Use of Hydroxychloroquine is Associated with Improved Lipid Profile in Rheumatoid Arthritis Patients

Abstract

Background/Purpose

We examined the association between hydroxychloroquine (HCQ) and plasma lipid and glucose levels in RA cohort.

Patients and Methods

We performed compared fasting lipid profiles and plasma glucose between patients that were and were not taking HCQ. We divided patients into three groups based on HCQ exposure during follow up : Those who had never taken HCQ, those who took it intermittently, and those who took it continuously. We used multivariable models and propensity scoring to compensate for the effect of non-random treatment assignment.

Results

We followed 1261 RA patients for a total of 4,605 observations between 1996 and 2014. After adjusting for age, sex, ethnicity, other DMARD, lipid lowering medications, bmi, and smoking, patients taking HCQ at baseline had significantly lower total cholesterol (TC) (P≤ 0.001), LDL (P ≤0.001), triglycerides (TG) (P =0.013), and lipid profile ratios TC/HDL (P ≤0.001) and LDL/HDL (P ≤0.001), as well as higher HDL (P ≤0.001).

In longitudinal analyses, after adjusting for confounders patients that continuously took HCQ showed significantly lower TC, LDL, TC/HDL, LDL/HDL, and higher HDL (P ≤ 0.01). Fasting plasma glucose levels were not significantly associated with HCQ exposure.

Conclusion

HCQ use was associated with lower lipid levels but not with the plasma glucose in this RA cohort. These findings support the need for a randomized trial to establish the role of HCQ in cardiovascular disease prevention in RA patients.

Introduction

Rheumatoid arthritis (RA) is a chronic disease characterized by systemic inflammation that leads to joint damage and disability, and may be associated with increased risk of cardiovascular disease (CVD) a major cause of morbidity and mortality in RA. [1] The relationship between RA disease activity, lipid levels and cardiovascular diseases is complex [2, 3] However, CVD risk factor modification, such as reducing total cholesterol and plasma glucose may be beneficial for preventing CVD. Hydroxychloroquine (HCQ), a disease modifying anti rheumatic drug (DMARD) with a good safety profile and low cost, has been reported to improve lipid profiles [4–6] and glucose level in RA. [7, 8] We aimed to examine the association between the use of HCQ and plasma lipid and glucose levels in a large RA cohort.

Patients and Methods

From 1996 to 2009, we recruited 1328 consecutive patients who met the 1987 criteria for RA[9] from private and public rheumatology clinics in San Antonio, Texas. We have described this cohort in previous publications.[10–12] All patients participated in a comprehensive baseline evaluation of their clinical and psychosocial characteristics conducted by a physician and trained research assistants. We followed patients prospectively for yearly follow up evaluations in which we assessed clinical features of RA, laboratory measures, as well as DMARD use. The study protocol was approved by our IRB, and all patients provided signed, informed consent.

Variables

Demographic information included age, sex, ethnicity and duration of RA. The duration of RA was considered to be the time between RA diagnosis and the evaluation date. For clinical purpose we include prednisone use, tender, swollen and deformed joint counts, disease activity score (DAS28ESR), and hand sharp score. [13, 14]

Laboratory Studies

The erythrocyte sedimentation rate (ESR) was measured by the Westergren technique. Total plasma cholesterol (TC), high-density lipoprotein (HDL), and low-density lipoprotein (LDL), and glucose were measured using a Synchron LX automated system (Beckman Coulter, Fullerton, CA). Patients were asked to fast overnight before laboratory testing.

Statistical analysis

We performed cross sectional analyses comparing fasting lipid profiles and plasma glucose between patients that were and were not taking HCQ at baseline. We subsequently divided patients into three groups based on HCQ exposure: Those who did not take HCQ during the time of our study, those who took it intermittently, and those who took it continuously throughout the study. Using these groups, we performed generalized estimated equation (GEE) regression models including all follow up visits to examine plasma lipid and glucose levels by HCQ exposure. We adjusted for age, sex, duration of RA, ethnicity, use of lipid lowering medications, anti-TNF alpha medications and other non-biologic DMARDs including methotrexate, sulfasalazine, leflunomide, azathioprine, and diabetes medication. We also compare visits with and without HCQ, using the GEE regression model. In an alternative approach to account for potential confounding by indication, we estimated the patient’s propensity to receive HCQ and use the propensity score as a covariate in regression model. The propensity score was the predicted probability that a patient would be taking HCQ at a given visit, which was calculated from a logistic regression model using the independent variables listed in tables 1 –3.

