Relation of Lesion Location to Verbal and Nonverbal Mood Measures in Stroke Patients
Background and Purpose The aim of the present study was to evaluate the relation between poststroke depression and lesion location, avoiding previous methodological shortcomings. In particular, we intended to determine whether patients with left frontal lesions showed the highest depression scores.
Methods Patients in the study, categorized on the basis of lesion location, included 149 stroke patients with lesions located in the anterior, central, or posterior regions of the right or left hemisphere. Verbal and nonverbal mood measures as well as the Hamilton Depression Scale Overall Score were the dependent measures of our investigation. Furthermore, the number of patients who could not be assessed or could be evaluated only with the nonverbal mood measure due to the presence of severe language disorders was recorded.
Results No significant relation was observed between depressed mood and lesion location. Approximately one quarter of the left brain–damaged patients were partially or totally excluded from the study because of severe language disorders.
Conclusions Our data appeared to show that when methodological pitfalls and selection bias are carefully controlled, left frontal lesions are not a major determinant of poststroke depression.
Depression, a very common condition after stroke, occurs in about the 30% to 50% of stroke patients.1 2 It can have a negative influence on recovery3 4 but can also be effectively treated with antidepressants.5 6
In a large and influential series of papers, Robinson and coworkers (Starkstein and Robinson,1 10 Lipsey et al,5 Robinson et al,7 8 and Starkstein et al9 ) have consistently claimed that (1) there are two forms (major and minor) of poststroke depression (PSD); (2) the major form is undistinguishable, from the clinical and neurochemical point of view, from the “endogenous” form of major depression, whereas the minor form can be considered as a form of reactive, “dysthymic” depression; and (3) the major form of PSD is caused by lesions encroaching on the left frontal lobe7 8 or the underlying basal ganglia,9 whereas the minor form has no similar specific anatomical correlate.
Two main methodological objections have been directed at the work of Robinson and colleagues. The first consists in the fact that the distinction between major and minor forms of PSD was substantially based in their studies on DSM-III-R11 diagnostic criteria, which are not necessarily valid in stroke patients. The second objection stems from the fact that because the clinical interview on which the diagnosis of PSD was based consisted mainly of a verbal interaction between examiner and patient, subjects with severe language disorders were necessarily excluded from the study. To circumvent these methodological objections, Stern and Bachman12 and Gainotti et al,13 14 respectively, have recently developed two new diagnostic instruments, the Visual Analogue Dysphoria Scale (VADS) and the Post-Stroke Depression Rating Scale (PSDRS): The former involves a 100-mm vertical line with a simple schematic “happy face” at the top pole and a “sad face” at the bottom. Patients are instructed, with either words or gestures, to place a mark on the line at the point that represents their degree of sadness. The scale is highly correlated with other more extensive and verbally demanding measures of mood state, and it could in principle be administered to even severely aphasic patients.
The PSDRS is a scale specifically constructed by having in mind symptoms usually observed in depressed stroke patients and problems typically met by these subjects. It comprises 10 sections, each of which aims to evaluate a specific aspect of the emotional, affective, and vegetative disorders of stroke patients. The first results obtained with the PSDRS and the VADS have been at variance with the model of PSD proposed by Robinson and coworkers because they have shown that the major form of PSD (defined on the basis of DSM-III-R diagnostic criteria) is phenomenologically much more similar to the minor form of PSD than to the major form of functional depression14 ; the motivated (reactive) aspects of depression are in the foreground both in the major and minor forms of PSD, whereas the unmotivated (biological) aspects of depression prevail in patients with major forms of functional depression14 ; and the relation between depressed mood and lesion location does not correspond to the schema proposed by Robinson and coworkers.8
The present study was devised to again take into account the problem of the relations between depressed mood and lesion location, trying in the process to avoid some of the previously mentioned methodological pitfalls. To be sure, the following precautions were taken: (1) patients were categorized on the basis of lesion location and not the presence of a “major” or “minor” form of PSD, because results of our previous study14 had shown that the distinction between major and minor depression may be misleading in stroke patients. (2) The evaluation of PSD was made using different verbal and nonverbal measures of depressed mood. We reasoned that if left frontal lesions play a critical role in PSD, patients with lesions located in the anterior regions of the left hemisphere should have the highest depression scores on these indexes. (3) The categorization of lesion location was made following a method similar to that proposed by Robinson et al.7 8 (4) Only patients who met strict inclusion criteria (and in particular those who had no history of previous depression) were enrolled in the study. (5) To control the bias introduced in studies on PSD by the verbal requests of a clinical interview, the number of RBDP and LBDP who could not be assessed or could be assessed only by means of the VADS, due to the presence of severe language disorders, was carefully recorded.
