Research into many brain diseases, including stroke,1 Alzheimer's disease,2 Parkinson's disease,3 epilepsy,4 Huntington's disease,5 and AIDS,6 has shown significant associations between these conditions and the presence of depressive disturbances. These associations are important for a variety of reasons. First, the occurrence of depression in neurologic disease is a natural experiment whose study continues to shed light on our understanding of the role of the brain in depressive illnesses. Second, the co-occurrence of depression and neurologic disease magnifies morbidity. Third, treatment of depression in neurologic disease may greatly improve prognosis.
Perhaps the most interesting issue regarding the association between depression and neurologic disease is whether depression is caused by neurologic disease. Two general types of causal links should be distinguished. In the first, depression arises as a psychological reaction to the impairments or social disruption produced by the neurologic disease, in the same way that depression might arise in any individual faced with the adversities produced by a serious disease. In the second, depression is a specific symptom of the neurologic disease and is intimately tied to the pathophysiology of the disease. The relationship between stroke and depression has received sufficient investigation to make it a suitable model in discussing such potential causal relationships.
In this article we review research on depression and stroke to illustrate how to approach causal links between neurologic disease and mental syndromes. Our major conclusion is that stroke lesions, under certain circumstances, cause depression through a direct but unknown pathophysiologic process.
Dr. Allan House recently has argued7 that there are few data to support the hypothesis that depression is a "specific complication of stroke." Although he did not overtly acknowledge his assumptions, he began with the premise that depression is a natural psychological reaction to the adversity of stroke.
Dr. House contested the existence of "a specific syndrome of poststroke depression," which has been proposed by Robinson and others.1 His argument against the poststroke depression hypothesis is based on the refutation of certain predictions that he believes flow naturally from the hypothesis.
Specifically, Dr. House has postulated7 that if there were such a specific complication of stroke as "poststroke depression," then three things would be expected. First, the clinical picture of depression after stroke (such as its constituent symptoms or frequency) would be different from depression in the absence of stroke or after other general medical conditions. Second, depression would have a different etiology after stroke. Third, depression associated with stroke would have "different characteristics in response to treatment." After reviewing the empirical evidence in these three areas, Dr. House concluded that none of these predictions has been consistently supported by the existing literature. He implied that we cannot, as yet, prove that depression is a specific complication of stroke.
We disagree with Dr. House's postulates and feel that he has finessed the central issue in this debate. The critical question regarding the association of depression and stroke is not whether poststroke depression fulfills Dr. House's postulates, but rather, "What is the cause of depression after stroke?" Dr. House's criteria to assess causality are inadequate to make a judgment about this issue.
For some patients, depression, anxiety, and other symptoms commonly follow stroke and are indicative of the individual coming to terms with an adverse event. These natural human emotions do not imply a particular pathophysiology of the stroke or the depression. In the sense that they resemble symptoms seen after any adverse event, they are not specific to stroke. Their cause is not the stroke per se, but rather the characteristics of the stroke that make it an adverse event to that particular individual, at that particular time, in his or her particular life.
In other patients, a different kind of depression emerges after stroke. This is major depression, a well-defined disturbance, which most psychiatrists operationalize as a syndromic category8 based on disjunctive criteria (5 of 9 possible symptoms). These criteria characterize severe depression due to genetic causes (probably multiple types), environmental stressors, personal vulnerability, or symptomatic response to physical factors ranging from reserpine medication to structural lesions. It is the syndrome of major depression after stroke that we propose might at times be caused directly by the pathophysiology of stroke. Dr. House suggested that clinical presentation, pathophysiological mechanisms, and response to treatment would all be unique if major depression was a specific consequence of stroke. However, the syndrome of major depression can probably be caused by a range of pathologies that affect the brain—as is suggested by work in other diseases2—6—as well as by genetic predispositions. One would not a priori predict that poststroke major depression would manifest itself in a unique clinical presentation, any more than one would presume that every form of pneumonia could be distinguished on clinical examination alone. In fact, it was shown by Lipsey et al.9 that poststroke major depression has a very similar clinical presentation to major depression in elderly patients without structural brain lesions.
Dr. House's criterion of unique response to treatment is similarly inappropriate. For instance, cases of congestive heart failure arising from different causes may respond as well to the same diuretic, just as antihypertensive agents are often effective across a wide etiologic spectrum. Antidepressant medication has been shown in controlled treatment trials to be effective in treating disorders ranging from pure familial depressive disorder to social phobia to obsessive-compulsive disorder. The fact that poststroke depression responds to antidepressant therapy provides no evidence for or against a causal relationship between stroke and depression.
