Serum Apolipoprotein B Levels and Apolipoprotein B/Apolipoprotein A1 Ratio as Predictors of Dyslipidemia Risk in Patients Treated with Second-Generation Antipsychotics: A Retrospective Cohort Study

Introduction

Second-generation antipsychotics (SGAs), also known as atypical antipsychotics, are a class of medications, such as olanzapine, clozapine, risperidone, aripiprazole, and amisulpride. Compared with first-generation antipsychotics (FGAs), SGAs demonstrate superior efficacy, greater improvements in quality of life, and a reduced incidence of extrapyramidal side effects, making them the first-line treatments for various psychiatric disorders.^1–3 However, treatment with SGAs is associated with the development of metabolic syndrome (MetS).⁴ The prevalence of MetS among patients treated with SGAs is 47.4%, which is significantly higher than that in first-episode patients and antipsychotic-naïve patients (both approximately 10%).^5,6

Patients with MetS face double the cardiovascular disease (CVD) risk and a more than fivefold increased risk of diabetes within 5–10 years.⁷ In contrast, those with schizophrenia experience a 16- to 19-year reduction in life expectancy compared with the general population, with CVD being the leading cause of death.^8,9 The pathogenesis of MetS is multifactorial, with dyslipidemia contributing to its development. Validated early-stage biomarkers for risk stratification of future dyslipidemia among patients receiving SGAs would allow pre-symptomatic intervention, thereby preventing the development of MetS. Nevertheless, such biomarkers are currently unavailable.

Apolipoproteins serve crucial regulatory roles in maintaining lipid metabolic homeostasis.¹⁰ Apolipoprotein A1 (ApoA1) is the main apolipoprotein component of high-density lipoprotein (HDL), which can promote the reverse transport of peripheral cholesterol to the liver, thereby exerting an anti-atherosclerotic effect.¹⁰ In contrast, apolipoprotein B (ApoB) is responsible for transporting cholesterol to peripheral tissues, which may enhance the formation of atherosclerotic thrombosis.¹¹ The ratio of ApoB to ApoA1 (ApoB/ApoA1 ratio) can reflect the balance between pro-atherosclerotic factors and anti-atherosclerotic factors in plasma. Several studies have found that serum ApoB levels and the ApoB/ApoA1 ratio can predict the risk of developing metabolic diseases, even outperforming traditional biomarkers.^12–15 However, to the best of our knowledge, no study has reported the predictive value of baseline serum ApoB levels and the ApoB/ApoA1 ratio for dyslipidemia risk associated with SGAs.

This study aimed to investigate the predictive value of baseline serum ApoB levels and the ApoB/ApoA1 ratio for the development of dyslipidemia within 4 months in patients initiating SGAs.

Materials and Methods

Study Population

This single-center retrospective cohort study was conducted at Zhongshan Third People’s Hospital in China, using medical records of patients hospitalized between March 2019 and March 2025. This study was approved by the Ethics Committee of the Zhongshan Third People’s Hospital (Ethics Approval No. SSYLL-KY-20231103). As a retrospective observational study with anonymized data, the ethics committee formally waived the requirement for informed consent in this study.

Inclusion criteria: (1) inpatients aged ≥18 years, (2) no antipsychotic treatment for at least 3 months prior to enrollment, (3) absence of dyslipidemia or hyperglycemia at admission.

Exclusion criteria: (1) endocrine disorders, severe hepatic/renal/cardiac/pulmonary dysfunction, autoimmune diseases, malignancies, allergies, or active infections, (2) use of glucocorticoids or medications for diabetes/dyslipidemia in the preceding 3 months, (3) pregnancy or lactation.

Variables

Data Collection

Demographic and clinical data were extracted from medical records, including: sex, age at admission, baseline weight, smoking status, psychiatric diagnosis (classified according to ICD-10 criteria), total cumulative dose and duration of SGA treatment, concomitant medications, baseline serum levels of ApoB and ApoA1, and serial serum lipid levels measured at baseline and monthly intervals for approximately four months post-admission. The follow-up period spanned from admission until the earliest occurrence of any of the following events: dyslipidemia diagnosis, hospital discharge, or the 4-month endpoint.

