Protective effect of hydroxychloroquine on renal damage in patients with lupus nephritis: LXV, data from a multiethnic US cohort

Abstract

Objective

To assess whether hydroxychloroquine can delay renal damage development in lupus nephritis patients.

Methods

Lupus nephritis patients (n = 256) from the LUpus in MInorities, NAture versus nurture study (n = 635), a multiethnic cohort of African Americans, Hispanics, and Caucasians, age ≥16 years with disease duration ≤5 years at baseline (T0) were studied. Renal damage was defined using the Systemic Lupus International Collaborating Clinics Damage Index (≥1 of the following lasting at least 6 months: estimated/measured glomerular filtration rate <50%, 24-hour proteinuria ≥3.5 gm and/or end-stage renal disease, regardless of dialysis or transplantation). Patients with renal damage before T0 were excluded (n = 53). The association between hydroxychloroquine use and renal damage (as defined, or omitting proteinuria) was estimated using Cox proportional regression analyses adjusting for potential confounders. Kaplan-Meier survival curves based on hydroxychloroquine intake or the World Health Organization (WHO) class glomerulonephritis were also derived.

Results

Sixty-three (31.0%) of the 203 patients included developed renal damage over a mean ± SD disease duration of 5.2 ± 3.5 years. The most frequent renal damage domain item was proteinuria. Patients who received hydroxychloroquine (79.3%) exhibited a lower frequency of WHO class IV glomerulonephritis, had lower disease activity, and received lower glucocorticoid doses than those who did not take hydroxychloroquine. After adjusting for confounders, hydroxychloroquine was protective of renal damage occurrence in full (hazard ratio [HR] 0.12, 95% confidence interval [95% CI] 0.02–0.97, P = 0.0464) and reduced (HR 0.29, 95% CI 0.13–0.68, P = 0.0043) models. Omitting proteinuria provided comparable results. The cumulative probability of renal damage occurrence was higher in those who did not take hydroxychloroquine and those classified as WHO class IV glomerulonephritis (P < 0.0001).

Conclusion

After adjusting for possible confounding factors, the protective effect of hydroxychloroquine in retarding renal damage occurrence in systemic lupus erythematosus is still evident.

INTRODUCTION

Systemic lupus erythematosus (SLE) is an autoimmune disorder that can affect multiple organs with varying degrees of severity. Renal involvement in SLE can range from silent disease to severe renal insufficiency and occurs in 50–70% of lupus patients depending on the population studied (1–3). Despite advances in therapy, the morbidity and mortality in patients with lupus nephritis remain high. Lupus nephritis leads to the development of end-stage renal disease in 17–25% of patients (4–6), decreased survival (7, 8), and higher health care costs (9). However, to date, none of the available therapies have been conclusively proven to protect against the development of renal damage.

Hydroxychloroquine is an antimalarial agent that traditionally has been used to treat mucocutaneous, musculoskeletal, serosal, and constitutional manifestations of SLE. Previous studies from our group and others have shown that hydroxychloroquine usage is associated with a reduced risk of damage accrual (10), improved survival (11, 12), a decreased frequency of lupus flares (13, 14), an increased probability of remission in patients with membranous nephritis treated with mycophenolate mofetil (15), and even a decreased probability of renal failure if used prior to the onset of lupus nephritis (16). Given these beneficial effects, hydroxychloroquine is gaining ground as an essential therapy in patients with SLE.

The aim of the present study was to investigate whether hydroxychloroquine use delays the development of overall renal damage as measured by the Systemic Lupus Collaborating Clinics/American College of Rheumatology Damage Index (SDI) renal domain (17) by studying patients from the LUpus in MInorities, NAture versus nurture (LUMINA) multiethnic US cohort who had renal involvement. Our working hypothesis is that the use of hydroxychloroquine will be associated with delaying such occurrence in these patients.

PATIENTS AND METHODS

Patients.

LUMINA is a longitudinal observational cohort that was established in 1994 as a multicenter collaborative effort between the University of Alabama at Birmingham, the University of Texas Health Science Center at Houston, and the University of Puerto Rico Medical Sciences Campus with the goal of understanding the clinical and genetic differences observed in SLE patients from various ethnic groups. The Institutional Review Board of all participating institutions approved the study according to the Declaration of Helsinki for research in humans.

Previously, the cohort has been described in relation to the study visits and variables that constitute this database (18, 19). Briefly, at the time this study was performed, the LUMINA cohort was comprised of 635 patients of Hispanic (n = 118 from Texas, n = 102 from Puerto Rico), African American (n = 234), and Caucasian (n = 181) ethnicities who met at least 4 of the updated and revised American College of Rheumatology (ACR) criteria for SLE (20, 21). In general, African Americans and Texan-Hispanics tended to have renal involvement more frequently than the Puerto Rican-Hispanics and Caucasians. African Americans and Texan-Hispanics also tended to accrue damage more rapidly than the other 2 groups. These patients were also, overall, of lower socioeconomic status (22, 23).

