Introduction
Suicide is a major public health concern and is considered as a complex phenomenon with biological, psychological, and social causes.Reference Turecki and Brent 1 Suicidal behavior includes suicidal ideation (SI), suicide attempts, and completed suicide. A history of suicide attempts or the presence of SI is an important risk factor and a major predictor of subsequent suicide.Reference Blanco and Olfson 2 The pathophysiological mechanisms of suicide had been investigated,Reference Niculescu, Le-Niculescu and Levey 3 , Reference Serafini, Canepa and Adavastro 4 but the mechanisms leading to suicidal behavior remain unclear. Because more than half of patients with depression present SI, and one-third of patients with SI progress to committing a suicidal act,Reference Nock, Hwang and Sampson 5 identifying potential biomarkers of SI in patients with major depressive disorder (MDD) is crucial for suicide prevention.
Patients with depression have cognitive deficits in executive function, memory and attention compared to healthy controls.Reference Rock, Roiser and Riedel 6 People with a history of suicide attempts or with current SI have also been reported to have cognitive deficits in the above-mentioned neuropsychological domains.Reference Keilp, Gorlyn and Russell 7 - Reference Westheide, Quednow and Kuhn 9 Specifically, depressed participants with SI were found to have higher cognitive inhibition deficit compared with depressed participants without SI, assessed by using Go/No-Go task.Reference Westheide, Quednow and Kuhn 9 Among executive functions, cognitive inhibition serves to restrict access to irrelevant information from working memory, and suppresses irrelevant stimuli for the ongoing task. With impaired cognitive inhibition, a person lacks capacity to reduce the intrusion of SI and retrieval of incorrect information from working memory, resulting in sustained negative affect.Reference Richard-Devantoy, Gorwood and Annweiler 10
In addition, accumulating evidence has indicated that dysregulated immune system and cytokines may contribute to depressive symptoms and suicide.Reference Brundin, Bryleva and Thirtamara Rajamani 11 , Reference Courtet, Giner and Seneque 12 For instance, patients with MDD have been reported to exhibit elevated serum levels of interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and C-reactive protein (CRP).Reference Dowlati, Herrmann and Swardfager 13 - Reference Howren, Lamkin and Suls 15 Interleukin-6 (IL-6), a cytokine secreted by T-cells and macrophages, can bind to a soluble form of the IL-6 receptor (sIL-6R). The IL-6/sIL-6R complex can stimulate cells that only express glycoprotein 130, a process that displays mainly proinflammatory functions.Reference Scheller, Chalaris and Schmidt-Arras 16 Therefore, serum levels of sIL-6R may be suited for assessing inflammatory activity in BD than IL-6.Reference Bai, Su and Li 17 TNF-α exerts its effects by binding two receptors, TNF-α receptor subtype 1 (TNF-αR1) and TNF-α receptor subtype 2, and the binding of TNF-α to TNF-αR1 has been related to apoptotic neuronal death.Reference Chen and Goeddel 18 The soluble forms of the TNF-α receptors prolong the half-life of TNF-α, and their circulating levels may be used to determine the overall production of TNF-α. Hence, the sTNF-αR1 is more reliable markers of proinflammatory activity.Reference Huang, Chen and Tu 19 Cytokine dysregulation has also been reported in the blood, cerebrospinal fluid, and postmortem brains of individuals with suicidal behavior compared with healthy controls.Reference Ducasse, Olie and Guillaume 20 - Reference Serafini, Parisi and Aguglia 22 However, previous investigations recruited participants with different psychiatric diagnoses, and participants with a history of suicide attempts and with the presence of SI were mixed in the analyses, which led to inconsistent findings.Reference Dowlati, Herrmann and Swardfager 13 Although patients with a history of suicide attempts have been included in many studies, only few studies have examined MDD patients with or without SI.Reference Gananca, Oquendo and Tyrka 23 Furthermore, most studies on the correlation between suicidal behavior and proinflammatory biomarkers did not control for cognitive function. There is need to investigate the correlation between SI and proinflammatory cytokines among patients with MDD, as well as the confounding role of cognitive deficits.
