Welcome to the Schizophrenia Resource Centre

Welcome, this website is intended for healthcare professionals in EMEA with an interest in the treatment of schizophrenia. By clicking the link below you are declaring and confirming that you are a healthcare professional

You are here

Frontal cortex control dysfunction related to long-term suicide risk in recent-onset schizophrenia

Schizophrenia Research, 1-3, 157, pages 19 - 25



Suicide is highly-prevalent and the most serious outcome in schizophrenia, yet the disturbances in neural system functions that confer suicide risk remain obscure. Circuits operated by the prefrontal cortex (PFC) are altered in psychotic disorders, and various PFC changes are observed in post-mortem studies of completed suicide. We tested whether PFC activity during goal-representation (an important component of cognitive control) relates to long-term suicide risk in recent-onset schizophrenia.


35 patients with recent-onset of DSM-IV-TR-defined schizophrenia (SZ) were evaluated for long-term suicide risk (using the Columbia Suicide Severity Rating Scale) and functional MRI during cognitive control task performance. Group-level regression models associating control-related brain activation with suicide risk controlled for depression, psychosis and impulsivity.


Within this group, past suicidal ideation was associated with lower activation with goal-representation demands in multiple PFC sectors. Among those with past suicidal ideation (n = 18), reported suicidal behavior was associated with lower control-related activation in premotor cortex ipsilateral to the active primary motor cortex.


This study provides unique evidence that suicide risk directly relates to PFC-based circuit dysfunction during goal-representation, in a major mental illness with significant suicide rates. Among those with suicidal ideation, the overt expression in suicidal behavior may stem from impairments in premotor cortex support of action-planning as an expression of control. Further work should address how PFC-based control function changes with risk over time, whether this brain–behavior relationship is specific to schizophrenia, and address its potential utility as a biomarker for interventions to mitigate suicide risk.

Keywords: Schizophrenia, Suicide, Cognitive control, fMRI, Frontal cortex.

1. Introduction

Suicide is a major public health problem worldwide. It is a leading cause of death, among the most common causes of death for young people, including young adults (Nock et al, 2008 and Hawton and van Heeringen, 2009), and confers an enormous public health impact (United States Public Health Service: Office of the Surgeon General, 1999, United States Public Health Service, 2001, and Goldsmith and Institute of Medicine (US), 2002). In schizophrenia, the risk for suicidal ideation, behavior and completed suicide is particularly high early in the illness course (Dutta et al, 2010, Pompili et al, 2011, and Nielssen et al, 2012), though suicide risk remains elevated for many years after a single suicide attempt ( Dutta et al., 2010 ).

Despite the increasing attention to clinical risk factors for suicide, how brain dysfunction confers this risk remains unclear. Post-mortem studies of suicide victims reveal many serotonergic disturbances in the lateral and medial PFC (reviewed in Mann, 2003 ). These findings are generally independent of co-morbid depression history, or psychiatric diagnosis per se. These studies suggest that the lateral and medial PFC are key loci of serotonergic dysfunction associated with suicide.

Nonetheless, it remains unclear how disrupted PFC-based circuit operation contributes to suicide risk in at-risk populations. The major cognitive neuroscience models of PFC function generally posit superordinate control processes, which support cognitive processes as diverse as attention, behavior, decision-making, thought/language, and emotion-regulation. These models include a role for lateral PFC subregions (especially dorsolateral PFC, or DLPFC) in goal-representation via the encoding and use of rules or strategies for decision-making, thereby biasing processing of attention, perception and action, to influence motor output via striato-thalamic circuits (Miller and Cohen, 2001 and Koechlin et al, 2003). In contrast, the posterior medial PFC monitors conflict between goals and tasks ( Botvinick et al., 2001 ) and other mismatches between the individual's status and goals, such as negative affect and pain ( Schackman et al., 2011 ). In these conditions, the medial PFC signals to the DLPFC the need to bolster control to optimize goal attainment.

Other closely-linked frontal cortical regions represent goal-relevant information, including rostral PFC, which represents hierarchical aspects of complex rules and actions ( Badre, 2008 ); and rostral medial PFC sectors (e.g. dorsomedial and ventromedial PFC), which represent self-referential aspects and valuate environmental stimuli. Disturbances in elements of these interacting networks may then manifest clinically as disturbances of the control of thought, behavior or emotion, observed as suicidal ideation or behavior. Considering that cognitive control performance is associated with serotonergic gene variation ( Strobel et al., 2007 ), frontal-based control processes may link serotonergic dysfunction to suicide.

