Dr. Wasko has received consultant fees (less than $10,000 each) from Centocor and UCB.
Special Theme Articles: Vascular Comordibidity in the Rheumatic Diseases
Hydroxychloroquine use associated with improvement in lipid profiles in rheumatoid arthritis patients
Article first published online: 30 MAR 2011
Copyright © 2011 by the American College of Rheumatology
Arthritis Care & Research
Volume 63, Issue 4, pages 530–534, April 2011
How to Cite
Morris, S. J., Wasko, M. C. M., Antohe, J. L., Sartorius, J. A., Kirchner, H. L., Dancea, S. and Bili, A. (2011), Hydroxychloroquine use associated with improvement in lipid profiles in rheumatoid arthritis patients. Arthritis Care Res, 63: 530–534. doi: 10.1002/acr.20393
- Issue published online: 30 MAR 2011
- Article first published online: 30 MAR 2011
- Accepted manuscript online: 15 NOV 2010 11:14AM EST
- Manuscript Accepted: 26 OCT 2010
- Manuscript Received: 29 JUN 2010
Cardiovascular disease (CVD) is the leading cause of death in patients with rheumatoid arthritis (RA). Disease-modifying therapies that improve risk factors for CVD, such as dyslipidemia, are desired. This study used an electronic health record to determine if hydroxychloroquine (HCQ) use was associated with an improvement in lipid levels in an inception RA cohort.
All adult individuals with the initial diagnosis of RA between January 1, 2001, and March 31, 2008, were identified (n = 1,539). Only patients with at least one lipid level post–RA diagnosis were included (n = 706). Information on demographics, medical history, body mass index (BMI), laboratory measures, and medications were collected at office visits. Potential risk and protective factors for dyslipidemia were controlled for in linear mixed-effects regression models for low-density lipoprotein (LDL), high-density lipoprotein (HDL), total cholesterol, triglycerides, LDL/HDL, and total cholesterol/HDL.
Patients were 69% women and 98% white, with a median age of 65 years and a median BMI of 29.8 kg/m2. In the adjusted regression models, HCQ use was associated with the following average differences in lipids: LDL decrease of 7.55 mg/dl (P < 0.001), HDL increase of 1.02 mg/dl (P = 0.20), total cholesterol decrease of 7.70 mg/dl (P = 0.002), triglycerides decrease of 10.91 mg/dl (P = 0.06), LDL/HDL decrease of 0.136 (P = 0.008), and total cholesterol/HDL decrease of 0.191 (P = 0.006), which were stable over time.
Use of HCQ in this RA cohort was independently associated with a significant decrease in LDL, total cholesterol, LDL/HDL, and total cholesterol/HDL. Considering these results, its safety profile, and low cost, HCQ remains a valuable initial or adjunct therapy in this patient population at high risk for CVD.
Cardiovascular disease (CVD) is the leading cause of mortality in patients with rheumatoid arthritis (RA) (1). Atherogenic dyslipidemia is recognized as an important risk factor for CVD. Data from the general population suggest that optimizing lipid levels in this group would help to minimize this risk (2). Patients with active RA have abnormal lipid profiles that are related to disease activity (3). A drug that both treats rheumatic joint disease and improves lipid levels would therefore be advantageous in managing patients with RA.
Previous studies have shown that hydroxychloroquine (HCQ), an antimalarial medication used in the treatment of autoimmune diseases, has modest favorable effects on the lipid profiles of patients with systemic lupus erythematosus (SLE) and RA. However, these studies were small and had short observation times.
The purpose of the present study was to investigate the relationship between HCQ use and lipid profiles in an RA inception cohort of 706 patients using electronic health records containing physician diagnoses, laboratory values, and medications, hypothesizing that HCQ will have a favorable effect on lipid levels.