In this study, we investigated the correlation between SI and proinflammatory cytokines among adult patients with MDD, with the adjustment of working memory and inhibitory control. We hypothesize that: (1) the levels of serum proinflammatory cytokines, namely CRP, sIL-6R, and sTNF-αR1, may be associated with SI after controlling for working memory and inhibitory control and (2) SI is correlated with impaired cognitive function in working memory and inhibitory control.
Methods
Our study was conducted in the psychiatric outpatient clinic of Taipei Veterans General Hospital. Consecutive outpatients aged between 20 and 64 years diagnosed with MDD based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, were enrolled. Age-, sex-, and body mass index (BMI)-matched normal controls were also recruited. Exclusion criteria included major physical disorder (ie, autoimmune/immune diseases, stroke, and epilepsy), a history of alcohol or substance misuse or dependence, and major psychiatric comorbidities such as schizophrenia, bipolar disorder, other major psychosis, intellectual disability, and organic mental disorder. We also excluded individuals who received electroconvulsive treatment in the past year before recruitment. All participants were not taking medications that could affect cytokine levels such as nonsteroidal anti-inflammatory drug or antibiotics. The healthy individuals were recruited by poster advertising in the clinic and in the community, and they underwent the Mini-International Neuropsychiatric Interview with a psychiatrist to rule out psychiatric illnesses. The detailed medical histories of all subjects were reviewed and physical examinations were taken to exclude any physical illness. Written informed consent was obtained from all patients prior to their inclusion in the study. The study was approved by the Institutional Review Board of the Taipei Veterans General Hospital and was conducted in accordance with the Declaration of Helsinki.
The severity of depression symptoms in all patients was assessed using the 17-item Hamilton Rating Scale for Depression (HAMD). Patients with MDD were classified as having SI if they scored ≥2 on item-3 of HAMD. The working memory (2-back) task and Go/No-Go task were the two cognitive tasks included in the computerized tests. In the working memory tasks, all participants were asked to respond as quickly as possible when they saw a number that appeared on a screen again separated by one other number (eg, if 31-45-31 was displayed in order, participants responded when 31 was displayed the second time). In the Go/No-Go task, participants were asked to respond as quickly as possible when the × symbol appeared. They were asked not to press the key when the + symbol appeared. After participants completed the pretest with all correct responses, formal tests were then administered to record their reaction times (mean), correct responses, commission errors, and omissions as performance parameters. Go/No-Go task was used to measure a participant’s capacity for sustained attention and response control. Commission errors are related to disinhibition and increased impulsivity, while omission errors reflect one’s deficiency in sustained attention to the task.
Proinflammatory cytokine levels of all subjects, including CRP, sIL-6R, and sTNF-αR1, were assessed using enzyme-linked immunosorbent assay (ELISA) kits (R&D systems, Minneapolis, MN). Fasting serum samples were collected in serum separator tubes, clotted for 30 minutes, and stored at −80°C until use. All assays were performed according to the manufacturer’s instructions. The final absorbance of the mixture was measured at 450 nm, and the mixture was analyzed using an ELISA plate reader with Bio-Tek Power Wave Xs and Bio-Tek’s KC junior software (Winooski, VT). The standard range was considered as per the manufacturer’s instructions. A linear regression R-square value of ≥0.95 represented a reliable standard curve.
Patients’ characteristics were analyzed using descriptive analysis. Chi-square tests were applied to compare categorical data. ANOVA with post hoc analysis was used to compare continuous data among the three groups (depressed patients without SI, depressed patients with SI, and controls). Linear regression analysis was used to examine the correlation between SI and inflammatory cytokines. Linear regression models were used to investigate the association between SI and cognitive function while controlling for confounding factors. Sensitivity analyses were performed with adjustment for age, sex, education, illness duration, and BMI in model 1, additional adjustment for the total HAMD score minus item 3 in model 2, and additional adjustment for cytokine levels in model 3. General linear models (GLMs) were used to assess proinflammatory cytokine levels between groups with the adjustment of age, sex, education, BMI, illness duration, and total HAMD score. The significance level was set at a P < .05. Statistical analysis was performed using SPSS 11.5 (SPSS Inc., Chicago, IL).
