Call to action regarding the vascular‐bipolar link: A report from the Vascular Task Force of the International Society for Bipolar Disorders

Abstract Objectives The association of bipolar disorder with early and excessive cardiovascular disease was identified over a century ago. Nonetheless, the vascular‐bipolar link remains underrecognized, particularly with regard to how this link can contribute to our understanding of pathogenesis and treatment. Methods An international group of experts completed a selective review of the literature, distilling core themes, identifying limitations and gaps in the literature, and highlighting future directions to bridge these gaps. Results The association between bipolar disorder and vascular disease is large in magnitude, consistent across studies, and independent of confounding variables where assessed. The vascular‐bipolar link is multifactorial and is difficult to study given the latency between the onset of bipolar disorder, often in adolescence or early adulthood, and subsequent vascular disease, which usually occurs decades later. As a result, studies have often focused on risk factors for vascular disease or intermediate phenotypes, such as structural and functional vascular imaging measures. There is interest in identifying the most relevant mediators of this relationship, including lifestyle (eg, smoking, diet, exercise), medications, and systemic biological mediators (eg, inflammation). Nonetheless, there is a paucity of treatment studies that deliberately engage these mediators, and thus far no treatment studies have focused on engaging vascular imaging targets. Conclusions Further research focused on the vascular‐bipolar link holds promise for gleaning insights regarding the underlying causes of bipolar disorder, identifying novel treatment approaches, and mitigating disparities in cardiovascular outcomes for people with bipolar disorder.


| INTRODUC TI ON
Vascular disease is an exceedingly common, yet arguably underrecognized source of morbidity and mortality in bipolar disorder. The association between bipolar disorder and vascular disease is large in magnitude, consistent across studies internationally, and independent of confounding variables where assessed. The vascular-bipolar link is multifactorial and is difficult to study given the latency between the onset of bipolar disorder, often in adolescence or early adulthood, and subsequent vascular disease, which usually occurs decades later. As a result, studies have among those with bipolar disorder, as well as aspects relevant to special populations. We endeavor to provide a broad overview of multiple domains, and future detailed reviews on subtopics herein are anticipated. mortality, which is widely recognized and comprises a core focus of prevention and treatment strategies, excess cardiovascular mortality among people with bipolar disorder is not widely appreciated and is not a core focus of prevention and treatment. Excessive risk of cardiovascular mortality has arguably been best demonstrated in Scandinavian countries with registry data that can be linked to mortality statistics, allowing a population-based assessment of risk.
Osby et al assessed mortality in a cohort of 15 386 inpatients with bipolar disorder from 1973 to 1995. 1 Bipolar disorder conferred a substantially increased risk of mortality with a Standardized Mortality Ratio (SMR) of 2.6 with 33% of the excess deaths being due to vascular disease (cardiovascular and cerebrovascular) for which the SMR was 2.2. 1 An elevated risk of cardiovascular mortality was also seen in a Danish registry-based study of 11 648 first admissions for bipolar disorder (SMR of 1.6). 2 A pooled estimate for the SMR for cardiovascular deaths in bipolar disorder for studies published between 1987 and 2007 was 1.9. 3 Cardiovascular SMR may differ according to age; for example, the cardiovascular SMR in a recent population-based cohort study in Sweden was 8 for adults <40 years old and 2.5-4 for adults ≥40 years old. 4 Key mortality studies from 2008 to present are highlighted in Table 1. [4][5][6][7][8][9][10][11][12] The highlighted study by Crump et al included both inpatients and outpatients in Sweden, although likely still underascertained bipolar disorder, with 6618 identified cases from a national cohort of 6.5 million. 6 Those with bipolar disorder died 9 years younger than expected and this elevated risk of premature (all cause) mortality persisted after adjusting for age, sociodemographic variables, and substance use. The fully adjusted estimates showed significant associations with death from CVD (HR 2.1, 95% CI 1.9-2.5, for women and 1.7, 95% CI 1.5-2.0, for men). Similar magnitude estimates were seen for mortality from ischemic heart disease and stroke. Interestingly, the risk factors for CVD were not as strongly associated in this sample (HR of 1.7 in women and 1.6 in men for diabetes mellitus, and there was no increase in the risk of hypertension or lipid disorders). When stratified by prior diagnosis of chronic diseases, such as heart disease and diabetes, the hazard ratio for mortality from these diseases associated with bipolar disorder was significantly lower, leading the authors to conclude "better provision of primary care may effectively reduce premature mortality. 6 " It has indeed been found that those with bipolar disorder appear undertreated for these CVRFs and that such undertreatment is associated with greater mortality risk. 7 The aforementioned studies were all based on cases of bipolar disorder identified based on health service utilization.