PFC-based cognitive control disturbances may therefore represent an important mechanism underlying suicide risk. Cognitive control-related medial and lateral PFC dysfunction is well-established in schizophrenia (Minzenberg et al, 2009, Lesh et al, 2011, and Lesh et al, 2013), including in the first-episode ( Yoon et al., 2008 ). Schizophrenia outpatients with past suicide attempts also have increased volumes of inferior PFC white matter relative to those without suicide attempts ( Rusch et al., 2008 ), as well as decreased orbitofrontal gray matter density ( Aguilar et al., 2008 ), and relatively higher-risk schizophrenia patients exhibit altered effective connectivity between the left-hemisphere posterior cingulate cortex and medial PFC, relative to lower-risk patients and healthy controls ( Zhang et al., 2013 ). Schizophrenia patients with past suicide attempts also have increased right amygdala volume ( Spoletini et al., 2011 ), which could in principle be associated with altered ascending influence on PFC circuit function. In addition, decreased fractional anisotropy in the cingulum bundle is observed in suicidal traumatic brain-injured patients ( Yurgelun-Todd et al., 2011 ). More diagnostically-heterogeneous populations with past suicide attempts show functional disturbances in medial and lateral PFC sectors during varied cognitive tasks (Audenaert et al, 2002, Amen et al, 2009, Jollant et al, 2010, and Reisch et al, 2010). These studies, while preliminary, suggest that patients with suicidal behavior exhibit dysfunction of PFC-based circuits during complex cognition, and may be impaired over and above those patients who share other clinical features (e.g. diagnosis or other symptoms).

We therefore tested the hypothesis that PFC-based circuit function with explicit control demands directly relates to suicide risk, in schizophrenia patients who are early in the illness course, when this risk is particularly elevated. We employed an emerging clinical standard for suicide risk assessment, the Columbia Suicide Severity Rating Scale ( Posner et al., 2011 ). Critically, in the analyses we accounted for major symptom domains previously identified as clinical risk factors for suicide in schizophrenia, including depression, psychosis and impulsivity. This allowed tests of the direct relationships of frontal circuit dysfunction to suicide risk, which are not simply accounted for by these clinical risk factors. Additionally, in the model of past suicidal behavior to brain dysfunction, we analyzed only those subjects who were positive for past suicidal ideation, allowing us to potentially disambiguate brain function associated with overt behavior from that associated with ideation.

2. Experimental/materials and methods

2.1. Subjects

The study was conducted at the Imaging Research Center at the University of California — Davis Medical Center. All procedures were approved by the UC Davis School of Medicine Institutional Review Board. Inclusion criteria included age 18–50 years, right-handedness (by Edinburgh Handedness Inventory), and diagnosis of 295.X (by DSM-IV-TR). Exclusion criteria included neurological illness (including head injury with loss of consciousness), uncorrectable visual problems or peripheral motor disturbance, full-scale IQ < 80 (by Wechsler Abbreviated Scale of Intelligence), active substance abuse or dependence in the 6 months prior to study, significant uncontrolled medical illness, and previously-known incompatibility with MRI procedures. All included subjects tested negative for illicit drugs in the urine at all study visits. After complete description of the study to the subjects, written informed consent was obtained.

All patients were recruited from the UCD Early Diagnosis and Preventive Treatment (of Psychosis) research clinic, as clinically-stable outpatients, with onset of psychotic symptoms within 2 years of study, and no hospitalizations or changes in medication regimen for at least two months prior to study. The frequencies of prescribed medications at study were antipsychotics (n = 32), anticonvulsants (n = 4) and antidepressants (n = 6). None were receiving lithium or clozapine. Patients were assessed with the Structured Clinical Interview for DSM-IV-TR. Diagnosticians were masters/doctoral-level, SCID-trained clinicians, with demonstrated reliability, defined by ≥ .80 intraclass correlation-coefficient (ICC) for continuous measures and kappa ≥ .70 for categorical measures. All diagnoses were confirmed via consensus conference, and monthly reliability interviews to prevent drift. Based upon 10 sessions during the course of this study, diagnostic reliability for the SCID was kappa ≥ 0.8, and for symptom total scores ICCs were ≥ 0.76. A subset of the present sample was previously reported for AX-CPT task-related fMRI effects compared to a healthy comparison group (not focused on suicide risk), in Lesh et al. (2013) . The present study represents a secondary analysis drawn from a subset of that sample, to specifically test relationships of brain function to suicide risk.

2.2. Clinical measures

2.2.1. Clinical measure of suicide risk — description and rationale

Suicide risk was rated with the Columbia Scale for the Rating of Suicide Severity (C-SSRS), a relatively brief, yet comprehensive, structured interview-based instrument with good validity and internal reliability in 3 multi-site studies with diverse clinical populations ( Posner et al., 2011 ). This scale is comprised of three subscales. Suicidal Ideation (SI) is an ordinal subscale containing items including the wish to be dead, the specificity of these thoughts, including whether they are “active”, with intent and plan. Intensity of Ideation (II) considers the frequency, duration, controllability, deterrents and reasons for thoughts of suicide. Suicidal Behavior (SB) is a nominal subscale that categorizes past actual, interrupted and aborted attempts to die, and preparatory acts, and for actual attempts, the potential or actual lethality or medical damage sustained in the attempt(s). Items on each subscale are associated with future suicidal behavior and/or completed suicide ( Posner et al., 2011 ). SI items are rated on a yes/no basis, II items on a 1–5 scale, and SB as yes/no, with counts of actual, interrupted and aborted attempts.