Results
In this study, 77 patients with MDD and 60 healthy subjects were recruited, and patients with MDD were divided into the SI group (n = 36) and the no SI group (n = 41). Patients with MDD had lower educational level than the controls (P < .001). Patients and SI had a longer duration of illness and higher total HAMD scores than controls and patients without SI (P < .001; Table 1). MDD patients with SI performed worse on 2-back task and Go/No-Go task than the other two groups (Table 1). GLMs for levels of proinflammatory cytokines with the adjustment of confounding factors revealed a higher level of CRP among patients with SI than the controls and patients without SI (P = .004; Figure 1).
Table 1. Demographic Data, Inflammatory Markers, and Cognitive Function Between MDD Patients With/Without Suicide Ideation and Controls
Note: Bold type indicates statistical significance.
Abbreviations: CRP, C-reactive protein; HAMD, 17-item Hamilton Depression Rating Scale; SD, standard deviation; SI, suicidal ideation; sIL-6R, soluble interleukin-6 receptor; sTNF-αR1, soluble tumor necrosis factor-α receptor type 1.
Figure 1. Levels (pg/mL) of cytokines between major depressive disorder (MDD) patients with or without SI and the controls with the adjustment of age, sex, education, BMI, illness duration, and total HAMD score. CRP, C-reactive protein; HAMD, 17-item Hamilton Depression Rating Scale; SI, suicidal ideation; sIL-6R, soluble interleukin-6 receptor; sTNF-αR1, soluble tumor necrosis factor-α receptor type 1.
Linear regression analyses showed that peripheral CRP level was positively associated with SI after controlling for demographic data, depression severity, working memory, and inhibitory control (β = 0.21, P = .037; Table 2). In addition, we found that the peripheral CRP level was negatively correlated with the correct (β = −0.001, P = .001) and positively correlated with the omission errors (β = 0.001, P = .001) in the working memory task. With respect to the neuropsychological correlates of SI, linear regression analyses adjusting for age, sex, education, illness duration, BMI, depression severity, and cytokine levels revealed that SI was positively associated with the omission errors (β = 0.96, P = .002) and negatively associated with the correct (β = −0.96, P = .002) in the working memory task (Table 3). SI was shown to be positively correlated with commission errors (β = 0.38, P = .035) in the Go/No-Go task (Table 3).
Table 2. Linear Regression Models for the Association Between Suicide Ideation and Cytokine
Note: Model 1: Adjusted by age, sex, education, illness duration, and total HAMD score minus item 3. Model 2: Model 1 additionally adjusted by working memory and cognitive control. Bold type indicates statistical significance.
Abbreviations: CRP, C-reactive protein; HAMD, 17-item Hamilton Depression Rating Scale; sIL-6R, soluble interleukin-6 receptor; sTNF-αR1, soluble tumor necrosis factor-α receptor type 1.
Table 3. Linear Regression Models for the Association Between Suicidal Ideation and Cognitive Function
Note: Bold type indicates statistical significance.
Abbreviations: BMI, body mass index; CRP, C-reactive protein; HAMD, 17-item Hamilton Depression Rating Scale; sIL-6R, soluble interleukin-6 receptor; sTNF-αR1, soluble tumor necrosis factor-α receptor type 1.
Discussion
We found that MDD patients with SI had higher levels of CRP than controls and patients without SI. Higher levels of serum CRP were found to be associated with higher SI after controlling for confounding factors. We also found that MDD patients with SI exhibited deficits in working memory and inhibitory control in the computerized 2-back task and Go/No-Go task, respectively. In addition, SI was found to be correlated with deficits in working memory and inhibitory control.
