Understanding HDL Cholesterol Abnormalities in Schizophrenia Patients

Introduction Schizophrenia is a major chronic psychiatric disease that profoundly impacts on various aspects of patients’ functional domains and results in their living with disability.1 In comparison to the general population, schizophrenia patients are also at an increased risk to comorbid with ageing-related physical illnesses, such as metabolic syndrome, diabetes mellitus, and cardiovascular disease, as well as premature mortality following deleterious physical health.2,3 Since cardiovascular disease contributes to the most essential cause for the premature mortality of schizophrenia,4 it is clinically important to early assess the risk of cardiovascular disease in order to help schizophrenia patients to decrease physical morbidities and related mortality gap. During the development course of metabolic syndrome, lipid abnormalities usually occurs earlier5 and have been suggested as important markers of cardiovascular disease.6,7 Among the abnormalities of lipid profiles, low high-density lipoprotein cholesterol (HDL-C) and high triglycerides (TG) contribute to the two major components of metabolic syndrome regarding atherogenic dyslipidemia.8 Low HDL-C is suggested to behave as the most powerful lipid parameter to predict the risk and the treatment outcome of coronary heart disease.9 On the other hand, the atherogenic index of plasma (AIP), defined as the ratio of TG to HDL-C post logarithmical transformation, acts as a marker for evaluating cardiometabolic risk10 and shows the strongest association with metabolic syndrome superior to other atherogenic indices in the schizophrenia population.11 On the other hand, schizophrenia patients tend to engage in sedentary lifestyles than general population, which is found to be associated with reduced functional exercise capacity and increased cardiometabolic concerns.12 One study suggested schizophrenia patients with metabolic syndrome presents poorer physical capacity measured by 6-minute walk test (6MWT) and manifests significantly different levels of four in five metabolic syndrome components, including waist circumference, blood pressure, glucose and TG than those without metabolic syndrome, except the levels of HDL-C.13 However, according to the criteria of low HDL-C in metabolic syndrome, females have higher HDL-C threshold levels than males.14 The prior work showed 6MWT differences by metabolic syndrome but not specified by low HDL-C based on sex-different thresholds of HDL-C levels. Therefore, the knowledge gap regarding whether physical capacity and sex have independent influences on low HDL-C remains unclear. Due to the important health impact of HDL-C and its multi-factorial mechanisms,15 further study is warranted to evaluate the possible modifiable risk factors with regard to low HDL-C to present the hints in order to correct the HDL-C abnormality is clinically important. To date, the predictors associated with low HDL-C in schizophrenia patients have not been established. The overall aims of the cross-sectional study were to adopt logistic regression analyses to 1) investigate the significant predictors of low HDL-C diagnosed by sex-specified HDL-C thresholds, 2) and compare three model performances by controlling for potential confounding factors. Methods Participant This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (IRB No. 202100168B0) and complied with the Declaration of Helsinki. After providing a detailed explanation of the study, written informed consent was obtained from all participants. Eligible patients with schizophrenia were recruited from Kaohsiung Chang Gung Memorial Hospital if they met the following criteria: (1) 20–60 years of age, (2) having a diagnosis of schizophrenia as defined by the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5),16 (3) being treated with an antipsychotic drug at a stable dose for at least 1 month before the study and with relatively stable psychotic symptoms, as defined by a total score on the Positive and Negative Syndrome Scale (PANSS) of <95,17 (4) no history of major physical illnesses, such as cardiovascular disease, neuromuscular disorders or renal disorders, for safe exercise participation,18 and (5) no physical disability that caused difficulty in walking. Sample Size Given effect size = 0.7 and α = 0.05, a minimum of 110 schizophrenia patients