TA B L E 1 Key studies on cardiovascular mortality in bipolar disorder from past decade
Representative population studies, that do not rely on or require treatment seeking, circumvent concerns regarding biased sampling. Such population-based studies also show evidence of excess incidence and premature onset of CVD. For example, a longitudinal study evaluating a 3-year period in a US representative sample (NESARC) showed an elevated incidence of CVD (self-reported physician diagnosis) in individuals with bipolar disorder compared to those with major depressive disorder and controls. After controlling for confounding factors, adults with bipolar I (n = 1047) and bipolar II (n = 392) disorder were more than twice as likely to report a CVD diagnosis. Onset of CVD occurred at much younger ages among people with bipolar disorder than people without mood disorders (17 years younger for bipolar I disorder, and 14 years younger for bipolar II disorder) and significantly younger even than people with major depressive disorder (n = 4396). 14 Of note, only one-quarter of adults with bipolar disorder in this study reported lifetime exposure to antimanic treatment, and such exposure was not associated with increased risk of CVD.

| Future directions
There is consistent and compelling evidence linking bipolar disorder to excess cardiovascular mortality, with the strongest evidence coming from northern European patient registry samples. There is also evidence of increased and premature CVD in representative population samples. There remains a need for mortality studies in bipolar disorder with a more global representation, as well as a need to examine mortality in representative population samples (ie, not solely samples based on health service utilization). Insights generated from such studies would be bolstered by inclusion of a broad spectrum of biological, environmental, and traditional CVRFs.

| CLINIC AL COHORT S TUD IE S OF C ARDIOVA SCUL AR MORTALIT Y IN B IP OL AR DISORDER
The above-described nationwide studies have high power to detect differences in cardiovascular mortality, but lack the detailed characterization of bipolar samples possible in clinical cohort studies. One of the earliest studies that reported on CVD mortality in a clinical sample of patients with bipolar disorder was by Tsuang et al (1980).

| Future directions
There is great need for long-term longitudinal studies of well-phenotyped prospective samples to better understand how the varied course of illness and treatments received by patients may influence the risk of CVD and related comorbidities. Due to the lag between exposure and outcome, incorporation of putative mediators and intermediate vascular phenotypes/ biomarkers into these study designs would be valuable.

| TR ADITIONAL C ARDIOVA SCUL AR RIS K FAC TOR S
Clinical and epidemiologic studies from multiple countries demonstrate that there is excess prevalence of traditional CVRFs among people with bipolar disorder. Indeed, CVD risk may even be elevated among unaffected first-degree relatives of persons with bipolar disorder. 21,22 Recent meta-analyses confirm increased prevalence and/or incidence rates of obesity, 23 hypertension, 24 diabetes, 25 and cigarette smoking 26 among individuals with bipolar disorder. Summary estimates from these meta-analyses for the associations of bipolar disorder with obesity, hypertension, and diabetes are reported in Table 2 44 and smoking. [45][46][47][48] CVRFs are also associated with reduced neurocognitive function, particularly in frontal-executive tasks. [49][50][51][52][53][54][55][56][57] The strength of this association may depend in part on the symptomatic status and on the illness duration/stage. 58,59 Prospective studies indicate that the association among CVRFs, particularly obesity, and impaired cognition is bidirectional. 53,59 The association of smoking with neurocognition is inconsistent, possibly because nicotine has dopaminergic properties that may counteract any vascular-related neurocognitive risk. [60][61][62] Finally, CVRFs are also associated with overall functioning.
Many, 42,52,[63][64][65] although not all, 66,67 cross-sectional studies have found that among individuals with bipolar disorder those with, vs without, CVRFs such as obesity, metabolic syndrome, and diabetes had poorer psychosocial functioning and higher rates of disability.
Obesity 67 and clinically significant weight gain 66 may predict less improvement in functioning following mood episodes, whereas weight loss may yield improved functioning. 68

