A composite peripheral blood gene expression measure as a potential diagnostic biomarker in bipolar disorder

In this study, Munkholm et al, out of the 19 genes they looked at, they have found that two genes (POLG and OGG1) were down regulated in bipolar patients versus normal controls. One gene (NDUFV2) was up regulated in depressed patients versus euthymic (in a normal state) patients. These genes are involved in mitochondrial function and DNA repair. very interesting indeed. Mitochondria are organelles which are the power houses of cells, providing all the energy a cell needs, through oxidative phosphorylation, in the form of adenosine triphosphate (ATP). If there is something wrong with mitochondria, then oxidative cellular energy production suffers. People with chronic fatigue syndrome are thought to have mitochondrial abnormalities. I can see how may be involved in bipolar disorder, as in mania, there is an overabundance of energy and in depression, the opposite. Not only that, but psychiatric symptoms are often seen in people who have mitochondrial diseases!

Of course, DNA repair is crucial for the survival of cells. If damage occurs in DNA and is not repaired, a mutation persists. Genes cannot function properly with mutations in them. Some mutations actually turn normal genes into oncogenes, meaning cancer causing genes. Many diseases are caused by single nucleotide mutations, for example cystic fibrosis, sickle cell anemia, and Tay Sachs disease are all caused by a change in one nucleotide of a gene. DNA repair can be a contributing factor to any disease.

These genes (POLG, OGG1, NDU1FV2) and their levels can be used as a composite biomarker for bipolar disorder. And studying the genes can also shed light as to the molecular mechanism of bipolar disorder.


A composite peripheral blood gene expression measure as a potential diagnostic biomarker in bipolar disorder


Gene expression in peripheral blood has the potential to inform on pathophysiological mechanisms and has emerged as a viable avenue for the identification of biomarkers. Here, we aimed to identify gene expression candidate genes and to explore the potential for a composite gene expression measure as a diagnostic and state biomarker in bipolar disorder. First, messenger RNA levels of 19 candidate genes were assessed in peripheral blood mononuclear cells of 37 rapid cycling bipolar disorder patients in different affective states (depression, mania and euthymia) during a 6–12-month period and in 40 age- and gender-matched healthy control subjects. Second, a composite gene expression measure was constructed in the first half study sample and independently validated in the second half of the sample. We found down regulation of POLG andOGG1 expression in bipolar disorder patients compared with healthy control subjects. In patients with bipolar disorder, up regulation of NDUFV2 was observed in a depressed state compared with a euthymic state. The composite gene expression measure for discrimination between patients and healthy control subjects on the basis of 19 genes generated an area under the receiver-operating characteristic curve of 0.81 (P<0.0001) in sample 1, which was replicated with a value of 0.73 (P<0.0001) in sample 2, corresponding with a moderately accurate test. The present findings of altered POLG,OGG1 and NDUFV2 expression point to disturbances within mitochondrial function and DNA repair mechanisms in bipolar disorder. Further, a composite gene expression measure could hold promise as a potential diagnostic biomarker.

In the present study, we investigated the expression of 19 candidate biomarker genes in the PBMCs in rapid cycling bipolar disorder patients longitudinally across different affective states and as repeated measures in healthy control subjects. We found downregulation of two genes, POLG and OGG1, in bipolar disorder patients compared with healthy control subjects after correction for multiple testing and adjusting for possible confounders. In comparisons between affective states, we found increasedNDUFV2 expression in a depressed state compared with a euthymic state. Further, a composite gene expression measure was constructed on the basis of individual gene expression levels and its discriminant capacity validated in an independent cohort. The composite gene expression measure for discrimination between bipolar disorder patients and healthy control subjects based on 19 genes generated an area under the ROC curve of 0.81 (P<0.0001) in sample 1, which was replicated with a value of 0.73 (P<0.0001) in sample 2. This corresponds with a moderately accurate test38 and surpassed that based on an abbreviated set of genes, identified by being more closely associated with a bipolar diagnosis.

OGG1 expression dysregulation is a novel finding in bipolar disorder. OGG1encodes the 8-oxoguanine DNA glycosylase, the primary enzyme responsible for the excision of 8-oxoguanine (8-oxodG), an oxidated DNA guanine nucleoside resulting from exposure to reactive oxygen species. In knockout mice, it has consistently been demonstrated that lacking an OGG1 repair system leads to increased accumulation of oxidative DNA lesions.39 Animal studies further suggest that OGG1 deficiency could increase susceptibility to neurodegeneration under conditions of increased oxidative stress.40Accumulation of oxidatively generated DNA damage has been associated with cardiovascular disease41 and diabetes,42 which are also associated with bipolar disorder. Further, oxidatively generated DNA damage may contribute to a shortened lifespan,43 also observed in bipolar disorder.44 Recently, we showed high levels of oxidatively generated damage to DNA in this cohort, for the first time demonstrating elevated levels of urinary excreted 8-oxodG in bipolar disorder patients through all affective phases (hypomania/mania, depression and euthymia) compared with healthy control subjects.21 It is thus possible that the OGG1 downregulation identified in the present study may lead to accumulation of oxidative DNA lesions and increased total levels of oxidatively generated damage to DNA, reflected by the observed high levels of 8-oxodG that was previously reported.21 The relationship between base excision repair and urinary excretion of oxidatively damaged nucleosides, however, is complex and incompletely understood,45 and a causal relationship cannot be established on the basis of our findings.