We used the following criteria to stratify patients in each group on each of the two non-continuous C-SSRS subscales (SI and SB): for SI, the presence of any past suicidal ideation (i.e. positive on any SI item), versus no past suicidal ideation; and for SB, the presence of any past behavior that can be considered the initiation of deliberate self-harm with intent to die as the critical discrete threshold (i.e., a positive response to actual attempt, interrupted attempt, or aborted attempt). For SB, we considered positive responses that were restricted to Non-Suicidal Self-Injurious Behavior and/or Preparatory Acts or Behavior, as not meeting this criterion. The target subgroups in the SI and SB models were then dummy-coded as e.g. “SI-positive” or “SI-negative”. In contrast, the II subscale contains items that are rated in a continuous manner and therefore were summed to establish a subscale total score for neuroimaging analyses.

Both past overt suicidal behavior and lifetime “worst-point” suicidal ideation were targeted on the basis of the following evidence. Overt suicidal and self-injurious behavior represents a proximal risk factor (or prelude to) eventual completed suicide for many individuals, as those who survive a suicide attempt subsequently remain at elevated risk for completion for many years ( Owens et al., 2002 ). However, many who complete suicide do so in a first attempt, and suicidal ideation even in the absence of overt suicidal behavior is a strong risk factor for eventual completion, especially at the “worst-point”, the time in the patient's life in which they experienced the most intense desire to commit suicide (Beck et al, 1999 and Joiner et al, 2003). These observations suggest that suicidal ideation in the absence of associated overt self-injurious behavior is an important risk factor, as it may herald future completed suicide without any interim suicide attempts prior to death.

We also evaluated the major symptom domains in schizophrenia that contribute to suicide risk, including psychosis (Scale for the Assessment of Positive Symptoms, SAPS), depression, with an instrument validated for psychotic illness (Calgary Depression Scale, CDS), and trait impulsivity (Barratt Impulsivity Scale, 11th version, BIS-11). All symptom measures were obtained within two weeks of MRI.

2.3. AX version of the Continuous Performance Task (AX-CPT)

fMRI was conducted while subjects performed the AX version of the Continuous Performance task (AX-CPT), a widely used cognitive control task ( Yoon et al., 2008 ). This task involves presentation of a “Cue” letter that provides the rule which guides the subsequent stimulus-response mapping to the Probe letter stimulus. The cue letter is presented for a 0.5 second duration, followed by a 3.5 second delay, and then the Probe letter is shown for a duration of 0.5 s, and finally a 9.5 second inter-trial interval (total trial duration 14 s). Subjects are instructed to make a “target” response (two-choice button-press) to the probe letter X, only when it follows the cue letter A. All other stimuli require a non-target response, including trials with probe letter Y, and trials in which the probe letter X is preceded by any letter other than A (collectively referred to as B Cues). Trials with target cue–probe pairings (e.g., AX) occur with high frequency (70%) to establish the pre-potent tendency for the subject to make a target response to the probe letter X. The major goal-representation demand occurs with B-Cues, in which subjects must represent and use the proper rule to overcome the trained pre-potent tendency to make a “target” response to probe letter X. Performance measures included accuracy and reaction time (RT) in each task condition (cue–probe pairs), and the decrement (“cost”) in accuracy and RT on BX trials compared to AX trials (see Table 1 ).

Table 1 Demographic and clinical characteristics, and task performance in schizophrenia group (n = 35).

Measure SI − (n = 17) SI + (n = 18) SI +/SB − (n = 10) SI +/SB + (n = 8)
  Mean SD Mean SD Mean SD Mean SD
Age 21.5 3.2 21.6 4.0 20.6 2.1 23.8 6.0
Sex (% M) 14 (80%)   15 (83%)   7 (70%)   5 (63%)  
IQ 99 10 105 14 106 15 104 11
Personal educ 12.3 1.7 13.3 2.1 12.4 1.3 15.2 2.4
Parental educ 14.1 4.2 15.6 3.1 16.6 3.2 13.6 1.7
C-SSRS: II total 0 0 13.4 ⁎⁎ 8.1 10.9 5.6 18.5 ⁎⁎⁎ 10.1
SAPS total 12.0 14.0 13.4 15.1 14.5 14.4 11.0 17.5
CDS total 0.5 1.0 2.3 3.2 1.9 2.2 3.2 4.8
BIS-11 total 51.2 10.1 52.9 9.1 54.5 9.3 50.0 8.4
AX accuracy (%) 94.8 4.9 94.1 7.1 93.1 8.4 96.0 3.0
AY accuracy (%) 86.4 15.6 83.7 19.5 85.5 21.5 80.0 15.9
BX accuracy (%) 80.1 20.8 86.4 14.2 86.8 16.3 85.5 10.2
BY accuracy (%) 97.3 5.6 95.0 8.9 97.1 4.3 91.0 14.0
Accuracy Cost (%) 14.7 16.8 7.7 12.7 6.3 14.6 10.5 8.2
AX RT (ms) 644 122 626 107 632 116 614 95
AY RT (ms) 822 119 795 149 781 146 823 165
BX RT (ms) 766 166 746 191 736 193 767 205
BY RT (ms) 674 120 693 154 677 142 726 187
RT Cost (ms) 123 92 120 105 104 102 152 113

p < 0.05.

