Early diagnosis of bipolar disorder

Abstract

Bipolar disorder (BD) is a severe mood disorder characterized by recurrent episodes of mania and depression, and it is prone to delayed diagnosis, which can lead to worsened outcomes, including more frequent mood episodes, greater functional impairment, and comorbidities. Early diagnosis of BD remains a significant challenge, although recent advances offer promising insights, such as research in molecular biomarkers, neuroimaging, exosomes, genetics, and epigenetics. This mini-review highlights their potential for providing earlier, more accurate identification of BD and discusses the underlying reasons why current research has not yet succeeded. For instance, the high heterogeneity of symptomatic presentations leads to low consistency in study participants; delayed BD diagnosis results in the inclusion of potential BD patients in the depression group; low specificity of biomarkers stems from limited understanding of BD pathophysiology; and there is a possibility that BD is not innate but develops over the course of the disease. Deepening our understanding of BD pathology, identifying more specific biomarkers, and integrating multiomics approaches for validation studies in well-defined homogeneous cohorts hold promise for significant breakthroughs.

Key Words: Bipolar disorder; Early diagnosis; Major depressive disorder; Biomarker; Exosome

Core Tip: Bipolar disorder (BD) is a severe mood disorder with delayed diagnosis, leading to worsened outcomes such as frequent mood episodes and comorbidities. Early diagnosis remains challenging, despite advances in biomarkers, neuroimaging, and genetics. This review explores these advances and highlights issues like symptom heterogeneity, delayed diagnosis, low biomarker specificity, and the evolving nature of BD. A deeper understanding of BD pathology, more specific biomarkers, and multiomics validation in homogeneous cohorts could lead to significant breakthroughs in early BD diagnosis.

INTRODUCTION

Bipolar disorder (BD) is a multifaceted psychiatric condition characterized by alternating episodes of depression and mania/hypomania, with a global incidence rate of 2%[1]. The first onset of BD most commonly occurs around the age of 15 years, with the median age of onset being ~20 years[2]; both of which are crucial stages for individual learning and development. The chronic course of BD and its high comorbidity rate are linked to increased morbidity and mortality, making it one of the leading causes of disability among young and working-age individuals[3]. Early diagnosis and optimal intervention are critical, as the response to treatment is more effective in the early stages[4]. However, the average delay in clinical diagnosis of BD is 9 years[5]. Approximately 50% of BD patients initially present with depressive symptoms, making the clinical manifestations closely resemble those of major depressive disorder (MDD)[6]. In addition, BD can vary greatly in terms of intensity and duration among infected individuals. This nature of BD contributes to the complexity of its clinical presentation and poses challenge for its early diagnosis. Despite advances in understanding the early diagnosis of BD, significant limitations remain. This article reviews the progress in research on the early diagnosis of BD and its limitations.

CLINICAL PRESENTATION AND DIAGNOSTIC CRITERIA

The concept of BD has evolved over time, beginning with Aretaeus of Cappadocia's early suggestion of the homogeneity between depression and mania[7], followed by the introduction of the concept of manic–depressive psychosis by Kraepelin[7], and later updates in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR) and the International Classification of Diseases, 11^th Revision (ICD-11). According to the DSM-5-TR, the criteria for BD require at least one episode of mania or hypomania. Mania is characterized by an expansive, elevated, or irritable mood lasting at least 1 wk, with significant impairment that often necessitates hospitalization (bipolar I disorder). Hypomania, which defines bipolar II disorder when paired with a previous depressive episode, is a milder form of mania, requiring at least 4 d of elevated mood and increased energy. The ICD-11 offers more flexibility in defining hypomania, no longer requiring it to last several days. A key distinction between the ICD-11 and DSM-5-TR lies in their approach to mixed symptoms. In DSM-5-TR, the criteria are expanded to include subsyndromal mixed symptoms – defined as at least three counterpolar symptoms –during manic, hypomanic, or depressive episodes. In contrast, ICD-11 maintains that mixed symptoms represent an episode, requiring several prominent symptoms from the opposing mood state; a less stringent criterion that aligns more closely with Kraepelin's broader view of mixed states, as shown in Figure 1 for symptomatic expressions of BD subtypes. However, traditional diagnostic tools largely rely on clinical assessments and patient history, which can be subjective and affected by recall bias. Therefore, objective, reliable biomarkers are urgently needed to help differentiate BD from other mental conditions at an early stage.

