Introduction

Bipolar disorder (BD) is defined by severe mood oscillations which alternate between mania and depression. The pathophysiology of BD is complex and presents many molecular and morphological alterations suggestive of impairment in cellular plasticity and resilience and also including a number of different factors such as oxidative stress, alterations to the expression of trophic factors, and inflammatory markers [1••]. A major target for the research community is to identify potential biomarkers that could be used for early diagnosis in patients susceptible to the disease and for its treatment [2••, 3]. Unfortunately, the current diagnosis of BD is based on information obtained by self-report and behavioral observations which lacks substantial biological validation and results in underdiagnoses and misdiagnosis therefore impacting treatment decisions [1••, 2••]. Due to the heterogeneity of BD, it is not expected to develop a general biomarker as patients in different stages (mania, euthymia, depression) present different profiles of the disease. Therefore, a set of different biomarkers that could identify the subgroups of patients could help to advance the treatment strategies and tailor a unique treatment for each stage of the disorder [4].

Biomarkers could improve diagnostic accuracy by indicating the specific features that are found in patients with BD and, more importantly, transform the treatment into a more personalized approach. Further, the use of biomarkers could help to identify the alterations found in patients with BD between manic and depressive stages and may also improve the treatment and quality of life for patients that have long been diagnosed with BD [1••, 2••, 3]. A recent report addressing the biomarkers network from the International Society for Bipolar Disorder (ISBD-BIONET) [1••] has provided powerful insights for the identification of possible biomarkers for BD. With this introduction in mind, we aimed to incorporate and critically evaluate the recently published literature regarding specific alterations to BD and the potential peripheral biomarkers that could reflect brain alterations in BD.

Potential Peripheral Biomarkers for Bipolar Disorder

Mental health research for BD has increasingly provided evidence that peripheral biomarkers are a very important approach that could provide many answers using simple tests, which are widely available and are relatively low cost. More important, considering the inaccessibility of the brain, the use of peripheral tissues such as blood, urine, or cerebrospinal fluid could present accessible markers that may be associated with brain alterations in BD. Upon reviewing the latest literature, four important areas are of particular interest in presenting potential biomarkers: mitochondrial dysfunction and oxidative stress, trophic factors, inflammatory cytokines, and urinary metabolites. Within these topics, focus has been placed on factors that have been consistently linked to BD.

Mitochondrial Dysfunction and Oxidative Stress

The mitochondrial electron transport chain (ETC.) is the principal source for the generation of reactive oxygen species (ROS) in mammalian cells [5]. Oxygen is involved in vital biological processes including the production of energy through oxidative phosphorylation [6]. Therefore, dysfunction of the ETC., which is consistently reported in BD [7••], is associated with increased levels of ROS production. When ROS production overwhelms the ability of the endogenous/exogenous antioxidant system, a state known as oxidative stress takes place [8]. In this state, the macromolecules including proteins, lipids, and DNA are targets for oxidative damage, which can affect the membrane fluidity, protein functionality, and DNA structure [9, 10]. These alterations to proteins, lipids, and DNA induced by oxidative stress are widely reported using post-mortem brain samples [11, 12, 13•, 14] and peripheral blood cells as well [13•, 15–17] from patients with BD, suggesting that mitochondrial dysfunction and oxidative stress play an important role in the pathophysiology of BD.

Upon reviewing the literature regarding microarray studies in BD, Scola et al. (2013) reported important alterations to the subunit levels of the mitochondrial ETC. The expression levels of the subunits NDUFV1, NDUSF1, NDUFS7, and NDUFS8 of complex I were found to be reduced in BD. This reduction may suggest that patients with BD are prone to deregulation of the ETC., which could increase the production of ROS [7••] leading to oxidative stress. Interestingly, a different pathway was found to be altered in the ETC. in schizophrenia (SCZ), in which the subunits altered in complex I are responsible for the proton pump as opposed to the electron transfer as found in BD [7••]. For this reason, we do not find the same findings regarding the oxidative stress parameters in BD and SCZ. The findings regarding reduced expression levels for the NDUFS7 were previously supported in BD [3, 13•].