were required to achieve 95% of power using G*Power 3.1. software. Baseline Assessments Clinical Characteristics The psychiatric diagnoses of the patients with schizophrenia were performed using the Mini International Neuropsychiatric Interview (MINI), which is a short and structured diagnostic interview for assessing psychiatric disorders.19 The MINI in Mandarin Chinese has been widely used in international researches with satisfactory reliability.20 Patients’ age at onset, duration of illness and hospitalization times were obtained through interviews with the patients and a review of their medical records. The psychopathology of patients with schizophrenia was evaluated using the PANSS, which consists of three subscales representing positive, negative, and general psychiatric symptoms.21 The use of hypolipidemic drugs use and diagnosis of diabetes mellitus were recorded. Antipsychotic Treatment The patients included were treated with a stable dose of an antipsychotic drug for at least 1 month before the study began. On the basis of the properties of the muscarinic-binding affinity antipsychotics were categorized into high (clozapine or olanzapine) and low (antipsychotics other than clozapine or olanzapine) groups.22 The dose of antipsychotic drugs was recalculated based on the defined daily dose recommended by the World Health Organization Collaborating Centre for Drug Statistics Methodology.23 Biochemical Measures Metabolic indices, including glucose and lipid profiles of HDL-C, TG, total cholesterol (TC) and low-density lipoprotein cholesterol (LDL), were measured by colorimetric method after 8–10 h of overnight fasting at the Laboratory of Medicine, Kaohsiung Chang Gung Memorial Hospital. Height (centimeter) and weight (kilogram) were measured by one research staff member. Body mass index (BMI, kg/m2) was calculated by weight (in kilogram) divided by height (in meter). Low HDL-C was defined as the presence the sex-specific HDL-C thresholds:14 HDL-C <40 mg/dL (1.0 mmol/L) in males or <50 mg/dL (1.3 mmol/L) in females. AIP was calculated using the following formula: log10 (TG/HDL-C). Patients with schizophrenia were further categorized into a group with normal HDL-C and a group with low HDL-C. 6-Minute Walk Test (6MWT) Measures The physical capacity was measured by 6MWT performed in a 25-m, flat and straight indoor corridor of the study hospital site to prevent external stimuli. 6MWT has been shown to be better tolerated and to demonstrate more daily activities than other exercise tests.24 All participants were instructed to walk back and forth for six minutes as fast as possible without jogging or running according to the recommendations of the American Thoracic Society.25 Although 30-m is standard corridor for 6MWT, non-standard 25-m corridor has good reliability of schizophrenia patients and is well adopted in schizophrenia researches.26,27 The total distances walked in meters were measured and recorded as the 6MWT distances for statistical analyses, with the greater 6MWT distances suggesting superior physical capacity. A cutoff of >366-meter was calculated by the mean value minus standard error. Statistical Analysis SPSS for Windows (version 21.0; SPSS, Inc., Chicago, IL, USA) and R software 4.5.2 were used to analyze data. Variables are presented as either mean ± standard deviation or frequency. The chi-square test was used to compare categorical variables between groups. We checked the normality of variables by using Kolmogorov Smirnov Test. Descriptive statistics were calculated using the independent t test or Mann Whitney u-test to compare the continuous variables of demographic data and disease characteristics. The area under the receiver operating characteristics curve (AUC) was calculated to examine the model performance. We adopted logistic regression to determine the significant predictors correlated with low HDL-C among all participants with schizophrenia. The logistic regression modeling included variables (sex, hospitalization times, 6MWT distances, TG, education years, height and BMI) which were significantly different between groups. After using Durbin–Watson statistics, the value 1.793 indicated there was no autocorrelation. We then used variation inflation factor <10 to test multi-collinearity and found no collinearity. The predictors included sex, hospitalization times, 6MWT distances and TG, and the primary outcome