Bipolar disorder is a major psychiatric illness characterised by its early onset, severity and alternating mood episodes.^{Reference Ferrari, Stockings, Khoo, Erskine, Degenhardt and Vos1} A considerable proportion of bipolar disorder cases have been significantly associated with cognitive deficits, although to varying degrees, even when the bipolar disorder is in remittance.^{Reference Bora, Hidiroglu, Ozerdem, Kacar, Sarisoy and Civil Arslan2} Studies have also suggested that patients with bipolar disorder can experience accelerated cognitive ageing, which presents as a steeper decline in function.^{Reference Seelye, Thuras, Doane, Clason, VanVoorst and Urosevic3,Reference Chen, Huang, Lee, Chang, Lin and Chiu4} Cognitive impairment in patients with bipolar disorder may worsen as the disease progresses, which may negatively affect patient recovery.^{Reference Duarte, Becerra and Cruise5} In addition, age can dynamically, neurodevelopmentally and neuroprogressively interact with the neuropathology of bipolar disorder.^{Reference Kohler and Marlinge6} Although cognitive dysfunction has increasingly been recognised as a key symptom in patients with bipolar disorder, the biomarkers related to cognitive dysfunction have rarely been investigated, especially in different age groups of patients with bipolar disorder.

Evidence suggests that vitamin D deficiency may be a key marker of cognitive decline.^{Reference Pavlovic, Abel, Barlow, Farrell, Weiner and DeFina7} Moreover, higher vitamin D levels and higher vitamin D supplement intake have been associated with a slower cognitive decline.^{Reference Goodwill, Campbell, Simpson, Bisignano, Chiang and Dennerstein8,Reference Beydoun, Hossain, Fanelli-Kuczmarski, Beydoun, Canas and Evans9} This indicates that vitamin D may play an essential role in brain health and function, and may exert various neuroprotective effects on numerous brain areas.^{Reference Mayne and Burne10} A survey revealed that more than a third of patients with severe mental disorders, including bipolar disorder, were deficient in vitamin D.^{Reference Boerman, Cohen, Schulte and Nugter11} In addition, clinical studies suggested that vitamin D may play a role in the pathophysiology or cognitive profile of bipolar disorder. For example, a negative and moderate correlation was identified between vitamin D levels and mania severity or disease severity in bipolar disorder.^{Reference Altunsoy, Yuksel, Cingi Yirun, Kilicarslan and Aydemir12} Polymorphism in the vitamin D receptor gene was associated with lower expression of the dopamine D1 receptor gene and may contribute to bipolar disorder development.^{Reference Ahmadi, Mirzaei, Hossein-Nezhad and Shariati13} Furthermore, vitamin D deficiency has been associated with poor cognitive functioning in individuals experiencing major depressive episodes and in psychotic disorders, including psychotic bipolar disorder.^{Reference Belzeaux, Annweiler, Bertrand, Beauchet, Pichet and Jollant14,Reference Nerhus, Berg, Simonsen, Haram, Haatveit and Dahl15} An image study suggested a significant and positive association between vitamin D and intracranial volume in psychotic bipolar disorder.^{Reference Gurholt, Nerhus, Osnes, Berg, Andreassen and Melle16} In summary, although bipolar disorder has been reported to be associated with altered cognitive functioning and hypovitaminosis D, the associations between vitamin D levels, bipolar disorder pathophysiology and cognitive features are not well understood.