Figure 1 Expression of mood states in different subtypes of bipolar disorder according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision. The symptoms of mania include elevated, expansive, or irritable mood, grandiosity, inflated self-esteem, racing thoughts, decreased need for sleep, increased talkativeness, increased sexuality, and distraction. The symptoms of depression include low mood, loss of interest and motivation, sadness, low self-esteem, feelings of guilt, concentration difficulties, sleep difficulties, loss of appetite, hopelessness, and suicidal ideation.

INSIGHTS DRAWN FROM PERIPHERAL BIOMARKERS

Peripheral biomarkers, which can be detected in blood, saliva, or other bodily fluids, offer a noninvasive means of identifying BD. One promising area of research focuses on inflammatory biomarkers. Studies have shown that proinflammatory cytokines are elevated in BD patients, particularly during manic and depressive phases[8]. These findings suggest that inflammation may be associated with the acute phases of BD, and monitoring these markers could aid in early diagnosis and management[9]. For instance, a systematic review and meta-analysis found that levels of interleukin (IL)-7 were significantly decreased in BD, while IL-9, C-C motif chemokine ligand (CCL)3, CCL4, CCL5, and CCL11 were significantly increased in BD compared to MDD, suggesting a role of inflammation and T-cell network dysregulation in these disorders[10]. Another study highlighted the differential expression of cytokines such as IL-4, which was found to have diagnostic value in distinguishing between active depressive episodes in MDD and BD[11]. A study on peripheral mitochondrial DNA suggested its potential as a neuroinflammatory biomarker for MDD, which could help distinguish it from BD[12]. Finally, a real-world retrospective study analyzed various hormones and inflammatory markers, finding that MDD patients had higher levels of IgA and IgM, while BD patients had elevated neutrophil and monocyte counts, providing a basis for differentiation[13]. Metabolic profiling also offers insights into the differentiation of BD from MDD. A study utilizing metabolomics-based urinary biomarker models identified distinct metabolic signatures for MDD and BD, with specific metabolites showing potential as diagnostic markers[14]. Similarly, serum metabolic profiles assessed through nuclear magnetic resonance spectroscopy revealed differences in metabolites such as pyruvate and pantothenic acid, which could serve as potential biomarkers for distinguishing BD from MDD[15]. Changes in insulin-signaling-related proteins have been observed in both manic and mixed mood states, suggesting their potential as diagnostic markers for distinguishing BD from MDD[16]. The exploration of homocysteine levels as a potential biomarker for BD has also been a focus of research. Although some studies suggest that high homocysteine levels may be a risk biomarker for BD, the association between homocysteine and MDD reduces the value of homocysteine as a diagnostic marker for BD. Further well-powered studies are needed to clarify its utility as a biomarker for early diagnosis[17]. Additionally, machine learning (ML) approaches have also been used to integrate peripheral biomarkers with clinical data, enhancing diagnostic accuracy. For example, an AI algorithm combined with RNA editing-based blood biomarkers demonstrated high accuracy in discriminating MDD from depressive BD in an external cohort, underscoring the potential of integrating biomarker data with advanced computational techniques for differential diagnosis[18].

EXOSOMES: A SOURCE OF PROMISE

Exosomes, small extracellular vesicles released by cells, carry molecular signatures reflective of their cell of origin, including proteins, lipids, and RNAs. The advantages of exosomes include their stability in bodily fluids, ability to cross the blood–brain barrier, and ease of isolation from blood samples[19]. For instance, brain-derived exosomal miRNA profiles have been analyzed to understand their correlation with peripheral exosomal miRNA, offering insights into the pathological processes of BD and the potential for early diagnostic assays[20]. Similarly, research into plasma extracellular vesicles enriched for neuronal origin has demonstrated their utility in providing a sensitive and accurate basis for biomarker discovery in neurological disorders, further supporting the concept of exosomes as a window into brain pathological processes[21]. Moreover, the exploration of noncoding RNAs, such as miRNAs and long noncoding RNAs (lncRNAs), within exosomes has gained attention. These molecules have regulatory roles in gene expression and are implicated in the pathophysiology of mood disorders. Exosomal miRNAs and lncRNAs could provide insights into the molecular mechanisms underlying BD and MDD, offering potential biomarkers for differential diagnosis[22]. In addition, the exploration of brain insulin resistance in adults with bipolar depression using extracellular vesicles of neuronal origin has provided evidence that these vesicles can reflect changes in insulin signaling pathways, which are implicated in BD pathogenesis. This underscores the potential of exosomal content to reveal underlying neurobiological mechanisms and guide therapeutic interventions[23], although this area of research remains underexplored. Collectively, these studies illustrate the promising role of exosomal content in capturing neurobiological changes associated with BD, offering a valuable tool for advancing our understanding and management of the disorder.