Oxidative damage to proteins includes the formation of 3-nitrotyrosine and increased protein carbonyl content. 3-Nitrotyrosine is an end product of nitrosative damage to a tyrosine residue induced by peroxynitrite/carbon dioxide-derived radicals. Equally, the formation of carbonyl content occurs when oxygen or peroxide radicals react with amine groups through a metal-cation-catalyzed reaction [2••]. Kapczinski and collaborators reported that increased serum carbonyl levels were also found in both manic and depressive stages of BD [18]. Andreazza et al. (2010 and 2013) showed that oxidative damage to proteins was also found in the post-mortem prefrontal cortex (PFC) of patients with BD, where elevated carbonyl levels were found in the synaptosomes, elevated 3-nitrotyrosine levels in the mitochondria, and globally elevated carbonyl group levels. More recently, Kim et al. found that increased oxidative and nitrosative modifications to dopamine-rich areas are present in the post-mortem PFC samples of patients with BD using immunohistochemistry and acceptor photobleaching Förster resonance energy transfer technique [19]. Another recent study reported that serum superoxide dismutase (SOD) levels, an endogenous antioxidant enzyme, were lower in patients with BD when compared to healthy controls [20]. Therefore, these findings suggest that oxidative damage to proteins could be important to the physiopathology of BD; however, less evidence has been found to support the use of protein alterations as a specific biomarker.

In a meta-analysis examining oxidative stress markers in BD, Brown et al. (2014) reported that lipid peroxidation showed a significant change in BD. Lipid peroxidation is a process initiated by the reaction of ROS; the initial phase of lipid peroxidation generates lipid hydroperoxides (LPH) that can be detoxified by the enzymatic antioxidant system. However, if the lipid peroxidation cascade is not stabilized, LPH reacts with other lipid molecules to form non-radical species including malondialdehyde (MDA), 4-hydroxy-2-nonenal (4-HNE), 8-isoprostane (8-ISO), and acrolein [21, 22]. Considering that lipids represent around 70 % of the main component of white matter, this oxidative alteration may play an important role in the pathophysiology of BD [2••]. Oxidative damage to lipids was also found higher in patients with BD in the myelin through the measurement of 4-hydroxynonenal and 8-isoprostrane [14]. Banerjee et al. also found higher serum levels of lipid peroxidation in patients with BD using the TBARS method [23]. More recently, plasma MDA levels were found to be higher in patients with BD when compared to healthy controls [20]. Interestingly, another study by our group through multivariate multiple regression analyses revealed that LPH correlates with brain white matter abnormalities [24•], supporting that lipid peroxidation may be an important biomarker for BD.

Additionally, nitric oxide (NO) levels and DNA/RNA damage also present an important correlation with BD [2••]. Several studies reported that plasma/serum NO levels were found to be higher in patients with BD compared to healthy controls [20, 25, 26] and reduced in pretreated patients with BD [27]. These findings could support the use of this molecule as a potential biomarker; however, more studies are required to substantiate this hypothesis.

Another important finding to be discussed is that oxidative stress could be responsible for epigenetic modifications in BD. Alterations in 5-methylcytosine levels have been reported at the BDNF promoter and the COMT gene in patients with BD [28, 29]. Global methylation levels were also found to be decreased in patients with BD and their relatives when compared to healthy controls [30]. Also, Scola and collaborators reported that mitochondrial dysfunction may lead to alteration on methylation and hydroxymethylation levels in cortical primary neurons [31]. Oxidative damage to guanosine, measured by 8-hydroxy-2-deoxyguanosine levels (8-OHdG), was found to be higher in patients with BD when compared to healthy controls, and, importantly, it was found to be positively correlated to the number of manic episodes [32]. The same study also found that patients with BD presented decreased global 5-methylcytosine (5mc) levels similarly to the findings reported by Huzayyin and collaborators (2014). In addition, Che et al. [6] found that oxidative damage to RNA was also increased in the post-mortem hippocampus of patients with BD [12]. Once again, the findings suggest that oxidative damage to DNA/RNA may be related to the pathophysiology of BD. Furthermore, alterations to DNA conformation and production of aberrations such as 8-OHdG and 5mc may be a potential biomarker for BD; however, more studies are required to verify the difference of expression of these alterations to DNA/RNA in the different stages of the illness.