| Future directions
While bipolar disorder is strongly related to a myriad of CVRFs, greater prospective research to discern temporality and mediators is needed. A better understanding of the bidirectional influence of these vascular comorbidities on bipolar disorder could help us understand how they may moderate outcomes and impact treatments. Whereas there are many cohort studies in bipolar disorder that carefully characterize the prospective symptomatic course using repeated measures, there is a paucity of studies that integrate similar repeated measures of CVRFs. Such studies are greatly needed, and would benefit from including measures of lifestyle (see Section 5) and biological (see Section 7) mediators.

| LIFE S T YLE MED IATOR S
Many behaviors associated with bipolar disorder may serve as mediators of risk for vascular disease among individuals with bipolar disorder. These are important to recognize as they may be potentially modifiable and targets for intervention.
Smoking is 2 to 3.5 times more common among persons with bipolar disorder than the general population, with smoking prevalence estimates ranging from 30% to 70%. 26,69 Indeed, smoking may be a risk factor for developing bipolar disorder. 70 Smoking cessation treatment, which may include counseling with or without adjunctive medication, has not been studied in randomized trials in samples with bipolar disorder alone. However, a large randomized controlled trial in which 70% of participants had a mood disorder found that varenicline was more effective than nicotine replacement or bupropion and all active treatments were more effective than placebo. 71 Adverse neuropsychiatric events were infrequent and not more common with active treatment relative to placebo. 71 Preliminary findings from a study of 65 outpatients with schizophrenia, schizoaffective disorder, or bipolar disorder found that varenicline improves Framingham estimates of cardiovascular risk, despite the fact that relapse to smoking is common as is substantial weight gain among those with sustained remission. 72 Sleep disturbances are a risk factor for CVRFs such as obesity and hypertension, and are also common in individuals with bipolar disorder in all phases of illness including remission. In one study, 55%

TA B L E 2 Summary results from meta-analysis on the associations between cardiovascular risk factors and bipolar disorder
of an interepisode sample with bipolar disorder met diagnostic criteria for insomnia. 73 Sleep disturbances are associated with weight gain, 74 and among individuals with bipolar disorder, with more severe mood symptomatology and poorer treatment response. 75 Thus, addressing sleep disturbances could be an adjunctive intervention to reduce cardiovascular risk, although this has not been studied. For patients with sleep disturbance, pharmacotherapy (ie, melatonin or melatonin agonists) 76,77 and psychosocial interventions can be helpful. For example, those whose sleep disturbances remain refractory could benefit from interpersonal and social rhythms therapy or a bipolar-specific modification of cognitive-behavioral therapy for insomnia (CBTi-BP), which involves regularizing bedtime and rise times, stimulus control, and cautious use of sleep restriction.
Individuals with bipolar disorder are also more likely to have poor dietary and nutritional habits. 78,79 Up to 30% of people with bipolar disorder report binge-eating behavior, which is associated with metabolic risk independent of its effect on obesity. 80 Moreover, binge-eating behavior is a predictor for weight gain following treatment with quetiapine. 81 In addition, several studies suggest that individuals with bipolar disorder are more likely to consume only one meal per day, 79 a high-fat and low-fiber diet, 82 more saturated fat, 83 more sucrose and sweetened beverages, 84 and add more salt to their food. 85 While the findings relating to nutrition are generally convergent, there are exceptions, such as a study reporting that people with bipolar disorder reported consuming fewer calories, including from carbohydrates and fats, and more fiber, than historical controls-despite having twice the prevalence of metabolic syndrome. 86 Mood states and mood symptom burden may explain in part these inconsistencies. In addition, evening chronotype may be associated with significantly higher rates of both binge eating and unhealthy eating habits. 87 The estimated prevalence of a sedentary lifestyle is between 40% and 65% in people with bipolar disorder. 88 Greater age and greater BMI are associated with particularly low levels of physical activity among individuals with bipolar disorder. 89 Samples with bipolar disorder also demonstrate a diminished functional exercise capacity, with poor exercise test performance related to depressive symptoms and musculoskeletal complaints. 90 Greater physical activity is also associated with a reduced burden of depressive symptoms, better quality of life, better overall function, but a greater burden of manic symptomatology. 88,91 There is reason to believe that exercise may improve mood. 92,93 In a sample with unipolar depression, exercising 45 min/day for 3 days/week was as effective in reducing depressive symptoms as 50-200 mg flexibly dosed sertraline at 16 weeks, with a lower risk of recurrence over long-term (10 month) follow-up. 94,95 Non-randomized pilot data in bipolar disorder are encouraging. 96 Exercise also appears to increase brain-derived neurotropic factor (BDNF) levels in women but not men with bipolar disorder, 97 leading some to hypothesize that exercise promotes neurogenesis. 98 While there exist a variety of pharmacological and lifestyle interventions seeking to optimize nutrition/diet and/or physical activity in the general population, attempts to study or adapt these interventions for those with bipolar or related disorders have been limited. A pilot study of integrated medical and outpatient psychiatric care for bipolar disorder that included focus on sleep/wake rhythms, nutrition, and exercise improved function and reduced psychiatric hospitalizations; however, this study did not report on the impact on risk factors for CVD. 99