INSIGHTS DRAWN FROM NEUROIMAGING

A large-scale meta-analysis of functional neuroimaging studies has identified condition-dependent differences in brain activation and connectivity in individuals with BD[24]. These studies have highlighted functional differences in specific brain regions, such as the prefrontal, parietal, and limbic areas, which are crucial for cognitive and emotional processing[24], providing a framework for identifying reproducible neural biomarkers that can guide diagnosis and treatment. The ENIGMA Bipolar Disorder Working Group has been instrumental in conducting large-scale neuroimaging studies that reveal widespread patterns of lower cortical thickness, subcortical volume, and disrupted white matter integrity associated with BD. These studies have provided further insights into the pathophysiological mechanisms of BD and thus help advance early diagnosis of the disorder[25]. Additionally, systematic reviews of neuroimaging studies have explored brain gyrification patterns in BD, revealing altered patterns that may reflect early neurodevelopmental processes and suggesting that brain gyrification could serve as a potential biomarker for optimizing the prediction and diagnosis of BD, particularly in its early stages[26]. Precision neuroimaging biomarkers are also being developed to offer personalized insights into brain alterations associated with BD. Techniques such as ML applied to neuroimaging data have been used to differentiate patients with BD from healthy individuals, providing mechanistic insights into the pathophysiology of the disorder. These advances underscore the potential of neuroimaging to contribute to precision medicine approaches in BD[27]. Endophenotypic studies that directly compare imaging differences in key brain regions across BD, MDD, and healthy populations could play a crucial role in identifying BD-specific alterations, thus contributing to early and accurate diagnosis. One of the key developments in this area is the integration of multimodal neuroimaging techniques, such as combining behavioral assessments or genomic approaches with magnetic resonance imaging (MRI) features[28,29], which has been shown to enhance diagnostic accuracy in adolescents at risk for BD.

INSIGHTS DRAWN FROM GENETICS

The advancement of genetic testing technology has significantly promoted the genetic research of mental disorders. The identification of genetic markers and the elucidation of the genetic architecture of BD have provided fresh insights into the disorder. Genome-wide association studies have been instrumental in uncovering numerous genetic loci associated with BD, which are crucial for understanding its pathogenesis and for developing early diagnostic tools[30,31]. These studies have revealed that BD is highly polygenic, with a substantial overlap in genetic risk factors with other psychiatric disorders, emphasizing the complexity of its genetic underpinnings[32]. For instance, research into the genetic association between personality traits and mood disorders has revealed that polygenic scores for neuroticism are significantly associated with MDD, while extraversion scores are associated with BD. This suggests shared genetic risk factors for neuroticism and MDD, and for extraversion and BD, providing another layer of genetic differentiation between these disorders[33]. The application of polygenic risk scores (PRSs) has also been investigated as a means to differentiate BD from MDD. A study found that PRSs for BD and MDD were associated with cortical alterations in different brain areas, suggesting that genetic risk for these disorders might affect neurodevelopmental processes. These findings highlight the potential of PRSs as a tool for distinguishing between BD and MDD based on genetic susceptibility[34]. BDNF gene expression has been highlighted for its capacity to distinguish between the two disorders, as it plays a fundamental role in brain function and mood regulation[35]. Another study explored the association of ANK3 variants with BD in the Korean population, suggesting that certain haplotypes may confer susceptibility to BD. This genetic association underscores the potential for specific genetic markers to serve as endophenotypes for BD, aiding in its differentiation from MDD[36]. The exploration of endophenotypes, which are heritable traits associated with BD, has been a promising approach to reduce phenotypic heterogeneity and improve our understanding of the neurobiology of the disorder. Identifying endophenotypes such as circadian rhythm instability and neuroimmune dysregulation can aid in the early detection of BD[37]. These insights underscore the importance of continued research into the genetic and biological foundations of BD to develop more effective diagnostic strategies. Additionally, as mentioned above, integrating genetic data with other parameters, such as neuroimaging and clinical data, associates genetic variations with brain structure and function, providing insights into the neurobiological mechanisms of BD[25]. These findings are crucial for early diagnosis, as they help identify individuals at risk before the full onset of hypomanic or manic episode.