MATERIALS AND METHODS
Data for this study were extracted by the Geisinger Health System Information Technology Department from the electronic health records (EPIC). Validation of the data was performed by the Information Technology Department. In addition, one of the investigators (AB) performed chart reviews on 18 known RA patients, documenting 100% concordance between the data received from the Information Technology Department and chart documentation in all 10 variables per patient, including RA diagnosis, laboratory values, and medication use patterns. The Rheumatology Department at Geisinger Health System entered the electronic health records system completely by January 1, 2001. All adult (age ≥18 years) individuals with an initial diagnosis of RA within the Geisinger Health System between January 1, 2001, and March 31, 2008, were identified (n = 1,539). The first date was considered the index date. RA was defined as a patient with a diagnosis with an International Classification of Diseases, Ninth Revision code of 714.0 at ≥2 outpatient encounters with a Geisinger Health System rheumatologist. The diagnosis of RA was validated against the American College of Rheumatology (ACR) criteria (4) by one of the authors (SD) who manually reviewed 100 random charts. Of these 100, 97 patients with the study definition of RA met the ACR criteria for a diagnosis of RA. Of the 3 patients who did not meet the criteria, one had the diagnosis of psoriatic arthritis, one had mixed connective tissue disease, and the third had juvenile idiopathic arthritis. This high concordance rate of 97% between ACR criteria and the study definition of RA was deemed sufficient to proceed with study analyses. Only patients with at least one lipid level post–RA diagnosis were included (n = 706).
Many of the 706 RA patients had repeated post-RA lipid measurements, resulting in 2,851 lipid-level results. The time in days from each patient's RA diagnosis date to each lipid result date was calculated (RA duration).
The primary outcomes are levels of total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), and triglycerides (cobas Roche/Hitachi Diagnostics) in HCQ users versus nonusers. In addition, two rates were included as outcomes, LDL/HDL and total cholesterol/HDL (the atherogenic index), that have recently been reported to be important predictors of increased cardiovascular events (5).
Demographics included age at RA diagnosis, sex, and race. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared for each patient. The maximum BMI over the observation period was used.
Hypertension was defined as 2 or more outpatient visits within 2 years with a diagnosis of hypertension or at least 2 consecutive measurements of elevated systolic blood pressure (>140 mm Hg) or elevated diastolic blood pressure (>90 mm Hg). Diabetes mellitus was defined as having 1 or more outpatient visits with a diagnosis of diabetes mellitus, a nonfasting blood glucose result of ≥200 mg/dl, a glycosylated hemoglobin result of ≥7%, or a hypoglycemic medication order (α-glucosidase inhibitors, amino acid derivatives, biguanides, insulin-sensitizing agents, meglitinide analogs, sulfonylureas, thiazolidinediones, exenatide, pramlintide, sitagliptin, antidiabetic combination medications, or insulin).
A typical dosage of 400 mg/day of HCQ (6.5 mg/kg) was identified as a time-varying primary predictor. Twenty-one patients who would be most likely to receive a lower dose of HCQ due to their weight were identified in the cohort with a BMI of <20 kg/m2. One of the investigators (SJM) manually reviewed electronic records. Of the 21 patients, 15 were receiving 400 mg/day, 2 were receiving 300 mg/day, and 4 were receiving 200 mg/day of HCQ. Additional medications captured as time varying included nonsteroidal antiinflammatory drugs (NSAIDs), glucocorticoids, methotrexate, and tumor necrosis factor α inhibitors (etanercept, adalimumab, infliximab), hydroxymethylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors, and other lipid-lowering medications (antihyperlipidemic combinations, bile sequestrants, fibric acid derivatives, intestinal cholesterol absorption inhibitors, nicotinic acid derivatives, and miscellaneous antihyperlipidemics). Exposure to medications included medication order start and stop dates to capture therapeutic segments during the observation period. Continuous HCQ use was defined as a discontinuation gap of <3 months.
Rheumatoid factor (by latex agglutination method, positive at ≥14 IU/ml; Roche Diagnostics), anti–cyclic citrullinated peptide antibodies (by enzyme-linked immunosorbent assay, positive at ≥20 units; Inova Diagnostics), erythrocyte sedimentation rate (ESR; by Westergren method), and C-reactive protein (CRP) levels (cobas Roche/Hitachi Diagnostics) were measured. Only results that were performed at a Geisinger Health System facility were utilized in this study.