| Future directions
There is a need to directly evaluate smoking cessation interventions in samples with bipolar disorder, which have unique risks such as treatment-emergent mania. Preliminary evidence of exercise as a treatment for bipolar disorder is a promising area for future research; it is remarkable that there are thus far no published clinical trials that target aerobic/cardiorespiratory fitness in bipolar disorder. Several effective treatments targeting lifestyle have been developed in the general population, but there is a great need to develop treatments tailored for bipolar disorder specifically, rather than "off the shelf" general population approaches that do not adequately integrate such themes as mood symptoms and medication side-effects.
From a methodologic perspective, studies that incorporate objective measures of sleep and physical activity (eg, accelerometry) and that move beyond simple self-reports of nutrition and diet are needed.

| Association of psychotropic medications with cardiovascular disease and cardiovascular risk factors
The medications commonly used to treat bipolar disorder adversely affect CVRFs with the potential for substantial changes in weight or other risk factors 103 (Note: while various psychotropic medications have cardiovascular side-effects and/or risks such as arrhythmia, hypertensive crisis, and orthostatic hypotension, 104 the focus of the current article is on atherosclerotic CVD). Widely considered the first-line mood stabilizer, lithium has been associated with weight gain, but is likely associated with less weight gain than valproate or antipsychotics. 105 Lamotrigine or carbamazepine do not commonly induce substantive changes in weight. 106,107 Antipsychotics, particularly second-generation antipsychotics (SGAs), are known to promote weight gain. Clozapine, olanzapine, iloperidone, chlorpromazine, sertindole, quetiapine, risperidone, and paliperidone appear to confer the greatest risk for weight gain. 103 The lowest potential for weight gain is seen with ziprasidone, lurasidone, and cariprazine. 103 Aripiprazole is sometimes also listed as lower in risk, although study results have varied. Aripiprazole is associated with at least double the weight gain observed for adult patients taking mood stabilizers, 108 and is associated with greater weight gain vs placebo in short-term studies of youth with bipolar disorder. 109 Overall, the general propensity of mood stabilizers and antipsychotics to increase the risk of CVRFs underscores the importance of routine mon-

| Potential role of vascular-related medications as repurposed treatments for bipolar disorder
It has been suggested that the bidirectional associations between CVD and bipolar disorder may represent a pathophysiologic nexus and thus a potential opportunity to repurpose medications traditionally used in the treatment of CVD for bipolar disorder. 118 Statins, a class of medications that are 3-hydroxy-3-methylglutaryl-CoA (HmG CoA)  well-being, 134 although these studies were not specifically designed for this purpose.

