Pharmacological Interventions to Treat Antipsychotic-Induced Dyslipidemia in Schizophrenia Patients: A Systematic Review and Meta Analysis

Introduction: Antipsychotic-induced dyslipidemia represents a common adverse effect faced by patients with schizophrenia that increases risk for developing further metabolic complications and cardiovascular disease. Despite its burden, antipsychotic-induced dyslipidemia is often left untreated, and the effectiveness of pharmacological interventions for mitigating dyslipidemia has not been well-addressed. This review aims to assess the effectiveness of pharmacological interventions in alleviating dyslipidemia in patients with schizophrenia. Methods: Medline, PsychInfo, and EMBASE were searched for all relevant English articles from 1950 to November 2020. Randomized placebo-controlled trials were included. Differences in changes in triglycerides, HDL cholesterol, LDL cholesterol, and VLDL cholesterol levels between treatment and placebo groups were meta-analyzed as primary outcomes. Results: Our review identified 48 randomized controlled trials that comprised a total of 3,128 patients and investigated 29 pharmacological interventions. Overall, pharmacological interventions were effective in lowering LDL cholesterol, triglycerides, and total cholesterol levels while increasing the levels of HDL cholesterol. Within the intervention subgroups, approved lipid-lowering agents did not reduce lipid parameters other than total cholesterol level, while antipsychotic switching and antipsychotic add-on interventions improved multiple lipid parameters, including triglycerides, LDL cholesterol, HDL cholesterol, and total cholesterol. Off label lipid lowering agents improved triglycerides and total cholesterol levels, with statistically significant changes seen with metformin. Conclusion: Currently available lipid lowering agents may not work as well in patients with schizophrenia who are being treated with antipsychotics. Additionally, antipsychotic switching, antipsychotic add-ons, and certain off label interventions might be more effective in improving some but not all associated lipid parameters. Future studies should explore novel interventions for effectively managing antipsychotic-induced dyslipidemia. Registration: PROSPERO 2020 CRD42020219982; https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020219982.


INTRODUCTION
Dyslipidemia refers to abnormalities in lipid levels such as increases in total and low-density lipoprotein (LDL) cholesterols, low concentrations of high-density lipoprotein (HDL) cholesterols, and high triglyceride levels. This metabolic abnormality causes almost a third of ischemic heart disease and a fifth of global cerebrovascular disease (1). Patients with schizophrenia are at an increased risk of developing cardiovascular disease in part due to the illness itself (2)(3)(4)(5)(6)(7), as well as a higher prevalence of well-known lifestyle factors that promote cardiovascular disease risk, namely sedentary lifestyle, poor diet, and smoking (8,9). Antipsychotics are the cornerstone of treatment in schizophrenia and are widely prescribed across other psychiatric conditions (10). However, their use is associated with severe metabolic adverse effects, including weight gain, dyslipidemia, insulin resistance, and risk of type 2 diabetes mellitus (T2D) in a population burdened with premature cardiovascular mortality.
While the prevalence of dyslipidemia and consequent effects on morbidity and mortality are high worldwide among the general population, particular subgroups may be at a greater risk. In particular, patients with schizophrenia are at an increased risk of dyslipidemia and its associated influence on cardiovascular disease and metabolic dysfunction (11,12). Despite its high prevalence and associated cardiovascular risk, dyslipidemia often goes untreated among patients with schizophrenia. Reported rates for non-treatment are almost 90% (13)(14)(15), and patients with schizophrenia are often medically underserved and disadvantaged in their physical health care (16)(17)(18). As shown by results from a study by the National Institute of Mental Health, namely the Recovery After an Initial Schizophrenia Episode-Early Treatment Program (RAISE-ETP), at baseline more than half of patients (161/394 or 56.5%) had dyslipidemia and only 0.5% were receiving treatment (16).
Given the variety of approaches used to address dyslipidemia in this patient population, as well as the absence of clear clinical guidelines, it is important to summarize the available evidence and guide clinical decision making. Hence, we performed a systematic review and meta-analysis of randomized controlled trials to compare the effects of pharmacological interventions vs. placebo treatment in antipsychotic-induced dyslipidemia in patients with schizophrenia.

METHODS
The protocol for the review is registered on PROSPERO (PROSPERO 2020 CRD42020219982; https://www.crd.york. ac.uk/prospero/display_record.php?ID=CRD42020219982). PRISMA guidelines were used for study design and reporting.