Each patient's postindex HCQ status was evaluated for each lipid measurement result date. Potential risk and protective factors for dyslipidemia were controlled for in a linear mixed-effects regression model for each lipid outcome. The time in years from each patient's RA diagnosis date to each lipid result date was calculated and included in all models. Nonlinear time trends and interactions between HCQ and time were checked for all models. In the model, HCQ and all other medications were treated as time-varying covariates as yes/no indicators evaluated at each lipid result date. A secondary analysis also was performed, in which HCQ was treated as cumulative duration of exposure. In this model, the continuous duration of HCQ up to the lipid result date was calculated. Results are expressed as the average fixed-effects estimates over time for HCQ.
To better understand the effect of statins, statin use was categorized as high potency, medium potency, and low potency (6) for the 359 patients that were ever receiving statins (using the closest statin order prior to the first lipid result). In order to investigate the possible impact of statin exposure, an additional, identical sensitivity analysis was performed only on patients that were not receiving statins.
Among the 706 incident RA patients, 98% had an LDL result, 96% had an HDL result, 97.5% had a total cholesterol result, and 95% had a triglycerides result; 95% had all 4 lipid panel results at least once. Approximately 30% had only 1 lipid panel result during observation, approximately 17% had 2, approximately 12% had 3, and 11% had 4; the remaining had 5 or more lipid panel results during the study timeframe, resulting in a total of 2,851 lipid results. The median time between results was 183 days (interquartile range [IQR] 103–338 days).
The general characteristics of the patients are listed in Table 1. A total of 256 patients (36%) were ever receiving HCQ. The median exposure time to HCQ was 1.98 years (IQR 1.00–3.85 years). Patients were predominantly women (69%) and 98% were white, with a median age of 65 years and a median BMI of 29.8 kg/m2. Rheumatoid factor (69% known) and anti–cyclic citrullinated peptide antibodies (33% known) were positive in 79% and 42%, respectively. The HCQ ever users were more likely to be women and younger, and less likely to have ever been treated with methotrexate compared with the HCQ never users. Statin use was similar in the two groups, including the categorization by statin potency, with the exception of the medium-potency statin category that was more prevalent in the HCQ never users. Lipid levels prior to inclusion in the study cohort were available in 185 HCQ ever users and 313 HCQ never users, and there were no differences in the median total cholesterol, LDL, HDL, and triglycerides levels or between the median LDL/HDL and total cholesterol/HDL ratios between the two groups.
|Overall (n = 706)||Never (n = 450)||Ever (n = 256)||P†|
|Women||486 (69)||298 (66)||188 (73)||0.047|
|White||689 (98)||441 (98)||248 (97)||0.349|
|Hypertension||485 (69)||314 (70)||171 (67)||0.412|
|Diabetes mellitus||262 (37)||168 (37)||94 (37)||0.871|
|Positive rheumatoid factor (n = 489 [69%])||385 (79)||226 (79)||159 (78)||0.853|
|Positive anti-CCP antibodies (n = 233 [33%])||98 (42)||54 (44)||44 (40)||0.624|
|Medication use (ever)|
|TNFα inhibitors||213 (30)||145 (32)||68 (27)||0.115|
|NSAIDs||537 (76)||341 (76)||196 (77)||0.814|
|Steroids||624 (88)||397 (88)||227 (89)||0.858|
|Methotrexate||418 (59)||286 (64)||132 (52)||0.002|
|Statins||359 (51)||238 (53)||121 (47)||0.151|
|Low potency||28 (12)||15 (10)||13 (17)||–|
|Medium potency||110 (49)||80 (54)||30 (39)||0.037|