⁎⁎ p < 0.005.

⁎⁎⁎ p < 0.10.

IQ, Intelligence Quotient (WASI 2-scale), C-SSRS, Columbia Suicide Severity Rating Scale, SI, Suicidal Ideation subscale, II, Intensity of Ideation subscale, SB, Suicidal Behavior subscale, SAPS, Scale for the Assessment of Positive Symptoms, CDS, Calgary Depression Scale, BIS-11, Barratt Impulsivity Scale, 11th version, AX-CPT, AX version of Continuous Performance Task. See Experimental/materials and methods section for definition of individual task conditions.

2.4. Functional neuroimaging

2.4.1. Acquisition and pre-processing

fMRI was conducted on a 1.5 T GE scanner and BOLD contrast was acquired during single-shot, echo-planar imaging (EPI), using a T2*-weighted sequence and whole-brain coverage. The parameters of the EPI sequence were TR 2000 ms, TE 40 ms, flip angle 90°, FOV 220 × 220 mm, with 24 contiguous slices, each 4.0 mm isotropic voxel size, matrix size 64 × 64 (from 80 mm above to 16 mm below the anterior commissure–posterior commissure plane). Preprocessing was conducted using SPM8, including 6-parameter linear motion correction (spatial alignment to the first scan in each time-series), co-registration to the subject's co-planar T2 image, normalization to the standard MNI template (non-linear warping, using parameters derived from the subject's co-planar T2 image), intensity normalization and smoothing with an 8 mm kernel. All subjects had cumulative head movement less than one voxel in each dimension.

2.4.2. Modeling of the BOLD signal and inferential statistics

We modeled task-related events by convolving the canonical hemodynamic response function (HRF) with a series of delta functions, placed at onset of each Cue and Probe event. Only correct trials were entered into statistical contrasts, to emphasize on-task trials. Scan-to-scan movement (in 6 dimensions) was entered as nuisance regressors. To evaluate goal-representation, neural responses were contrasted between [High-control Cues (B Cues) minus Low-control Cues (A Cues)]. These were conducted as voxel-wise, fixed-effects analysis at the subject-level, then random-effects analysis at the group-level. The group-level regression model for each task contrast included total scores on scales for Depression (Calgary Depression Scale), Impulsivity (Barratt Impulsivity Scale, 11th version), and Psychosis (SAPS), and either A) C-SSRS II subscale totals, B) SI + vs. SI −, or C) SB + vs. SB − variables, in parallel analyses. Each reported contrast is for SI, II or SB, showing brain regions where task-related BOLD signal change during goal-representation is significantly related to SI, II or SB, each controlling for depression, impulsivity and psychosis. The model that evaluated SB was conducted on the 18-patient subsample who reported past suicidal ideation, in order to evaluate how suicidal behavior relates to brain dysfunction, beyond that of suicidal ideation alone. Eight of these 18 SI + patients (44%) reported past suicidal behavior. For the determination of statistical significance at the group-level, we conducted a Monte Carlo simulation, using the Alpha-Sim program in AFNI. Contrast maps were first masked to the frontal cortex (derived from PickAtlas library) and thresholded at p < 0.005, and the simulation then returned cluster sizes that comprise the empirically-determined statistical threshold of p < 0.05. The cluster-thresholds in the contrast maps were as follows for each model: SI, 32 voxels (256 mm3), II, 31 voxels (248 mm3) and SB, 30 voxels (240 mm3).

3. Results

3.1. Clinical measures

Please see Table 1 for subject demographic and clinical characteristics.

3.2. Neuroimaging results

Table 2 and Fig 1, Fig 2, and Fig 3 show the results of group-level contrasts indicating brain regions where the three subscales of suicide risk were significantly associated with brain activity during goal-representation, independent of depression, psychosis or impulsivity. Suicidal Ideation ( Table 2 and Fig. 1 ) was inversely related to neural activity during goal-representation in three major clusters. The first was centered on medial frontal sectors, including the pregenual anterior cingulate gyrus and adjacent ventromedial PFC, extending to bilateral ventrolateral PFC. A second cluster was centered on the left rostral pole, extending into the ipsilateral left dorsomedial PFC. A third major cluster was observed in the right dorsal anterior cingulate gyrus. The Intensity of Ideation scores ( Table 2 and Fig. 2 ) were significantly inversely associated with an overlapping set of frontal regions, including a large, bilateral midline PFC cluster which included dorsomedial and pregenual PFC. However, in addition, several lateral PFC regions were observed, including bilateral DLPFC and VLPFC, right rostrolateral PFC, and right supplementary motor area. For each of these clusters, the test statistics were negative in sign, indicating that higher suicide risk scores were associated with relatively lower goal-related activity. With the Suicidal Behavior scale ( Table 2 and Fig. 3 ), among those subjects who reported past suicidal ideation, additional positive past report of suicidal behavior was inversely associated with brain activation during goal-representation in the left dorsal premotor cortex, ipsilateral to primary motor cortex that was active during task performance.

