Another way to examine the suicidal brain is through morphometric analysis. Here, we try to make it more directly accessible to the clinician. This modality measures the physical makeup of brain structures by examining measures such as cortical thickness, cortical surface area, and/or cortical volume of the relevant brain areas. The implication is that the neural systems which may be hypo/hyperactive during functional processing may have altered physical attributes. Many morphometric studies have been done to find alterations in those with STBs.

The current review identified such studies which reported associations between these morphometric measures and suicidal behavior.

One of the earlier studies addressing the association between history of suicidal attempts in elderly, depressed individuals and both cortical and sub-cortical abnormalities was attempted by Hwang et al. (55). In that study, authors used voxel-based analysis and were able to show widespread gray matter volume (GMV) reduction in the frontal (i.e., left medial, bilateral superior, right middle, right inferior, and left posterior frontal cortices), parietal (i.e., left superior, right inferior, and left lateral parietal cortices), occipital gyrus (i.e., left cuneus), left STG, and sub-cortical (i.e., bilateral insula, left lentiform nucleus, right claustrum, bilateral midbrain, bilateral culmen, and right inferior and bilateral superior semilunar lobules) brain regions in late-onset geriatric depressed individuals with a history of suicide attempts compared to those without previous attempts. Here, volume reduction was most prominent within the dmPFC, consistent with impairment in reward-based learning and top-down executive control.

Wagner et al. reported cortical thinning in the left dorsolateral, ventrolateral prefrontal, and ACC in depressed adults with documented suicidal behavior, i.e., high-risk group of suicide as compared to depressed adults with a non-high risk for suicide (56).

A study by Huber et al. specifically found white-matter volume differences in the left ACC between veterans with a history of attempting suicide relative to veterans with a history of SUI (57). Along with RSFC data previously mentioned, the work by Barredo and colleagues found that cortical thickness of the ACC/PCC was shown to predict the functional connectivity between the lateral pars orbitalis and anterior cingulate/paracingulate control regions (46).

Wang et al. specifically reported significant differences in GMV in the bilateral MFG across patients with mood disorders and suicidal behavior, patients with mood disorders without suicidal behavior, and healthy controls (58). However, they did not find significant differences between participants with SUI and those with a history of actual suicide attempts.

Ding et al. used a region-specific approach to study differences between suicide attempters with a past history of mood disorders and suicidal behavior, participants with a mood disorder but not suicidal behavior, and healthy controls with neither (59). Reduced cortical volume was observed within the left ventrolateral prefrontal cortex in suicide attempters compared to both control groups. In addition, the orbitofrontal and dorsal prefrontal cortices (but not medial prefrontal cortex) also showed reduced cortical measures in suicide attempters compared to healthy controls. This is consistent with task-based and resting-state fMRI studies showing generally both reduced top-down executive control and impairment in reward-based learning regions and networks. Structurally, there were significant GMV decreases among suicide attempters across clinical conditions compared to non-attempters. Diffusion Tensor Imaging (DTI) findings also showed significantly reduced fractional anisotropy among those with past attempts versus those without (60). In a study of future suicide attempters (i.e., individuals attempting suicide between baseline and follow-up assessment) with mood disorders, participants showed lower baseline ventral and rostral prefrontal GMV compared to those who did not attempt (61). Besides the studies examining participants with depression and/or previous suicide attempts, distinct markers that included an involvement of frontal regions, particularly reduced cortical volume within the left MFG and cortical thinning within the posterior frontal lobe including the bilateral precentral gyrus, were also found in individuals with current SUI as compared to healthy controls without even a family history of psychiatric disorders or suicide attempts (62). Interestingly, greater cortical surface area and cortical volume within the left dorsolateral prefrontal gyrus were reported to be associated with reduced SUI in a non-clinical population with mild levels of stress and perceived lack of social support (63). This would be consistent with increasing behavioral control and top-down influence on impulsivity with decreasing pathology.

In a recent study by Yang et al. authors found that suicidal depressed patients had reduced GMV in the right IFG and left caudate but increased GMV in the left calcarine fissure (35), areas associated with visual processing not irrelevant to potential affective or social stimuli.

A recent study by Kang et al. involving individuals with depression with and without a history of suicide attempts showed altered morphometry in the lateral parietal and occipital brain regions along with frontal areas. In that study, depressed patients with past suicide attempts were found to have larger surface area within the left postcentral and left lateral occipital areas and large cortical volume within the left postcentral and left lateral orbitofrontal areas, whereas smaller surface area within the left SFG was found (64). In another recent study by Harenski et al. criminal offenders with a history of suicide attempts had widespread decreased gray matter within both cortical and sub-cortical regions, including the PCC/precuneus, IPC, dorsal prefrontal cortex, amygdala, insula, superior occipital gyrus, cuneus, and cerebellum (65). Using local shape volume analysis, researchers specifically found significant volumetric differences between suicidal and non-suicidal depressed individuals in the left amygdala, left hippocampus, left putamen, bilateral pallidum, and bilateral thalamus (66). In another study however, it was only the reduced volume of the right hippocampus that was most prominent in participants with a recent history of suicide attempts within the past 2 months as compared to healthy individuals (67). This may represent limited or impaired processing of new information relative to old beliefs and memories, consistent with Van Heeringen’s framework. Interestingly, in a postmortem study, compared to healthy controls, the suicidal depressed participants had an increased neuron number in CA2/3 subregions of the hippocampus gyrus (68).