| Future directions
Several medications used to treat bipolar disorder carry a substan-

| B I OLOG I C AL MED IATOR S
There have been numerous peripheral biomarker (ie, derived from blood samples) studies in bipolar disorder, as comprehensively summarized in the ISBD Biomarker Task Force paper. 135 The most frequently studied peripheral biomarkers in bipolar disorder, 135,136 namely, inflammatory markers, oxidative stress markers, and

| Inflammation
Inflammation is associated with incident CVD, traditional CVRFs, 142 and endothelial dysfunction (ED). 143,144 Markers of inflammation, most consistently c-reactive protein (CRP), contribute to the prediction of CVD beyond Framingham risk scores. 142,145 Putative mechanisms of the association between mood disorders and inflammation include glucocorticoid resistance, blood-brain barrier disruption, altered neurotransmitter metabolism, impaired functional connectivity, astrocyte and microglia activation, neuronal damage and degeneration, and reduced neurotrophic support. 146 In adult bipolar disorder, increased frontal cortical inflammation has been observed in postmortem samples 147 and pro-inflammatory genotypes have been implicated. 148 Numerous studies on the topic, including several meta-analyses, confirm that peripheral pro-inflammatory markers are elevated in bipolar disorder, particularly during symptomatic intervals of mania and depression. 149,150

| Oxidative stress
Oxidative stress, reflecting an imbalance between oxidants and antioxidants, is relevant to vascular function, atherosclerosis, and CVD. 151,152 The interaction of reactive oxygen species with endothelium-derived nitric oxide contributes to ED which is in turn a precursor to atherosclerosis. 153 Oxidative stress-related endothelial damage contributes both to atherosclerotic disease and to metabolic conditions that confer risk for CVD, such as diabetes, dyslipidemia, and hypertension. 154 These effects are, in part, related to shared genetic susceptibility to oxidative stress and CVD. 155 As is common in articles regarding novel therapeutic approaches in bipolar disorder, the concept of therapeutically targeting mitochondrial function for the purpose of reducing oxidative stress is also relevant to CVD. 156 Postmortem studies in bipolar disorder have shown elevated oxidative stress in the brain, especially in frontal regions. 157,158 Similar to the inflammation literature, studies consistently demonstrate elevations of peripheral oxidative stress markers in bipolar disorder, particularly during symptomatic intervals. Oxidative stress has been posited to contribute to neuroprogression.

| BDNF
Whereas the links between CVD and inflammation/oxidative stress are more widely recognized, and perhaps more self-evident, there is also a substantial literature regarding the implications of BDNF. 159 Reduced BDNF levels have been observed in adults suffering from acute coronary syndrome. 160 Serum BDNF levels were inversely associated with CVD and CVD mortality in a prospective study of 3687 Framingham participants, and BDNF genotypic analysis in a related study suggests that the protective effect may be causal. 161 Even in young adults, BDNF levels are associated with vascular function and functional capacity for exercise. 162 Deficiency of BDNF may be implicated in reduced endothelial integrity and increased endothelial cell apoptosis. 163 The BDNF valine to methionine substitution at

| Future directions
While several putative biomarkers have been identified, little is known about the temporal nature of these relationships and/or whether they are causal. Prospective study would be important to discern the relation between these biomarkers and the course of BD and atherosclerosis/CVD. Finally, target-engagement clinical trials are warranted to evaluate whether modifying these biomarkers has an influence on the development and/or progression of CVD.

| Leukoaraiosis/white matter hyperintensities
Hyperintensities in brain MRI are thought to reflect focal ischemic damage. Increased rates of white matter hyperintensities (WMH), also described as leukoaraiosis, is among the most consistent structural neuroimaging findings in bipolar disorder. 167,168 While described as non-specific with regard to underlying cause, leukoaraiosis is often considered a manifestation of cerebral small vessel disease, and is an independent predictor of stroke and dementia. 169 Meta-analysis of 27 studies found significantly increased rates of deep WMH among those with bipolar disorder compared to controls (OR 3.2, 95% CI 2.2, 4.5). 170 Rates of overall MRI hyperintensities were numerically but non-significantly higher in bipolar disorder than in major depressive disorder (OR 1.6, 95% CI 0.9-2.7) and schizophrenia (OR 1.5, 95% CI 0.9, 2.7). The risk of hyperintensities relative to controls was greatest in youth (OR 5.7, 95% CI 2.3, 13.7). There is evidence that WMH burden is related to familiality and bipolar disorder subtypes, 171 although findings in relatives of people with bipolar disorder are inconsistent. 172 Despite questions regarding whether the association between bipolar disorder and WMH is explained in part by comorbidities such as migraine, developmental disorders, and/or CVRFs, 173 the consistency of increased WMH in bipolar disorder alongside the association of WMH with systemic and cerebral vascular disease indicates the need for continued research on this topic.

