Changes in Regional Cerebral Blood Flow following Light Treatment for Seasonal Affective Disorder: Responders versus Nonresponders

Changes in Regional Cerebral Blood Flow following Light Treatment for Seasonal Affective Disorder: Responders versus Nonresponders Russell G. Vasile, Gary Sachs, Janis L. Anderson, Beny Lafer, Elizabeth Matthews, and Thomas Hill Background: Several brain imaging studies of antidepressant pharmacologic treatment utilizing single photon emission computed tomography (SPECT) have reported a normalization of deficits in cerebral blood flow (CBF) associated with recovery; other studies report no change, or a reduction in CBF following successful treatment. There have been no published SPECT studies of seasonal affective disorder (SAD) assessing response to light treatment in relation to changes in regional CBF (rCBF). In this study, we sought to test the hypothesis that increases in rCBF would be observed in SAD patients who responded to light treatment. Methods: Ten depressed patients with SAD underwent functional brain imaging studies with 99m Tc-hexamethylpropyleneamine oxime SPECT before and after light treatment. Results: Relative increases in rCBF were observed in all brain regions compared to cerebellum in treatment responders, whereas nonresponders showed no change or decreases in rCBF relative to cerebellum. Significant differences in mean percentage change in rCBF between responders (n 5 5) and nonresponders (n 5 5) were detected in frontal and cingulate cortex, and thalamus. Conclusions: These findings provide preliminary support for the hypothesis that an increase in rCBF is associated with recovery from depression in SAD. © 1997 Society of Biological Psychiatry Key Words: Cerebral blood flow, seasonal affective disorder, light treatment BIOL PSYCHIATRY 1997;42:1000 –1005

Introduction The majority of single photon emission computed tomography (SPECT) and positron emission tomography (PET) investigations of depressed patients report deficits in From the Department of Psychiatry (RGV, GS, JLA, BL) and Department of Radiology (EM, TH), Harvard Medical School; Deaconess Hospital (RGV, EM, TH); Massachusetts General Hospital (GS, BL); and Department of Medicine, Division of Psychiatry, Brigham and Women’s Hospital (JLA), Boston, Massachusetts. Address reprint requests to Dr. Russell G. Vasile, Department of Psychiatry, Deaconess Hospital, Palmer Baker Span 4, One Deaconess Road, Boston, MA 02215. Received October 31, 1995.

© 1997 Society of Biological Psychiatry

cerebral activity in the frontal and prefrontal cortex, and somewhat less consistently in limbic structures and basal ganglia (Lesser et al 1994; George et al 1993). Several functional imaging studies of antidepressant pharmacologic treatment have described a normalization of deficits in cerebral activity associated with recovery, although other reports have reported no change or a reduction in cerebral blood flow (CBF) or cerebral metabolic rate following successful treatment (Rubin et al 1994). Findings, however, have varied depending on subclassification of depression, treatment used, and differences in methodology. 0006-3223/97/$17.00 PII S0006-3223(97)00155-8

Regional Cerebral Blood Flow in SAD

BIOL PSYCHIATRY 1997;42:1000 –1005

1001

We found no published SPECT studies of seasonal affective disorder (SAD) assessing response to light treatment in relation to changes in regional CBF (rCBF). One PET study assessed 6 SAD patients who responded to light treatment. A within-subject comparison before and after treatment revealed one significant finding—an increase in glucose metabolic rate in the right posterior frontal cortex (Cohen et al 1992). In the present study, we utilized high-resolution SPECT to test the hypothesis that increases in cortical rCBF would be observed in SAD patients who responded to light treatment.