Table 2 Frontal brain regions with a significant association of neural activity during goal-representation with long-term suicide risk in recent-onset schizophrenia.

Brain region Brodmann area Volume (mm3) Peak t score Peak coordinates
Suicidal Ideation
Left superior frontal gyrus 9 2200 3.84 − 22, 40, 38
      3.74 − 16, 54, 40
Left medial frontal gyrus 9   3.73 − 4, 40, 28
Right middle frontal gyrus 10 528 3.72 34, 54, 4
Left medial frontal gyrus 10 1152 3.48 − 12, 44, 14
      3.40 − 2, 55, 12
      3.36 − 12, 48, 12
Right medial frontal gyrus 10 1624 3.40 6, 50, 10
      3.32 15, 42, 15
Right dorsal anterior cingulate gyrus 24 440 3.38 10, 0, 44
  32   3.37 15, 4, 42
Intensity of Ideation
Left superior frontal gyrus 9 6184 5.10 − 26, 50, 40
      4.53 − 20, 42, 38
      4.28 − 20, 52, 38
      4.10 − 8, 54, 40
Left medial frontal gyrus 10   4.00 − 12, 48, 12
Left middle frontal gyrus 6 968 4.57 − 26, − 12, 58
      4.48 − 30, − 8, 50
Right dorsal anterior cingulate gyrus 32 920 4.52 16, 8, 40
      4.29 15, 4, 42
Right middle frontal gyrus 10 712 4.21 36, 56, 4
Right dorsal anterior cingulate gyrus 32 2200 4.18 6, 46, 12
Right superior frontal gyrus 6 896 4.13 24, − 6, 58
Left medial frontal gyrus 6 1448 4.04 − 8, − 14, 62
      4.04 − 8, 0, 50
  32   3.74 − 10, 10, 52
Right precentral gyrus 6 1224 3.80 38, − 8, 38
      3.57 48, − 5, 32
Left inferior frontal gyrus 44 352 3.77 − 42, 8, 10
Right superior frontal gyrus 8 256 3.59 24, 42, 52
Left middle frontal gyrus 9 456 3.55 − 36, 8, 40
Right dorsal anterior cingulate gyrus 32 464 3.50 4, 32, 30
Right middle frontal gyrus 9 560 3.33 34, 26, 40
Right inferior frontal gyrus 45 248 3.09 48, 24, 8
Suicidal Behavior
Left middle frontal gyrus (dorsal premotor cortex) 6 280 4.58 − 16, − 10, 66

Fig. 1 SPM depicting frontal brain regions where lower activation during goal-representation is associated with lifetime worst-point suicidal ideation in recent-onset schizophrenia. Map rendered at p < 0.005 and frontal clusters corrected to p < 0.05.


Fig. 2 SPM depicting frontal brain regions where lower activation during goal-representation is associated with lifetime worst-point intensity of suicidal ideation in recent-onset schizophrenia. Map rendered at p < 0.005 and frontal clusters corrected to p < 0.05.


Fig. 3 SPM depicting region in left dorsal premotor cortex where lower activation during goal-representation is associated with lifetime suicidal behavior in recent-onset schizophrenia. Map rendered at p < 0.005 and frontal cluster corrected to p < 0.05.

4. Discussion

This study represents a first effort to test a contemporary model of prefrontal cortical function, derived from a cognitive neuroscience perspective, to understand how frontal dysfunction may underpin suicide risk in patients with schizophrenia. We found several frontal sectors that showed strong relationships between neural activity and measures of suicide risk. Importantly, these associations were independent of the major established clinical risk factors for suicide in these groups, including depression, impulsivity and psychosis, which themselves are associated with disturbed brain function. It therefore appears that the present findings identify brain dysfunction that directly relates to suicide risk in these patients.

The schizophrenia group showed clear associations of suicidal ideation with impaired goal-representation in several frontal regions that have been implicated in both a meta-analysis of functional neuroimaging studies of PFC-dependent task performance by schizophrenia patients ( Minzenberg et al., 2009 ), as well as in a recent paper from our group that reported a subset of the present sample, in comparison to a matched healthy control group ( Lesh et al., 2013 ). These results suggest that the failure to maintain a proper context for action, with the rule-representation that supports this function mediated by frontal-based networks, may lead to repeated goal-failure as a contributor to suicidal thoughts in patients with schizophrenia.

In schizophrenia patients such as the sample under study, disruptions in goal-representation may lead to failures to select the thoughts, emotions, and behaviors necessary to attain these goals. These cognitive failures could have a cumulative effect over time, leading to suicidal thoughts and behaviors, perhaps in a manner analogous to how chronic unremitting depression or physical pain culminates in suicidal thoughts and behaviors in other clinical populations. In contrast, a more time-independent scenario could be manifest as isolated “moments in time” when the profound lack of experienced self-efficacy of thoughts, emotions or behaviors (due to PFC dysfunction) triggers suicidal thoughts or actions, perhaps in an automatic, associative manner. While these might also occur with less severely ill populations (e.g. those with anxiety disorders) under certain (typically highly stressful) circumstances, schizophrenia patients may be relatively less capable of “unselecting” these thoughts or actions as an expression of intact self-regulation. While this account is somewhat speculative, these various phenomena (e.g. stress-response, self-regulation and agency) are also distinctly related to control processes, and highly-dependent on the PFC (Haggard, 2005, Maier et al, 2006, and Arnsten, 2011). Heterogeneity in the PFC regions/circuits subserving these psychological functions could then form the basis for considerable clinical variation in suicide phenomenology observed across disorders as diverse as psychotic disorders, late-life depression, chronic pain, personality disorders and impulse control disorders.