Our study demonstrated a positive correlation between serum CRP levels and SI. The result corroborated previous observations supporting the relationship between CRP and SI.Reference O’Donovan, Rush and Hoatam 24 - Reference Park and Kim 26 O’Donovan et al found that MDD patients with SI had higher CRP levels than those with low SI and healthy controls.Reference O’Donovan, Rush and Hoatam 24 By contrast, Courtet et al found no significant differences in CRP levels between depressed patients with high SI and those with low SI in an inpatient population.Reference Courtet, Jaussent and Genty 27 However, they did not compare CRP levels between patients and healthy controls. Increased expression of proinflammatory mediators may contribute to the pathophysiology of depression and suicidal behavior through increased tryptophan metabolism via the kynurenine pathway, HPA axis dysregulation and altered monoamine metabolism.Reference Brundin, Bryleva and Thirtamara Rajamani 11 , Reference Raison, Capuron and Miller 28 Inflammatory cytokines are able to activate the indoleamine-2,3-dioxygenase, decreasing the concentrations of tryptophan and increasing the concentrations of kynurenine, and changes in the kynurenine pathway have been proposed to be involved in MDD and suicidal behavior.Reference Brundin, Bryleva and Thirtamara Rajamani 11 , Reference Raison, Capuron and Miller 28 However, the literature is ambiguous regarding the change in peripheral tryptophan and kynurenine levels in depressed individuals. While some reported no association between kynurenine levels and MDD,Reference Bradley, Case and Khan 29 others found lower concentrations of kynurenine.Reference Pompili, Lionetto and Curto 30 More investigations are needed to examine the kynurenine pathway in the pathophysiology of MDD. Furthermore, we found an association between serum CRP level and working memory, and it has been speculated that working memory may mediate increased SI and negative affect.Reference Richard-Devantoy, Gorwood and Annweiler 10 , Reference Bauer, Gustafsson and Nigg 31 Additional studies are necessary to elucidate the underlying mechanisms of CRP for suicidal behavior and cognitive dysfunction among patients with MDD.
We found no significant differences between groups in the levels of sIL-6R and sTNF-αR1 in this study. High circulating levels of IL-6 have been reported to be associated with increased SIReference O’Donovan, Rush and Hoatam 24 and suicide attempts.Reference Lindqvist, Janelidze and Hagell 32 However, the results of previous studies were contradictory. Some studies found no differences in IL-6 levels between individuals with and without a history of suicide attempts.Reference Isung, Mobarrez and Nordstrom 33 , Reference Vargas, Nunes and Pizzo de Castro 34 Contrarily, Kim et al found a lower IL-6 production in MDD patients who had recently attempted suicide than those without history of suicide within 1 year.Reference Kim, Lee and Kim 35 The inconsistent findings may be due to many factors. For example, elevated levels of IL-6 may be related to depression severity instead of suicidal behavior, and genotypic differences may interfere with individual inflammatory responses.Reference Kim, Hong and Hwang 36 The relationship between SI and peripheral TNF-α levels remains unclear. Li et al found that the plasma TNF-α level equally increased in 33 suicidal and 19 nonsuicidal MDD patients compared with 64 healthy subjects. They suggested that the elevated levels of TNF-α may be associated with depression instead of suicide.Reference Li, Qi and Chen 37
To our knowledge, no study has compared working memory between patients with and without SI by using the n-back test. We found that adult depressed patients with SI had less correction and more errors in working memory 2-back task, compared to controls and depressed participants without SI. Participants with SI also had more commission and omission errors in Go/No-Go task, indicating their disinhibition and increased impulsivity. The results are consistent with those of previous studies, which evidenced the relationship between SI and cognitive deficits.Reference Westheide, Quednow and Kuhn 9 , Reference Kim, Kang and Cho 38 , Reference Kasckow, Youk and Anderson 39 Working memory comprises a range of cognitive and affective functions, including emotional regulation and optimal affective information processing.Reference Schmeichel, Volokhov and Demaree 40 MDD patients with a history of suicide attempts have been shown to perform worse on n-back task than nonattempters.Reference Keilp, Gorlyn and Russell 7 Future studies should generalize findings from the n-back test to other stimuli and from working memory to other cognitive processes. By using the Go/No-Go paradigm similar to that in the present study, Westheide et al found that depression inpatients with recent suicide attempts and current SI had more commission errors and fewer correct responses in the Go/No-Go task compared with patients without SI.Reference Westheide, Quednow and Kuhn 9 Cognitive deficits have been demonstrated to play a role in the progression from SI to suicide attempts.Reference Huber, Hodgson and Yurgelun-Todd 41 Moreover, our results may add to the evidence for impairments in inhibitory control in each stage of suicide: depression without SI, depression with SI, and suicide attempt.