Table 1.

Baseline characteristics of 1,261 RA patients by HCQ exposure.

	Currently taking HCQ (n=254)	Not Currently Taking HCQ (n=1,007)	Padj*
Demographics
Age, mean (SD)	57.2 (11.1)	57.4 (11.8)	0.8
Female, n (%)	200 (78%)	738 (73%)	0.07
Ethnic			0.6
Hispanic, n (%)	146(57)	598(59)
White, n (%)	92 (36)	356 (35)
Black n (%)	14(6)	37(4)
Other, n (%)	2(1)	16(2)
Duration of RA, years     Mean ± SD	10.7 ± 9.0	12.2 ± 10.6	0.04
Laboratory
Total Cholesterol mg/dl,     mean (SD)	182.3 ± 35.6	190.5 ± 39.8	≤0.001
LDL mg/dl, mean ± SD	99.4 ± 30.6	109.3 ± 34.4	<0.001
HDL mg/dl, mean ± SD	58.3 ± 18.2	53.4 ± 16.0	<0.001
Triglycerides mg/dl,     mean ± SD	123.6 ± 68.1	136.7 ± 79.4	0.009
Total Cholesterol/HDL,     mean ± SD	3.4 ± 1.2	3.8 ± 1.5	<0.001
LDL/HDL, mean ± SD	1.9 ± 0.9	2.2 ± 1.1	<0.001
Plasma glucose mg/dl,     mean ± SD	98.5 ± 36.6	102.6 ± 40.0	0.2
ESR mg/dl, mean ± SD ++	26.1 ± 25.1	30.0 ± 25.7	0.1
Clinical
Prednisone use, n(%)	101 (39.7)	392 (38.9)	0.5
Tender joints, mean(SD)**	8.2(8.5)	8.6(8.9)	0.5
Swollen joints, mean(SD)**	2.6(3.9)	3(4.5)	0.1
Deformed joints, mean(SD)**	6.73 (7.9)	6.97 (8.0)	0.23
DAS28ESR, mean(SD)$	3.94 ± 1.67	4.20 ± 1.66	0.04
Sharp Score, mean (SD)#	60.6 ± 85.3	65.6 ± 86.3	0.4
Diabetes Med, n(%)	31(12.2)	127(12.6)	0.8
Current lipid lowering,n(%)	47(18.5)	171(16.9)	0.5
Any Other DMARD, n (%)	211(83)	772(76.6)	0.03

^*P-values were adjusted (P_adj) for age, sex, duration, ethnicity and if the patients were currently on lipid lowering medications, bmi, smoking or other DMARDS. P-value model examining plasma glucose additionally adjusted for diabetes medications.

⁺⁺Erythrocyte sedimentation rate (ESR) was unavailable for 20 patients; of these 4 were currently taking HCQ and 16 were not.

^**Articular data was unavailable on 13 patients; of these 12 were not currently taking HCQ and 1 was not.

^$DAS28ESR was unavailable for 29 patients. Of these 5 were currently taking HCQ and 24 were not.

^#Sharp score was unavailable for 394 patients. Of these 73 were currently taking HCQ, and 321 were not.

Table 3.

Association of HCQ with lipids in 5045 study visits

	Currently taking HCQ (n=929)	Not Currently Taking HCQ (n=4116)	P*
Laboratory
Total Cholesterol mg/dl,     mean (SD)	178.9 ± 38.0	185.6 ± 41.4	≤0.0001
LDL mg/dl, mean ± SD	98.0 ± 30.3	106.7 ± 33.2	<0.0001
HDL mg/dl, mean ± SD	58.9 ± 18.6	52.6 ± 16.4	<0.0001
Triglycerides mg/dl,     mean ± SD	115.7 ± 62.2	135.7 ± 93.0	<0.0001
Total Cholesterol/HDL,     mean ± SD	3.3 ± 1.3	3.8 ± 1.5	<0.0001
LDL/HDL, mean ± SD	1.8 ± 1.0	2.2 ± 1.1	<0.0001

^*P-Value were obtained through GEE regression and was adjusted for age, sex, duration, ethnicity, and if the patients were currently on lipid lowering medications or other DMARDS.