Though many disorders present mitochondrial dysfunction, several different pathways can be involved, and subsequently, the findings in this section are more specific to BD. Together, the findings highlighted above suggest that oxidative damage specifically for lipid peroxidation and DNA is a potential peripheral marker and may be associated with brain alterations in BD.

Trophic Factors

Neurotrophins (NT) and other trophic factors are responsible for the regulation of cell dynamics and are expressed in the brain and in the peripheral tissues in a region-specific manner. Neurotrophins, such as brain-derived neurotrophic factor (BDNF), and other trophic factors including insulin-like growth factor-1 (IGF-1) and vascular endothelial growth factor (VEGF) are capable of signaling neurons, glial cells, and other cellular systems to enable survival, differentiation, and growth [33–35]. The abovementioned NTs were selected specifically in order to explore the unique alterations found in BD. Attention has been focused on gaining a better understanding of the role these factors play in the different stages of BD and their potential as biomarkers. For a detailed review of trophic factors, refer to Scola and Andreazza (review submitted).

Most published reports on BDNF have suggested a strong correlation between low levels of this NT and BD [36–39]. Recently, a study describing two different cohorts (Sahlgrenska and Karolinska sets) reported the ratio of proBDNF to mature BDNF levels was higher in patients in comparison to healthy controls; however, serum proBDNF levels were found to be lower in patients when compared to controls, suggesting the alteration in the conversion of pro to mature BDNF may be associated with the pathophysiology of BD [40••]. These findings suggest that the decrease in the levels of global/mature BDNF found may impair long-term memory and may be less effective in supporting normal cell regulation, both of which are found to be altered in BD.

Similar to BDNF, IGF-1 and VEGF presented a strong association within the pathophysiology of BD. Drug treatment has an important regulatory effect on the expression levels of both trophic factors [41, 42]. Peripheral levels of IGF-1 were lower in patients with BD compared to patients with schizophrenia [43]. Studies have shown a potential effect of IGF-1 signaling cascade in regulating stress, depression, and hippocampal neurogenesis [43, 44]. Therefore, the findings provided regarding the effects of IGF-1 suggest the importance of this NT in the pathophysiology of BD and its potential to be used as a biomarker should be explored in greater detail.

VEGF has an important role in regulating neuron dynamics, and it is also implicated in neuronal protection and cellular differentiation [45, 46]. Studies have shown a correlation between higher levels of this factor and major depressive disorder [47, 48], while lower levels of VEGF may be linked with suicidal ideation [49]. Moreover, plasma VEGF levels were found to be increased in patients in manic [50] and depressive [51] stages; however, lithium treatment was able to decrease VEGF mRNA levels. The data shows that VEGF may be altered in a state-dependent manner in patients with BD, suggesting this factor may be associated with BD.

Alterations in trophic factor levels in the brain and periphery in BD may have an important pathophysiological implication, which could lead to the morphological alterations found in patients with BD. Abnormalities of the PFC white matter in patients with BD were already reported by our group [24•], and these alterations may be, in part, related to dysregulation of these trophic factors. Further studies will be required to elucidate and support that trophic factors may be an important tool for the diagnosis of BD.

Inflammation

Some studies have shown the importance of inflammation in the brain and periphery in the pathophysiology of BD [52–54]. A possible cause of the alterations in the inflammatory cytokines in this disorder could be related to the chronic activation of the immune system during the manic or depressive episodes [55]. However, just a few studies have described assays of chemokines in subjects with BD [54]. The deregulation between the different inflammatory markers may be related to the alterations found in this disorder [53]. Understanding the role of these inflammatory factors in the different stages of BD could lead to the development of inflammatory biomarkers.