| Neuroimaging correlates of cardiovascular risk factors
Obesity among otherwise healthy people is associated with significantly lower total brain volume, particularly lower grey matter volume, [174][175][176][177][178][179] and with greater brain volume loss over time. 180 These findings have led a number of research groups to investigate whether obesity is associated with brain changes in bipolar disorder. In a firstepisode mania sample, overweight/obese bipolar disorder patients had lower white matter volumes than normal-weight patients, particularly in the frontal and temporal lobes and subcortical white matter. 181,182) A second study, also in first-episode mania patients, reported lower temporal-parietal-occipital white matter integrity in overweight/ obese compared to normal-weight patients. 183 BMI-related gray matter reductions have also been detected early in the course of bipolar disorder. In an adolescent bipolar disorder sample, higher BMI was associated with lower orbitofrontal cortical and prefrontal cortical thickness. 184 Brain chemistry has been less well studied. In one firstepisode mania sample, higher BMI predicted greater hippocampal glutamate and lower hippocampal N-acetylaspartate (NAA) in bipolar disorder patients. 185,186) Interestingly, this was independent of any effect of BMI on hippocampal volumes, since two separate groups reported that hippocampal volume did not vary with BMI in bipolar disorder patients. 186,187 That BMI-related brain changes were located primarily in limbic brain areas, and that they were detected in bipolar disorder patients but not age-and gender-matched control subjects, suggests that higher BMI is associated with unique brain changes in bipolar disorder, such that frontal/temporal limbic brain changes characteristic of bipolar disorder are more pronounced with higher BMI. In keeping with this hypothesis, the only prospective study to date found that first-episode mania patients with clinically significant weight gain (≥7% of baseline weight) over 12 months also had significantly greater volume loss in the left anterior cingulate gyrus, orbitofrontal cortex, and middle temporal gyrus over the same time period. 188 However, the observational nature of the study precludes definitive conclusions regarding the direction of the observed associations. Moreover, "third variable" causes such as genes that impact both BMI and limbic brain structure/function cannot be excluded. [189][190][191] Comparatively fewer studies have examined dysglycemia (ie, diabetes or insulin resistance/glucose intolerance) in relation to neuroimaging phenotypes in bipolar disorder.
Hajek and colleagues reported that dysglycemia among adults with bipolar disorder is associated with lower hippocampal and insular cortex volumes compared to euglycemic adults with bipolar disorder and healthy controls, and is associated with stronger aging effects on hippocampal volumes. 192 A subsequent MR spectroscopy study from the same group found that dysglycemia among adults with bipolar disorder was associated with reduced prefrontal levels of NAA and total creatine compared to euglycemic adults with bipolar disorder and healthy controls; lower levels of these markers were in turn related to poorer global functioning. 193 The relationship between cigarette smoking and neuroimaging findings among individuals with bipolar disorder has also received relatively little attention. We are aware of only one study investigating this relationship, which reported that in a mixed sample of bipolar disorder and schizophrenia patients, smokers had reduced cortical thickness in the left anterior cingulate cortex and insula. 194

| Cerebral blood flow
An extensive literature has used neuroimaging to examine the function of large-scale neural networks in bipolar disorder, identifying consistent patterns of altered neural response to cognitive and emotional tasks and abnormal coordination of brain activity within and between neural circuits. 195 However, the integrity of the cerebrovascular system, which provides the brain with oxygen and energy substrates, has been comparatively less studied in bipolar disorder. Cerebral blood flow (CBF) is considered an indicator of brain health 196 that is associated with WMH and with systemic vascular risk factors. 197,198 Indeed, CBF has been the most common vascular-related neuroimaging phenotype studied in bipolar disorder. A recent systematic review identified 33 studies with a total of 508 bipolar disorder participants and 538 controls. 199  to healthy controls. 207 Most studies that evaluated CBF in bipolar vs unipolar depression have not found significant differences, although one study using pattern recognition analysis found that, despite lack of differences from healthy controls, subgenual anterior cingulate cortex CBF classified unipolar vs bipolar depression in females with 81% accuracy. 208 Finally, with the exception of a single lithium discontinuation study, 209 there is a paucity of data regarding bipolar disorder treatment and treatment response in relation to CBF and changes therein.