Dr. House's criterion for specificity is somewhat closer to the mark. We agree that if a specific form of poststroke depression exists, it by definition results from some mechanism directly related to the brain injury. However, it is clear that depression following stroke might at times have other (or even multiple) causes.
It is precisely the similarity of major depression after stroke to major depression in other settings that raises doubts about whether major depression after stroke is always simply a human reaction to the adversity brought on by illness. A human reaction to adversity is typically universal and nonspecific. Major depression after stroke has not been shown to respond to nonspecific psychological or psychosocial interventions,7 but it has been shown to respond to specific pharmacologic treatment for depression.10,11 If a patient developed chest pain after stroke, along with joint pain, anxiety, and indigestion, we might think of the chest pain as a "somatization" of his or her distress and attempt to treat it with a psychological intervention. If, however, a patient developed chest pain with radiation to the left hand, accompanied by nausea, we might recognize the syndrome of angina and be concerned that the patient has heart disease. We make a similar argument about depressed mood (or anhedonia) as part of a major depressive syndrome, as opposed to depression as part of nonspecific distress and demoralization.
If Dr. House's criteria assessing causal associations are inadequate, what criteria are more appropriate? Susser12 has persuasively argued that the demonstration of causality rests on three principles linking the cause to the effect: association, temporal order, and direction. The proposed criteria, as applied for neuropsychiatric disorder by Lyketsos and Treisman,13 could be expressed for depression after stroke as follows:
1. Under the appropriate circumstances, depression and stroke occur together the great majority of the time.
What is the evidence pertaining to this criterion? Starkstein et al.14 reported that 7 of 8 patients (88%) with left-sided basal ganglia lesions showed major depression, compared with 1 of 7 patients (14%) with right-sided basal ganglia lesions and 0 of 10 patients with thalamic lesions, either right or left side. The patients were between 1 and 2 weeks post stroke. Among patients with cortical lesions, Starkstein et al.15 found that 3 of 5 patients with left anterior cortical lesions had major depression, compared with 0 of 9 patients with anterior or posterior right hemisphere cortical lesions. These patients were also studied between 1 to 2 weeks after stroke.
Astrom et al.16 found 12 of 14 patients (86%) with left anterior lesions had major depression, compared with 2 of 7 (29%) with left posterior lesions and 2 of 23 (9%) with right hemisphere lesions. These patients were also studied 4 to 5 days after hospital admission. House et al.17 found that 1 of 8 patients with left anterior lesions and 0 of 5 patients with left posterior lesions had major depression, compared with 0 of 7 with right anterior and 3 of 5 with right posterior lesions. These patients were studied 1 month after stroke. Dam et al.,18 who studied patients ranging from <90 to >720 days after stroke, did not report the frequency of major depression with left anterior lesions but found that depression scores were higher in patients with right compared with left hemisphere stroke.
Morris et al.19 reported that 9 of 12 patients with left prefrontal or left basal ganglia lesions had major or minor depression, compared with 1 of 12 (8%) with left hemisphere lesions in other locations and 5 of 17 (29%) with right hemisphere lesions. Patients were examined approximately 8 weeks following stroke. Herrmann et al.20 found that 8 of 14 patients (57%) with left anterior lesions and aphasia had major (n=5) or minor (n=3) depression, compared with 0 of 7 with fluent aphasia and posterior left hemisphere lesions. These patients were all less than 3 months post stroke. Thus, 4 of 6 groups studying independent samples of patients during the first few weeks to a few months after stroke found significant associations of left anterior lesions with depressive disorder. The later the patients were studied after the acute stroke, the less likely investigators were to find this laterality phenomenon.
2. There is no other coherent explanation for the depression, such as other illness, medications, substance use, or a personal or family history of major depression.
No study has demonstrated that major depressions following stroke may be explained by concomitant physical illness, medications, or substance abuse. Morris et al.21 have demonstrated, however, that risk factors that would be expected to influence the expression of major depression—such as recent life events, premorbid personality vulnerability, impaired social function, and family history of psychiatric disorder—all appear to have additive effects on the likelihood of developing depression following stroke. Thus, the available data suggest that poststroke depressions, like primary depressive disorder, may be multifactorial in cause and do not represent a homogeneous population. These factors do not militate against a causal relationship between stroke and depression but only identify that the risk of a stroke's leading to depression increases with greater numbers of vulnerability factors.
3. The onset of depression is within one month of the stroke.
The discussion above pointed out that the association of left anterior brain injury with poststroke depression could not consistently be demonstrated after the first few weeks following stroke. This finding provides support for the hypothesis of a causal relationship between stroke and depression.