Methods and Materials Subjects between 18 and 65 years of age were recruited from clinical trials offering light therapy to consenting SAD patients. Prospective patients were diagnosed using the Structured Clinical Interview for DSM-III-R and a modified seasonality module. Depressive symptoms were rated by trained psychiatrists using the Structured Interview Guide for the Hamilton Depression Rating Scale– Seasonal Affective Disorders Version (SIGH-SAD), which includes atypical vegetative symptoms (Williams 1988). Inclusion criteria included: meeting DSM-III-R criteria for major affective disorder seasonal type or bipolar II not otherwise specified, seasonal pattern; and attaining a SIGH-SAD score greater than 20. Scans were performed during the months associated with the onset of seasonal affective disorder; subjects were studied between October and December. Exclusion criteria were: failure to respond to a previous trial of adequate light treatment; having a history of psychosis, epilepsy, mania, or alcohol or drug abuse in the past 3 months; being pregnant or acutely suicidal; or having begun psychotherapy in the preceding 4 weeks. Patients with a history of cocaine abuse were also excluded. After complete description of the study to the subjects, written informed consent was obtained. A total of 10 patients (8 women and 2 men) participated in the study. Their mean age was 33.5 years (SD 5 11.3 years). More detailed assessment of the medication and medication withdrawal status of the patients revealed that 5 patients had never been on antidepressant medication. Of the 5 patients who had been treated with antidepressant medication in the past, 3 had been off their medication for 1 month or longer prior to entry in the study, 1 had been off for 1 week, and 1 patient was subsequently found to have continued fluoxetine 20 mg/day throughout the study. Prior to light treatment, patients were reassessed on the SIGH-SAD, underwent a SPECT scan, and were given light boxes and instructions for use beginning the day after the baseline SPECT study. Treatment consisted of . 2500

Figure 1. Template for cerebral regions of interest. Regions of interest chosen for analysis included the following: A, cingulate; B, C, frontal; D, E, thalamus; F, G, temporoparietal; H, I, basal ganglia; J, K, occipital. Right and left brain regions were combined for purposes of analysis.

lux delivered via commercially available fluorescent light boxes used at home for 1/2 to 2 hours daily. After approximately 1 week (6 –10 days) of light treatment, patients returned for a follow-up SIGH-SAD and a second SPECT study. Patients were considered responders if their 1-week SIGH-SAD indicated improvement of at least 60%. The psychiatrist (RGV) rater of the SIGH-SAD remained blind to the SPECT results. We utilized a dedicated high-resolution SPECT head scanner (Strichman SME810) with single slice reconstructed in-plane resolution of 8 mm and axial resolution of 12 mm (Stoddart and Stoddart 1992). Twenty axial slices were obtained sequentially (spacing 6 mm; matrix, 128 3 128 pixels, with the slices aligned parallel to the canthomeatal line). Counts were attenuation corrected. A region of interest (ROI) was drawn around an entire axial slice, which was then used to calculate the whole brain/ cerebellum ratio. Whole brain was calculated as a separate measure reflecting the mean value of the separate ROIs expressed as a ratio to cerebellum—a relative, not absolute measure of whole brain activity. In obtaining whole brain data, whole brain volume was realigned with automated software that controlled pitch, roll, and yaw. This whole

1002

R.G. Vasile et al

BIOL PSYCHIATRY 1997;42:1000 –1005

Table 1. Clinical and Treatment Characteristics of Study Subjects

Patient Responders 1 2 3 4 5

Age

Sex

Lux

Entry Hamilton score

47 27 27 40 24

F M F F F

10,000 10,000 10,000 2,500 2,500

28 23 22 39 27

7 3 9 16 2

, 1 month 1–3 months , 1 month 3– 6 months . 6 months

None None None Prozaca Anafranil, elavil, pamelor

No No No Yes Yes

10 7 12 11 16

No No Yes Yes No

F F F F

10,000 2,500 2,500 2,500

20 29 23 38

32 16 21 33

3– 6 months 1–3 months 3– 6 months 3– 6 months

Yes Yes No Yes

11 16 3 31

No No Yes No

M

2,500

27

25

1–3 months

Norpramin None None Prozac, parnate, cylert None

No

23

No

Nonresponders 6 26 7 36 8 22 9 55

10

36

Exit Hamilton score

Duration of episode prior to study

Treatment this episode

Treatment ever?

