young RC, et al. Clinically defined vascular depression. Am J Psychiatry. 1997;154(4):562-565.

31. Navarro V, Gasto C, Lomena F, et al. Prognostic value of frontal functional neuroimaging in late-onset severe major depression. Br J Psychiatry. 2004;184:306-311.

32. Fujikawa T, Yokota N, Muraoka M, Yamawaki S. Response of patients with major depression and silent cerebral infarction to antidepressant drug therapy, with emphasis on central nervous system adverse reactions. Stroke. 1996;27(11):2040-2042.

33. Simpson S, Baldwin RC, Jackson A, Burns AS. Is subcortical disease associated with a poor response to antidepressants? Neurological, neuropsychological and neuroradiological findings in late-life depression. Psychol Med. 1998;28:1015-1026.

34. Alexopoulos GS, Kiosses DN, Choi SJ, Murphy CF, Lim KO. Frontal white matter microstructure and treatment response of late-life depression: A preliminary study. Am J Psychiatry. 2002;159(11):1929-1932.

35. Baldwin R, Jeffries S, Jackson A, et al. Treatment response in late-onset depression: relationship to neuropsychological, neuroradiological and vascular risk factors. Psychol Med. 2004;34(1):125-136.

36. Wu JC, Bunney WE. The biological basis of an antidepressant response to sleep deprivation and relapse: review and hypothesis. Am J Psychiatry. 1990;147(1):14-21.

37. Wu J, Buchsbaum MS, Gillin JC, et al. Prediction of antidepressant effects of sleep deprivation by metabolic rates in the ventral anterior cingulate and medial prefrontal cortex. Am J Psychiatry 1999;156(8):1149-1158. Erratum in: Am J Psychiatry. 1999;156(10):1666.

38. Wu JC, Gillin JC, Buchsbaum MS, et al. Effect of sleep deprivation on brain metabolism of depressed patients. Am J Psychiatry. 1992;149(4):538-543.

39. Volk SA, Kaendler SH, Hertel A, et al. Can response to partial sleep deprivation in depressed patients be predicted by regional changes of cerebral blood flow? Psychiatry Research. 1997;75(2):67-74.

40. Smith GS, Reynolds CF 3rd, Pollock B, et al. Cerebral glucose metabolic response to combined total sleep deprivation and antidepressant treatment in geriatric depression. Am J Psychiatry. 1999;156(5):683-689.

41. Speer AM, Kimbrell TA, Wasserman EM, et al. Opposite effects of high and low frequency rTMS on regional brain activity in depressed patients. Biol Psychiatry. 2000;48(23):1133-1141.

42. Loo CK, Sachdev PS, Haindl W, et al. High (15 Hz) and low (1 Hz) frequency transcranial magnetic stimulation have different acute effects on regional cerebral blood flow in depressed patients. Psychol Med. 2003;33(6):997-1006.

43. Nahas Z, Teneback CT, Kozel AF, et al. Brain effects of transcranial magnetic delivered over prefrontal cortex in depressed adults: the role of stimulation frequency and distance from coil to cortex. J Neuropsychiatry Clin Neurosci. 2001;13(4):459-470.

44. Kozel FA, Nahas Z, DeBrux C, et al. How the distance from coil to cortex relates to age, motor threshold and possibly the antidepressant response to repetitive transcranial magnetic stimulation. J Neuropsychiatry Clin Neurosci. 2000;12:376-384.

45. McConnell KA, Nahas Z, Shastri A, et al. The transcranial magnetic stimulation motor threshold depends on the distance from coil to underlying cortex: a replication in healthy adults comparing two methods of assessing the distance to cortex. Biol Psychiatry. 2001;49(5):454-459.

46. Mosimann UP, Marre SC, Werlen S, et al. Antidepressant effects of repetitive transcranial magnetic stimulation in the elderly: correlation between effect size and coil-cortex distance. Arch Gen Psychiatry. 2002;59:560-561.

47. Paus T, Jech R, Thompson CJ, et al. Dose-dependent reduction of cerebral blood flow during rapid-rate transcranial magnetic stimulation of the human sensorimotor cortex. J Neurophysiol. 1998;79(2):1102-1107.

48. Nadeau SE, McCoy KJ, Crucian GP, et al. Cerebral blood flow changes in depressed patients after treatment with repetitive transcranial magnetic stimulation: evidence of individual variability. Neuropsychiatry Neuropsychol Behav Neurol. 2002;15(3):159-175.

49. Mervaala E, Kononen M, Fohr J, et al. SPECT and neuropsychological performance in severe depression treated with ECT. J Affect Disord. 2001;66(1):47-58.

50. Michael N, Erfurth A, Ohrmann P, et al. Metabolic changes within the left dorsolateral prefrontal cortex occurring with electroconvulsive therapy in patients with treatment resistant unipolar depression. Psychol Med. 2003;33(7):1277-1284.

51. Dougherty DD, Weiss AP, Cosgrove GR, et al. Cerebral metabolic correlates as potential predictors of response to anterior cingulotomy for treatment of major depression. J Neurosurg. 2003;99(6):1010-1017.