POLG downregulation in bipolar disorder has previously been demonstrated in lymphoblastoid cells;8 however, we believe our study is the first to demonstratePOLGdownregulation in PBMCs of bipolar disorder patients. Mutations in thePOLG gene encoding the catalytic gamma subunit of mitochondrial DNA polymerase cause multiple deletions or depletion of mitochondrial DNA alone or in combination and are associated with mitochondrial diseases with a wide range of clinical manifestations.46 Interestingly, transgenic mice with brain-specific expression of mutant POLG exhibit a phenotype resembling bipolar disorder with antidepressant-induced mania-like behavior and periodic activity related to estrous cycle in female animals.47 The mood-stabilizer valproate was additionally demonstrated to alter POLG gene expression in vitro.48Mitochondrial dysfunction has been linked with the pathophysiology of bipolar disorder49and clinically, high rates of comorbidity between mitochondrial disorders and bipolar disorder, with psychiatric symptoms often being the prominent and presenting feature of mitochondrial disorders.50 Mice expressing a proof-reading-deficient version ofPOLGdisplay features of accelerated aging and a shortened lifespan51 as well as gender-dependent hypertension,52 which is noteworthy considering that bipolar disorder is associated with cellular signs of accelerated aging53 and a high occurrence of cardiovascular comorbidity.54 Our finding of aberrant gene expression of POLG lends further support to a role for POLG in bipolar disorder pathophysiology.


NDUFV2 expression has not previously been described in PBMCs of bipolar disorder patients and state-related alterations of NDUFV2 specifically have not been investigated. The nuclear gene NDUFV2 encodes the NADH dehydrogenase (ubiquinone) flavoprotein 2a subunit of the mitochondrial complex I, which is involved in oxidative phosphorylation and proton transport. Several lines of evidence implicateNDUFV2 in bipolar disorder. NDUFV2 is located at 18p11, a reported susceptibility locus for bipolar disorder55 and polymorphisms in the upstream region of NDUFV2 have also been associated with bipolar disorder.56,57 Further, upregulation of NDUFV2 expression in postmortem brain samples from bipolar disorder patients compared with healthy control subjects have been described.58 Three studies have investigated NDUFV2expression in lymphoblastoid cell lines, with inconsistent findings of both downregulation ofNDUFV2 in bipolar I patients20, 57 and upregulation in bipolar II patients,20whereas one study found no differences between bipolar disorder patients and healthy control subjects.59 One possible reason for the discrepant findings may be that the previous studies included patients in various affective states, not having characterized the affective state of participants. Our finding of upregulation of NDUFV2 expression in a depressed state compared with a euthymic state could indicate that alterations ofNDUFV2 expression are state related, suggesting a possible role for NDUFV2 as a state biomarker.

Of note, our finding of upregulation of just one gene in primary analysis, thePGAM1, in bipolar disorder patients compared with healthy control subjects mirrored previous findings in lymphoblastoid cells.19

Our investigation of a composite gene expression measure yielded somewhat promising results. The likelihood ratios for the composite gene expression scores were overall modest (<3 and >0.3 for LR [+] and LR [], respectively), which indicates a relatively small effect on posttest probability corresponding to a limited value as a diagnostic test by itself. This indicated a relatively small shift in the probability of a correct diagnosis using the full gene set, however, not excluding a somewhat useful property for the test in certain situations. Choosing a cutoff on the composite measure that placed equal value on sensitivity and specificity, a sensitivity of 78% and specificity of 60% was obtained in the first sample with values of 62 and 75% in the second sample. Although the values obtained in the first sample are likely inflated by nature, the sensitivity and specificity values obtained in the second sample are comparable to tests in the other areas of medicine such as the prostate-specific antigen test for prostate cancer (sensitivity of 21% and specificity of 91%)60 and the MagStream HemSp fecal immunochemical test for the detection of colonic neoplasms (sensitivity of 23.2% and specificity of 87.6%).61The superior discriminant capacity of the composite measure based on the full set of genes as compared with the abbreviated set is indicative of the importance of including several individual potential biomarkers, which by themselves may contribute only discretely. Further, it is possible that the additional inclusion of laboratory values on a protein level, that is, inflammatory markers and markers of oxidative stress could increase the strength of the composite measure as a useful diagnostic test.