| Future directions
Thus far, there is a gap with regard to the link between neuroimaging phenotypes and hypertension/blood pressure or lipids levels, 55 and related studies are warranted. There are opportunities to capitalize on the advantages of ASL, which magnetically tags blood as a non-invasive endogenous tracer. This offers advantages in general, were also no differences in the reactive hyperemia index or AIX as measured by finger plethysmography in a larger sample of 56 participants with a mean age of 40 years. 211 Another group reported albuterol-induced vasoreactivity deficits among 31 adults (mean age 43 years) with depression; subgroup analyses found significantly lower vasoreactivity among those with unipolar depression (n = 11, P = .005) but only nominal lower vasoreactivity among those with bipolar depression (n = 20, P = .086). 212 In another study focused on identifying age-related effects, the older-but not the younger-half of the sample (median split on 32 years) had greater than expected PWV and AIX relative to normative values and this was related to the retrospectively estimated burden of mood symptoms. 213 In an older sample with mixed mood disorders, lower FMD was seen in those who had a greater burden of mania/hypomania over more than a quarter century of prior follow-up in a prospective cohort. 214 After an initial study regarding retinal vascular photography in adolescents with bipolar disorder, 215 there have since been several studies on this topic in adults. [216][217][218][219] One study found that both adults with bipolar disorder and those with schizophrenia had narrower arterioles and wider venules as compared to healthy volunteers, and adults with bipolar disorder had narrower arterioles and wider venues even as compared to adults with schizophrenia, 216

| Bipolar disorder in youth
Similar to adults, the prevalence of traditional CVRFs, and their clustering, is increased in youth with bipolar disorder. 33,34,221,222 There is elevated cardiovascular risk among first-and second-degree relatives of youth with bipolar disorder, particularly among relatives of youth with familial bipolar disorder. 223 As with adults, obesity among youth with bipolar disorder is associated with greater psychiatric complexity, particularly increased suicide attempts in both clinical and largely untreated epidemiologic samples. 33,34 With regard to CVD-related lifestyle variables, there is increased cigarette smoking, suboptimal nutrition, and eating behaviors, and reduced physical activity, in comparison to healthy youth. [224][225][226] Traditional CVRFs are negatively associated with executive function among youth with bipolar disorder, an association not observed among healthy youth. 56,57 As with adults, there is evidence that inflammation is elevated and BDNF is reduced in youth with bipolar disorder. [227][228][229] Studies have also found that BDNF, inflammatory markers, and/or oxidative stress markers are associated with traditional CVRFs as well as with non-invasive ultrasound measures of vascular structure and function. 227,230,231 Preliminary findings from the first retinal photography study in bipolar disorder suggest retinal vascular caliber is associated with blood pressure, peripheral endothelial function, and mood symptoms in youth with bipolar disorder but not among healthy controls. 215 In terms of neuroimaging findings, similar to adults, BMI is negatively associated with cortical volumes and structure, particularly in frontal regions. 184 In contrast to adult studies, there appears to be anomalous elevated CBF in midline and frontal regions among youth with bipolar disorder, although acute aerobic exercise appears to have a temporarily normalizing effect on CBF in youth with bipolar disorder. 184 Finally, there is evidence of reduced cerebrovascular reactivity, as measured using a standard breath-hold paradigm during fMRI, among youth with bipolar disorder, and this is especially evident in periventricular and deep white matter, regions where WMH are commonly observed later in life. 232