The mechanism underlying vitamin D's influence on brain function is complex and likely to involve direct induction of neurotrophic factors and indirect anti-inflammatory and metabolic effects.^{Reference Cui, Gooch, Petty, McGrath and Eyles17} Research suggests a connection between inflammation and the pathogenesis of bipolar disorder.^{Reference Jones, Vecera, Pinjari and Machado-Vieira18} In addition, inflammatory markers were identified, suggesting that inflammation could contribute to the neurocognitive deficits observed in bipolar disorder.^{Reference Poletti, Mazza, Calesella, Vai, Lorenzi and Manfredi19,Reference Castaño-Ramírez, Sepúlveda-Arias, Duica, Díaz Zuluaga, Vargas and López-Jaramillo20} Therefore, the anti-inflammatory neuroprotective role of vitamin D may preserve axonal integrity in the face of neuro-damage, which can be identified as neurofilament light chain (NfL) levels.^{Reference Mayne and Burne10,Reference Naifar, Maalej Bouali, Guidara, Ellouze, Jmal and Omri21} The NfL is a unique biomarker related to axonal damage and neural cell death. It demonstrates promise as a potential marker of neuropsychiatric disease activity and progression associated with cognitive deficit.^{Reference Bavato, Cathomas, Klaus, Gütter, Barro and Maceski22,Reference Jakobsson, Bjerke, Ekman, Sellgren, Johansson and Zetterberg23} Patients with bipolar disorder have been reported to have higher NfL levels, both in their central nervous systems and peripheral blood, compared with healthy controls.^{Reference Jakobsson, Bjerke, Ekman, Sellgren, Johansson and Zetterberg23,Reference Aggio, Fabbella, Finardi, Mazza, Colombo and Falini24} However, studies analysing the association between vitamin D, NfLs and the cognitive profile of bipolar disorder have not been conducted.

Results

Patient characteristics and age effect

We recruited 100 patients (48 men and 52 women) with bipolar disorder who were in a euthymic state. The mean age was 47.49 years (s.d. 11.71), and the mean age at onset was 24.78 years (s.d. 9.79). The median and mean NfL levels were 9.26 and 11.10 pg/mL, respectively. The mean vitamin D level was 16.46 ng/nL, with 72% of the patients exhibiting vitamin D deficiency. A significant positive association was identified between age and NfL levels (r = 0.377, P < 0.001), and a nonsignificant negative association was identified between age and vitamin D levels (r = −0.63, P = 0.266; scatterplot in Supplementary Fig. 1 available at https://doi.org/10.1192/bjo.2022.608). Age also had a significant negative correlation with composite neurocognitive score (r = −0.031, P < 0.001) and the neurocognitive domains of verbal memory (r = −0.16, P = 0.019), working memory (r = −0.039, P < 0.001), processing speed (r = −0.031, P < 0.001) and executive function (r = −0.028, P = 0.001) as determined through the BAC-A. With a cut-off value of 45 years, significant differences were identified in the composite neurocognitive score, working memory and executive function scores between age subgroups (Supplementary Table 1).

Comparison of vitamin D status

Table 1 presents the sociodemographic and clinical characteristics of the patients with bipolar disorder with and without vitamin D deficiencies, stratified by age. In the younger age group, 62.79% exhibited vitamin D deficiency; in the older age group, 78.95% exhibited vitamin D deficiency. No significant differences were identified in the disease course, amount of physical activity, NfL levels or DDD of the younger patients with and without vitamin D deficiency. In the older age group, we noted significantly higher numbers of hospital admissions in the patients with versus without vitamin D deficiency (8.49 ± 8.55 v. 2.92 ± 2.88; P = 0.031). Furthermore, the patients without vitamin D deficiency reported more physical activity than those with (1552.67 ± 429.54 v. 666.56 ± 603.16; P = 0.016). No differences were noted in the ages, premorbid estimated IQs, gender, marital or job statuses, smoking or alcohol use habits, nutrient supplement intakes or proportions of physical comorbidities of the patients with and without vitamin D deficiency in both age groups. Moreover, the differences in cognitive function in the patients with and without vitamin D deficiency and in those with higher and lower median NfL levels were nonsignificant for both age groups (Supplementary Tables 2 and 3).

Table 1 Sociodemographic and clinical characteristics of patients with bipolar disorder with and without vitamin D deficiency, stratified by age

HRSD, Hamilton Rating Scale for Depression; YMRS, Young Mania Rating Scale; IPAQ, International Physical Activity Questionnaire; NfL, neurofilament light chain; DDD, defined daily dose.

* P < 0.05.