Results

A total of 1328 patients were recruited into the study. However, 67 lacked complete lipid profiles, so our final sample included 1261 patients (938 female, 323 male) with a mean ± SD age of 59.6±11.5 years. At baseline, 254 patients were on HCQ and 1007 were not. After adjusting for age, sex, ethnicity, lipid lowering medications, bmi, smoking and use of any other DMARDs, patients taking HCQ had significantly lower TC (p≤ 0.001), LDL (p < 0.001), TG (p=0.009), and lipid profile ratios TC/HDL (p < 0.001) and LDL/HDL (p < 0.001) compared to those who were not taking HCQ. Furthermore, HDL was significantly higher in patients taking HCQ (p < 0.001). Plasma glucose level was not significantly associated with HCQ use (p=0.2). Patients taking HCQ at baseline also had significantly lower ESR (p=0.042), though this association lost statistical significance after the adjustment of other DMARD use (p=0.1).

With respect to clinical analysis including prednisone use, swollen, tender joints, DAS28 ESR, and Sharp score, there were no statistically significant differences between those who were currently taking HCQ versus those who were not, except for DAS28 ESR and any other DMARD (Table 1).

When taking into account follow-up visits, we divided patients based on HCQ exposure: those who had never taken HCQ during the time of our study (used as referent group), those who took it intermittently, and those who took it continuously throughout their time in the study. After adjusting for confounders, the patients who had never taken HCQ had significantly higher TC (p = 0.03, p= 0.01), LDL (p= 0.03, p< 0.001), TG (p < 0.01, p < 0.01)) and lipid profile ratios TC/HDL (p = 0.001, p < 0.001) and LDL/HDL (p=0.04, p < 0.001) when compared to those who took the drug intermittently or continuously, respectively. HDL was significantly higher in patients who took HCQ continuously versus those who have never taken it (p < 0.001). HCQ use was associated with lower plasma glucose concentration, but the difference was not significant after adjusting for the propensity score, (Table 2). Patients who had never taken HCQ had significantly higher ESR compared to those who took it continuously throughout the study (p =0.04), though there was no significant difference when compared to those who took the drug intermittently (Table 2). In all models, compared to patients with no HCQ exposure, patients taking HCQ continuously showed a more significant improvement in lipid profiles, glucose and ESR than those who took it intermittently.

Table 2.

Pooled follow-up visits of 1,261 RA patients divided by HCQ exposure

	None	Some Visits	All Visits	Padj*
No of patients/No of visits	836/2911	316/1394	109/300
Female, n(%)	2127 (73)	1,049 (75)	239 (80)
Non-Hispanic White, n(%)	1,059 (36)	372 (26)	123 (40)
Duration of RA, mean ± SD	15.3 ± 10.9	13.3 ± 9.3	11.9 ± 9.5	≤0.001, ≤0.001
Laboratory
Total Cholesterol mg/dl,     mean (SD)	185.7 ± 40.5	182.0 ± 38.0	181.0 ± 35.8	0.031, 0.01
LDL mg/dl, mean (SD)	106.5 ± 33.7	103.0 ± 30.7	97.22 ± 29.8	0.026, <0.001
HDL mg/dl, mean (SD)	52.8 ± 16.3	54.5 ± 17.2	60.6 ± 18.4	0.121, <0.001
Triglycerides mg/dl,     mean (SD)	131.5 ± 80.6	123.1 ± 63.0	114.7 ± 56.9	<0.001, <0.001
Total Cholesterol/HDL,     mean (SD)	3.80 ± 1.5	3.60 ± 1.3	3.21 ± 1.1	0.001, <0.001
LDL/HDL, mean (SD)	2.21 ± 1.1	2.09 ± 1.1	1.76 ± 0.8	0.04, <0.001
Plasma glucose mg/dl,     mean (SD)	103.0 ± 38.7	101.9 ± 37.6	95.1 ± 30.4	0.77, 0.09
ESR mg/dl,mean (SD)	29.0±25.2	32.5±26.9	21.9±23.4	0.01, 0.04
Clinical
Tender joints, mean (SD)**	7.45 ± 8.4	8.5 ± 8.7	6.36 ± 7.2	0.48, 0.31
Swollen joints, mean (SD)**	2.04 ± 3.7	2.51 ± 3.8	1.43 ± 2.8	0.39, 0.007
Deformed joints, mean(SD)**	8.42 ± 8.4	9.70 ± 8.7	5.69 ± 6.7	0.003, 0.066
DAS28ESR $	3.99 ± 1.6	4.22 ± 1.6	3.48 ± 1.61	0.7, <0.001
Sharp Score #	82.5 ± 95.2	94.9 ± 101.3	46.1 ± 71.8	0.013, 0.009
Diabetes Med, n (%)	448 (15.4)	249(17.9)	47(15.7)	0.56, 0.67
Current Lipid Lowering, n(%)	755 (25.9)	288 (20.7)	67 (22.4)	0.15, 0.62
Any Other DMARD, n (%)	2141 (73.5)	1069 (76.7)	253 (84.3)	0.54, 0.07

