Association of bone turnover markers and cognitive function in Chinese chronic schizophrenia patients with or without vitamin D insufficiency

We enrolled 118 inpatients with chronic schizophrenia from the Chaohu Hospital of Anhui Medical University. All subjects matched the following criteria for inclusion: (1) consistent with the diagnosis of schizophrenia by two clinical psychiatrists according to the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5); (2) Han Chinese, 18 to 60 years of age, education years ≥ 5 ; (3) had at least three years of disease duration, and received a fixed-dose of antipsychotic medications in at least 12 months prior to being recruited [21]; (4) were able to understand and sign the informed consent; (5) had no known history of alcohol dependence or drug abuse.

Subjects who meet the following criteria have been excluded: (1) met the DSM-5 diagnostic criteria for serious psychiatric disorders other than schizophrenia; (2) had serious physical diseases such as heart disease, cerebral infarction, and chronic liver or kidney disease; (3) had known endocrine disorders that affect bone metabolism, like diabetes or thyroid disorders [11]; (4) had used drugs that clearly affect bone metabolism or vitamin supplements in addition to antipsychotic drugs within three months; (5) pregnant or lactating women.

This study had the approval of the Ethics Committee of Chaohu Hospital, affiliated with Anhui Medical University (Approval No. kyxm-202211-002), with all participants signed informed consent.

All subjects completed an interview. General demographics and clinical characteristics of all participants were collected by the researchers. PANSS was used by two trained psychiatrists to assess psychiatric symptoms in schizophrenia patients. Antipsychotic treatment doses were converted to chlorpromazine equivalent doses through the internationally recognized defined daily doses (DDDs) method [22].

We used the Chinese version of Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) for measuring the cognitive function of patients. The scale consists of five indicators: immediate memory, visuospatial/constructional, language, attention, and delayed memory [23]. The scale has shown adequate validity and repeat measurement reliability in Chinese populations [24].

All subjects were collected 20ml of fasting venous blood from 6:00 am to 8:00 am on the day of the interview assessment. Fasting blood glucose (FBG), triglyceride (TG), total cholesterol (TC), low-density lipoprotein (LDL), and high-density lipoprotein (HDL) were measured. FBG was measured by the oxidase method. TG was determined by the TG_2 reagent method. The Cholesterol_2 Reagents method was used to detect TC. LDL was measured using the LDL Cholesterol Direct method. HDL was measured using the Direct HDL Cholesterol Reagents method. We used 10 ml of blood samples to detect BTMs and vitamin D levels. The plasma separated by centrifugation is stored in a -80 °C refrigerator until the time of the assay. We detected CTX, PINP, OCN, OPN and 25-OH-D by enzyme-linked immunosorbent assay (ELISA) method on a Rayto RT-6100 microplate analyzer.

Then, in the total sample and in each subgroup, Spearman correlations analysis was performed between BTMs and cognitive function. Multivariate stepwise regression analysis was performed with the RBANS scores for each domain and the total score as dependent variables to further determine the relationship between BTMs and cognition in the VDS, VDI group and total sample, and some potential confounders were included in the analysis, such as age, gender, years of education, BMI, PANSS score, drug equivalents, FBG, blood lipids and whether the patient is taking benzodiazepines and Benzhexol.

In total, 118 subjects completed the entire study process. Among them, VDS has 72, and VDI has 46. Table 1 shows that there were no significant differences between VDS and VDI in terms as to age, sex, years of education, BMI, drug equivalent, and disease duration. The PANSS total score, FBG, TG, TC and LDL of VDI were slightly higher than that of VDS, and the high-density lipoprotein was slightly lower than that of VDS, but there was no statistical significance.

In Table 2, compared with VDS group, the bone formation index in VDI group was significantly lower, with a statistically significant difference in OCN (t = 2.74, p = 0.007), while the bone resorption index was higher in the VDI group, with a statistically significant difference in CTX (t = -3.49, p = 0.001). But there was no significant difference in cognitive function between the VDS and VDI groups.

Due to the significant differences in BTMs between VDS and VDI groups, we analyzed the correlation between BTMs and cognitive function in the two groups and total sample. Table 3 shows that in the total sample, CTX and language function showed a positive correlation (r = 0.229, p < 0.05), while PINP was negatively correlated with language function (r = -0.189, p < 0.05). This correlation is consistent with the VDS group (all p < 0.05), whereas only in the VDS group, PINP and RBANS total scores showed a negative correlation (r = -0.240, p < 0.05). In the VDI group, PINP was positively associated with delayed memory (r = 0.320, p < 0.05), in contrast to OPN, which was negatively associated with delayed memory (r = -0.330, p < 0.05). In both groups and total sample, there was no significant correlation to be found between OCN and any cognitive function variable.

In order to further clarify the correlation between the above BTMs and cognitive function, we performed stepwise multiple linear regression and corrected potential confounding factors (age, gender, years of education, BMI, PANSS score, drug equivalents, FBG, blood lipids and whether the patient is taking benzodiazepines and Benzhexol) in each group and the total sample. Table 4 shows that in the total sample, CTX and language function exhibited independent positive correlation (B = 0.050, p = 0.027, 95%CI = 0.006 ~ 0.093, R² change = 0.042). In the VDS group, there was an independent negative correlation between PINP and language function (B = -0.250, p = 0.031, 95%CI = -0.477~-0.024, R² change = 0.065), while in the VDI group, the positive correlation was still present between OPN and delayed memory. (B = -0.586, p = 0.036, 95%CI = -1.132~-0.041, R² change = 0.083).