Interestingly, the analysis restricted to patients with past suicidal ideation revealed that past suicidal behavior was specifically associated with impaired control-related activity in dorsal premotor cortex, above that of suicidal ideation alone. Premotor cortex subserves complex aspects of action control, such as planning, selection and online control of action (Scott, 2004, Chouinard and Paus, 2006, and Hoshi and Tanji, 2007). This suggests that, among those who experience conscious thoughts of suicide, an important threshold for conversion to overt suicide-related behavior may be disturbances in premotor mediation of these higher-order aspects of action control. Premotor function could then constitute a specific anatomic target for the remediation of overt suicidal behavior, as distinct from interventions that target suicidal ideation.

The present results also demonstrate more generally that testing sophisticated models of PFC function in severely ill clinical populations at risk for suicide is tractable, with potential to develop candidate biomarkers to study novel interventions to mitigate suicide risk. There is preliminary evidence that suicide risk may be lower with lithium treatment in bipolar disorder ( Baldessarini and Tondo, 2008 ) and clozapine treatment in schizophrenia ( Meltzer et al., 2003 ); however, even for promising pharmacological (or psychological ( Brown et al., 2005 )) treatments, it remains essentially unknown what the neurobiological mechanism might be that confers protection against suicide, especially at the level of neural system function. The present study is limited in the capacity to evaluate potential psychotropic medication effects on the brain–behavior relationships observed here, though it is worth mentioning that none of the patients were taking either lithium or clozapine (at or prior to study). The issue of how lithium and clozapine might act to mitigate suicide risk and whether other medications (such as other atypical antipsychotics, and antidepressants) affect suicide risk are important questions that warrant further study. Knowledge of the sites and mechanisms of brain circuit dysfunctions which confer suicide risk is critical to advance the development of treatments, both biological and psychological, to mitigate this tragic, prevalent outcome.

Role of funding source

This work was supported by an American Foundation for Suicide Prevention Young Investigator Award, and the Doris Duke Charitable Foundation Grant # 2009045, both to MJM, and MH059883 to CSC.


MJM and CSC designed the study and wrote the protocol. TAL, TAN and RNR managed the data acquisition. MJM managed the literature searches. MJM, JHY and RNR undertook the statistical analysis, and MJM wrote the first draft of the manuscript. All authors contributed to and have approved the final manuscript.

Conflicts of interest

The authors have no conflicts to declare.


We thank Madison Titone, BA, Sandra Garcia, BA, and Taylor Salo, BS for their assistance with the study data acquisition and management.


