Background
Although the dynamics that contribute to a person dying by suicide are multifactorial, highly complex and vary between individuals [
1], studies consistently suggest that psychiatric disorders confer an increased risk of suicide. For example, a Danish study found that nearly 50% of people who died as a result of suicide either had been or were psychiatric inpatients [
2]. A retrospective study in Hong Kong discovered that 80% of the people who died by suicide had a psychiatric disorder, with 29% of these having a co-morbid diagnosis [
3]. This study also showed the highest risk of suicide was in people with mood disorders (50%), with suicides in psychotic disorders showing a gender difference (females: 35.1%; males: 19.8%). Most significantly, the study showed that a psychiatric diagnosis had an adjusted odds ratio of 28.67 for risk of suicide.
The concept that psychiatric disorders significantly increase the risk of suicide is highlighted by a systematic literature review which showed that, of all factors studied, psychiatric disorders had the strongest association with suicide (all psychiatric disorders 91%; of which mood disorders accounted for 55% and co-morbid diagnoses 44%) [
4]. A recent meta-analysis, which used data from both case–control studies and cohort studies, reported that the relative risk of suicide that could be attributed to any psychiatric disorder was 7.5% for males and 11.7% for females [
5]. Together, these data demonstrate that there is an urgent need to understand the drives to suicidal tendencies in people with psychiatric illnesses; the ultimate goals of such investigations being to identify markers for such tendencies and/or to develop interventions to reduce this behaviour.
There is now a body of literature which supports the hypothesis that neurochemical changes in the brain are associated with suicidal behaviour, which include roles for the central serotonergic system [
6], the signal transducer protein kinase C [
7] and the glutamatergic system [
8]. There are also studies linking the endogenous opioidergic system to the major psychiatric disorders [
9,
10] and suicide [
11]. In the case of the mu opioid receptor (OPRM), it has been reported that these receptors are increased in the inferior frontal, cingulate, postcentral, medial temporal and lateral occipitotemporal gyri as well as the thalamus [
12] of young (<41 years) people who had died as a result of suicide. OPRM were also reported to be increased in the frontal cortex and caudate [
13] from people who had died as a result of suicide compared to those from age and sex matched control subjects. Significantly, one of these studies [
12] only included people with depression among those who had died by suicide, whilst the second study predominantly consisted of people with undetermined diagnoses or depression (one subject had schizophrenia) [
13]. Thus, it is not clear if there is an interaction between psychiatric disorder and OPRM in people who died by suicide. Finally, in a study that did not assign psychiatric diagnoses to the cohorts, it has been reported that OPRM in the prefrontal cortex from people who died by suicide bound agonists with a higher affinity than that for people who did not die by suicide. However, a decreased binding affinity was reported for the pre- and post-central gyri from people who died as a result of suicide [
14].
Given the potential for changes in OPRM to be important in the predisposition of an individual to attempt suicide and the relative paucity of studies on OPRM, we decided to measure the density of OPRM using [3 H][D-Ala2, N-MePhe4, Gly-ol]-enkephalin (DAMGO) binding in BA24 from people with no history of psychiatric disorders (controls), people who had died by suicide and had no discernable history of psychiatric disorders (suicides) and people with schizophrenia, major depressive disorder (MDD) or bipolar disorder (BP); some of whom had died by suicide.
Following our initial studies we went on to measure [3 H]DAMGO binding in BA 9 and the caudate-putamen (CPu) from people with schizophrenia who had or had not died as a result of suicide and compared these data to those from age/sex matched controls (non-suicides) to begin to determine the extent of altered OPRM binding in these people. Finally, to determine if changes in radioligand binding were associated with changes in levels of OPRM protein, we used Western blotting and an anti-human OPRM antibody to measure protein levels in BA9, 24 and the caudate from the schizophrenia study cohorts.
Discussion
This study has shown decreased
3 H]DAMGO binding in BA24, BA9 and the CPu from people with schizophrenia who died by suicide which, in BA24, was first revealed as a significant decrease in
3 H]DAMGO binding in larger cohort where people who had died by suicide were compared to those who had not. Significantly there were no differences in
3 H]DAMGO binding in BA 24 from people with either MDD or BP who had died as a result of suicide, making this a specific association with suicide in schizophrenia. However, this study did show an increase in
3 H]DAMGO binding in BA24 from people with BP who did not died by suicide. Some caution is needed in the interpretation of this finding in BP because of the relatively small number of individuals making up this cohort, particularly given that lithium treatment has been shown to increase OPRM expression in rats [
34]. Although this novel finding deserves further investigation, it was not pursued in this study as it was not central to investigating the interactions between suicide x psychiatric disease.
Our study showed that
3 H]DAMGO binding increased with age, replicating previous studies that have reported this relationship [
12,
13]. Significantly, although DAMGO binds to all OPRM splice variants with similar affinities [
35] our data showed that levels of OPRM protein did not vary with age. Thus, our data is the first to indicate that the number of available
3 H]DAMGO binding sites increases with age in the absence of changes in levels of OPRM, which would be consistent with decreasing levels of endogenous ligand occupancy with age. This hypothesis is supported from previous data on endogenous opioids in rat CNS [
36] and in pain responsiveness in humans [
37]. Together, these data suggest levels of endogenous opioid ligands decrease with age.