Subjects and Methods
Subjects were selected from among inpatients consecutively admitted from June 1994 to June 1996 at either the Neurology Clinic of the Catholic University of Rome or the Rehabilitation Center “Clinica Santa Lucia” in Rome. Before inclusion in the study and after explanation of the procedures, all subjects or their representatives gave informed consent. All subjects were right-handed persons who had suffered a unilateral cerebrovascular accident between 2 weeks and 6 months before their inclusion in the study and who met the following criteria: (1) a single monohemispheric stroke, documented by CT scan or MRI data; (2) age ranging between 35 and 75 years; (3) no history of previous stroke, previous depressive disorders, or more generally, previous important psychiatric disturbances. The 149 stroke patients who fulfilled all these inclusion criteria were assessed for evaluation of mood disorders.
Assessment of Mood Disorders
Evaluation of mood disorders was part of a larger psychiatric assessment conducted by two examiners who were blinded with respects to the results of the CT and MRI data. The examination was always conducted in the late morning to minimize any possible effects of diurnal mood variations. The psychiatric assessment included a structured psychiatric interview, devised to categorize patients as affected by a “major depression,” a “minor depression,” or “no depression,” according to DSM-III-R diagnostic criteria; the VADS; the PSDRS; and the HDRS.15
The following measures of depressed mood drawn from the more general psychiatric assessment were considered in the present study: (1) scores obtained on the VADS (expressed in number of millimeters from the “happy” pole); (2) scores obtained on the “Depressed Mood” item of the HDRS, ranging from 0 (no depression) to 4 (extremely sad mood); and (3) scores obtained on the “Depressed Mood” section of the PSDRS, ranging between 0 (well-balanced mood) and 5 (gloomy, black mood).
The focus of our attention was on mood measures because a sad mood is the key symptom of depression in general, mood is the only component of depression taken into account by the VADS, and other symptoms listed by the DSM-III-R as indicative of major depression can be due to the brain lesion per se in a stroke patient.14 However, to be certain that inferences drawn from the assessment of mood could be generalized to depression in general, the Overall Score (ranging from 0 to 40) obtained on the HDRS was also taken into account in our study.
Evaluation of Stroke Location
To check the relations between depressed mood and anatomical locus of lesion, one of us (C.M.), blinded with respect to the psychiatric assessment, evaluated all patient MRI or CT scan data, using all sections showing the lesion to obtain estimates of its size and location. The first step of this analysis consisted of tracing lesions into schematic brain templates, following the guidelines of Damasio and Damasio.16 These diagrams were then used to obtain a categorical evaluation of lesion location with a method very similar to that proposed by Robinson et al.8 According to this method, on any scan cut two lines are traced perpendicular to the major anteroposterior axis. One is set at a distance of 40% from the frontal pole and the other at a distance corresponding to the 60% of this axis. A lesion is considered anterior if its center falls on any scan cut rostrally to the anterior line, posterior if its center consistently falls caudally to the posterior line, and central if its center lies between the 40% and the 60% lines on any cut or if the lesion is very large with borders crossing both lines.
Patients Who Fulfilled the Inclusion Criteria but Could Not be Reliably Assessed Because of Severe Language Disorders
One aim of our study was to obtain information concerning the bias introduced in research on poststroke depression by the presence of severe language disorders and the possibility of overcoming this bias by means of nonverbal tests, such as the VADS. The number of patients who, although fulfilling the inclusion criteria, could not be assessed at all or could be only partially assessed was therefore carefully recorded. Within the group of 81 LBDP who fulfilled the inclusion criteria, 14 (17%) showed a language disorder so severe as to preclude not only the verbal interview or administration of the (verbal) depression rating scales but also the administration of the VADS. Nine additional aphasic patients (11% of the initial sample of LBDP) could perform the VADS but not the clinical interview or the depression rating scales. On the contrary, within the group of 68 RBDP who fulfilled the same criteria, none was discarded or was only partially assessed because of language disorders.
Correlations Between Verbal and Nonverbal Mood Measures Scores and Overall Depression Score on the HDRS
Because measures obtained in our study were partly verbal and partly non-verbal, dealing mainly with mood disorders but also considering the global depression score of the HDRS, we thought it useful to check the correlations existing among these different measures. Furthermore, since it has been suggested that RBDP could verbally deny their depressive symptoms,17 this set of correlations was studied both in the whole sample of stroke patients and separately in patients with right- and left-brain damage. Table 1⇓ shows the correlation matrixes between these various measures computed in the whole patient population and separately by hemisphere.