was low HDL-C. Dummy codes were set for sex (female = 0, male = 1) and the group of HDL-C (normal HDL-C = 0, low HDL-C = 1). Male sex and the group of normal HDL-C were category references to female sex and the group of low HDL-C, respectively. Variables of education years, height and BMI were selected as covariates. The model fit was verified using the Hosmer–Lemeshow goodness-of-fit test, with a p-value >0.05 indicating a good fit. On the other hand, we used R software 4.5.2 with 1000 bootstrap resamples for model validation, assumption check, and calibration; accuracy, sensitivity and specificity were assessed. In Model 1, no covariates were selected to control. In Model 2, the covariate of education years was selected to control. In Model 3, the covariates of education years, height and BMI were selected to control. All three models had a good model fit according to the results of the Hosmer–Lemeshow goodness-of-fit test (all p-value > 0.05). The calibration curves of all three models also had a good fit between the predicted and actual probabilities (Supplementary Figure 1). All continuous variables had liner fit with the log-odds of the primary outcome (Supplementary Figure 2). Unstandardized coefficients (B value), standard errors (SE) and Odds ratios (OR) with 95% confidence intervals (CIs) were computed. The False Discovery Rate (FDR) correction was adopted to adjust for multiple comparisons of all p-value. P value of FDR less than 0.05 was considered statistically significant after correcting for multiple comparisons. Standardized effect size was reported as the absolute value of Cohen’s d. Results Between-Group Differences of Subjective Characteristics and Metabolic Indices A total of 133 patients with schizophrenia were recruited; 12 patients refused to participate in the study. One patient had missing data and was removed from statistical analysis. The remaining 120 participants were categorized to the group with normal HDL-C (n = 73, 60.8%) or group with low HDL-C (n = 47, 39.2%) (Figure 1). Table 1 presents a summary of the characteristics of the participants in the two groups. The group with low HDL-C had higher female proportion (76.6% vs 45.2%, p < 0.01), higher education years (13.9 ± 2.0 years vs 13.0 ± 2.2 years, p < 0.05, Cohen’s d = 0.391), shorter height (161.6 ± 7.2 cm vs 164.9 ± 9.5 cm, p < 0.05, Cohen’s d = 0.386), higher BMI (28.8 ± 6.0 kg/m2 vs 26.2 ± 5.4 kg/m2, p < 0.05, Cohen’s d = 0.464) and higher hospitalization times (6.1 ± 6.5 vs 3.5 ± 4.5, p < 0.01, Cohen’s d = 0.481) than the normal HDL-C group. Low HDL-C group (normal HDL-C = 0, low HDL-C = 1) negatively associated with height. Demographic characteristics included age and cigarette use, and other disease characteristics included the antipsychotics dose did not differ significantly between the groups. There were no significant differences of hypolipidemic drugs use and diabetes mellitus between groups. Between-Group Differences of Metabolic Indices and AIP Figure 2 shows that the group with low-HDL had higher TG values (156.7 ± 103.3 mg/dL vs 111.0 ± 75.8 mg/dL, p < 0.01, Cohen’s d = 0.522) and higher AIP (0.5 ± 0.2 vs 0.2 ± 0.3, p < 0.001, Cohen’s d = 1.297) than the group with normal HDL-C, while there were no significant differences of TC (181.9 ± 32.6 mg/dL vs 189.4 ± 36.1 mg/dL, Cohen’s d = 0.215) and LDL (107.1 ± 35.2 mg/dL vs 105.1 ± 29.8 mg/dL, Cohen’s d = 0.063) between groups. Between-Group Differences of 6MWT Distances Figure 3 presents that the group of low HDL-C had shorter 6MWT distances (426.6 ± 93.0 m vs 470.4 ± 80.4 m, p < 0.01, Cohen’s d = 0.511) than the group with normal HDL-C. By using the value of 366-meter as the cut-off point of the 6MWT distances, higher proportion of shorter 6MWT distances in the group of low HDL-C (19.1% vs 6.8%, p < 0.05) was statistically significant. Logistic Regression Analysis for Examining the Most Correlation with Low HDL-C Model 1 (AUC = 0.79, 95% CI = 0.716–0.864, accuracy/sensitivity/specificity = 0.73/0.60/0.81) revealed that female sex (B = 1.382, SE = 0.460, OR = 3.982, 95% CI = 1.617–9.808), hospitalization times (B = 0.080, SE = 0.040, OR = 1.083, 95% CI = 1.002–1.171), 6MWT distances (B = −0.005, SE = 0.003, OR = 0.995, 95% CI = 0.990–1.000) and TG (B = 0.007, SE = 0.003, OR = 1.008, 95% CI = 1.002–1.013) significantly correlated with low HDL-C after FDR correction (all p < 0.05). The above-mentioned