|High potency||88 (39)||53 (36)||35 (45)||–|
|Other lipid-lowering agents||100 (14)||64 (14)||36 (14)||0.953|
|Age, median (IQR) years§||65.0 (57–75)||66.0 (59–76)||61.5 (54–72)||< 0.001|
|BMI (n = 676 [96%]), median (IQR) kg/m2||29.8 (26–34)||29.7 (26–34)||30.1 (26–36)||0.579|
|Total cholesterol, median (IQR) mg/dl¶||194 (170–222)||194 (169–218)||0.914|
|LDL, median (IQR) mg/dl¶||111 (88–132)||108 (88.5–131)||0.580|
|HDL, median (IQR) mg/dl¶||51 (42–61)||52 (44–64)||0.233|
|Triglycerides, median (IQR) mg/dl¶||144 (101–206)||132 (104–191)||0.246|
|LDL/HDL, median (IQR)¶||2.13 (1.61–2.80)||2.13 (1.51–2.76)||0.499|
|Total cholesterol/HDL, median (IQR)¶||3.81 (3.15–4.67)||3.67 (2.7–4.57)||0.302|
|CRP level (n = 234 [33%]), median (IQR) of maximum mg/liter||7.7 (2.7–20.0)||7.2 (2.8–21.0)||8.0 (2.5–19.8)||0.986|
|ESR (n = 513 [73%]), median (IQR) of maximum mm/hour||31.0 (16–56)||33.0 (18–58)||29.0 (14–54)||0.259|
In the regression models, after adjusting for demographic features, BMI, ESR, CRP level, rheumatoid factor, anti–cyclic citrullinated peptide, diabetes mellitus, hypertension, and use of glucocorticoids, NSAIDs, anti–tumor necrosis factor α therapy, methotrexate, and lipid- lowering medications, HCQ was associated with the following average differences in lipids: LDL decrease of 7.55 mg/dl (P < 0.001), HDL increase of 1.02 mg/dl (P = 0.20), total cholesterol decrease of 7.70 mg/dl (P = 0.002), triglycerides decrease of 10.91 mg/dl (P = 0.06), LDL/HDL decrease 0.136 (P = 0.008), and total cholesterol/HDL decrease of 0.191 (P = 0.006) (Figures 1 and 2). These effects were found to be stable over time, as demonstrated by a lack of a significant interaction between HCQ use and time since diagnosis. Cumulative exposure to HCQ models resulted in the same directional and significant findings.
The sensitivity analysis in the 347 patients that were not receiving statins after RA diagnosis found very similar results. HCQ use was associated with the following average differences in lipids: LDL decrease of 10.36 mg/dl (P < 0.001), HDL decrease of 1.89 mg/dl (P = 0.161), total cholesterol decrease of 13.87 mg/dl (P = 0.001), triglycerides decrease of 17.43 mg/dl (P = 0.048), LDL/HDL decrease 0.150 (P = 0.041), and total cholesterol/HDL decrease of 0.126 (P = 0.222).
These results suggest that HCQ use is independently associated with lower LDL, total cholesterol, LDL/HDL, and total cholesterol/HDL in patients with RA. Patients taking HCQ also had an increase in HDL and a decrease in triglycerides, although these findings were not statistically significant.
The observations from this study are consistent with the previously reported association of antimalarial agents with a favorable lipid profile in smaller studies, although this study did not find a significant association of HCQ use with increased HDL. In a study of 155 patients with RA or SLE, an association was found between HCQ treatment and low serum levels of cholesterol, triglycerides, and LDL (7). In another study of 66 patients with RA, an increase in HDL in patients taking HCQ was reported compared to an increase in total cholesterol and triglycerides and a decrease in HDL in patients taking gold (8). These results were confirmed in a study of 100 RA patients that compared the effects of gold and HCQ on lipids. Patients treated with gold had a decrease in HDL and an increase in triglycerides. Patients in the HCQ group had an increase in HDL and no significant changes in total cholesterol and triglycerides. There were no significant differences in efficacy for disease response between the two groups (9). In a longitudinal cohort of 264 patients with SLE, HCQ therapy was associated with lower serum cholesterol (10).