We have shown that decreased
3 H]DAMGO binding is detectable across several CNS regions in people with schizophrenia who died as a result of suicide, even when the effects of confounding factors such as age and tissue pH are taken into account, with no difference in binding levels between controls and people with schizophrenia. Our data partly replicates a previous study that showed no changes in binding to OPRM in people with schizophrenia who “had died from….. natural causes” [
38] but contrasts with another report showing decreased binding to OPRM in the caudate from people with the disorder [
39]. The latter study used tissue from people who had died as a result of either suicide or pneumonia but did not explore whether there was an association between low levels of binding and suicide; thus, it is possible that the decrease reported was due to suicide. However, our finding of low
3 H]DAMGO binding density in suicide
per se does not agree with previous studies which report increased OPRM binding in the inferior frontal, cingulate, postcentral, medial temporal and lateral occiptotemporal gyri as well as the thalamus of young suicides [
12] and in the frontal cortex and caudate [
13] from people who had died by suicide or unaltered levels of radioligand binding in prefrontal cortex and prepost central gyri from people who suicided [
14]. Notably, in the earliest studies investigating suicide and OPRM, tissue was obtained from individuals who had died by suicide and predominantly had MDD or BP; thus, these data are in contrast with our findings of no difference in levels of OPRM binding in people with MDD or BP who had died by suicide. By contrast, our finding of no change in
3 H]DAMGO binding in people with mood disorders or no history of psychiatric illness who had died by suicide is consistent with a previous study [
14], which did not report on psychiatric diagnoses.
As stated earlier, our findings of decreased
3 H]DAMGO binding with no changes in OPRM protein would be consistent with a change in receptor availability rather than receptor number. There are three scenarios that would account for this decrease in agonist binding without a change in overall levels of protein. Firstly, it is possible that there is a change in the confirmation of the receptor which affects the binding of the ligand. Secondly, the biochemical modifications that are required for the insertion of the receptor into the membrane could be aberrant, reducing the availability of the receptor but not affecting the overall number of receptors. Finally, the simplest explanation is that the decrease in binding is due to competition for the binding site between the radioligand and an endogenous ligand. The latter hypothesis has been argued previously in the report of altered affinity for
3 H]DAMGO in tissue from people who died as a result of suicide [
14] as changes in binding affinities are usually due to competition from endogenous ligands [
40,
41]. More recently, a positron emission tomography (PET) study reported that people with high impulsivity scores released higher levels of endogenous opioids in response to stress [
42], resulting in lower
11C]carfentanil non-displacable binding potential. Impulsivity has been proposed as a candidate endophenotype for suicidal behaviour [
43]; thus, it is possible the decreased OPRM availability in people with schizophrenia who died by suicide is due to increased endogenous opioid levels in response to external stressors. Two such molecules, endomorphin 1 and endomorphin 2, have high affinities and selectivity for OPRM (see [
44]) making them prime candidates for future studies.
As with most approaches to the study of the neurobiology of psychiatric disorders, this study has some potentially confounding factors. In particular, we and others [
12,
13] have shown that
3 H]DAMGO binding increases with age. In our study the changes in
3 H]DAMGO binding remained when the effect of age was taken into account using an ANCOVA. Another confound is that all people with psychiatric disorders received psychotropic drugs prior to death. Overall, the literature suggests that treatment of rodents with antipsychotics results in either a decrease in the levels of OPRM [
45,
46] or no change [
47] although clozapine, which none of the subjects with schizophrenia received (see Additional file
1: Table S1 for details), has been reported to cause decreases [
48]. Evidence against the change in
3 H]DAMGO binding in tissue from people with schizophrenia who died by suicide being related simply to the effects of antipsychotic drugs are i) people with schizophrenia who did or did not die by suicide had been treated with antipsychotic drugs, receiving similar levels of final recorded drug doses expressed as chlorpromazine equivalents (Mann Whitney U = 131.0, medians: schizophrenia_non_suicide = 1216, schizophrenia_suicide = 1418;) and the difference in
3 H]DAMGO binding is present in all regions between these groups, ii) some of the people with BP had received antipsychotic drugs before death but there was no change in
3 H]DAMGO binding in people with this disorder who had died by suicide compared to control subjects and an increase in the BP_non_suicide group.
One potential methodological confound in our study was that the cohort size of people with schizophrenia who died as a result of suicide was relatively small. However, a power analysis (
http://www.biomath.info/power/ttest.htm) showed our study was appropriately powered in BA 9 and 24 (power level 0.8) but was underpowered for the mean differences observed in the CPu (9 additional non-suicide people were needed).
Competing interests
There are no competing financial interests in relation to the work described in this paper. TTM, GP & JN report no competing interests. The following authors have received remuneration in the past: ES received honorarium from Astra-Zeneca and travel support from GSK. BD received travel support from GSK, honorarium from Pfizer, Eli Lilly and MSD.
Authors’ contributions
ES was involved in the conceptualisation, design, supervision of research and analysis of the project and had the primary responsibility for drafting the manuscript. TTM undertook the radioligand binding and Western blots in the schizophrenia cohort and had input into the manuscript. GP and JN were responsible for the radioligand binding in the suicide cohort and had input into the manuscript. BD was involved in the conceptualisation, design and analysis of the project as well as having significant input into the manuscript. All authors read and approved the final manuscript.