predictors were significant correlation after controlling for covariates of education in Model 2 (AUC = 0.82, 95% CI = 0.740–0.883, accuracy/sensitivity/specificity = 0.75/0.66/0.81). Model 3 (AUC = 0.82, 95% CI = 0.742–0.892, accuracy/sensitivity/specificity = 0.76/0.64/0.84) controlled for education, height and BMI, and found that female sex (B = 1.382, SE = 0.648, OR = 3.983, 95% CI = 1.118–14.185), hospitalization times (B = 0.086, SE = 0.042, OR = 1.090, 95% CI = 1.003–1.184), 6MWT distances (B = −0.007, SE = 0.003, OR = 0.993, 95% CI = 0.987–0.999) and TG (B = 0.008, SE = 0.003, OR = 1.008, 95% CI = 1.002–1.014) (all p < 0.05) remained significant (Table 2). Both Model 2 and 3 had superior performance than Model 1, while Model 3 showed relatively optimal accuracy and specificity with similar performance of Model 2. Discussion To our knowledge, this is the first study to date to evaluate the predictors of low HDL-C in schizophrenia patients. The study provides the following insights in schizophrenia patients: (1) the low HDL-C group had higher female proportion, higher education years, shorter height, higher BMI, higher hospitalization times, higher TG values, higher AIP level and poorer physical capacity, (2) female sex, higher hospitalization times, poor physical capacity and high TG levels had increased risks of low HDL-C in all three models, in which Model 3 had superior accuracy and specificity, but with similar performance of Model 2. The Association Between Sex and Low HDL-C The results regarding the effect of sex on serum lipids remain inconsistent.28 The prevalence of low HDL-C in this study was 39.2%, which is higher than that (25%) in the general population,29 and lower than previous finding in treated patients (44.7%).30 Our results show that female schizophrenia is predominant in higher risks of manifesting low HDL-C is supported by previous studies.29–31 The overall prevalence of metabolic syndrome is increasing worldwide, and the trend is higher in females than males, especially in younger age. Obesity is suggested to play an important role in this trend.28 The difference in the distribution of metabolic syndrome components between male and female contributes to the different trend of metabolic syndrome.31 Lipids accumulation patterns differ between females and males. Although peripheral and central obesity could be found in both sex, females are more often to develop peripheral obesity, while males are more frequent to develop central obesity. After menopause, the lipid accumulation shifts from peripheral to central obesity.28–31 However, the studies focusing on investigating the effect of estrogen on the prevalence of metabolic syndrome components reveal mixed findings. Other sex-related disparities such as shorter height in females and various patterns of lifestyle activities and nutrition intake between males and females are also considered.29–33 In our study, we suggest that female is a significant predictor of low HDL-C in schizophrenia patients after controlling for height as well as education years and BMI. This finding gives an important hint to screen HDL-C levels in order to assess if the HDL-C meeting the diagnosis of low HDL-C in schizophrenia, especially in female patients. The Association Between Hospitalization Times and Low HDL-C Psychiatric hospitalizations is an important factor in the illness course in schizophrenia, which exerts an unfavorable effect on treatment outcomes and likelihood to recover from acute phase, especially re-hospitalizations.34 However, other researchers found that deinstitutionalization may be associated with increasing cardiovascular mortality ratios for all causes of death.35 Despite this, chronic schizophrenia with antipsychotics treatment has higher risk of cardiometabolic abnormalities than those in the first episode through lifestyle interventions.30 In our study, we found that increased hospitalization times after psychosis relapsing is a significant predictor of low HDL-C in schizophrenia patients, who also manifested higher cardiometabolic risk indexed by AIP levels. This exploratory result hints that the treatment to prevent relapses is essential for reducing cardiometabolic comorbidity in schizophrenia patients. More investigations are warranted to explore the underlying mechanisms regarding the relationship between hospitalization times and low HDL-C. The