Laboratory Measurements

Overnight (≥ 8 hours) fasting venous blood samples were collected from participants. Serum glucose and apolipoprotein levels, along with lipid profiles (including the levels of total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), HDL cholesterol (HDL-C), and triglycerides (TG)), were subsequently analyzed using a Beckman Coulter AU680 automated analytical system (Beckman Coulter, Inc., Fullerton, CA, USA). The instrument was used in strict accordance with protocols following daily quality control inspections.

Diagnostic Criteria

According to the Chinese guidelines for the prevention and treatment of dyslipidemia in adults (2016 revision),¹⁶ dyslipidemia was defined as meeting any of the following criteria: (1) TC ≥ 5.2 mmol/L, (2) LDL-C ≥ 3.4 mmol/L, (3) HDL-C < 1.0 mmol/L, or (4) TG ≥ 1.7 mmol/L; hyperglycemia was defined as fasting blood glucose (FBG) > 6.1 mmol/L.

SGA Metabolic Risk

SGAs were stratified into three metabolic risk categories:¹⁷ (1) high risk: olanzapine, clozapine; (2) intermediate risk: risperidone, paliperidone, quetiapine; (3) low risk: amisulpride, aripiprazole, ziprasidone, lurasidone. Consistent use of a single SGA was defined as SGA monotherapy, while SGA polytherapy referred to either a combination of multiple SGAs or a switch between SGAs. Patients receiving SGA polytherapy spanning multiple metabolic risk categories were classified into the highest corresponding metabolic risk tier covered by their regimen.

SGA Dosage Standardization

Individual SGA dosages were standardized into olanzapine-equivalent units using the World Health Organization’s Defined Daily Dose (DDD) methodology,¹⁸ with the following conversion ratios: 1 mg olanzapine = 30 mg clozapine = 0.5 mg risperidone = 0.6 mg paliperidone = 40 mg quetiapine = 40 mg amisulpride = 1.5 mg aripiprazole = 8 mg ziprasidone = 6 mg lurasidone. The average daily dose of SGA was derived from the total dose divided by the total days of treatment.

Concomitant Medication Classification

Concomitant medications were categorized into five pharmacological categories: anxiolytics, mood stabilizers, hypnotics, antidepressants, and FGAs. No data were collected for medications used for ≤ 7 days or those not expected to impact serum lipid levels.

Statistical Analysis

Continuous variables were expressed as median (interquartile range [IQR]) and compared using the Mann–Whitney U-test. Categorical variables were expressed as numbers (%), and compared using Pearson chi-square or Fisher’s exact test, as appropriate.

To comprehensively evaluate the predictive value of baseline serum ApoB levels and the ApoB/ApoA1 ratio, we employed a multi-faceted analytical strategy designed to yield complementary insight: (1) Identification of Clinically Actionable Thresholds: A maximally selected log-rank statistic analysis was applied to identify the optimal cut-off points for baseline serum ApoB levels and the ApoB/ApoA1 ratio to predict dyslipidemia within 4 months of initiating treatment with SGAs. (2) Risk Assessment Using Dichotomized Variables: Multivariable Cox proportional hazards models were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) for the risk of dyslipidemia, comparing patients above versus below the identified cut-off points. (3) Risk Assessment Using Continuous Variables: To assess the dose-response relationship without relying on arbitrary thresholds, we also analyzed baseline serum ApoB levels and the ApoB/ApoA1 ratio as continuous variables in separated multivariable Cox models. To account for potential confounding, models incorporated clinically relevant covariates: age, sex, baseline weight, smoking status, antipsychotic-naïve status, SGA monotherapy, SGA metabolic risk category, and concomitant medication use. The Kaplan-Meier method with Log rank testing was used to compare the cumulative incidence of dyslipidemia across groups defined by the optimal cut-off points for baseline ApoB levels and the ApoB/ApoA1 ratio.