Duration of illness (years)

History of substance abuse

a

Continued during light treatment.

brain volume was then resliced for purposes of data analysis (Hill et al 1993a). For each scan 20 mCi of 99m Tc-hexamethylpropyleneamine oxime was administered intravenously, and images were obtained 20 –30 min later. Patients were required to lie still and silent in a quiet dimly lighted room. A template prepared by drawing ROIs from a standard brain atlas (Mills et al 1988) was used for every patient and control. The template could be linearly and symmetrically modified by eye to fit different brains. The ROIs that were selected outlined bilateral cortical areas in the cingulate, basal ganglia, frontal, occipital, temporoparietal, and thalamus brain regions. The ROIs chosen were large enough to provide reliable quantitative information based on the limits of resolution of the SPECT instrumentation (Figure 1). This ROI methodology has been demonstrated to have excellent test–retest reliability. A similar ROI methodology has been utilized by Mayberg et al (1994) in SPECT investigation of depressive disorders. We analyzed five middle transaxial slices (8 through 12), 30 mm thick, approximately 4 cm above the canthomeatal line. For each scan, the mean activity in each ROI was normalized to cerebellum (ROI/cerebellum). Normalization of ROI data to cerebellum, a standard methodologic strategy in brain SPECT studies, has been utilized in studies of Alzheimer’s disease and depressive disorders (Holman et al 1992; Sackheim and Prohovnik 1993). An automated program realigned and coregistered paired SPECT images obtained before and after light treatment (Hill et al 1993). The percentage change after treatment was determined for the normalized (n) data, ([nROI(postRX) 2 nROI(pre-RX)])/[nROI(pre-RX)]. The radiologist

who analyzed the pretreatment and posttreatment SPECT studies was blind to each subject’s clinical status and whether the scan being analyzed was a pre- or posttreatment study.

Results Study entry and exit SIGH-SAD score, clinical characteristics of the subjects, and treatment history are presented in Table 1. In the treatment-responsive patient group (n 5 5) (4 women, 1 man), pretreatment SIGH-SAD score 5 32.6 (SD 5 4.5) and posttreatment score 5 8 (SD 5 5), reflecting a mean reduction in SIGH-SAD scores of 76% (p , .001, paired t test). Five patients (4 women, 1 man) failed to respond to light therapy; these patients had a pretreatment SIGH-SAD score 5 30.6 (SD 5 6.3) and a posttreatment SIGH-SAD score 5 25.2 (SD 5 6.7) (NS, paired t test). Comparisons of the percentage changes in rCBF before and after light treatment were made for each group on each of the seven brain regions analyzed. Treatment-responsive patients showed a trend toward relatively increased rCBF after light treatment in all regions examined, whereas nonresponsive patients showed either a relative decrease or no change. Based on two-tailed t tests, the differences between responders and nonresponders reached statistical significance (p , .05) in frontal and cingulate cortex, and thalamus (Figure 2). Nonparametric analysis utilizing the Wilcoxon–Mann–Whitney test also found significant differences in the frontal and cingulate cortex, and thalamus. Having determined these differences between responders and nonresponders, we sought to assess whether there

Regional Cerebral Blood Flow in SAD

BIOL PSYCHIATRY 1997;42:1000 –1005

1003

Figure 2. Percentage change in rCBF after light treatment for SAD. Treatment-responsive patients (dark shading) demonstrate a consistent trend toward increase in rCBF in all brain regions, whereas nonresponders show no change or decrease. Differences in percentage change in rCBF between responders and nonresponders following light treatment attained statistical significance (p , .05) in the following brain regions: *frontal (t 5 2.3197, df 5 8, p 5 .0489); **cingulate (t 5 2.4031, df 5 8, p 5 .0430); ***thalamus (t 5 2.9227, df 5 8, p 5 .0192). Nonparametric analysis revealed significant differences after treatment in the identical regions: frontal (exact NP 5 .03), cingulate (exact NP 5 .03), and thalamus (exact NP 5 .01).

were any correlations between change in SPECT parameters and change in mood. Change in SIGH-SAD score was well correlated with changes in whole brain blood flow and rCBF in several areas of interest including left basal ganglia, right occipital, left and right frontal, and left and right thalamus (Table 2). Consideration of responders and nonresponders separately shows the correlations in the above ROIs are largely accounted for by the relationship between rCBF and improvement in the responders. Data analysis revealed no interaction between the time of year SPECT scans were performed and response or nonresponse to light treatment.