Association Between Physical Capacity and Low HDL-C Enhancing physical activity and reducing sedentary patterns are suggested as essential strategies in modifying lifestyle to prevent cardiovascular disease.36 Some studies reported that physical activity has beneficial effect on improving HDL-C and lowering TG.37–39 On population cohort study observed that physical activity at different ages in adulthood has positive effect on HDL-C; though the association between physical activity and lipid profile became nonsignificant after controlling for lifestyle factors and BMI, except males with activity higher than moderate intensity.40 On the contrary, one population cohort study found that, after controlling for confounding factors, physical activity correlated with increasing HDL-C level significantly in female, but not in male.41 Another study observed that physical activity has no significant correlation with HDL-C level in female.42 Previous study observed that schizophrenia patients with metabolic syndrome achieve shorter 6MWT distances than those without metabolic syndrome though there is no difference of HDL-C levels between groups.13 In our study, we observed that patients with shorter 6MWT distances showed significantly higher risks of low HDL-C than those with normal HDL-C. The result provides important information to fill the prior knowledge gap; if taking into account of sex-specific HDL-C threshold to assess the diagnosis of low HDL-C, the association between physical capacity and low HDL-C is significant. The Association Between TG and Low HDL-C HDL-C particles consist of various subgroups of proteins with abundant apolipoprotein A-I, cholesterol, and phospholipids and present various functions, including antioxidant, anti-inflammatory, and immunomodulatory effects.43 HD-C metabolism is closely linked with TG metabolism. During the HDL particle maturation process, the hydrolysis of TG is suggested to play an important role in transferring cholesterol and apolipoproteins to HDL particles as surface components.44 Alterations in plasma TG levels may result in a significant change in HDL-C catabolic pathways.15 It has been suggested that high TG levels which contribute to TG-rich lipoproteins of HDL-C particles could increase the catabolism rate of HDL-C.15–45 Increased TG together with decreased HDL-C concentrations are associated with increased risk of atherosclerosis and the following cardiovascular disease.46 In our study, the result that the group with low HDL-C had higher AIP level provides information that schizophrenia patients with low HDL-C have more severe inflammation and higher risks in manifesting atherosclerosis. On the other hand, we observed that higher TG level demonstrates an independent factor in correlation with low HDL-C, which is in agreement with the above-mentioned metabolism process of HDL-C and TG. The conclusion implies clinicians keep in mind the modifiable factors regarding elevated TG levels in schizophrenia patients to prevent low HDL-C and the following cardiovascular disease risks in schizophrenia. Limitations This study has some limitations that should be considered. First, this work suggests a cross-sectional correlation among the independent factors and low HDL-C in patients with schizophrenia and therefore does not provide information regarding the possible directionality of any associations. Second, this design did not control for the diet, adiposity distribution, medication effects, and menopausal status, which may be associated with lipid profiles. For example, high percentage of carbohydrate intake is associated with decreased HDL-C and increased TG levels and lower HDL levels.47 Further research is warranted to include the variables for further clarification. Finally, the relatively small sample size may limit the interpretation of this study, and the results may not be generalized to all patients with schizophrenia. Conclusion In summary, the results provide an important clinical utility for physicians to assess low HDL-C, especially for female schizophrenia with higher hospitalization, shorter 6MWT distances or higher TG due to increased susceptibility to cardiometabolic risks. Future prospective studies with larger samples are necessary to further clarify the causal relationship among the various metabolic indices and physical capacity and low HDL-C during their treatment with antipsychotics.