Interaction effect of vitamin D and NfL level on cognition

General linear models adjusted for age and gender, and models further adjusted for years of education, psychotropic DDD, number of hospital admissions and physical activity levels, were used to determine the effects of vitamin D, NfLs and their interaction on various cognitive domains in younger (Table 2) and older patients (Table 3) with bipolar disorder. In the younger age group, we identified a significant association between serum vitamin D, NfLs and their interaction to composite neurocognitive score. The association was particularly noted on verbal fluency and processing speed domains (model 2 in Table 2). In the younger age group, NfL levels had a considerable negative influence and vitamin D had a minor negative influence. Furthermore, the interaction of NfL and vitamin D levels positively affected cognitive function. In the older age group, the interaction between vitamin D and NfLs significantly and negatively affected composite neurocognitive score. The association was also particularly noted on verbal fluency (model 2 in Table 3). Moreover, vitamin D had a positive association with composite neurocognitive score; in addition, both vitamin D and NfLs had a significant positive association with verbal fluency in the older age group. In summary, we identified interaction effects of vitamin D and NfLs on composite neurocognitive function and verbal fluency domain for both age groups. Furthermore, age influenced the direction of the interaction effects. The crossed lines in the interaction plot (Fig. 1) indicate that the association between vitamin D status and cognition was influenced by NfL levels for both younger and older patients with bipolar disorder. In the younger patients with bipolar disorder with vitamin D levels <10 ng/mL, those with lower NfL levels exhibited more favourable cognitive function. However, when vitamin D levels were higher, the negative effect of NfLs was not present (Fig. 1(a)). In the older patients with bipolar disorder, the patients with lower NfL levels generally demonstrated increased cognitive function as their vitamin D status improved. However, the plot for the older patients with bipolar disorder with higher NfL levels revealed no notable cognitive changes in vitamin D levels (Fig. 1(b)).

Fig. 1. Interaction plots of effect of vitamin D and neurofilament light chain levels on cognition. (a) Younger patients with bipolar disorder and (b) older patients with bipolar disorder. BAC-A, Brief Assessment of Cognition in Affective Disorders; NfL, neurofilament light chain.

Table 2 Associations between total serum vitamin D and neurofilament light chain and their interaction effect on cognitive domains in younger patients (aged 20–45 years) with bipolar disorder

Model 1: age, gender, vitamin D level, NfL level and interaction of vitamin D and NfLs. Model 2: age, gender, defined daily dose of total psychotropic medications, years of education, number hospital admissions, level of physical activity determined with the International Physical Activity Questionnaire, vitamin D level, NfL level and interaction of vitamin D and NfLs. NfL, NfL, neurofilament light chain.

* P < 0.05.

** P < 0.01.

*** P < 0.001.

Table 3 Association between total serum vitamin D and neurofilament light chain and their interaction effect on cognitive domains in older patients (aged 46–65 years) with bipolar disorder

Model 1: age, gender, vitamin D level, NfL level and interaction of vitamin D and NfLs. Model 2: age, gender, defined daily dose of total psychotropic medications, years of education, number of hospital admissions, level of physical activity determined with the International Physical Activity Questionnaire, vitamin D level, NfL level and interaction of vitamin D and NfLs. NfL, neurofilament light chain.

* P < 0.05.

** P < 0.01.

*** P < 0.001.

Discussion

Main findings

In the younger patients with bipolar disorder, we noted a negative association between NfLs and vitamin D, and composite neurocognitive score and two neurocognitive domains (verbal fluency and processing speed). In the older patients, we noted a positive association between NfLs and vitamin D, and the cognitive domain of verbal fluency. Our results revealed a consistently significant antagonistic interaction effect on composite score and verbal fluency for both age groups. In addition, age was likely an effect-modified factor in the association. If NfLs can be regarded as a marker of neuroaxonal change over the course of bipolar disorder, clarifying the association of an intervening factor, such as vitamin D status, at different ages could considerably influence future intervention designs.