CLINICAL BEHAVIORAL AND SYMPTOM PREDICTION MODELS

Behavioral and clinical data analysis using advanced computational techniques has shown promise in diagnosis of BD. ML approaches, including support vector machines and random forests, have been increasingly applied to differentiate between psychiatric disorders and identify predictors for early onset of BD[38,39]. For instance, engagement of ML techniques in analyzing electroencephalogram data helps to differentiate between BD and MDD, as well as other psychiatric conditions[38]. The integration of passive sensing and deep anomaly detection in a contactless cohort study aimed to understand individual illness trajectories and predict relapses in BD. This approach involves continuous monitoring of physiological and behavioral data, which can provide insights into the dynamic changes associated with episode generation in BD[40]. Use of smartphone data to predict and monitor symptoms in patients with depression, including those with comorbid BD, has been explored. Behavioral markers derived from smartphone usage, such as screen-off events and communication patterns, have been correlated with changes in depressive states. This method of utilizing digital behavioral markers offers a supplementary tool for clinical evaluations, potentially aiding in the detection and monitoring of mood disorders[41]. These behavioral and psychological parameters associated with manic or hypomanic episode could aid in early diagnosis of BD. Lastly, the integration of clinical, cognitive, and MRI data has been explored to predict the transition to psychosis in individuals at ultra-high risk. Although the addition of neuroimaging and cognitive data provided marginal predictive value, these studies highlight the potential of multimodal modeling in improving the accuracy of clinical risk assessments and early diagnosis[42].

EXPLANATIONS FOR THE POOR PERFORMANCE OF STUDY

The early diagnosis of BD remains a significant challenge, despite advances in the field of psychiatry[43]. First, one of the primary reasons is the inherent heterogeneity of BD. BD is a complex illness with variability at the level of symptom presentation, severity, clinical course, cognitive capacity, and everyday function, which complicates the identification of reliable diagnostic markers[44]. Second, genetic studies have identified several potential candidate genes associated with an increased risk for developing BD, but these findings have not yet translated into practical diagnostic tools. The genetic underpinnings of BD are complex and involve multiple genes, each contributing a small effect, which makes it challenging to pinpoint specific genetic markers that could be used for early diagnosis[45]. BD shares risk genes with other mental illnesses, such as schizophrenia[46], adding another layer of complexity for identifying genetic markers specific to BD[47]. Third, neuroimaging study, especially integrating with other approaches, has shown promise in understanding the pathophysiological processes underlying BD[28]. However, the variability in neuroimaging results across studies has limited their utility in early diagnosis. For instance, while some studies have identified structural and functional abnormalities in specific brain regions, these findings are not consistent across all individuals with BD, and similar abnormalities can also be found in other psychiatric disorders, such as schizophrenia[48]. This overlap in neuroimaging findings between different disorders further complicates the use of these techniques for early diagnosis of BD. Fourth, relevant studies are mostly retrospective or cross-sectional, focusing on statistical differences in manifestation of indicators, but lack prospective studies and diagnostic validation research, which makes them far from clinical application. Last, around 50% of BD patients initiate with depressive symptoms, which closely resemble those of MDD[6] and lead to 9 years in an average delay of clinical diagnosis of BD[5]. This symptom overlap could lead to the inclusion of potential BD patients in the depression cohort in clinical studies. In this regard, an ongoing study we are conducting, focusing on the different expression of peripheral biomarkers between patients with MDD and BD, and healthy controls recruits MDD patients with at least 5 years of depression history to minimize the possibility of including potential BD patients in the depression group. Furthermore, we plan to validate the biomarkers with predictive property for BD diagnosis in samples of another longitudinal study started 6 years ago, which included first-episode, medication-naive MDD patients and collected plasma samples[19]. The targeted biomarkers will be tested in the baseline plasma of those who ever shifted to bipolar episodes, and who were not during the past 5 years at least, tested for the predictive capacity of targeted biomarkers.

CONCLUSION

Although significant efforts have been made, limited progress has been achieved in the early diagnosis of BD. The clinical diagnosis of BD mainly relies on the presentation of signs and symptoms, which can lead to misdiagnosis due to recall bias from patients and physicians' varying experiences. Future research should focus on integrating multiple diagnostic approaches, including genetic, biomarker, neuroimaging, and clinical assessments, with more homogeneous participant groups to validate their effectiveness, which could improve the accuracy and timeliness of BD diagnosis in clinical practice.