^*P-values were obtained from GEE regression models for clinical comparison, models were adjusted for age, sex, RA duration, ethnicity, bmi, smoking, and lipid and DMARD medications. For laboratory comparisons p-value were adjusted for propensity scores.

^**We were unable to obtain articular data at all visits on 147 patients. of these 116 patients were never exposed to HCQ; 24 were intermittently exposed; and 7 were continuously taking HCQ.

^$We were unable to obtain both articular data and ESR at all visits on 177: of these 134 patients were never exposed to HCQ; 33 were intermittently exposed; 10 were continuously taking HCQ.

^#We were unable to obtain hand radiographs at all visit on 987: of these 670 were never exposed to HCQ; 273 were intermittently exposed; 44 were continuously taking HCQ.

We found the following average differences in lipids comparing the never exposed versus continuously exposed group in the pooled follow up visits: LDL decrease of 9.3 mg/dl (p < 0.001), TG decrease 16.8 mg/dl (p < 0.001), TC decrease 4.7 mg/dl (p < 0.01), HDL increase 7.8 mg/dl (p < 0.001), LDL/HDL decrease 0.45 (p <0.001), and TC/HDL decrease 0.59 (p < 0.001).

When we compare follow up visits of patients with and without HCQ treatment, we found also statistically significant differences in all lipids level (p< 0.0001).Table 3.

With respect to clinical findings, swollen joints, DAS 28 ESR and Sharp Score were less severe in the group who take continuously HCQ versus those who never took HCQ (p < 0.01) and deformed joints and sharp score were also significantly different in the intermittently taken HCQ group compared to those who had never taken it (p ≤ 0.01) (Table 2).

We tested a HCQ X ethnic group interaction in the models for lipids measurements and it was not significant. This finding suggests that the effect of HCQ does not vary between ethnic groups.

Discussion

Despite recent advances in medical treatment, high cholesterol remains a significant public health problem in the United States, with more than one-quarter of adults aged 40–74 having high LDL–cholesterol.[15] Increased risk in CVD morbidity and mortality have been described in RA patients, independent of traditional risk factor like hypertension, diabetes, hyperlipidemia, smoking or increased age. [16, 17]

Our results suggest that HCQ is associated with significant lower in TC, TG, and LDL, as well as a significant increase in HDL in RA patients. These findings are particularly important because RA patients are predisposed to CVD.

In 1997, a prospective randomized clinical trial comparing the effect of intramuscular (IM) gold and HCQ on lipid profiles of RA patients suggested that HCQ has beneficial effects on lipids. Patients in the HCQ group had significantly higher HDL and lower TC compared to the IM gold group. In a cohort of 706 RA patients, Morris et al demonstrated that use of HCQ was associated with a significant decrease in LDL of 7.55mg/dl, decrease in TC of 7.70 mg/dl, decrease in LDL/HDL of 0.136, and decrease in TC/HDL of 0.191, with a non-significant increase of 1.02mg/dl in HDL. Similarly, we found a significantly lower average LDL, TG, TC, LDL/HDL and TC/HDL, as well as a significant increase in HDL in patients taking HCQ. However, the average differences were greater or similar in our cohort with respect to LDL, LDL/HDL, TG and TC/HDL. The significant increase in HDL is important to note because HDL is considered a powerful predictive tool, independent of other known coronary heart disease risk factors among postmenopausal women.[18] In another study of 155 women with RA or SLE was found an association between HCQ treatment and low serum level of TC, LDL, and TG level, but with no increased level of HDL.[6] In a more recent and smaller study of 72 lupus patients following for two year levels of TC, TG, LDL and HDL were significantly improved after treatment with HCQ.[19]