Search
We searched for studies published between 1950 and November 2020 using Medline, PsychInfo and EMBASE databases, with the following search string: psychotic disorder OR schizophrenia OR schizoaffective OR schizophreniform OR psychosis OR first episode AND hyperlipidemia OR triglycerides OR cholesterol OR lipid OR LDL cholesterol OR VLDL cholesterol OR HDL cholesterol. The search was limited to studies conducted in human participants and published in English. References cited in previously published literature reviews and meta-analyses pertaining to interventions for metabolic disturbances in the schizophrenia population were reviewed for additional studies.

Inclusion Criteria
Articles were initially screened using title and abstract, based on the study's relevance to our meta-analysis. Thereafter, articles were further screened to ensure that studies met the following inclusion criteria: (a) randomized placebo-controlled trial; (b) diagnosis of schizophrenia spectrum disorders comprising the majority (>50%) of study populations; (c) patients with current metabolic abnormalities; (d) an active pharmacological intervention used to improve metabolic abnormalities or an antipsychotic switching/add-on method if the antipsychotic change is aimed to improve metabolic parameters; and (e) primary outcome listed as lipids or other metabolic measures if lipid outcomes were included in the list of metabolic measures.
Studies were excluded from analysis during the final screening stage if (a) not aimed at improving metabolic measures; (b) comparing different modes of antipsychotic administration (i.e., deltoid, sublingual, gluteal etc.); (c) comparing effectiveness between different antipsychotics; (d) evaluating non-pharmacological intervention (e.g., behavioral interventions, dietary modulations etc.); or, (e) evaluating strategies to prevent dyslipidemia (i.e., patients did not have metabolic abnormalities or dyslipidemia at baseline).

Outcomes Extracted
The primary outcomes included the following lipid parameters: total cholesterol, triglycerides, LDL cholesterol, HDL cholesterol, and very low-density lipoprotein cholesterol (VLDL) cholesterol. Additional secondary outcome data were also extracted including body weight, body mass index (BMI), waist circumference, waist to hip ratio, fasting blood glucose, fasting insulin, hemoglobin A1c (HbA1c), diastolic blood pressure, systolic blood pressure, the homeostatic model assessment of insulin resistance (HOMA-IR), and total positive and negative symptom scale (total PANSS). Outcomes were extracted for both the intervention and placebo groups, where the placebo groups were used as comparators. Outcomes were extracted by two authors (PK and KC-D) and were checked by authors, FP and JL. For studies that examined multiple doses of the same intervention, the data pertaining to the higher dose were extracted.

Data Analysis
Review Manager 5.4 (Revman 5.4.0 (Mac Version) Cochrane Collaboration, Oxford) was used to analyze the data extracted from the final list of included articles. Continuous outcomes were reported using mean differences (MD) with 95% confidence intervals (CIs), following the inverse variance statistical method and random effects model to account for study heterogeneity. Missing standard deviations (SDs) were calculated using other available statistics that were reported. Endpoint data were primarily used unless not available, in which case mean change data were used. Endpoint and change data were combined during the analysis, as we used mean difference rather than standardized mean difference (39). For Emsley et al. which was a double-blind trial with an open-label extension (27), data were extracted at the endpoint of the double-blind phase. Study heterogeneity was calculated using the I 2 statistic, with significant heterogeneity being classified as I 2 ≥ 50%. Significant statistical differences were classified as p < 0.05. Changes in lipid profiles (i.e., HDL cholesterol, LDL cholesterol, VLDL cholesterol, triglycerides, and total cholesterol) were assessed for all interventions pooled and for the following 4 subgroups: lipid lowering agents; antipsychotic switching or antipsychotic add-on interventions; the off-label lipid lowering agent metformin; and other off-label lipid lowering agents.
All included studies were judged for risk of bias in random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias using the Cochrane Risk of Bias tool (40). Studies were judged to have either a low, high, or unclear risk of bias. Sensitivity analyses were conducted after removing studies found to be at high risk of bias to examine their impact on findings.

Antipsychotic Switching/Add-on Interventions
Antipsychotic switch/add-on strategies were associated with significant decreases in triglycerides (Figure 2; N = 7, n = 800;

Secondary Outcomes: Additional Metabolic Measures
Cumulatively, the pharmacological interventions reviewed in this paper were associated with significant reductions in body weight

Risk of Bias
Risk of bias in random sequence generation was deemed to be low in 29 studies, high in 1, and unclear in 18 (Supplementary Figure 11). Outcomes did not change significantly after the study with high risk of bias was removed.