  • Aguilar et al., 2008 E.J. Aguilar, G. Garcia-Marti, L. Marti-Bonmati, J.J. Lull, D. Moratal, M.J. Escarti, M. Robles, J.C. Gonzalez, M.I. Guillamon, J. Sanjuan. Left orbitofrontal and superior temporal gyrus structural changes associated to suicidal behavior in patients with schizophrenia. Prog. Neuropsychopharmacol. Biol. Psychiatry. 2008;32:1673-1676 Crossref
  • Amen et al., 2009 D.G. Amen, J.R. Prunella, J.H. Fallon, B. Amen, C. Hanks. A comparative analysis of completed suicide using high resolution brain SPECT imaging. J. Neuropsychiatry Clin. Neurosci.. 2009;21(4):430-439 Crossref
  • Arnsten, 2011 A.F. Arnsten. Prefrontal cortical network connections: key site of vulnerability in stress and schizophrenia. Int. J. Dev. Neurosci.. 2011;29(3):215-223 Crossref
  • Audenaert et al., 2002 K. Audenaert, I. Goethals, K. Van Laere, P. Lahorte, B. Brans, J. Versijpt, et al. SPECT neuropsychological activation procedure with the Verbal Fluency Test in attempted suicide patients. Nucl. Med. Commun.. 2002;23(9):907-916 Crossref
  • Badre, 2008 D. Badre. Cognitive control, hierarchy, and the rostro-caudal organization of the frontal lobes. Trends Cogn. Sci.. 2008;12(5):193-200 Crossref
  • Baldessarini and Tondo, 2008 R.J. Baldessarini, L. Tondo. Lithium and suicidal risk. Bipolar Disord.. 2008;10(1):114-115 Crossref
  • Beck et al., 1999 A.T. Beck, G.K. Brown, R.A. Steer, K.K. Dahlsgaard, J.R. Grisham. Suicide ideation at its worst point: a predictor of eventual suicide in psychiatric outpatients. Suicide Life Threat. Behav.. 1999;29(1):1-9
  • Botvinick et al., 2001 M.M. Botvinick, T.S. Braver, D.M. Barch, C.S. Carter, J.D. Cohen. Conflict monitoring and cognitive control. Psychol. Rev.. 2001;108(3):624-652 Crossref
  • Brown et al., 2005 G.K. Brown, T. Ten Have, G.R. Henriques, S.X. Xie, J.E. Hollander, A.T. Beck. Cognitive therapy for the prevention of suicide attempts: a randomized controlled trial. JAMA. 2005;294(5):563-570 Crossref
  • Chouinard and Paus, 2006 P.A. Chouinard, T. Paus. The primary motor and premotor areas of the human cerebral cortex. Neuroscientist. 2006;12:143-152 Crossref
  • Dutta et al., 2010 R. Dutta, R.M. Murray, M. Hotopf, J. Allardyce, P.B. Jones, J. Boydell. Reassessing the long-term risk of suicide after a first episode of psychosis. Arch. Gen. Psychiatry. 2010;67(12):1230-1237 Crossref
  • Goldsmith and Institute of Medicine (U.S.), 2002 S.K. Goldsmith, Institute of Medicine (U.S.). Committee on pathophysiology & prevention of adolescent & adult suicide. Reducing Suicide: A National Imperative (National Academies Press, Washington, D.C., 2002)
  • Haggard, 2005 P. Haggard. Conscious intention and motor cognition. Trends Cogn. Sci.. 2005;9(6):290-295 Crossref
  • Hawton and van Heeringen, 2009 K. Hawton, K. van Heeringen. Suicide. Lancet. 2009;373(9672):1372-1381 Crossref
  • Hoshi and Tanji, 2007 E. Hoshi, J. Tanji. Distinctions between dorsal and ventral premotor areas: anatomical connectivity and functional properties. Curr. Opin. Neurobiol.. 2007;17:234-242 Crossref
  • Joiner et al., 2003 T.E. Joiner Jr., R.A. Steer, G. Brown, A.T. Beck, J.W. Pettit, M.D. Rudd. Worst-point suicidal plans: a dimension of suicidality predictive of past suicide attempts and eventual death by suicide. Behav. Res. Ther.. 2003;41(12):1469-1480 Crossref
  • Jollant et al., 2010 F. Jollant, N.S. Lawrence, E. Olie, O. O'Daly, A. Malafosse, P. Courtet, et al. Decreased activation of lateral orbitofrontal cortex during risky choices under uncertainty is associated with disadvantageous decision-making and suicidal behavior. NeuroImage. 2010;51(3):1275-1281 Crossref
  • Koechlin et al., 2003 E. Koechlin, C. Ody, F. Kouneiher. The architecture of cognitive control in the human prefrontal cortex. Science. 2003;302(5648):1181-1185 Crossref
  • Lesh et al., 2011 T.A. Lesh, T.A. Niendam, M.J. Minzenberg, C.S. Carter. Cognitive control deficits in schizophrenia: mechanisms and meaning. Neuropsychopharmacology. 2011;36(1):316-338 Crossref
  • Lesh et al., 2013 T.A. Lesh, A.J. Westphal, T.A. Niendam, J.H. Yoon, M.J. Minzenberg, J.D. Ragland, M. Solomon, C.S. Carter. Proactive and reactive cognitive control and dorsolateral prefrontal cortex dysfunction in first episode schizophrenia. Neuroimaging Clin. N. Am.. 2013;2:590-599 Crossref
  • Maier et al., 2006 S.F. Maier, J. Amat, M.V. Baratta, E. Paul, L.R. Watkins. Behavioral control, the medial prefrontal cortex, and resilience. Dialogues Clin. Neurosci.. 2006;8(4):397-406
  • Mann, 2003 J.J. Mann. Neurobiology of suicidal behavior. Nat. Rev.. 2003;4:819-828 Crossref
  • Meltzer et al., 2003 H.Y. Meltzer, L. Alphs, A.I. Green, A.C. Altamura, R. Anand, A. Bertoldi, et al. Clozapine treatment for suicidality in schizophrenia: International Suicide Prevention Trial (InterSePT). Arch. Gen. Psychiatry. 2003;60(1):82-91 Crossref