Complex interaction of NfLs and vitamin D in bipolar disorder

In interpreting the positive correlation between NfLs and verbal fluency in older patients with bipolar, researchers must consider that older patients have a longer disease course; therefore, early axonal damage could later have been affected by remyelination and axonal regeneration. This would be consistent with neuroimaging findings indicating that NfL level is positively associated with axial diffusivity in bipolar disorder.^{Reference Aggio, Fabbella, Finardi, Mazza, Colombo and Falini24} Our finding of a negative correlation between vitamin D and cognition in young patients with bipolar disorder contrasts with that of most studies on neuropsychiatric diseases.^{Reference Byrn and Sheean29} However, the parallel association (both negative or positive) identified between NfLs and vitamin D and cognition was consistent with the results of trials involving patients with multiple sclerosis, which has a similar neuropathology to bipolar disorder.^{Reference Magioncalda, Martino, Conio, Piaggio, Teodorescu and Escelsior30,Reference Magioncalda, Martino, Tardito, Sterlini, Conio and Marozzi31} In the aforementioned trials, disease status modified the association between vitamin D and NfL levels. In patients with active multiple sclerosis, the theoretical negative correlation of vitamin D and NfLs (non-parallel) have been reported to affect their brain function. However, in patients with multiple sclerosis in remittance or who received disease-modifying therapy, a parallel correlation or no correlation has been reported between vitamin D and NfLs and brain function.^{Reference Hänninen, Jääskeläinen, Herukka and Soilu-Hänninen32,Reference Røsjø, Lindstrøm, Holmøy, Myhr, Varhaug and Torkildsen33} In this study, the patients with bipolar disorder were in a euthymic state and were receiving medication. This may explain the parallel association between vitamin D, NfLs and cognitive function in the two age groups, similar to the aforementioned multiple sclerosis findings. Our findings may further be explained by the following. First, in the early stages of the disease in younger patients with bipolar disorder, NfLs play a significant role in determining disease severity, which negatively influences cognition. Meanwhile, the influence of an individual's vitamin D status on cognition may be modified by other factors, such as medication. For example, although mood stabilisers are generally provided when treating bipolar disorder, they are frequently associated with reduced vitamin D levels.^{Reference LoPinto-Khoury, Brennan and Mintzer34} Second, in the older patients with bipolar disorder in our study, a positive association between vitamin D and cognitive function was identified, and vitamin D may enhance the remyelination process. This, together with the elevated NfL levels, led to a positive association between vitamin D and NfL levels and the verbal fluency domain. In addition, the age-modified interaction effect could contribute to evidence supporting the neuroimmune development of bipolar disorder, in which, at the onset of the illness, individuals experience an increased proinflammatory state in adolescence, an anti-inflammatory state during young adulthood and a nearly normalised immune state in later stages of the disease course.^{Reference Guglielmo, Miskowiak and Hasler35} The interaction plot revealed an antagonistic interaction effect of vitamin D and NfLs, indicating that in the early stages of bipolar disorder in younger patients, NfL level may serve as a marker for neuro-damage resulting from neuroinflammation, and cognition differed in patients with higher and lower NfL levels, particularly those with extremely low vitamin D levels. In older patients with bipolar disorder who have a longer disease course, neuroinflammatory processes become stable, and the positive effects of vitamin D become more prominent in patients with lower rather than higher NfL levels. Therefore, our findings are consistent with the most current overview of the immune-inflammatory model in bipolar disorder,^{Reference Magioncalda and Martino36} and further add the role of neuroaxonal damage and vitamin D status (Fig. 2).

Fig. 2. Inflammatory model and age-modified interaction effects in treated euthymic bipolar disorder. NfL, neurofilament light chain.