All p values were two-tailed, with statistical significance defined as p < 0.05. Statistical analyses were performed using IBM SPSS (version 25.0) and R software (version 4.3.0).

Results

Baseline Characteristics of Subjects

Baseline characteristics of patients are summarized in Table 1. A total of 311 patients were enrolled in this study. Among them, thirty-three patients (10.6%) with missing baseline ApoA1 measurements were excluded from the ApoB/ApoA1 ratio analyses. Based on a maximally selected log-rank statistic analysis (Figure 1), the optimal cut-off points for stratifying patients into two distinct groups were determined to be 0.70 g/L for baseline serum ApoB levels and 0.45 for the ApoB/ApoA1 ratio. Compared with the high ApoB group (> 0.70 g/L), patients in the low ApoB group (≤ 0.70 g/L) were younger, had a lower proportion of smokers, and exhibited lower baseline levels of TC, LDL-C, and TG but higher HDL-C. Separately, when stratified by the ApoB/ApoA1 ratio, the low ratio group (≤ 0.45) exhibited lower baseline weight and levels of TC and LDL-C but higher levels of HDL-C and FBG than the high ratio group (> 0.45).

Table 1 Baseline Characteristics of Patients Grouped by the Optimal Cut-off Points for Baseline ApoB Levels and the ApoB/ApoA1 Ratio

Figure 1 Determination of the optimal cut-off values of 0.70 g/L for baseline serum apolipoprotein B (ApoB) levels and 0.45 for the ratio of apolipoprotein B to apolipoprotein A1 (ApoB/ApoA1 ratio) using a maximally selected log-rank statistic analysis.

SGA Treatment of Subjects

SGA treatment and concomitant medications of patients are presented in Table 2. The most commonly prescribed SGAs were olanzapine (182, 58.5%), risperidone (119, 38.3%), quetiapine (49, 15.8%), aripiprazole (30, 9.6%), paliperidone (14, 4.5%), clozapine (12, 3.9%), amisulpride (12, 3.9%), ziprasidone (2, 0.6%), and lurasidone (1, 0.3%). Except for a significantly higher proportion of antidepressant use in the high ApoB/ApoA1 ratio group, no significant differences were observed in other medication usage (including SGA treatment and concomitant medications) between groups stratified by ApoB levels or the ApoB/ApoA1 ratio.

Table 2 SGA Treatment and Concomitant Medications of Patients Grouped by the Optimal Cut-off Values for Baseline ApoB Levels and the ApoB/ApoA1 Ratio

Baseline Serum ApoB Levels and the ApoB/ApoA1 Ratio in Relation to Dyslipidemia Risk

In the overall cohort, the proportion of patients with new-onset dyslipidemia at months 1, 2, 3, and 4 was 28.0% (87/311), 25.7% (80/311), 7.7% (24/311), and 1.6% (6/311), respectively. Corresponding proportions in the ApoB/ApoA1 ratio subgroup were 28.4% (79/278), 25.5% (71/278), 7.6% (21/278), and 2.2% (6/278).

Table 3 presents the association of baseline serum ApoB levels and the ApoB/ApoA1 ratio with incident dyslipidemia, analyzed as continuous variables. Elevated baseline serum ApoB levels and a higher ApoB/ApoA1 ratio were associated with an increased risk of dyslipidemia. For every 0.1 g/L increase in baseline serum ApoB levels, the risk of incident dyslipidemia increased by 34% (adjusted HR 1.34, 95% CI: 1.21–1.48, p < 0.001). Similarly, for every 0.1 unit increase in the baseline ApoB/ApoA1 ratio, the risk of dyslipidemia increased by 20% (adjusted HR 1.20, 95% CI: 1.10–1.30, p < 0.001).