Our study benefitted from several methodological aspects. We applied careful standardization of blood sampling conditions, adhering to a short interval during the morning and obtaining samples in a fasting state. We further ensured blinding of laboratory staff to participant status and, crucially, we measured the expression of several candidate reference genes and evaluated their stability in contrast to previous studies7, 8, 12 that included only one reference gene, which is not recommended.62 We further used a split sample design in the evaluation of the full composite gene expression measure, allowing for testing this in independent samples. Finally, we assessed gene expression prospectively in patients during depressive, manic and euthymic states, which no other study has done.

Some limitations apply to the present study. First, the sample size was relatively small, and because not all patients experienced episodes of all polarities, the amount of between-subject variation relative to within-subject variation was therefore relatively large. Future studies should include larger sample sizes that would potentially allow for strict within-subject analyses and a further exploration of biomarker candidates to function in a personalized manner. Second, our findings primarily relate to mitochondrial function, which is influenced by lithium, mood-stabilizers and antipsychotics,63 although the direction and nature of the association is not uniform and knowledge about the effect of medication on gene expression in peripheral blood is limited. As the included bipolar disorder patients were medicated, we cannot entirely rule out the possibility that differences in the gene expression between bipolar disorder and healthy control subjects were due to, or at least partially explained by, an effect of medication. The effect of medication on OGG1 and POLG expression in bipolar disorder patients in vivo has not previously been investigated. POLG expression has been demonstrated to increase in vitro after valproate administration,48 potentially indicating, that the downregulation we observed was not due to mood-stabilizing medication. Findings of NDUFV2 expression in lymphoblastoid cells that are likely free of influence of medication are inconsistent, showing both elevated and decreasedNDUFV2 gene expression in bipolar disorder patients compared with healthy control subjects,20 and one small study (n=4) foundNDUFV2 upregulated after the administration of valproate but unaltered after lithium administration.20 The effect of medication on NDUFV2 expression is thus unclear, not giving specific indication as to the potential influence of medication on the finding of upregulated NDUFV2 expression in a depressed state compared with a euthymic state in bipolar disorder patients in the present study.

In comparisons between affective states within bipolar disorder patients, however, medication likely did not influence results to a large degree, as majority of the patients did not change medication during the study. Along these lines, exploratory analyses did not indicate an influence of medication on the composite gene expression measure discriminating between affective states in bipolar disorder patients. In future studies, it will be valuable to study unmedicated patients in comparison with healthy control subjects. However, for comparisons between affective states, it is likely not feasible to study unmedicated rapid cycling bipolar disorder patients longitudinally, due to the severity of illness. Third, the abbreviated composite gene expression measure was developed in the entire sample and the split sample design, therefore, did not constitute a genuine replication in the abbreviated gene set. Finally, the mean duration of illness for the bipolar disorder patients was relatively long and because neurobiological mechanisms potentially differ depending on the illness stage,64 findings may not be generalizable to all the bipolar disorder patients.

An issue that applies to studies investigating gene expression in peripheral blood in general pertains to the relationship between gene expression in the brain and that of peripheral blood. Although it is unclear to what extent peripheral blood gene expression patterns reflect those of the brain,65 peripheral blood cells express a large proportion of the genes in the human genome66 and a significant proportion of SNP-expression relationships are conserved between the brain and peripheral blood lymphocytes.67 The peripheral blood transcriptome may thus reflect system-wide biology and as such be a relevant tissue source for biomarker candidates. However, it is not clear whether it is a relevant surrogate tissue in relation to the brain.68

Candidate gene expression markers for the present study were selected a prioriusing a hypothesis-driven and transparent approach on the basis of previous gene expression findings and current hypotheses regarding the pathophysiology of bipolar disorder. The method involved combining potential biomarkers within multiple pathways in an effort to capture some of the complexity involved in the pathophysiology of bipolar disorder. Biomarker discovery in neurodegenerative69and medical disorders such as cancer,70diabetes and cardiovascular disease71have used both a hypothesis-driven and a hypothesis-free, data-driven approach. Although facing the challenge of identifying clinically meaningful biomarkers,72 a systems-based approach integrating hypothesis-free biomarker discovery and networks is, by itself, likely superior, given its ability to better interrogate the multivariate and combinatorial characteristics of cellular networks, that are implicated in complex disorders,73 and a combination of both data-driven methods and knowledge-based hypotheses-driven methods appear promising.74In this regard, our strictly hypothesis-driven approach could be considered a limitation.

In conclusion, our results suggest a potential for a composite gene expression measure as a diagnostic biomarker of bipolar disorder. In addition, we demonstrated aberrant regulation of the POLG, NDUFV2 and, for the first time, the OGG1 gene, pointing to disturbances within mitochondrial function and DNA damage repair mechanisms as pathophysiological mechanisms in bipolar disorder. The findings need replication in larger samples.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s