Patients with active RA have altered lipid metabolism characterized by hypocholesterolemia with a global reduction of total cholesterol, LDL, HDL, very LDL, and triglycerides (11, 12). RA patients have been shown to have low levels of HDL and higher lipoprotein(a) (13) and increased total cholesterol/HDL, which is associated with an increased risk of CVD (14). It has been reported that higher inflammatory marker levels correlate inversely with total cholesterol and HDL levels, and with related levels of apolipoproteins B and A-I. Reduced HDL and elevated lipoprotein(a) correlate with elevated CRP levels in patients with RA (15). There is some evidence that controlling disease activity may improve lipid profiles (15, 16).
HCQ has a relatively weak antiinflammatory effect; therefore, it is unlikely that the beneficial effect on lipids is conferred solely by controlling systemic inflammation. The mechanism by which this effect is mediated is uncertain. It has been reported that chloroquine, a medication structurally similar to HCQ, is an inhibitor of cholesterol biosynthesis in rat hepatocytes (17). In cultured human fibroblasts and mouse adrenal cells, chloroquine has multiple effects on cholesterol metabolism. Chloroquine also inhibits lysosomal hydrolysis of cholesteryl esters through an increase in the pH within lysosomes and inactivates acid proteases. Additionally, it stimulates the capacity of LDL receptor and the activity of HMG-CoA reductase, and slows the degradation of HMG-CoA reductase (17).
A significant concern regarding the present study is confounding by indication. Since the allocation of a patient to treatment with HCQ was not randomized, it is possible that patients with less severe or better controlled disease were treated with HCQ, thus confounding the results. While it is impossible to eliminate confounding by indication in observational studies, there were several safeguards to limit its influence in the results of the present study. Although the patients in the HCQ ever users group were younger and less likely to have ever been treated with methotrexate, they were not more likely to have positive rheumatoid factor or have been treated with steroids or tumor necrosis factor α inhibitors (surrogate markers of disease severity). Furthermore, the two groups had similar median levels of ESR and CRP (surrogate markers of disease activity). Median lipid levels prior to inclusion in the study cohort and statin use were similar in the two groups, including categorization by statin potency, with the exception of the medium-potency statin category that was more prevalent in the HCQ never users, which, if any, would bias the results toward favorable lipid profiles in the HCQ never users group. At the analysis level, a linear mixed-effects regression model controlled for characteristics that were different between the groups and could potentially be associated with lipid profiles.
The present study reports on an incident RA cohort using electronic health records with physician diagnoses, laboratory values, and medications. Adjustment was performed for major confounders that affect lipid profiles; however, other risk factors for dyslipidemia, such as family history, central adiposity, and physical activity, were not available in this cohort. Additional limitations include that the disease activity was not measured directly, although surrogate measures as CRP levels and ESR are reported. The data were extracted from documentation for routine medical care and not systematically collected for research purposes, resulting in missing laboratory values in a number of patients. More than 50% of patients were excluded due to the fact that lipid profiles were not available. This likely resulted from patients having primary care physicians outside of the Geisinger Health System. Finally, the study cohort is ethnically homogeneous, limiting the generalizability of our findings.
The findings of this study support the potential benefit of HCQ in improving the atherogenic lipid profiles in patients with RA. This is particularly important in this patient population that has a higher risk for CVD. Although HCQ has only modest disease-modifying effects in the treatment of RA, its apparent effects on lipid profiles, and elsewhere reported inhibition of platelet aggregation and reduction in diabetes mellitus risk, along with its excellent safety profile and low cost, make it a beneficial first-line or adjunct therapy for RA patients, particularly those with traditional CVD risk factors.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Bili had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Morris, Wasko, Kirchner, Dancea, Bili.
Acquisition of data. Morris, Antohe, Kirchner, Dancea, Bili.
Analysis and interpretation of data. Morris, Wasko, Sartorius, Kirchner, Bili.
- 8A comparison of hydroxychloroquine and myocrisin in rheumatoid arthritis [abstract]. Br J Rheumatol 1994; 33 Suppl: 149., , , .