Table 3 Association of Baseline Serum ApoB Levels and the ApoB/ApoA1 Ratio with Incident Dyslipidemia

Table 4 presents the risk of dyslipidemia stratified by the optimal cut-off points for baseline serum ApoB levels and the ApoB/ApoA1 ratio. According to multivariate Cox proportional hazards model 1 (adjusted for age and sex), compared with their respective low groups, the high ApoB group exhibited a significantly elevated risk (HR = 2.48, 95% CI: 1.74–3.54, p < 0.001), while the high ApoB/ApoA1 ratio group showed an even greater risk (HR = 3.29, 95% CI: 1.73–6.23, p < 0.001). These significant associations persisted in the fully adjusted model 2 (ApoB: HR = 2.98, 95% CI: 2.05–4.32, p < 0.001; ApoB/ApoA1 ratio: HR = 3.17, 95% CI: 1.62–6.22, p = 0.001).

Table 4 Risk of Dyslipidemia Stratified by the Optimal Cut-off Values for Baseline Serum ApoB Levels and the ApoB/ApoA1 Ratio

Figure 2 shows Kaplan-Meier curves indicating a significantly higher cumulative incidence of dyslipidemia among participants in the high ApoB group and those in the high ApoB/ApoA1 ratio group compared with their respective low-level counterparts (Log rank test, both p < 0.001).

Figure 2 Kaplan-Meier cumulative incidence curves for dyslipidemia incidence, stratified by the optimal cut-off points for baseline apolipoprotein B (ApoB) levels and the ratio of apolipoprotein B to apolipoprotein A1 (ApoB/ApoA1 ratio).

Discussion

SGAs are widely and effectively used to treat mental disorders, including schizophrenia, bipolar mania, depression, anxiety, sleep disorder, obsessive-compulsive disorder, autistic disorder, Tourette’s disorder, posttraumatic stress disorder, and delirium.¹ However, their clinical utility is frequently accompanied by metabolic adverse effects, such as dyslipidemia and subsequent MetS.⁴ As demonstrated in our study, over 50% of patients developed dyslipidemia within two months of initiating SGAs. In addition to elevating CVD and diabetes risk, these metabolic side effects have also been associated with poorer performance on specific cognitive domains in patients with schizophrenia and comorbid MetS, compared with those without MetS.^7,19 As patients with mental disorders typically require long-term antipsychotic treatment, it is critical to balance efficacy and safety. Long-term antipsychotic use demonstrates no elevated hospitalization risk from severe physical morbidity while correlating with significantly reduced mortality.²⁰ Conversely, antipsychotic tapering or discontinuation over several months elevated relapse risk versus maintenance treatment, without improving social functioning or altering clinical/social outcomes at two-year follow-up.²¹ These collective findings pose a major therapeutic challenge, as effective pharmacotherapy must achieve dual objectives: controlling psychiatric symptoms while simultaneously preventing dyslipidemia and subsequent MetS.

SGAs exerted significantly differential effects on serum levels of TC, LDL-C, HDL-C, and TG, manifested as three distinct metabolic risk profiles: (1) clozapine and olanzapine had the highest metabolic risk; (2) risperidone, paliperidone, and quetiapine showed intermediate metabolic risk; (3) amisulpride, aripiprazole, brexpiprazole, cariprazine, lurasidone, and ziprasidone had the lowest metabolic risk.^17,22 The identification of biomarkers capable of predicting antipsychotic-induced dyslipidemia risk would allow optimizing SGA selection, balancing symptom control with prevention of lipid abnormalities and subsequent MetS, thereby advancing personalized antipsychotic therapy. However, as far as we know, no study has reported such candidate biomarkers. The absence of validated biomarkers complicates SGA selection in mental disorders.

