Deficits in attention were apparent in all depressed subjects, but significantly greater in past suicide attempters and primarily attributable to their greater susceptibility to interference on the Stroop. In our previous study with a much smaller sample (Keilp et al., 2001
), Stroop performance was poor in past suicide attempters, but only distinguished them from non-patients. Increased statistical power in this study enabled us to demonstrate that this impairment in attention is worse in those with a past history of more severe suicidal behavior, relative to other depressed patients. Moreover, deficits were evident without any type of explicit emotional provocation or emotionally biased stimuli, suggesting that cognitive control mechanisms themselves are dysfunctional, and not simply susceptible to specific types of emotional arousal. Stroop performance did not clearly distinguish all attempters from non-attempters, but the general trend of this difference was consistent with an association with suicide attempt.
Stroop performance is typically impaired in depression, most likely because it relies on many of the same brain areas affected by depression (Mayberg, 2003
). The circuitry of the emotion regulation system, involving the amygdala, anterior cingulate, and prefrontal cortex (Ochsner and Gross, 2005
), overlaps considerably with the selective attention system and is affected by genetic risk for depressive disorders (Pezawas et al., 2005
). Wagner and colleagues (2006)
found that overactivation of rostral cingulate and dorsolateral prefrontal cortex during Stroop performance distinguished depressed subjects from healthy volunteers. However, metabolic deficits in prefrontal cortex and cingulate are accentuated in high lethality suicide attempters (Oquendo et al., 2003
). Posner and colleagues (2002)
suggest that deficits on attentional conflict tasks may be characteristic of conditions where emotional dysregulation and suicide risk is prominent, such as Borderline Personality Disorder. These deficits are apparent on Stroop tasks (LeGris and van Reekum, 2006
). Our data are consistent with this hypothesis, but suggest these deficits are most pronounced among those with a past history of suicidal behavior - even among those without BPD. Poor performance on conflict tasks, then, may be more closely associated with suicide risk than with personality disorder per se.
Deficits related to past history of suicidal behavior were also comparable in unipolar and bipolar depression, and were not affected by bipolar disorder subtype. In a previous report of a preliminary analysis of past suicide attempters with Bipolar I and Bipolar II disorder from a separate cohort (Harkavy-Friedman et al., 2006
), we had found differences in Stroop performance between these subtypes, with Bipolar II subjects worse. However, the preliminary nature of that analysis, differences in enrollment criteria (intercurrent substance abuse and medication both allowed in that study) and the greater representation of higher lethality subjects in this study eliminated these differences here.
Neither past suicide attempters nor depressed subjects in general were significantly impaired on the CPT used in this study. This is consistent with one other study that used the 4-digits fast condition of the CPT-IP in a depressed sample, although, in that same study, deficits were found on a similar task using spatial stimuli (Cornblatt et al., 1989
). The sensitivity of the CPT task used in this study is well established in psychotic and pure attention disorders, both in the published literature (Cornblatt and Keilp, 1994
) as well as in our own laboratory. Clinically stable schizophrenic subjects (n=106) administered the same task in our lab performed significantly more poorly than any group in this study (z=−1.1 ± 1.0; raw d’=1.28 ± 0.83; Keilp, unpublished data). However, our review of the literature suggests that deficits on CPT tasks are most likely to be found in depression when tasks use spatial stimuli (Cornblatt et al., 1989
; Liu et al., 2002
) or run longer than 7 minutes (Cohen et al., 2001
; Koetsier et al., 2002
; Egeland et al., 2003
), over twice the length of the task used here. Deficits in depressed patients, then, only tend to emerge when stimuli are more difficult to verbalize or in later blocks of trials; whereas deficits in schizophrenic subjects are often evident in the earliest blocks of trials and consistent over all blocks. This coincides with theories linking cognitive impairments in depression to failures of right-sided cortical and subcortical arousal mechanisms (Liotti and Mayberg, 2001
) rather than in the more basic information processing capabilities that are impaired in psychosis. Depressed suicide attempters clearly performed no differently than other depressed subjects, and very much unlike the subjects with psychosis in our other studies.
It is possible that subtle brain injuries sustained by higher lethality attempters may have contributed to poorer Stroop performance in this study. Impairments of selective attention in depression have been linked to white matter abnormalities (Videbech et al, 2004
), but appear related to global slowing. The lack of impairment in CPT performance, and in reaction times to Word and Color conditions of the Stroop indicate that is not the case here. Similarly, impairments of Stroop performance in depression have been linked to intercurrent psychosis (Nelson et al., 1998
; Politis et al., 2004
) and psychomotor retardation (Benoit et al., 1992
; Lemelin and Baruch, 1998
; Narita et al., 2004
), but neither type of symptom was more pronounced in our past attempter samples nor correlated with performance.
Overall, data suggest that subjects with a past history of suicide attempt are more severely impaired in aspects of attention affected by depression, despite no differences in observer-rated symptom severity (i.e. Hamilton). Risk for suicidal behavior, therefore, may be associated with more extensive dysfunction in cortical control mechanisms despite no differences in overt depressive symptomatology. Less obvious, subjective aspects of depression, however, may be related to these impairments. Psychological aspects of depression are related to elevated glucose metabolism in dorsal cingulate (Milak et al., 2005
), a region commonly activated by Stroop tasks (Posner & DiGirolomo, 1998
; Cohen et al., 2004
; Wagner et al., 2006
At a neuropsychological level, data support Posner and colleagues’ hypothesis (2002)
that defects in emotion regulation networks, which are most severe when risk for suicidal behavior is high, will be manifest in persistent impairments of attention control. At a phenomenological level, impairments on conflict tasks such as the Stroop suggests a difficulty shifting attention from compelling but inappropriate stimuli. These impairments may predispose suicidal individuals to attend to prepotent emotional states, such as intense feelings of pessimism (Oquendo et al., 2004
) or self-blame (Grunebaum et al., 2005
), and to formulate actions based on a narrow view of their condition. Impaired selective attention performance, then, may underlie the “cognitive rigidity” that is a common clinical feature of suicide attempters (Pollock & Williams, 1998
). Attention control is a prominent feature of current psychotherapeutic treatments for suicidality, such as Dialectical Behavior Therapy (Linehan et al., 2006
) and Mindfulness-based Cognitive Therapy (Williams et al., 2006
), which train patients to redirect attention at moments of crisis.
Previous research suggests that decision-making is impaired in suicidal individuals (Jollant et al., 2005
), but decision processes may be corrupted by a failure to adequately focus attention and control the information on which they are based. At least one study of subjects with Human Immunodeficiency Virus found that poor decision-making task performance (Iowa Gambling Task) was associated with impairments of selective attention and memory (Hardy et al., 2006
). Further work is needed to understand the relationship between these measures - particularly how the narrowing of the focus of attention affects decision processes leading to suicidal acts - and to examine the neural circuitry common to selective attention processes and suicidal behavior.