In a study of suicide attempters with a family history of suicide and personal history of violent suicide attempts, Jollant et al. found an association between family history of suicide and reduced volume within the bilateral temporal regions, right dlPFC, and left putamen, as well as between violent methods of attempt and increased bilateral caudate and left putamen volumes (69). Reduced GMV in the dorsal striatal structures, particularly bilateral putamen and caudate, were associated with greater implicit SUI observed from suicide-related outcomes from the death version of the Implicit Association Test (IAT) (70, 71).

Several studies also reported cortical alterations within the temporal and parietal lobes but not the frontal lobe.

Reduced GMV within the right STG was observed in adolescents with MDD and a history of suicide attempts compared to adolescents with MDD but without any history of suicide attempts (72), whereas care-giver reported STBs were also associated with decreased volume at the left bank of the superior temporal sulcus in children (73). In another study on adolescents, McLellan et al. reported reduced volume of the right STG in adolescents with treatment-resistant depression and a history of suicide attempts as compared to healthy adolescents (74). Compared to healthy controls, patients with MDD and a history of suicide attempts in Peng et al.’s work showed decreased GMV within the right MTG and increased GMV within the right parietal lobe (75). Authors reported decreased GMV in the left limbic cingulate gyrus for the depressed suicidal group compared to the depressed non-suicidal group. Somewhat contrary to Peng’s work, patients with an attempt history have also been shown to have decreased GMV in the left anterolateral region of the parietal lobe as well as in the right cerebellum (76).

A recent study by the ENIGMA-MDD working group of over 18,925 participants examined morphometric differences between healthy controls, depressed participants, and participants with a history of attempted suicide and found multiple differences between the groups (77). Regarding volumetric differences, the thalamus and right pallidum were significantly smaller in depressed attempters compared to depressed and healthy controls. Regarding cortical surface area, depressed attempters had smaller cortical surface area of the left cuneus, left inferior parietal, left rostral middle frontal, and right pericalcarine cortex compared with healthy controls, but only the inferior parietal cortex was clinically distinct from depressed clinical controls. Lastly, in terms of cortical thickness, although there was not a significant difference between depressed attempters and clinical controls, attempters did display significantly lower cortical thickness in the left rostral middle frontal region. The authors concluded that these findings suggested impairment in decision making, impulsivity, and planning as well as attention and the concept of self (77). However, please note that a significant difference in such large studies does not mean the effect size is clinically meaningful. In other words, a clinically insignificant volumetric difference could be statistically significant due to the large number of subjects studied. For example, in this study, for the left pallidum and right nucleus accumbens subcortical volumes, the difference between clinical and healthy controls did not reach statistical significance after correction for multiple comparisons. Therefore, statistically significant differences may not necessarily translate to clinical significance, but it can inform next steps and build a future-focused plan for translational researchers.

Most recently, Sarkinaite et al. (78) published findings that showed volumetric differences between participants with past suicide attempts and healthy controls in the frontal and temporal cortex thickness and volume of the hippocampus. Notably, with number of attempts as a covariate, participants with increasing number of past suicide attempts showed decreasing thickness of temporal cortex in the inferior middle and temporal cortex.

In Figures 1, 2, we provide an overview of brain regions (Figure 1) and networks (Figure 2) that are most commonly found to be involved in functional MRI and brain morphometry research of suicidal thoughts and behavior. Both the figures are generated through FreeSurfer 7.3.2 (82, 83) and Yeo’s 17-network atlas (84).

The fourth neurobiological modality that we included in this study was Diffusion Tensor Imaging (DTI). This method analyzes the robustness of water diffusion within white-matter structures of the brain to find associated differences between patient populations and healthy controls or other cohorts. DTI also examines the physical “highways” within and between significant structures. While fMRI gives information on “in the moment” electrical communication that could conceivably indicate functional robustness of structures and pathways, DTI is a direct measure of the physical robustness of those structures and pathways. To that end, it may be considered as a more concrete/persistent measure to estimate the difference between healthy controls and the suicidal patients. Again, most of the modalities can be accomplished in a relatively short scanning session and are more potentially accessible to the working clinician.

Recent advancements in imaging have allowed the scientists to study differences in white-matter integrity, compactness, or structural connectivity in clinical populations – the “quality of the highways” so to speak. These studies use parameters such as apparent diffusion coefficient (ADC-speed of water flow regardless of direction), fractional anisotropy (FA-diffusion of water molecules in a particular direction), and edge-weight (white-matter structural connectivity) to measure connections between regions of interest (85, 86). Edge weight has been considered a potentially more appropriate parameter to measure the strength of structural connectivity because it takes into account both the number of white-matter fibers and the size of the regions of interest (87). Our focused review identified studies which reported associations between these measures and suicidal behavior.