Most recently published evidence by Rao et al. and Brietzke et al. showed that pro-inflammatory factors (interleukin (IL)) including IL-1, IL-2, IL-4, and IL-8 were increased in peripheral blood and port-mortem brain of subjects with BD [56, 57]. Other important inflammatory markers found to be consistently increased in patients with bipolar disorder include IL-6, IL-1β, and tumor necrosis factor alpha (TNF-α) [52, 58, 59]. Supporting these studies, Munkholm and collaborators compiled a meta-analysis examining studies regarding peripheral samples, where IL-4 and TNF-α were increased in patients when compared to healthy controls, identifying their possible role as biomarkers for BD [55]. However, inflammatory responses are dependent on mood states and the findings are still controversial [55, 58]. Therefore, more studies regarding the examination of these factors in the different stages and their relationship with the illness will be necessary [54]. Nevertheless, studies as of yet have shown that peripheral inflammation is increased in the manic state [57, 58].

Moreover, studies have shown that IL-13, apolipoprotein A1, and TNF-α are strongly correlated with BD [60]. Also, Goldstein et al. reported that high-sensitivity C-reactive protein (hsCRP) is correlated with manic symptoms in adolescents with BD. In addition, IL-6 levels were also found to be negatively correlated with BDNF levels [61•], suggesting that inflammation could be associated with the progression of illness [1••]. A number of abnormalities observed in BD could be explained by alterations in the expression of these cytokines, since they were found to play an important role in the mitochondria [62] leading to an increased production of oxidative stress in the brain [63]. Another important target of abnormal cytokine expression is the catecholamine system, where TNF-α and IL-1Ra were found to disrupt norepinephrine and dopamine signaling, which could lead to mood oscillations through alterations of the neurotransmitter levels [58]. Similarly, the mechanism of action of these cytokines remains to be fully understood, since all current evidence suggests that inflammation has an important role in BD. Importantly, Alsaif et al. (2012) reported that studies using plasma and serum presented important differences which could impact the identification of biomarkers using multiplex immunoassay analysis. The study also reported that in the BD group, six proteins were significantly changed in serum and ten in plasma, and it emphasized that each biofluid can provide independent information regarding the biological pathway targeted [64]. This area of research may open up more venues for the development of biomarkers and may also provide potential therapeutic targets for existing and novel drugs for BD treatment.

Urinary Metabolites

Metabolomic approaches, such as gas chromatography-mass spectrometry (GC-MS), to examine potential biomarkers have been increasingly used to capture disease-specific metabolic signatures [65]. Studies evaluating possible markers in the urine of patients with BD have shown that metabolites including choline, α-hydroxybutyrate, N-methylnicotinamide, and isobutyrate differed from healthy controls [66]. In addition, another study showed that a single compound, 2,4-dihydroxypyrimidine, may be a validated biomarker in BD [67].

Neuroimaging

Improvements in neuroimaging techniques and image analysis have allowed a better understanding of the neuronal changes in BD. Moreover, these techniques, which include diffusion tensor imaging (DTI), functional magnetic resonance imaging (fMRI), and proton magnetic resonance spectroscopy (H-MRS), are effective tools in examining and identifying changes in the brain anatomy which subsequently help in diagnostic and prognostic processes as well as allowing for the determination of potential biomarkers [1••, 68].

Recent findings in the literature have shown that euthymic patients with bipolar disorder present abnormalities in the prefrontal white matter through DTI [24•]. Interestingly, the data obtained from this study revealed that the white matter abnormalities are correlated with the peripheral lipid peroxidation levels, highlighting the importance of this technique for the diagnostics of BD. Another recent review examining the data of late-onset disease concluded that the current findings in the literature do not support consistent differences between early and late onset in BD [68], suggesting that more studies are required to elucidate these differences. However, Huang and collaborators reported that elderly patients with early- and late-onset mania presented a substantial reduction in various cortical regions [69]. Another study using DTI reported abnormalities in gray matter, where it was found to be reduced in different structures of the brain of patients with BD [70]. Moreover, Phillips and Swartz reported that BD presents abnormalities in the neural circuits regarding emotion processing, emotion regulation, and reward processing [71•]. For more information regarding neuroimaging studies, refer to Frey et al. (2013) and Phillips and Swartz (2014).