  • Miller and Cohen, 2001 E.K. Miller, J.D. Cohen. An integrative theory of prefrontal cortex function. Annu. Rev. Neurosci.. 2001;24:167-202 Crossref
  • Minzenberg et al., 2009 M.J. Minzenberg, A.R. Laird, S. Thelen, C.S. Carter, D.C. Glahn. Meta-analysis of 41 functional neuroimaging studies of executive function in schizophrenia. Arch. Gen. Psychiatry. 2009;66(8):811-822 Crossref
  • Nielssen et al., 2012 O.B. Nielssen, G.S. Malhi, P.D. McGorry, M.M. Large. Overview of violence to self and others during the first episode of psychosis. J. Clin. Psychiatry. 2012;73(5):e580-e587 Crossref
  • Nock et al., 2008 M.K. Nock, G. Borges, E.J. Bromet, C.B. Cha, R.C. Kessler, S. Lee. Suicide and suicidal behavior. Epidemiol. Rev.. 2008;30:133-154 Crossref
  • Owens et al., 2002 D. Owens, J. Horrocks, A. House. Fatal and non-fatal repetition of self-harm. Systematic review. Br. J. Psychiatry. 2002;181:193-199 Crossref
  • Pompili et al., 2011 M. Pompili, G. Serafini, M. Innamorati, D. Lester, A. Shrivastava, P. Girardi, et al. Suicide risk in first episode psychosis: a selective review of the current literature. Schizophr. Res.. 2011;129(1):1-11 Crossref
  • Posner et al., 2011 K. Posner, G.K. Brown, B. Stanley, D.A. Brent, K.V. Yershova, M.A. Oquendo, et al. The Columbia-Suicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults. Am. J. Psychiatry. 2011;168(12):1266-1277
  • Reisch et al., 2010 T. Reisch, E. Seifritz, F. Esposito, R. Wiest, L. Valach, K. Michel. An fMRI study on mental pain and suicidal behavior. J. Affect. Disord.. 2010;126(1-2):321-325 Crossref
  • Rusch et al., 2008 N. Rusch, I. Spoletini, M. Wilke, G. Martinotti, P. Bria, A. Trequattrini, et al. Inferior frontal white matter volume and suicidality in schizophrenia. Psychiatry Res.. 2008;164(3):206-214 Crossref
  • Scott, 2004 S.H. Scott. Optimal feedback control and the neural basis of volitional motor control. Nat. Rev. Neurosci.. 2004;5:532-546 Crossref
  • Shackman et al., 2011 A.J. Shackman, T.V. Salomons, H.A. Slagter, A.S. Fox, J.J. Winter, R.J. Davidson. The integration of negative affect, pain and cognitive control in the cingulate cortex. Nat. Rev. Neurosci.. 2011;12(3):154-167 Crossref
  • Spoletini et al., 2011 I. Spoletini, F. Piras, S. Fagioli, I.A. Rubino, G. Martinotti, A. Siracusano, C. Caltagirone, G. Spalletta. Suicidal attempts and increased right amygdala volume in schizophrenia. Schizophr. Res.. 2011;125:30-40 Crossref
  • Strobel et al., 2007 A. Strobel, G. Dreisbach, J. Muller, T. Goschke, B. Brocke, K.P. Lesch. Genetic variation of serotonin function and cognitive control. J. Cogn. Neurosci.. 2007;19(12):1923-1931 Crossref
  • United States. Public Health Service, 2001 United States. Public Health Service. National Strategy for Suicide Prevention: Goals and Objectives for Action: Summary. (U.S. Dept. of Health and Human Services, Public Health Service, Rockville, MD, 2001)
  • United States. Public Health Service: Office of the Surgeon General., 1999 United States. Public Health Service: Office of the Surgeon General. The Surgeon General's Call to Action to Prevent Suicide, 1999. (Dept. of Health and Human Services, U.S. Public Health Service, Washington, D.C., 1999)
  • Yoon et al., 2008 J.H. Yoon, M.J. Minzenberg, S. Ursu, B.S. Ryan Walter, C. Wendelken, J.D. Ragland, et al. Association of dorsolateral prefrontal cortex dysfunction with disrupted coordinated brain activity in schizophrenia: relationship with impaired cognition, behavioral disorganization, and global function. Am. J. Psychiatry. 2008;165(8):1006-1014 Crossref
  • Yurgelun-Todd et al., 2011 D.A. Yurgelun-Todd, C.E. Bueler, E.C. McGlade, J.C. Churchwell, L.A. Brenner, M.P. Lopez-Larson. Neuroimaging correlates of traumatic brain injury and suicidal behavior. J. Head Trauma Rehabil.. 2011;26(4):276-289 Crossref
  • Zhang et al., 2013 H. Zhang, X. Wei, H. Tao, T.E. Mwansisya, W. Pu, Z. He, A. Hu, L. Xu, Z. Liu, B. Shan, Z. Xue. Opposite effective connectivity in the posterior cingulate and medial prefrontal cortex between first-episode schizophrenic patients with suicide risk and healthy controls. PLoS One. 2013;8(5):e63477 Crossref


a Department of Psychiatry, University of California, San Francisco School of Medicine, San Francisco, CA, United States

b Veterans Affairs Medical Center, San Francisco, CA, United States

c Department of Psychiatry, University of California, Davis School of Medicine, Sacramento, CA, United States

d Center for Neuroscience, University of California, Davis, CA, United States

e Department of Psychiatry and Behavioral Sciences, Stanford School of Medicine, Palo Alto, CA, United States

f VA Palo Alto Health Care System, Palo Alto, CA, United States

Corresponding author at: Outpatient Mental Health, 116C, San Francisco Veterans Affairs Medical Center, 4150 Clement Street, San Francisco, CA 94121, United States. Tel.: + 1 415 221 4810x6554.