DISCUSSION
In this systematic review and meta-analysis, we examined different pharmacological interventions used to treat antipsychotic-induced dyslipidemia in schizophrenia spectrum disorders. The 29 pharmacological interventions analyzed were cumulatively effective in lowering total cholesterol, LDL cholesterol, and triglycerides, while increasing HDL cholesterol. However, improvements were not significant with VLDL cholesterol. Amongst the subgroups analyzed, we found that antipsychotic switching/add-on proved most effective in improving lipid parameters commonly dysregulated in schizophrenia, namely triglycerides and HDL cholesterol (77). Notably, the off-label lipid lowering agent metformin was more promising than approved lipid lowering agents in decreasing triglycerides and total cholesterol levels. However, other off label agents only showed a trend in improving lipid parameters.
Our findings suggest that off label strategies can be effectively employed to ameliorate antipsychotic-induced dyslipidemia. In particular, metformin shows considerable promise, improving lipid parameters and showing consistent association with a decrease in triglycerides and total cholesterol levels. Similar findings for triglycerides and total cholesterol levels were previously demonstrated in a review by Jiang et al. (78), and in the context of schizophrenia would benefit through evidence specific to long-term outcomes. Prior studies indicate that a 40 mg/dL reduction in LDL cholesterol and triglycerides translates into a 20% and 4-5% decrease in risk for developing cardiovascular disease, respectively, independent of baseline risk (79). Given this, our review suggests that the available strategies for targeting dyslipidemia are inadequate, reinforcing the need for novel, more effective interventions. Furthermore, while our findings provide strong evidence for antipsychotic switch/add-on interventions, study duration ranged from 6 to 24 weeks, which does not provide adequate time to assess the long-term effects of these treatments on dyslipidemia. While aripiprazole and quetiapine are both second generation antipsychotics with less severe metabolic side-effects compared to others like olanzapine and clozapine, they have their own metabolic burden that cannot be ignored and needs to be better understood over the longer term (80,81). Similarly, while our findings suggest that lipid lowering agents are not effective in improving dyslipidemia, these results may have been limited by the short duration of the included studies. The small number of studies also did not permit examination of the possible effects of dose of lipid-lowering agents.
Current studies provide general support for the potential effectiveness of pharmacological interventions, but further research is warranted to refining recommendations pertaining to individual treatments. Currently in many studies, lipid management was not a primary focus; of the 48 reviewed studies, only 26 identified lipid profile as a primary outcome measure, in contrast to 22 where it was positioned as a secondary outcome. More studies focused on this area of research sets the stage for additional insights and the increased power necessary to detect not only beneficial outcomes, but also the elucidation of specific variables contributing to effective treatment. At present, the significant heterogeneity among studies within intervention categories limit generalizations that can be made with respect to mechanisms or interacting variables. Factors affecting outcomes may include specific antipsychotic treatment, diagnosis and stage of illness, co-morbid health conditions, concomitant medications, and duration of antipsychotic and/or lipid intervention treatment. Differential pharmacological interventions may, in fact, vary as a function of patient population. Moreover, to date, many interventions are confined to a single study, precluding pooled data or comparisons between interventions. Our review restricted the population to schizophrenia patients, even though antipsychotics are used to treat patients with other psychiatric illnesses such as affective disorders who also share the metabolic burden (82) and may benefit from the reviewed interventions. Finally, while behavioral and lifestyle interventions remain first-line treatments for dyslipidemia, our review restricted the search to pharmacological interventions.
Given the prevalence of dyslipidemia in schizophrenia, along with the associated increased risk of metabolic complications and cardiovascular disease, it is imperative that such studies be undertaken. At present, dyslipidemia is often left untreated (13)(14)(15); indeed, the physical health of this population is generally overlooked while the focus is directed to managing psychotic symptoms (83,84). The integration of psychiatric and medical care falls short at present (85); however, this overview of dyslipidemia, its prevalence and current treatment underscores the need to ensure a more comprehensive model of care be implemented.

DATA AVAILABILITY STATEMENT
The original contributions generated for the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
PK, SMA, and MKH contributed to developing the original protocol. PK and KC-D contributed to the original screening, data extraction, risk of bias assessments, and writing the first draft of the manuscript (introduction, methods, and results). FP and JL wrote and registered the protocol with PROSPERO, re-ran the search, updated study selection and risk of bias, and contributed to final data extraction and synthesis prior to manuscript submission, as well as updating the first draft. LH assisted with editing and writing the final draft. SMA was involved in supervising all aspects of the review. GR and MKH contributed to editing the final draft. All authors contributed to the article and approved the submitted version.