In this retrospective study, baseline serum ApoB levels and the ApoB/ApoA1 ratio demonstrated strong and consistent predictive value for dyslipidemia risk. Using a maximally selected log-rank statistic analysis, we identified the optimal cut-off values of 0.70 g/L for serum ApoB levels and 0.45 for the ApoB/ApoA1 ratio to stratify patients into distinct risk groups. Kaplan-Meier curves confirmed a significantly higher cumulative incidence of dyslipidemia in both the high ApoB and the high ApoB/ApoA1 ratio groups compared to their low-level counterparts (Log rank test, both p < 0.001). Accordingly, patients exceeding these thresholds showed substantially elevated risks: the high ApoB group exhibited a 2.98-fold higher risk (95% CI: 2.05–4.32, p < 0.001), and the high ApoB/ApoA1 ratio group had a 3.17-fold higher risk (95% CI: 1.62–6.22, p = 0.001) compared with their low-risk counterparts, according to multivariate Cox proportional hazards models. Moreover, when analyzed as continuous measures, each 0.1 g/L increase in baseline serum ApoB levels was associated with a 34% (adjusted HR 1.34, 95% CI: 1.21–1.48, p < 0.001) increased risk of incident dyslipidemia, and each 0.1 unit increase in the baseline ApoB/ApoA1 ratio conferred a 20% (adjusted HR 1.20, 95% CI: 1.10–1.30, p < 0.001) increased risk.

The predictive value of ApoB and the ApoB/ApoA1 ratio for dyslipidemia in patients treated with SGAs aligns with the central roles these apolipoproteins play in lipid metabolism, which are pathways that are directly disrupted by SGAs. ApoB serves as the core structural protein of pro‑atherogenic lipoproteins, including LDL and very low-density lipoproteins (VLDL), which deliver cholesterol to peripheral tissues, whereas ApoA1 is the principal component of anti‑atherogenic HDL particles that mediate reverse cholesterol transport.^10,11 SGAs perturb intracellular cholesterol homeostasis through multiple interconnected mechanisms.²³ First, they impair the trafficking of LDL‑derived cholesterol out of endolysosomal compartments, resulting in intracellular accumulation and compensatory upregulation of LDL receptor‑mediated uptake. Second, SGAs influence the secretion and metabolism of ApoB‑containing VLDL and ApoA1‑containing HDL particles, which promotes ApoB secretion and VLDL production while often reducing plasma HDL levels. Collectively, these disruptions shift lipid metabolism toward a pro‑atherogenic state, characterized by elevated ApoB and an increased ApoB/ApoA1 ratio. Therefore, elevated baseline levels of these biomarkers likely identify individuals with pre‑existing susceptibility in lipid regulatory pathways, which becomes markedly exacerbated upon SGA treatment. This is supported by our data: compared with the low-level group, both the high ApoB group and the high ApoB/ApoA1 group showed significantly higher baseline levels of TC and LDL-C, as well as significantly lower HDL-C. The high ApoB group also demonstrated significantly higher TG levels. Although the high ApoB/ApoA1 group showed a non-significant increase in TG (IQR: 0.7–1.1 vs 0.6–1.0, p = 0.199), the overall profile aligns with a trend toward dyslipidemia.¹⁶ This mechanistic understanding further reinforces the utility of these metrics in clinical practice. Notably, ApoA1 and ApoB levels can be reliably measured under non-fasting conditions, as their concentrations remain unaffected by elevated TG levels.²⁴ This property, combined with their strong pathophysiological basis, makes ApoB levels and the ApoB/ApoA1 ratio more reliable predictors of dyslipidemia risk in patients treated with SGAs. These findings are consistent with previous research which has identified ApoB levels and the ApoB/ApoA1 ratio as independent and powerful predictors of MetS.^14,25,26 Therefore, ApoB and the ApoB/ApoA1 ratio serve as practical tools to facilitate individualized antipsychotic therapy selection guided by clinical presentation, patient tolerance, and metabolic risk profiles.