Table 2. Correlation between Percent Change in rCBF and Change in Hamilton Depression Rating Scale Score with Light Treatment Brain region Cingulate gyrus Left basal ganglia Right basal ganglia Left frontal Right frontal Left occipital Right occipital Left temporal parietal Right temporal parietal Left thalamus Right thalamus Whole brain

Entire sample

Responders

Nonresponders

.48 .59 .35 .51 .58 .47 .51 .44 .26 .57 .65 .51

.87 .76 2.02 .96 .96 .58 .70 .90 .96 .34 .71 .93

2.99 .55 .24 2.78 2.54 2.20 2.29 2.71 2.94 2.03 2.10 2.78

Discussion This pilot study, the first high-resolution SPECT investigation of changes in rCBF in SAD patients as a function of response to light treatment, provides preliminary support for the hypothesized association between relative increases in rCBF following light treatment and recovery from depression. There were no significant differences in pretreatment rCBF between responders and nonresponders. The percentage change in rCBF observed in this study may be compared with a 7% increase in rCBF observed in visual cortex following full field visual activation. Random rCBF change between studies has been found to be in the 2–3% range (Hill et al 1993b). It is important to underline specific methodologic issues in this study: 1) At study entry, no significant differences were found between responders and nonresponders in regards to antidepressant medication usage. Specifically, of the 2 responders previously treated with antidepressant medication, 1 had been off for 7 months, and the other was maintained on fluoxetine throughout the study. (A separate analysis of our data, excluding the patient who was determined to have been on fluoxetine throughout the study, did not affect the study findings.) Similarly, of the 3 nonresponders who had been treated with antidepressant medication, 2 had been off medication for more than 1 month, and the other had been off medication for 1 week. 2) We choose the criterion of at least a 60% reduction in SIGH-SAD score. This implied that a patient with a SIGH-SAD score of 20 would have to attain a posttreat-

1004

BIOL PSYCHIATRY 1997;42:1000 –1005

ment SIGH-SAD score of 8 to be considered a responder. 3) The duration and intensity of light treatment administered to each patient was not exactly identical. This methodological issue did not affect the focus of our findings, as duration and intensity of light did not correlate with response or nonresponse in the patients studied. 4) Although subjects were studied in a dimly lighted, quiet low-stimulation environment and asked to lie still and silent, our study could not control for a possible reduction in anxiety associated in subjects who may have felt more familiar with the scanning environment on the second scan. Consistent with our findings regarding changes in rCBF in cingulate and thalamus following successful light treatment in SAD, Goodwin et al (1993), utilizing comparable high-resolution SPECT methodology, reported an increase in rCBF in cingulate and thalamus regions in non-SAD depressed patients following successful antidepressant medication treatment. One study utilizing the planar xenon SPECT method reported a significant increase in global CBF in 4 SAD patients as compared to 4 normal controls following exposure to 1500 lux for 2 hours, but clinical response to light was not assessed (Murphy et al 1993). A PET study (Cohen et al 1992) of patients with SAD reported that occipital cortex glucose utilization was significantly increased in patients who responded to light treatment. In the present study, occipital rCBF increased in treatment responders, although the difference did not achieve statistical significance. Limitations of this study, in addition to its small sample size, include the following: 1) This study does not include a control group of SAD subjects who had two SPECT scans and no light treatment. Further, the absence of a healthy control group makes it impossible for us to comment on specific brain features of SAD. 2) Although our data enable us to comment on the relationship of clinical improvement in depressive symptoms to brain