Patients were ≥16 years of age and had a disease duration of ≤5 years. Each patient had a baseline or enrollment visit (T0) followed by a 6-month visit and subsequent yearly visits. Time of diagnosis (TD) was defined as the time when each patient met 4 ACR criteria. Each visit included an interview, physical examination, and laboratory tests. Additional clinical information covering the period between scheduled visits as well as data for missed study visits were obtained by review of all available medical records.

For the purpose of these analyses, disease duration was defined as the time period between T0 and the last visit (TL). TL was truncated at the first occurrence of renal damage for those patients who had developed it. For those patients who had not developed renal damage, TL was the time of their last visit, as already noted. Patients who had developed renal damage on or before T0 were not included in these analyses. Cumulative hydroxychloroquine intake was recorded up to the time of renal damage occurrence or to TL if renal damage had not occurred.

Variables.

Lupus nephritis was defined as a renal biopsy result demonstrating World Health Organization (WHO) class II–V histopathology; and/or proteinuria ≥0.5 gm per 24 hours or 3+ proteinuria attributable to SLE; and/or one of the following features also attributable to SLE and present on 2 or more visits performed at least 6 months apart: proteinuria ≥2+, serum creatinine level ≥1.4 mg/dl, creatinine clearance ≤79 ml/minute ≥10 red blood cells or white blood cells/high-power field (hpf), or ≥3 granular or cellular casts/hpf (2). Creatinine level was measured at the laboratories of the participating institutions by the modified Jaffe method in which creatinine combines with picrate in an alkaline solution to form a creatinine-picrate complex (22–24).

Renal damage, the outcome of interest, was defined by the SDI as ≥1 of the following manifestations lasting for at least 6 months: estimated or measured glomerular filtration rate <50%, 24-hour proteinuria ≥3.5 gm, and end-stage renal disease (regardless of dialysis or transplantation). Of note, as per the SDI scoring instructions, all variables are clinically ascertained and the maximum number of possible points for renal damage is 3. Therefore, for those patients who presented with end-stage renal disease without previous points on the renal domain, the maximum score of 3 was given. The independent variable for this study was the cumulative use of hydroxychloroquine prior to renal damage if it had occurred or before TL for those who had not developed renal damage. If otherwise, patients were categorized as those who did not take hydroxychloroquine.

Variables from the different domains were selected based on those known to differ between those who took hydroxychloroquine and those who did not using as a reference our previously published experience, and that of others (10, 25, 26). Variables from the socioeconomic-demographic domain, all obtained at T0, included age, sex, ethnicity, education, poverty (as defined by the US Federal Government, adjusted for the number of subjects in the household) (27), health insurance, and smoking.

Variables selected from the clinical domain are noted. Clinical manifestations (according to the ACR criteria) included arthritis, photosensitivity, malar rash, serositis, as well as renal, hematologic, and neurologic disorders (20, 21), the number of visits to the emergency room and hospitalizations, and whether the patients were alive or dead. Disease activity was assessed using the Systemic Lupus Activity Measure-Revised (SLAM-R) (28, 29), while damage was ascertained with the SDI (renal domain excluded), as already noted. Other variables such as hypertension (≥2 separate abnormal readings of systolic blood pressure ≥140 mm Hg and/or diastolic blood pressure 90 mm Hg, and/or the use of antihypertensive medications), diabetes mellitus (use of insulin and/or hypoglycemic agents), and abnormal creatinine values (per categories of the SLAM-R) were also included because of their potential relationship with the outcome of interest. All variables included in these analyses were measured at T0.

Medication variables studied included the cumulative exposure to cyclophosphamide, azathioprine, mycophenolate mofetil, glucocorticoids (as prednisone equivalent), angiotensin converting enzyme (ACE) inhibitors, and receptor blockers and statins.

Autoantibodies obtained at T0 including anti–double-stranded DNA (by immunofluorescence against Crithidia luciliae [normal <1:10]) (30) and antiphospholipid antibodies (aPL; abnormal >13 IgG phospholipid [GPL] units/ml and/or >13 IgM phospholipid [MPL] units/ml) by enzyme-linked immunoabsorbent assay technique or a positive lupus anticoagulant test (LAC; Staclot Test Diagnostica Stage 92600, Asniére-Sur-Seine, France) (31) were also examined. Patients were considered to be aPL positive if they exhibited abnormal levels of IgM and/or IgG aPL antibodies (>13 GPL units/ml and/or >13 MPL units/ml, respectively) or LAC positivity. Total cholesterol and low-density lipoprotein (LDL) cholesterol calculated using the Friedewal formula in sera and obtained at T0 were also included.

From the behavioral and psychological domain, abnormal illness-related behaviors as assessed by the Illness Behavior Questionnaire (32) were included.

Genetic polymorphisms previously found to be associated with hydroxychloroquine intake (HLA–DRB1*08) (10) or with some items of the renal domain damage (worsening proteinuria, HLA–DRB1*1503) (33), were also examined.

Statistical analysis.