| Bipolar disorder in the elderly
Patients with bipolar disorder begin to have increased risk of cardiovascular and metabolic morbidities in midlife and may die mainly from CVD before reaching geriatric age, with the average cardiovascular mortality occurring 10 years earlier than the general population. 4 In light of this premature mortality, patients with early-onset bipolar disorder who survive into old age probably represent a healthy survivor bipolar disorder subpopulation with favorable outcome.
Therefore, studies that focus only on individuals in their 60s and beyond may not be truly representative of the larger bipolar disorder population. 233 A limited number of studies have focused specifically on bipolar disorder in late life. The prevalence of cardiovascular morbidity including hypertension in older-age (>60 years) bipolar disorder (OABD) in existing reports appears to be similar to that in communitybased geriatric samples or patients with depressive disorder. [235][236][237][238] Furthermore, comparable percentages (about 25%) of older patients with bipolar disorder have coronary heart diseases in Asian and Western reports. 238 However, OABD had significantly higher BMI and a greater burden of endocrine/metabolic diseases. 234,235,237 Consequently, approximately two-thirds of OABD are reported to have a previous stroke or silent cerebral infraction (SCI). 239,240 In particular, SCIs are detected in 59.5% of OABD, which is dramatically higher than the 20% detected in healthy elderly people. 239 Patients with bipolar disorder, including those with early-onset illness, are at higher risk of cerebrovascular disease in late life. The number of recurrent affective episodes may increase the risk of stroke in OABD with typical-onset age. 239 Cardiovascular morbidity, particularly stroke, may increase with age in OABD and optimal treatment requires awareness of SCI and prevention against recurrent episodes.
Among the older (aged >50 years) community-dwelling patients with serious mental illness, given better social functioning and comparable cognitive function in older bipolar patents, they may remain at higher risk for obesity and medical morbidity than schizophrenic ones.
Treatments targeting cognitive impairment and CVDs across their life span are necessary. 241  Several studies have sought to evaluate the effectives of integrating physical and mental health within a unified treatment model. [260][261][262][263] However, such models have not yet been disseminated and/or implemented beyond the original context of a given study, and are therefore not available to most people with bipolar disorder.

| Future directions
The aforementioned disparities are multifactorial, and related to a combination of patient factors, provider factors, and system-level factors. Although preponderance of evidence supports the existence of disparities in care, exceptions exist, 264 and understanding the mechanisms and factors underlying more equitable service delivery models may yield valuable insights. There is an urgent need for more studies evaluating integrated care settings and pathways, particularly in real-world clinical and community settings. Finally, a factor not commonly invoked in discussions of treatment disparities is the premature onset of CVD among people with bipolar disorder; simply shifting the frame of cardiovascular focus earlier by a decade would better align prevention and early intervention strategies with the epidemiology of CVD in bipolar disorder. One example of such an approach is that of the American Heart Association, which includes bipolar disorder among conditions that confer increased risk for early CVD, and advises earlier screening for, and tighter control of, CVRFs. 140 Another example is the PRIMROSE research program, which demonstrated the superiority of modified risk prediction models for people with severe mental illness over models based solely on traditional CVRFs. 110

| CON CLUS IONS
The title of this study declares a call to action. The hope of the authors is that this article has served to make explicit the need for such a call (Figure 1), and to illuminate potential future directions.
In an era when there has been a marked reduction in cardiovascular mortality in the general population, the excess of burden of cardiovascular mortality among people with bipolar disorder has increased.
We view this discrepancy as outrageous, not because there is a way to address it but because there are numerous ways to address it.
But if we focused this call to action solely on the topic of excess CVD and mortality, we would be missing an opportunity to highlight the relevance of blood vessels to the psychiatric and neurocognitive symptoms that contribute to suffering, functional impairment, and In conclusion, progress regarding the vascular-bipolar link will require a more deliberate and systematic integration of vascular metrics in research, treatment approaches, healthcare systems, and education of healthcare professionals and the public. We hope that this article will serve to mobilize a broader spectrum of stakeholders with a vested interested in the vascular-bipolar link, and as such serve as an impetus to accelerate progress. We intend to continue distilling and disseminating the research literature on this topic, including future focused subtopic reviews as the literature and knowledge base continues to accumulate. Finally, we invite any stakeholders who are interested in the vascular-bipolar link to connect with us in order to share ideas and discuss ways to collaboratively encourage and enable action.