4. Resolution of the active phase of stroke is followed by resolution of depression in close temporal proximity.
Because stroke is a disorder that does not have a resolution, it is difficult to be sure how to apply this criterion to poststroke depression. Numerous studies have demonstrated that major depression following stroke resolves within the first year after stroke in the majority of cases.1,17,22,23 One might conclude from these data that major depression following stroke has a natural course, usually less than 1 year. This does not argue either for or against a casual connection between stroke and depression but only supports what is known about major depression in other circumstances: that there is spontaneous remission in most cases.
5. There is specific evidence outside the clinical association to indicate a pathophysiological mechanism through which the stroke may cause depression.
Apart from the association of left anterior lesions with poststroke depression during the acute stroke period, probably the strongest evidence for a causal relationship between depression and stroke is the clinical—pathological correlation between the proximity of the lesion to the frontal pole and severity of depression. With the exception of the neuropathology of depression in Alzheimer's disease, this correlation of severity of depression with proximity of the stroke lesion to the frontal pole is perhaps the most consistently replicated clinical—pathological correlation in psychiatry. Although the strength of the correlation has varied from study to study and some investigators have found correlations only in the left hemisphere while others have found correlations in both the left and right hemispheres, every investigator who has examined this relationship (even House et al.17) found that the severity of depressive symptoms increased as the lesion was closer to the frontal pole if patients were studied within the first 6 months following stroke.20,22,24
There has been no other explanation besides a causal connection of stroke and depression to explain this clinical—pathological correlation. This correlation has not been explained, for example, by a relationship to a third variable such as cognitive impairment, physical impairment, or aphasia. If Dr. House argues that there is no causal relationship between stroke and depression, how does he explain his own data showing that the location of injury within the anterior-to-posterior axis correlates significantly with severity of depressive symptoms as measured by the Present State Examination score at 1 month after stroke and with both the Beck Depression Inventory and the Present State Examination score at 6 months?
Other published data also suggest a basis for a causal relationship between stroke and depression that involves the biologic changes brought on by the stroke lesion. Several studies have demonstrated abnormalities of dexamethasone suppression in patients with poststroke major depression.25—27 Barry and Dinan28 reported that patients with poststroke depression had a blunted growth response to desipramine compared with nondepressed stroke patients or normal control subjects. This finding suggested subsensitivity of the alpha-2 adrenergic receptor as a basis for poststroke depression. Bryer et al.29 found reductions in spinal fluid levels of 5-hydroxyindoleacetic acid in patients with poststroke depression compared with nondepressed control subjects. Mayberg et al.30 demonstrated a significant correlation between the amount of serotonin S2 receptors in the left temporal cortex and severity of depression in patients with stroke. Taken together, these findings of specific and potentially pathophysiological explanations for poststroke depression provide a very persuasive case for stroke and depression being causally related.
We do not argue that depression after stroke is always a "specific complication" of stroke. Rather, there are sufficient data to support the hypothesis that, in some circumstances, stroke causes depression through a pathophysiological process that may some day be identified and specifically treated. Although House argues that poststroke depression is no more than an understandable response to life-threatening physical illness, the cumulative data speak for themselves.
What is needed now is not an argument about whether stroke can cause depression, but further study of how to identify the patients who are most at risk, what mechanism may mediate this depressive disorder, and whether specific targeted treatments can be developed to ameliorate the depression as well as the associated increased mortality and delayed physical recovery that are associated with poststroke depression.31—33
Future research addressing causal links of stroke to other neuropsychiatric disorders (for example, anxiety disorders) should also use operational criteria, such as the ones proposed here, to assess causality between stroke and these other disorders. It would also be important to assess the mechanisms of poststroke major depression in patients without left anterior lesions and whether these overlap with the depression provoked by strategic lesion location.
Research should also examine whether stroke could be used as a model to identify the risk factors and anatomical substrates of depression in other physical disorders, such as traumatic brain injury, multiple sclerosis, epilepsy, Huntington's disease, AIDS, and Parkinson's disease, to mention only a few. Psychiatrists and neuroscientists have been working for many years to identify the structural and physiological basis of primary depressive disorder. Functional imaging studies30,34 have frequently identified left prefrontal cortical abnormalities of metabolic activity in patients with primary depression. Structural imaging studies of depression in elderly patients have identified hyperintensities on MRI scan in subcortical white matter and basal ganglia.35 These findings suggest overlap in the anatomic substrates of primary and poststroke depression. The insights that may be gained from studying patients with symptomatic depressions may shed light on the pathophysiological mechanisms and anatomical substrates of depressive disorder in patients without known neuropathology.