R.G. Vasile et al

rCBF changes, the absence of a control treatment group prevents us from definitely attributing this association to a response to light. 3) The SPECT methodology utilized in this study provides information about relative activity in one brain region compared to another, not absolute quantitative data. Without elaborate arterial sampling and modeling, absolute rCBF data cannot be obtained with SPECT utilizing hexamethylpropyleneamine oxime (HMPAO). We are obliged, therefore, to use ratios to account for variation in the active component of each HMPAO kit. 4) Since all patients were studied off light treatment first and on light treatment second, an ordering effect cannot be excluded. 5) As the data presented is normalized to cerebellum, an underlying assumption in this report is that there are no specific cerebellar changes with light treatment that might account for our findings. It is theoretically possible that light treatment reduced cerebellar activity (the denominator for the normalization) in treatment responders. To this point there have been no published reports of changes in cerebellar rCBF associated with response to antidepressant treatment, although several reports have suggested such changes in frontal, temporal, and limbic brain regions (Rubin et al 1994). In conclusion, the hypothesized association between resolution of depression in SAD patients and increased rCBF was supported in this pilot study. These preliminary findings require confirmation in a larger sample of SAD patients. The pathophysiologic significance of the changes in rCBF in SAD patients following light treatment remains to be explored.

Support for this study was provided by the A. John Erdman, III Fund, Massachusetts General Hospital. We wish to thank Dr. Ron Bosch of the Harvard School of Public Health for his consultation regarding statistical methodology.

References Cohen RM, Gross M, Nordahl TE, Semple WE, Oren DA, Rosenthal N (1992): Preliminary data on the metabolic brain pattern of patients with seasonal affective disorder. Arch Gen Psychiatry 49:545–552. George MS, Ketter TA, Post RM (1993): SPECT and PET imaging in mood disorders. J Clin Psychiatry 54:6 –13. Goodwin GM, Austin MP, Dougall N, Ross M, Murray C, O’Carroll RE, et al (1993): State changes in brain activity shown by the uptake of 99mTc-exametazime with single photon emission tomography in major depression before and after treatment. J Affect Disord 29:243–253. Hill TC, Stoddard H, Matthew E (1993a): Comparisons of manual and automated coregistration of SPECT brain images

without the use of external fiducial markers. Radiology 189:392. Hill TC, Matthew E, Martin D, Biegel JD (1993b): Cortical activation studies with Tc-99m ethyl cysteinate dimer. Radiology 189:115. Holman BL, Johnson KA, Gerada B, Carvalho PA, Satlin A (1992): The scintigraphic appearance of Alzheimer’s disease: A prospective study using technetium 99m-HMPAO SPECT. J Nucl Med 33:181–185. Lesser IM, Mena I, Boone KB, Miller BL, Mehringer CM, Wohl M (1994): Reduction of cerebral blood flow in older depressed patients. Arch Gen Psychiatry 51:677– 686. Mayberg HS, Lewis PJ, Regenold W, Wagner HN Jr (1994):

Regional Cerebral Blood Flow in SAD

Paralimbic hypoperfusion in unipolar depression. J Nucl Med 35:929 –934. Mills CM, de Groot J, Posin JP (1988): Magnetic Resonance Imaging: Atlas of the Head, Neck and Spine. Philadelphia: Lea and Febiger. Murphy DGM, Murphy DM, Abbas M, Palzidou E, Binnie C, Arendt J, et al (1993): Seasonal affective disorder: Response to light as measured by electroencephalogram, melatonin suppression, and cerebral blood flow. Br J Psychiatry 163: 327–331. Rubin E, Sackeim HA, Nobler MS, Moeller JR (1994): Brain

BIOL PSYCHIATRY 1997;42:1000 –1005

1005

imaging studies of antidepressant treatments. Psychiatr Ann 24:653– 658. Sackeim HA, Prohovnik I (1993): Brain imaging studies of depressive disorders. In: Mann JJ, Kupfer DJ, editors. Biology of Depressive Disorders, Part A: A Systems Perspective. New York and London: Plenum Press, pp 205–258. Stoddart HA, Stoddart HF (1992): New multidimensional reconstructions for the 12-detector scanned focal point, single photon tomograph. Phys Med Biol 37:579 –586. Williams J (1988): A structured interview guide for the Hamilton Depression Rating Scale. Arch Gen Psychiatry 45:742–747.