Similar findings, showing the benefit of HCQ as a lipid lowering medication, have been shown in systemic lupus erythematosus patients. [6, 15, 19–22]

In a cross sectional study, Urowitz et al [23] studied 123 patients with lupus and found that TC, VLDL, and LDL levels were significantly lower in patients taking antimalarial including patients taking concomitant prednisone. The cholesterol lowering effect of HCQ was found more pronounced in those SLE patients who are in stable doses of steroid or those who are initiating treatment with steroids.[24] However, in our cohort we didn’t find any differences when comparing HCQ in patients taking or not taking corticosteroids, probably because the prednisone use was similar in patients taking HCQ and those who were not taking it.

Considering that patients with RA have more difficulty in achieving a goal LDL when treated with statins compared to the general population, [25] HCQ is a reasonable and inexpensive adjuvant alternative to control the high TC or LDL level. Although the absolute differences in lipids level are not great, the statistical significance is high. Considering that small decrease in lipids level may be a protective factor in major cardiovascular disease, especially in secondary prevention, the differences that we found may still have clinical relevance.

The exact mechanism by which HCQ can lower plasma lipid levels is still unknown. It has been reported that chloroquine, which is related to HCQ, is a potent inhibitor of cholesterol biosynthesis by isolated rat hepatocytes, [26] and may increase LDL receptors, activity of HMG-Co reductase, and slow degradation of this enzyme. [27] The beneficial effect on glucose metabolism may also play an important role in decreasing lipid level. [28] In addition to lowering cholesterol, HCQ may improve plasma glucose levels. Insulin resistance is highly prevalent in RA patients and this condition may play an important role in subclinical atherosclerosis. [29] Although not all studies confirm that HCQ improve the sensitivity to insulin, [30] others have shown improvement in insulin sensitivity, and a decrease in TC and TG level.[31]

The glucose lowering efficacy, responsiveness, duration of action and impact on quality-of-life of HCQ, when added as an anti-hyperglycemic agent to patients with sulfonylurea-refractory type 2 diabetes was demonstrated in a study with 135 obese patients with type 2 diabetes, especially in those with glycosylated hemoglobin <13.5%. [32] In our study, we found no difference in the plasma glucose level at baseline and the difference during the follow up visits was not significant after adjusting for confounding.

Our study addressing HCQ as a lipid lowering medication in RA patients has strengths and limitations. Among its strengths, our cohort is large enough to provide the power to detect small effects. It is a prospective observational study in which a rheumatologist examined the patients, systematically asking questions about any medications that they were taking, and drawing blood for complete metabolic profile during each visit. We also have the benefit of a diverse cohort, recruited from a variety of health systems and socioeconomic strata. Furthermore, compared to patients taking HCQ, those who were not had similar ESR and DAS28ESR at baseline, thus decreasing the likelihood of an indication bias. We also used these data to score the patient’s propensity to be receiving HCQ at each visit. These propensity scores, which were used as covariates in multivariable models, compensate for the effect of nonrandom treatment assignment. [33] The association of HCQ with lower lipid profile remained significant after adjustment for propensity score. One possible limitation is the fact that the recruitment period was quite long (1996–2009) during which there may have been changes in the management of patients, beyond pharmacologically, but also in the delivery of service. There may have been changes in health care policy, which may have affected the outcomes we measured. Moreover, the pattern of use of HCQ regimen could have also changed over time.

Conclusion

HCQ use was associated with significantly lower TC, LDL, TG, and TC/HDL and LDL/HDL ratios, and with higher HDL. The association of HCQ with plasma glucose was not statistically significant after adjusting for confounding. Minimizing risk of CVD in RA patients remains a major concern in treating these patients. These findings suggest HCQ may help in reducing some traditional CV risk factors. Large randomized trials are needed to establish the potential of HCQ in CVD prevention in RA patients.