87 Mood and cerebral perfusion revisited Klaus P. Ebmeier studies or symptom provocation with tryptophan de- University Department of Psychiatry, Royal pletion, where deterioration in clinical state can be pre- Edinburgh Hospital, Morningside Park, Edinburgh, dicted [5,20,31]. Prospective single-scan designs have EH10 5HF, UK been used, trying to validate pre-treatment brain activ- ity maps by treatment response or outcome measure- ments [28]. While such longitudinal studies are an im- Twenty patients with major depression and observed diurnal provement on simple cross-sectional association stud- variations of mood were examined using clinical and neu- ies, their main disadvantage is that effects are based ropsychological measures and perfusion HMPAO-SPECT at on comparison between-subjects, which in selected 8 a.m. and 8 p.m. In thirteen patients depression scores varied samples of depressed patients are liable to confound- more than 15% between both measures, although 4 patients ing with other spurious differences of the compared with reverse diurnal variation caused mean group depres- groups. Mood induction experiments can be conducted sion scores to be not different between morning and evening. There was an overall trend for higher depression scores to be in healthy volunteers with due control of possible order associated with higher perfusion in posterior cingulate. This effects. Although the logical step from induced mood was mainly accounted for by significant positive correlations in healthy volunteers to mood changes in affective dis- in the morning scan in posterior, but also anterior cingulate orders may appear tenuous, remarkable similarities in and medial prefrontal cortex compared with evening scans. brain-behaviour relationships have been described [2, This means that morning regression slopes were steeper than 3,8,25,29,36]. Finally, psychological tasks during im- evening slopes. This result is discussed with regard to pos- age acquisition have been employed in depressed pa- sible interpretations, such as adaptive or habituating changes tients and during mood induction. If there is a diag- during the day that may occur in depressed patients. nosis or mood specific activation pattern, it tends to be associated with poor performance of the task [2,15,16]. Consequently, the localisation of change is determined 1. Introduction as much by the psychological task demand as by the mood change. In particular, no localisation of mood Evidence accumulated in affective disorder research related changes could be derived from such studies. over the last five years points towards the limbic frontal In the present study, which has been reported in lobes, including the (anterior) cingulate, the basal gan- less detail previously [13], we exploit the natural di- glia and parts of the temporal lobe (such as the amyg- urnal variations in symptom severity that occur be- dala) as structures underlying symptom formation [24]. tween morning and evening in some depressed patients. This evidence partly derives from cross-sectional as- Such measurable variations in mood are accompanied sociation studies. However, in order to demonstrate by parallel changes in neuropsychologicalperformance the association between symptoms and brain activity and even motor strength, which suggest that pervasive more convincingly, repeat-measures analyses that can changes in brain activity take place within 12 hours [30, link brain activity with symptom changes are neces- 34]. The advantage of this design is that patients are sary. Unfortunately, such studies tend to suffer from their own controls as far as socio-demographic vari- order effects that occur, for example, when patients are ables are concerned. The same is true for medication examined before and after treatment, i.e. first ill and and severity of depression. Finally, the order of exam- then recovered [22]. As a matter of principle, changes ination can be balanced as far as morning or evening in such studies can be due to regression to the mean or first examination is concerned. habituation effects that confound the observed associ- ations between symptoms and brain activity patterns. 2. Methods For economic reasons, balancing studies for order by examining subjects first when well, is impossible apart Twenty unipolar patients with a diagnosis of major from very rare occasions, such as lithium withdrawal depressive episode with melancholia [1] were exam- Behavioural Neurology 12 (2000) 87–92 ISSN 0953-4180 / $8.00 2000, IOS Press. All rights reserved 88 K.P. Ebmeier / Mood and cerebral perfusion revisited ined. There were 8 men and 12 women,with a mean age intervals above this level. Further details of the method of 44 years (SD: 12). Their premorbid IQ, estimated have been described previously [13,14]. with the National Adult Reading Test-Revised [32], Images were processed following this pattern: was 109 (SD: 10). Three patients were left-handed, 1. Manual yaw, roll and pitch correction of im- 17 right-handed as determined with the Annett Hand- ages was followed by an automated least-mean- edness Scale [6]. Illness severity was estimated with square co-registration of image-pairs from the the Hamilton Depression Rating Scale (17-item [23]) same subject in the SME scanning software and was 26 (SD: 6). The mean duration of episode at (Neuro 900 version 2.92). Images were then the time of examination was 44 days (SD: 62). Five edited to ensure they contained the same number patients were medication free at the time of imaging, of slices over equivalent brain volumes. 