Possible causal inferences of the interaction

Studies involving animal models have reported effects of vitamin D on reversing neuro-cytoskeleton imbalances, but not memory deficits, and have reported the protective effects of vitamin D to be more favourable before and during, rather than after, the demyelination process.^{Reference Nystad, Torkildsen and Wergeland37} The cross-sectional design of this study prevented us from identifying any temporal association between vitamin D and NfLs. Therefore, we could not determine whether neuroinflammation with high NfL levels caused vitamin D depletion or vitamin D deficiency aggravated axonal abnormality over the disease course. In addition, some studies have suggested that the effects of vitamin D should be identified through vitamin D receptor polymorphisms rather than serum levels.^{Reference Oliveira, Magalhães, Loures, Fraga, de Souza and Guimarães38} Because vitamin D receptor polymorphisms also contribute to bipolar disorder, gene polymorphisms may influence both neuroaxonal integrity and the effects of vitamin D on brain function in bipolar disorder.^{Reference Eghtedarian, Ghafouri-Fard, Bouraghi, Hussen, Arsang-Jang and Taheri39} Future longitudinal studies are required to investigate potential temporal associations between these factors. We also analysed the association between vitamin D, NfL levels and the interaction and clinical variables, such as clinical course and DDD, for each pharmacological category. We found only one positive association, between first-generation antipsychotic dose and NfL levels (data not shown). We need further studies to capture the whole picture of antipsychotics used during the disease course to understand the impact of antipsychotics on neuroaxonal integrity. Furthermore, few trials have investigated the influence of vitamin D levels on cognitive outcomes in psychiatric patients, with only one study involving patients with schizophrenia^{Reference Krivoy, Onn, Vilner, Hochman, Weizman and Paz40} and one study on multiple sclerosis.^{Reference Darwish, Haddad, Osman, Ghassan, Yamout and Tamim41} Although both of the trials suggested a potential influence of vitamin D supplements on cognitive performance, the scarcity of studies on the subject prevents us from determining whether vitamin D can improve cognition in bipolar disorder. Our review of the literature revealed only two studies investigating the influence of vitamin D on bipolar disorder. One trial investigated bipolar depression with vitamin D deficiency, and revealed no impact of vitamin D intervention on depressive symptoms.^{Reference Marsh, Penny and Rothschild42} Another open-label trial involving patients aged <18 years with bipolar disorder with mania revealed that vitamin D3 supplementation improved mood symptoms and brain neurochemistry.^{Reference Sikoglu, Navarro, Starr, Dvir, Nwosu and Czerniak43} Neither trial included cognitive outcome measurements or changes in NfL levels. Further clinical trials focused on the influence of vitamin D intervention on cognitive outcomes in bipolar disorder and the influence of NfLs are required to verify our findings.

Strengths and weaknesses

To our knowledge, this study is the first to provide evidence of the interaction effect of vitamin D and NfLs on neurocognitive domains in patients with bipolar disorder of different ages. Our study has several limitations. First, our study sample was recruited from a tertiary psychiatric hospital. Therefore, the patients likely had a more severe degree of illness and cognitive dysfunction. In addition, only the euthymic patients with bipolar disorder were recruited in this study, who were screened for their mood status and then referred for cognitive assessment within 1 week; therefore, some of the patients may be in early remission. This potential bias in our sample ascertainment may limit the generalisability of the current findings to the entire bipolar disorder population. Second, we do not have the longitudinal cognitive profile to ensure cognitive stability across episodes. Third, we did not have a healthy control to determine whether vitamin D deficiency or the age-modified interaction effect of NfL and vitamin D levels on cognitive function were unique to bipolar disorder. Fourth, the age restrictions we used in our sample limited our findings to older and teenaged patients with bipolar disorder. Additional bipolar disorder subgroups with new onset and longer disease duration should be included in future samples. Fifth, vitamin D status was assessed by measuring total serum 25(OH)D concentrations. However, only free (unbound) and albumin-bound vitamin D reflect biological activity;^{Reference Speeckaert and Delanghe44} in addition, the effects of vitamin D should be identified through vitamin D receptor polymorphisms. Nevertheless, a study involving an Asian sample reported that total 25(OH)D concentrations were closely correlated with bioavailable 25(OH)D.^{Reference Kim, Chung, Lê, Na and Cho45} Therefore, 25(OH)D can be regarded as a primary marker of vitamin D status. Sixth, peripheral NfL levels do not represent total NfL levels in the central nervous system. However, serum measurements of NfL levels have been considered reliable in many studies.^{Reference Travica, Berk and Marx46} Finally, as the cross-sectional study design, we precluded the causal effect of our findings, and further studies to clarify the linking mechanism are needed.

Implications

We identified the interaction effect of vitamin D and NfLs on cognitive function in bipolar disorder. In addition, we discovered that age might play the role of an effect modifier. The effects of vitamin D on serum NfLs and cognitive outcomes in bipolar disorder warrant further study, with inclusion of patients with bipolar disorder with a range of mood statuses and disease courses. Longitudinal or randomised controlled trial designs are required to validate our findings and achieve more robust causal inferences and clinical implications.