In a study of young adult healthy controls and young adults with MDD with and without a history of suicide attempts conducted by Jia et al. reduced FA was found in the (a) left anterior limb of the internal capsule (ALIC) for suicide attempters compared to non-attempters and healthy controls, (b) right frontal lobe (subgyral) for suicide attempters compared to healthy controls, and (c) right lentiform nucleus (putamen) for suicide attempters compared to non-attempters (88). In another similar study by Jia et al. it was also found that compared to healthy controls, depressed suicide attempters had significantly lower mean percentage of fibers projecting from the ALIC to the left medial frontal cortex, left OFC, and left thalamus. Compared to depressed non-suicide attempters, depressed suicide attempters had significantly lower mean percentage of fibers projecting from the ALIC to the left OFC and left thalamus (89). However, in a study involving panic disorder and suicide attempt, several regions, including the retrolenticular part of the internal capsule, splenium of the corpus callosum, superior and posterior corona radiata, posterior thalamic radiations, sagittal stratum (including the inferior longitudinal fasciculus and inferior fronto-occipital fasciculus), and superior longitudinal fasciculus, showed increased FA in individuals with panic disorder and history of suicidal attempt (PD + SA) compared to individuals with panic disorder but without any history of suicidal attempt (90). For the PD + SA group, Kim et al. also found that for two regions (i.e., right retrolenticular part of the internal capsule and bilateral posterior thalamic radiations), there was a significant positive association between FA and SUI.

In another study, lower baseline FA was found in the left ALIC, bilateral dmPFC, and right dorsal cingulum for future suicide attempters (i.e., who attempted suicide between the baseline and follow-up assessment) compared to non-attempters (61). In that study, compared to the non-future suicide attempt group, both future suicide attempters with or without a history of suicide attempts had lower FA for the left dmPFC, right dlPFC, and left ALIC. Future suicide attempters with a history of suicide attempts also showed lower FA in the right dmPFC and right dorsal cingulum. Authors found that after an exclusion of four participants with alcohol/substance use disorder, the left ventral prefrontal cortex also had lower FA for the future suicide attempters relative to non-attempters.

Fan et al. also found lower FA in the dorsal and ventral frontal regions that included the uncinate fasciculus for individuals with MDD and a history of suicide attempts as compared to non-suicide attempters (60).

In terms of structural connectivity parameters, it was determined that compared to individuals with MDD without SUI, individuals with MDD and SUI had reduced structural connectivity/edge weights across cortical (i.e., rostral middle frontal cortex, superior parietal cortex), subdivisions of the inferior frontal cortex (i.e., pars triangularis and pars orbitalis, frontal pole, and lateral occipital cortex), and sub-cortical (i.e., pallidum, thalamus, putamen, and caudate) regions in the left hemisphere (91). In terms of FA also, the frontal areas, especially right dmPFC and white-matter bundles in several regions, including the bilateral inferior fronto-occipital fasciculus, bilateral uncinate fasciculus, body of corpus callosum, right anterior limb of internal capsule, right external capsule, left posterior thalamic radiation, and bilateral posterior corona radiata, showed lower FA amongst suicide attempters with MDD compared to non-attempters with MDD and healthy controls (92, 93).

Another structural connectivity study showed that compared to euthymic non-attempters and healthy controls, there was significantly decreased structural connectivity in euthymic suicide attempters in the connections between the left olfactory cortex and left anterior cingulate gyrus, as well as a clear trend of decreased connectivity between the right medial orbital, SFG, and the right rectal gyrus and between the right calcarine fissure and both the left superior and middle occipital gyrus (94). Hwang et al. reported greater edge weight in the left PCC and greater structural connectivity strength of local connections amongst participants who were military veterans with prior suicide attempts in comparison to those with SUI only and with no suicidal behavior (95).

A 2021 study by Chen et al. evaluated white-matter integrity (generalized fractional anisotropy) and white-matter compactness (normalized quantitative anisotropy) among depressive patients with and without past suicide attempts (96). On a voxel-based (region of interest) analysis, participants with past suicide attempts had significantly lower white matter compactness and integrity in the corpus callosum, cingulate gyrus, and caudate than both the depressed non-attempt and the healthy control groups, with differences between the attempt group and healthy control group reaching statistical significance (96).

The same researchers evaluated white matter density and integrity among depressed patients with suicidal ideation but no history of attempts (97). In the voxel-based analysis, white matter integrity was found to be decreased in the suicidal ideation group, specifically in the corpus callosum and the ACC compared to depressed, non-SI, and healthy control participants. On RSFC, the suicidal ideation group had weaker connections within subnetworks of the frontal lobe compared to healthy controls but did not find differences between suicidal ideation and depression in suicidal participants.