This study observed that the low-risk group stratified by the ApoB/ApoA1 ratio had a significantly smaller sample size (n = 30) than the group stratified by ApoB levels alone (n = 96), reflecting the inherent biological properties of this composite metric. Achieving an ApoB/ApoA1 ratio ≤ 0.45, which requires both low ApoB levels and high ApoA1 levels, is challenging given the significantly lower serum ApoA1 levels observed in patients with schizophrenia compared with healthy controls.²⁷ Consequently, a low ApoB/ApoA1 ratio represents a stricter criterion for metabolic health than isolated low ApoB levels. Consistent with these findings, a systematic review and meta-analysis reported that children and adolescents with MetS showed elevated ApoB levels, an increased ApoB/ApoA1 ratio, and reduced ApoA1 levels.²⁸ Multivariable Cox proportional hazards models further demonstrated that, compared with their respective low-risk groups, the high ApoB/ApoA1 ratio group had a significantly higher hazard ratio for dyslipidemia than the high ApoB group. This pattern was consistent in both the age-sex-adjusted model (High ApoB/ApoA1 ratio: HR = 3.29, 95% CI: 1.73–6.23, p < 0.001; High ApoB: HR = 2.48, 95% CI: 1.74–3.54, p < 0.001) and the fully adjusted model (High ApoB/ApoA1 ratio: HR = 3.17, 95% CI: 1.62–6.22, p = 0.001; High ApoB: HR = 2.98, 95% CI: 2.05–4.32, p < 0.001). The threshold of ApoB/ApoA1 ratio ≤ 0.45 highlights the clinical value of this ratio: it may identify a truly low-risk, rare subgroup that can safely tolerate high metabolic-risk SGAs, such as olanzapine. Conversely, patients with baseline serum ApoB levels > 0.70 g/L and an ApoB/ApoA1 ratio > 0.45 should be administered high metabolic-risk SGAs with caution and receive intensified metabolic monitoring.

To our knowledge, this is the first study evaluating the predictive value of baseline serum ApoB levels and the ApoB/ApoA1 ratio for dyslipidemia risk in psychiatric patients treated with SGAs. Notably, previous research indicated that patients with normal baseline metabolic parameters experienced greater adverse changes in all metabolic indices compared with those with baseline abnormalities.²⁹ This finding underscored the clinical significance of our study, which specifically enrolled patients without dyslipidemia or hyperglycemia at admission.

This study has several limitations. First, as a retrospective study, it was susceptible to selection bias and incomplete data. To minimize residual confounding, we performed covariate adjustments in our analytical models. Consequently, our findings reflect real-world associations of baseline serum ApoB levels and the ApoB/ApoA1 ratio with the risk of dyslipidemia following SGA initiation, providing clinically relevant insights. Second, the strict inclusion and exclusion criteria limited the sample size. Therefore, while the study enabled group-level comparisons of metabolic risk across SGA classes, it precluded both agent-specific analysis and an investigation into formulation differences (oral vs long-acting injectable (LAI)). Given the differing pharmacodynamic profiles of SGAs, as well as emerging evidence^30,31 that LAI formulations may have distinct efficacy and adverse effects, dyslipidemia risk is likely to vary across specific agents and formulations. Thus, caution is warranted in applying the cut-off value to individual SGAs or specific formulations, especially those with high metabolic risk. Future studies should therefore investigate whether the predictive power of ApoB and the ApoB/ApoA1 ratio for dyslipidemia risk differs across specific agents and formulations. Such research could further refine personalized treatment strategies. Third, as our findings were derived exclusively from a Chinese population, extrapolation to other populations should be made with caution. Finally, key lifestyle determinants of metabolic health (eg, dietary patterns and physical activity) were not systematically assessed or included in the analytical models, which could introduce unmeasured confounding. However, as participants were inpatients, their diet and physical activity were uniformly regulated by the hospital, likely minimizing variability in these factors in our study.

Conclusion

This study demonstrates that baseline serum ApoB levels and the ApoB/ApoA1 ratio are valuable risk markers for dyslipidemia in patients treated with SGAs. However, given the retrospective design and inherent heterogeneity of the cohort, further validation through large prospective studies combined with molecular investigations is warranted. Such research should confirm the predictive utility of these biomarkers, establish optimal cut-off values for clinical implementation, and investigate whether combining these markers with traditional lipid profiles or other alternative laboratory parameters can improve the accuracy of predictive models for dyslipidemia risk in patients treated with SGAs.