Data reported in Table 1⇑ indicate the following: All the mood measures are strongly reciprocally correlated, irrespective of their verbal or nonverbal nature; however, the correlations between purely verbal measures are stronger than those between verbal and nonverbal measures. All the mood measures are strongly correlated with the overall depression score of the HDRS, suggesting that the severity of the depressed mood faithfully reflects the severity (although not the clinical form) of PSD. The strength of correlations obtained with verbal and nonverbal mood measures in RBDP and LBDP is quite comparable, which suggests that the communication of a depressive state is not dependent on the modality (verbal versus nonverbal) used to communicate it in RBDP or LBDP.
Scores on Various Mood Measures by Right- and Left-Stroke Patients With Lesions in Anterior, Central, and Posterior Brain Areas
Table 2⇓ shows results obtained on the various mood measures and on the overall depression score of the HDRS by RBDP and LBDP who could complete the psychiatric assessment and whose lesions encroached on the anterior, central, and posterior parts of the brain.
To check the significance of differences observed among the various groups (or the more global differences concerning the side or the intrahemispheric locus of lesion), each set of measures was analyzed with a 2 (hemispheric: right or left) × 3 (Intrahemispheric: anterior, central, or posterior) ANOVA. In none of the mood or depression measures taken into account in our research was a significant relation found between mood disorders and lesion location. To be sure, the highest mood and depression scores were consistently shown by patients with lesions located in the right hemisphere and the anterior parts of the brain, but neither the main factors nor the interaction reached the level of statistical significance. The situation did not change when, in a separate analysis of data obtained on the VADS, we also included in the study scores obtained by the 9 LBDP who could perform the VADS but not the verbal scales (as shown in the lower part of Table 2⇑). From the clinical point of view, these patients usually showed extensive brain lesions (5 of 9 had central lesions whose borders crossed both the anterior and the posterior cutting lines), important language disorders, and signs of severe depression. In fact, they obtained a VADS mean score of 66, which is higher than that obtained by any other group of brain-damaged patients.
Patients With Lesions in Different Brain Regions Whose Score Suggested Major Depression on the HDRS
Results of our study had indirectly falsified the Robinson model, showing that severity of depression is not higher in patients with a left frontal lesion than in those with damage located in other brain regions. To more directly compare the results of our investigation with those of studies conducted by Robinson and colleagues, we operationally defined as affected by major depression those patients who had obtained a score higher than 18 on the HDRS. This criterion was chosen because it is considered valid by other authors (for review, see Reference 1818 ) and because it allowed us to avoid the distinction between the major and minor form of PSD, based on DSM-III-R diagnostic criteria, that we have considered misleading in stroke patients.14
Table 3⇓ reports the number of right- and left-stroke patients with lesions located in the anterior, central, and posterior parts of the brain who scored above or below the cutoff point of 18 on the HDRS. These data clearly confirm that no relation exists between major depression and lesion location in stroke patients.
The scope of the present investigation consisted of checking, in strictly controlled conditions, the claim repeatedly made by Robinson and coworkers1 5 7 8 9 10 that the major forms of PSD are due to lesions encroaching on the anterior areas of the left hemisphere. In addition to the adoption of strict inclusion criteria (such as the exclusion from the study of patients with a history of previous important psychiatric disorders), we tried to avoid two main methodological shortcomings met by previous studies in this area. The first methodological problem lies in the fact that the distinction between major and minor depression may be improper in stroke patients, because some symptoms considered by the DSM-III-R as diagnostic of major depression can be a result of the brain lesion per se in these patients. The second methodological pitfall stems from the fact that subjects with severe language disorders were presumably excluded from previous studies in which the diagnosis of PSD was based on a clinical interview or on the use of verbal depression rating scales.
To overcome the first methodological problem, we avoided categorizing depressed patients as having major or minor forms of PSD and instead simply took measures of sad mood and the Global Depression score of the HDRS15 as measures of the severity of PSD. This methodological option was also justified by the fact that in a previous study14 we had shown that a continuum exists between the so-called major and minor forms of PSD.
To overcome the second methodological problem, we adopted two complementary strategies: we used a recently developed nonverbal measure of mood disorders, namely, the VADS;12 and we carefully recorded the number of stroke patients who, although fulfilling the inclusion criteria of our study, could not be assessed at all or could be evaluated only by means of the VADS.