15 were medicated: 6 on neuroleptics, 2 on endocrine 2. Images were exported as binary files and con- replacement (thyroxin), 1 on lithium carbonate, 14 on verted to Analyze format. antidepressants, 1 on hypnotics and 2 on anxiolytics. 3. The first scan was decay-corrected and sub- The drug regime had been stable in the previous two tracted from the second scan, to produce an ac- weeks. tivity map reflecting the brain state at the time Patients were examined twice within 12 hours, half of the second injection. at 8 a.m. first and 8 p.m. second, the other half in 4. The scan pairs were spatially normalised using reverse order. At both times, the following assessments a 12-point linear affine transformation into Ta- were carried out: lairach space (using SPM 95, cf. [13]). The transformation elements were derived from the (1) The Befindlichkeitsskala (BFS [38]) is an adjec- average of both scans from each subject and ap- tive check-list designed to measure short-term plied to each identically [18]. The images were variations of mood, with two parallel versions then smoothed with a (12 mm FWHM) Gaussian and subscales for depression and fatigue. filter in three dimensions in order to reduce the (2) The Alderley Park State Anxiety Questionnaire error variance associated with individual vari- (APSAQ [39]) is a self-rating scale, which was ability in gyral anatomy and to improve the sig- administered immediately after the injection of nal to noise ratio. tracer. 5. In order to take advantage of the improved treat- (3) The Stress Arousal Inventory (SAI [27]) is an ment of multiple comparisons in SPM96, data adjective checklist with two separate scales for were re-analysed with SPM96 using a blocked stress and arousal. ANCOVA design (one block per subject) with (4) The Digit Symbol Substitution Test (DSST [40]) two conditions (morning/evening). Global brain is a test of psychomotor speed. Two parallel ver- activity and age effects were removed using anal- sions were constructed from the original form. ysis of covariance. The psychometric covariate (5) The Auditory Verbal Learning Test [26,35] is a (BFS) was modelled separately for each condi- verbal memory test using a read 15-item word- tion. Tests were computed for main covariate ef- list, which has to be recalled immediately, and fects, i.e. significant correlations of BFS scores recalled and recognised after 30 min. Two par- with cerebral perfusion across conditions (8 a.m. allel versions were used [7]. and 8 p.m.) and for the interaction between co- (6) Each subject was therefore asked to squeeze a variate and condition, i.e. testing the hypothe- dynamometer as hard as they could; the mea- sis that regression slopes of regional perfusion sures for three attempts were averaged. on BFS scores were different in the morning and evening (not testing for significant correla- Subjects were imaged with the Neuro 900 scanner tions of symptom changes with change in per- (Strichman Medical Equipment Inc., Boston, USA). fusion as erroneously claimed in the 1997 pa- Each subject received 2 × 250 MBq of 99mTc- per [13]). The hypothesis that changes in perfu- exametazime (HM-PAO). During the injection, patients sion were correlated with changes in mood was were comfortably resting on the imaging table with tested by removing between-subject variability eyes closed and covered, and environmental noises kept from BFS scores (by subtracting subjects’ mean to a minimum. Slices were acquired parallel to the scores from morning and evening scores), and orbito-meatal plane starting at a level approximately modelling the interaction between this differ- 2 cm above the orbito-meatal (OM) line and at 1cm ence score and condition (8 a.m./8 p.m.). K.P. Ebmeier / Mood and cerebral perfusion revisited 89 Effects are reported as significant if they are con- The only significant effect after correction for multi- sidered to be so a posteriori using the algorithms of ple comparisons was the interaction of condition with SPM96. These take into account the total volume covariate, in the sense that correlations between BFS tested, the smoothness of the data, the magnitude of and perfusion were significantly greater in the morning effects (expressed as z-values) and the spread of ef- than the evening. This effect was observable in poste- fect, expressed as volume with z above a certain value rior cingulate (−2, −50, 16; Z =3.66; p =0.009, (usually with p<0.01).