The results of our study indicate that dysregulated proinflammatory cytokines, especially serum CRP levels, may play a crucial role in the identification of individuals at suicide risk. Therefore, patients with SI may be biologically distinguished from other subgroups of depressive patients. These results may have implications for the classifications of depression into more homogeneous groups and for developing different treatment strategies; future studies are warranted to investigate the efficacy of cytokine measures in suicide risk-assessment. For many individuals, suicidal behavior and SI denote impairment in executive function, which includes working memory and inhibitory cognitive control. A thorough understanding of this neuropsychological framework may suggest novel targets for treatment approaches to suicide, such as executive function training and problem-solving techniques.Reference Hopko, Funderburk and Shorey 42
Some study limitations should be addressed. First, our study results are limited by the cross-sectional design, and individuals with depression included were not medication‐free. The enrolled patients were receiving psychotropic medications during the cognitive function assessment and cytokine examination, which could have influenced the results, because some studies have suggested the anti-inflammatory properties of antidepressant drugs.Reference Li, Qi and Chen 37 , Reference Hiles, Baker and de Malmanche 43 Allowing the patients to continue medications was more ethical and may prevent relapse and recurrence. Because our study had a cross-sectional design, a causal link between proinflammatory cytokine levels and suicide risk could not be established. Second, given that both MDD and suicide are associated with emotion, the enrolled patients who were sufficiently calm to complete the assessment might have caused recruitment bias and an overestimation of cognitive deficits. Third, we only assessed neuropsychological domains using the go/no-go and 2-back tasks. Future studies should use other neuropsychological measures used in previous studies or include more than one measure to allow the comparison of findings. For example, people with major affective disorders have been shown to have extreme sensory processing patterns, assessed by using Adult/Adolescent Sensory Profile, and the involvement of altered sensory perception may be linked with higher depression, impulsivity, and hopelessness, which are major indicators of suicidal behavior.Reference Serafini, Gonda and Pompili 44 Fourth, the levels of proinflammatory cytokines, including TNF-α, may have diurnal variations.Reference Naidu, Morgan and Bailey 45 We did not collect blood samples of the participants at the same time of the day, which may confound our findings. Fifth, the participants’ SI in our study was not assessed through standardized scales that measure the attitudes, preparations, and behaviors to suicide, such as the Scale for Suicide Ideation and the Beck Scale for Suicide Ideation.Reference Beck, Kovacs and Weissman 46 , Reference Beck, Steer and Ranieri 47 Future studies may consider applying the abovementioned widely used scales when investigating suicidal behavior in order increase the validity of assessment, and to maintain comparability with previous work. Finally, information on the history, recency, and frequency of suicide attempts was not collected. Individuals with recent suicide attempts have been found to self-report more cognitive impairment than individuals with nonrecent suicide attempts.Reference Saffer and Klonsky 48 In addition, individuals with multiple suicide attempts have been reported to have higher comorbid health risks than those with a single suicide attempt or those without suicide attempts.Reference Rosenberg, Jankowski and Sengupta 49 The differences in cognitive function and levels of inflammatory cytokines between groups in our study may have thus been distorted.
Conclusion
In summary, the present study indicates that serum levels of CRP have distinct differences between MDD patients with or without SI. Higher levels of serum CRP and deficits in working memory and inhibitory control may be associated with higher SI among patients with MDD.
Acknowledgment
We sincerely thank the participants and the research assistants who made this study possible.
Author Contributions
Conceptualization: M.-H.C., Y.-M.B.; Data curation: M.-H.C., J.-W.H., K.-L.H., Y.-M.B.; Formal analysis: M.-H.H., M.-H.C.; Funding acquisition: Y.-M.B.; Methodology: M.-H.H., M.-H.C.; Supervision: M.-H.C., Y.-M.B.; Writing—original draft: M.-H.H.; Writing—review and editing: M.-H.C., J.-W.H., K.-L.H., S.-J.T., T.-P.S., Y.-M.B.
Disclosure
The authors have no conflict of interest and have no financial relationships relevant to this article to disclose.
Funding
The study was supported by grant from Taipei Veterans General Hospital (V103E10-001, V104E10-002, V105E10-001-MY2-1, V105A-049, V106B-020, V107B-010, V107C-181, and V108B-012), Yen Tjing Ling Medical Foundation (CI-109-21 and CI-109-22), and Ministry of Science and Technology, Taiwan (107-2314-B-075-063-MY3 and 108-2314-B-075-037). The funding source had no role in any process of our study.