Nonetheless, there are still some gaps in the literature regarding neuroimaging studies; no longitudinal studies were found using different points in the course of the illness and the examination of treatment response [1••] and neural circuitry [71•]. Together, the findings suggest that neuroimaging techniques are powerful tools that could be used to search for functional modifications in the neuroanatomical structures among children, adolescents, and adults in the early and late onset of the different stages of BD. Further studies are encouraged to examine potential markers that may be associated with peripheral biomarkers.

Genetic Findings

Most published findings suggest an association with the BDNF, COMT, and 5-HTT genes, which is also found within other psychiatric disorders. Using the genome-wide association studies (GWAS) and whole-genome sequencing in BD, the strongest evidence related to the pathophysiology is the CACNA1C and ANK3 polymorphisms [1••, 72–74]. Moreover, CACNA1C has been implicated with alterations of hippocampal, pregenual anterior cingulate cortex, and dorsolateral PFC functions, suggesting a dysfunctional regulatory circuit including limbic and prefrontal brain regions in patients with BD [75]. Through a meta-analysis, Nurnberger and collaborators (2014) reported that CACNA1C, DTNA, FOXP1, GNG2, ITPR2, LSAMP, NPAS3, NCOA2, and NTRK3 genes differed in their expression levels in the dorsolateral PFC in patients with BD. The same study suggests that pathways underlying the predisposition to BD include alterations to the calcium channels, glutamate signaling, hormonal regulation, and second messenger systems [72]. Also, genetic polymorphisms in the progranulin gene (GRN), which is related to fronto-temporal lobar degeneration, were found to be involved in many pathophysiological processes in the brain and periphery and were also found to be decreased in the plasma of patients with BD [4]. Importantly, it was found that drug-free manic patients presented downregulation in the expression levels of miRNA [76], which has an important role in inflammation and gene expression. These findings tend to reinforce specific hypotheses regarding the pathophysiology of BD and may provide new venues for new approaches to treatment and moderation of the illness progression. For more information regarding genetic studies in BD, refer to Frey et al. (2013).

High-Risk Studies

Another important topic that deserves mention concerns the offspring of probands with BD, which present an increased risk of developing BD or any major affective disorder [77, 78]. A recent meta-analysis of family high-risk studies from Rasic and collaborators (2013) addressed that familial transmission of risk is partially diagnosis specific and, more importantly, of higher risk than previously thought [79]. Dr. Duffy also states that high-risk studies have contributed to understanding the heterogeneity of BD and suggests that more studies are necessary to refine the characterization of the early stages of BD [78]. In an overview, high-risk family studies are very important for identifying the subphenotypes and also to better understand the cohort-specific genetic risk markers. In accordance with these statements, it is necessary to highlight the necessity of studies that could address and identify early clinical stages in the development of BD as well as to determine the molecular targets involved in this process. This information could have a tremendous role in the understanding of the pathology and the development of stage-specific treatments with illness onset and progression.

Conclusions

In summary, this review aimed to provide an update of the potential biomarkers which may enhance diagnosis accuracy, to examine the progression of the illness, and to improve the treatment for BD. The main findings highlighted in this review were as follows: (1) oxidative damage, specifically lipid peroxidation and DNA, is a potential peripheral marker; (2) BDNF may be used for diagnosis as well as observation of illness progression; (3) IL-4 and TNF-α were identified as possible inflammatory targets; (4) 2,4-dihydroxypyrimidine metabolite may be a validated urinary biomarker; (5) CACNA1C and ANK3 gene polymorphisms are the main findings of genetic studies of patients with BD; (6) BD presents abnormalities in the prefrontal white matter which is related with the peripheral lipid peroxidation levels and also presents abnormalities in the neural circuits regarding emotion processing, emotion regulation, and reward processing; and (7) high-risk family studies could identify the early clinical stages and also determine the molecular targets involved in BD. These findings emphasize the complexity of BD yet also present possible approaches that may help disease diagnosis and improve the development of novel therapies on those targets. Importantly, future studies are required to elucidate the usefulness of the potential biomarkers described in this review.