Results of our study can be summarized as follows: (1) in studies of PSD based on clinical interviews or depression rating scales, a bias between RBDP and LBDP usually exists. The bias results from the fact that about one quarter of the LBDP (namely, those affected by severe language disorders) cannot be properly evaluated, whereas all or almost all of the RBDP can be included in the study. (2) The bias can be only partially counterbalanced using simple nonverbal procedures such as the VADS, because the majority of severely aphasic patients are unable to understand not only the verbal questions but also the nonverbal requests of the examiner. (3) No significant relation can be found between depressed mood and lesion location when the influence of these main methodological shortcomings is carefully controlled.
As a matter of fact, not only did we fail to observe a significant relation between depressed mood and left frontal lesions, but even the nonsignificant trends observed in other directions in our study seemed due to the above-mentioned sampling bias. That depression measures were consistently higher in patients with lesions located in the right hemisphere and the anterior parts of the brain could certainly be due to the exclusion of patients with severe language comprehension disorders. In fact, most of the patients who could not be properly evaluated showed either a global aphasia, due to a large infact in the whole left sylvian territory, or a severe Wernicke’s aphasia, with lesions located in the posterior part of the left temporal lobe. If we consider that the 9 LBDP who could perform only the VADS had on this scale a mean score higher than that of any other brain-damaged group, it seems reasonable to infer that the exclusion of patients with severe language disorders has unduly decreased the depression scores of LBDP with central and posterior lesions. This hypothesis must certainly be considered with caution, because the VADS results suggest more depression in the excluded patients but not enough to change the overall results. However, this hypothesis of a selection bias could also account for the results of previous investigations4 19 20 in which the highest depression scores had been obtained by patients with right-hemisphere lesions. On the other hand, the pattern of results observed in RBDP, in which patients with anterior lesions consistently tended to obtain the highest depression scores, cannot be attributed to a selection bias. This observation is of interest because a similar pattern has been obtained in previous studies4 20 21 in which anterior lesions tended to be associated with the highest depression scores, irrespectively of the damaged hemisphere. The relation between PSD and frontal lobe injury therefore remains unchallenged by the present results and seems worthy of further investigation.
Selected Abbreviations and Acronyms
|HDRS||=||Hamilton Depression Rating Scale|
|PSDRS||=||Post-Stroke Depression Rating Scale|
|VADS||=||Visual Analogue Dysphoria Scale|
- Received April 24, 1997.
- Accepted July 25, 1997.
- Copyright © 1997 by American Heart Association
Starkstein SE, Robinson RG. Affective disorders and cerebral vascular disease. Br J Psychiatry.. 1989;154:170-182.
Gainotti G. Poststroke depression: psychological and biochemical interpretations. In: von Steinbuchel N, von Cramon DY, Poppel E, eds. Neuropsychological Rehabilitation. Heidelberg, Germany: Springer: 1992:215-225.
Synyor D, Jacques P, Kaloupek DG, Becker R, Goldenberg M, Coopersmith H. Poststroke depression and lesion location. Brain.. 1986;109:537-546.
Andersen G, Vestergaard K, Lauritzten L. Effective treatment of poststroke depression with the selective serotonin reuptake inhibitor citalopram. Stroke.. 1994;25:1099-1104.
Robinson RG, Kubos KL, Rao K, Price TR. Mood disorders in stroke patients: importance of location of lesion. Brain.. 1984;107:81-93.
Starkstein SE, Robinson RG, Price TR. The comparison of cortical and subcortical lesions in the production of post-stroke mood disorders. Brain.. 1987;110:1045-1059.
Starkstein SE, Robinson RG. Aphasia and depression. Aphasiology.. 1988;2:1-20.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders.. (DSM-III-R), 3rd ed. Washington, DC: American Psychiatric Press Inc; 1987.
Damasio H, Damasio AR. Lesion Analysis in Neuropsychology. New York, NY: Oxford University Press; 1989.
Salzman C, Scnheider LS, Alexopoulos ES. Pharmacological treatment of depression in late life. In: Bloom FE, Kupfer DJ, eds. Psychopharmacology: 4th Generation of Progress. New York, NY: Raven Press Publishers; 1994:1471-1477.
Lim ML, Ebrahim SBJ. Depression after stroke: a hospital treatment survey. Postgrad Med.. 1983;59:489-491.
House A, Dennis M, Warlow C, Hawton K, Malyneux A. Mood disorders after stroke and their relation to lesion location: a CT scan study. Brain.. 1990;113:1113-1190.