The significant impacts of vitamin D on bone metabolism are well established, and these effects are reflected in BTMs. To meet the calcium balance in the body, vitamin D and its metabolites enhance mineral deposits in the bone matrix when sufficient calcium is available, while in calcium deficiency, vitamin D promotes bone resorption while inhibiting bone mineralization [25]. This partly explains that in our study, patients with chronic schizophrenia who had vitamin D concentrations greater than 30 ng/ml had lower CTX and OPN levels but higher PINP and OCN levels. However, the results of various studies are inconsistent regarding the relationship between vitamin D concentrations and bone turnover marker levels. A large cross-sectional study of Chinese adults found that different vitamin D concentrations were negatively associated with bone metabolism markers, whether it was osteogenic or osteoclastic markers [26]. Another study, also conducted in Chinese adults, found that supplementation of vitamin D did not significantly affect CTX and PINP levels, while subjects with vitamin D concentrations above 30 ng/mL had increased CTX but no difference in PINP [14]. In contrast, it has been shown that in young postmenopausal women, improved vitamin D status can result in a significant decrease in serum PINP and CTX levels [27].

The association between vitamin D levels and cognitive function is inconsistent across studies. In healthy individuals, there may be an association between low vitamin D levels and cognitive function [15], and several RCT studies have shown that vitamin D supplementation positively affects cognitive function in people whose vitamin D levels were low at baseline [28‐30]. However, in a cross-sectional study of bipolar patients, there was no relationship between cognitive function and vitamin D levels in bipolar patients [31]. Consistent with this finding, our findings showed no difference between subject RBANS scores for VDS and VDI in patients with chronic schizophrenia.

Our study found that in chronic schizophrenic patients with vitamin D insufficiency, OPN concentrations were inversely correlated with delayed memory scores. OPN and vitamin D are not only closely related to the dynamic changes of bones but also can be used as markers of oxidative stress or inflammation [32]. The relationship between OPN and cognitive function has been verified in multiple studies. A study in a mouse model of AD found that the level of OPN-producing CD11c + microglia was strongly correlated with the extent of cognitive impairments and AD neuropathology [33]. In a cross-sectional survey of patients with cerebrovascular disease and neurodegenerative dementia, increased OPN was significantly related to vascular cognitive deficits and significantly related to neuroimaging markers of cerebrovascular disease and neurodegeneration [34]. One study found an association between OPN and schizophrenia and that OPN gene expression was upregulated in first-episode psychosis (FEP) patients [35]. Although current studies point to OPN affecting cognitive function by acting in oxidative stress-related pathways, Although the current study points to OPN affecting cognitive function by acting on oxidative stress-related pathways, the interaction between bone metabolism and OPN and, eventually, its impact on cognitive function also deserves attention.

Interestingly, our study found that in patients with sufficient vitamin D levels, PINP and language function were independently associated even after accounting for confounding factors, which was rarely reported in previous studies. And correspondingly, when vitamin D levels were not considered, CTX levels correlated with language function in the total sample. In an RCT of 103 community-dwelling older adults, no association was found between PINP and domains of cognitive function at baseline and after 18 months, whereas CTX, although associated with overall cognitive change at 18 months, was not correlated with any one specific cognitive domain and did not independently predict cognitive decline [36]. Although PINP cannot directly cross the BBB to affect the brain, the pathogenesis of schizophrenia appears to affect bone metabolism. The dopamine hypothesis is the most classic hypothesis on the pathogenesis of schizophrenia in the past forty years [37], and dopamine also affects bone metabolism. For example, dopamine inhibits osteoclast differentiation through D2 receptors [38], and activation of dopamine receptor D1 not only stimulates bone differentiation but also reduces bone loss [39], while PINP reflects the level of bone formation. Our research shows that there is a correlation between PINP and the cognitive function of patients with schizophrenia, which may be produced through the different effects of dopamine on various receptors, but the specific causal relationship still needs further study.

It is worth mentioning that OCN has proved its correlation with cognitive function in many previous studies [8, 40], but it was not found in our study. As for why our results did not show the positive associations shown in previous studies, we speculate that the previous associations were found in more homogeneous populations, such as elderly women and obese patients. Results of an RCT of community-dwelling older adults support our negative findings [36].

Importantly, our study found that the correlations between the levels of bone turnover markers and cognitive function appeared to be reversed at different vitamin D concentrations. This is possible because there is some threshold effect between vitamin D and BTMs, thus exhibiting different effects on cognition. In previous studies, different cut-off points of vitamin D levels could be used as markers of changes in BTMs. This potential threshold effect seems to result from the effect of vitamin D on PTH. In various previous studies, it was found that a rapid increase in PTH levels was observed when vitamin D concentrations were below 20 ng/ml [26], while PTH levels showed a steady state when vitamin D concentrations were above 30 ng/ml [41]. And in addition to the familiar effects of PTH on bone, there are many studies that have found a correlation between PTH and cognitive impairment. Possible mechanisms of action include PTH binding to receptors in the brain and inducing apoptosis through calcium overload [42]; or affecting cerebral blood flow through vasoactive effects, resulting in decreased cognitive function [43].

Our study has some limitations. Firstly, our current study presents only a cross-sectional result, and further longitudinal studies to clarify the causal relationship between bone metabolism and cognitive function are needed. Secondly, this study lacked a healthy control group to determine the differences in bone metabolism between schizophrenic and normal populations. Thirdly, our subjects are only inpatients. Although their diet and daily routine are relatively consistent—these are considered to have an impact on bone metabolism and vitamin D levels, the results may still be biased by the selection of samples. Finally, we could not rule out the effects of different drugs on skeletal or cognitive performance.