This was a prospective study (from the attribute or hydroxychloroquine intake to the event or renal damage) in the context of a longitudinal observational cohort study. As noted before, patients who developed renal damage on or before T0 (n = 53) were not included in these analyses. The study was limited to those patients with lupus nephritis at the baseline or T0 visit (n = 203). First, all variables listed above were compared between those who took hydroxychloroquine and those who did not using either Student's t-tests or chi-square tests. Those variables significant at P ≤ 0.10 (except death) or clinically relevant (HLA–DRB1*1503, ACE inhibitors and receptor blockers use) were included into a Cox proportional hazards regression model with renal damage, as defined, being the end point. In an alternative model, proteinuria was excluded from the end point. Disease duration from TD to T0 was adjusted for in these models. In both cases, the variables included in the reduced models were selected using a stepwise procedure. Each variable was systematically entered (or omitted) from the model contingent upon it being associated with the dependent variable with P ≤ 0.10 until a final, parsimonious model was obtained. Results are presented as hazard ratios (HRs) with their corresponding 95% confidence intervals (95% CIs). HRs >1 indicate a shorter time to the event (renal damage), while values <1 indicate a longer time. Kaplan-Meier survival curves as a function of hydroxychloroquine intake and WHO class IV glomerulonephritis were also examined with renal damage being the end point.

RESULTS

As seen in Figure 1, 256 of the 635 LUMINA patients had renal involvement at T0, but 53 were excluded because they had already accrued renal damage. Therefore, 203 patients were included in these analyses. Of those 203 patients, 63 (31.0%) had developed ≥1 renal damage domain items over a mean ± SD disease duration of 5.2 ± 3.5 years, the cumulative incidence being 20.2% and 30.0% at 5 and 10 years, respectively. As expected, patients were predominantly women (92.1%) and middle age (mean ± SD 36.9 ± 12.6 years). All ethnic groups were represented with 54 (26.6%) Texan-Hispanics, 26 (12.8%) Puerto Rican-Hispanics, 100 (49.3%) African Americans, and 23 (11.3%) Caucasians. The ethnic distribution among those patients who developed renal damage was 39.7% for the Texan-Hispanics, 3.2% for the Puerto Rican-Hispanics, 47.6% for the African Americans, and 9.5% for the Caucasians (P = 0.0051). The mean ± SD age for patients who developed renal damage was 30.4 ± 9.2 years, while disease duration at TL was 4.0 ± 2.6 years. Regarding the renal domain items of the SDI, 24-hour proteinuria ≥3.5 gm was recorded in 60.3% of the patients, while an estimated or measured glomerular filtration rate <50% occurred in 12.7%, and end-stage renal disease in 6.4% of the patients. In the remaining 20.6% of the patients, 2 (14.2%) or 3 (6.4%) damage items were recorded concomitantly.

Figure 1. LUMINA (LUpus in MInorities, NAture versus nurture) patients included in this study and all others. T0 = baseline or recruitment visit; TL = last visit or development of renal damage.

Hydroxychloroquine use.

Of the 203 patients, 79.3% had taken hydroxychloroquine prior to the event or to TL for those who had not experienced renal damage, while 20.7% had never taken hydroxychloroquine. For those patients taking hydroxychloroquine, the average daily dose (200 mg) was comparable regardless of renal damage occurrence. The highest daily average dose was, however, somewhat higher for those who did not develop renal damage versus those who did (384.0 mg versus 331.1 mg; P = 0.0008). The total duration of hydroxychloroquine intake was comparable in both groups. Table 1 shows the different characteristics of the patients by hydroxychloroquine category. Those who had taken hydroxychloroquine were more likely to have higher frequencies of arthritis, malar rash, photosensitivity, and less severe disease overall. On the other hand, those who had not taken hydroxychloroquine were more likely to be smokers and had more severe disease with higher frequencies of serositis, WHO class IV glomerulonephritis, and higher SLAM-R scores. They also had higher levels of LDL cholesterol, and had received higher doses of glucocorticoids but had been exposed less frequently to azathioprine. There were a higher number of deaths among those who did not take hydroxychloroquine. There were no differences between those who took hydroxychloroquine and those who did not in terms of the frequency of diabetes mellitus, hypertension, abnormal serum creatinine values, and the use of mycophenolate mofetil, cyclophosphamide, statins, and ACE inhibitors and receptor blockers.

Table 1. Baseline socioeconomic-demographic, clinical, serologic, genetic, and treatment characteristics in LUMINA patients with renal involvement as a function of the cumulative hydroxychloroquine intake*

Variable	Hydroxychloroquine use		P
	Yes (n = 161)	No (n = 42)
* Values are the mean ± SD unless indicated otherwise. LUMINA = LUpus in MInorities, NAture versus Nurture; SLE = systemic lupus erythematosus; WHO = World Health Organization; SLAM-R = Systemic Lupus Activity Measure-Revised; SDI = Systemic Lupus Collaborating Clinics/American College of Rheumatology Damage Index; anti-dsDNA = anti–double-stranded DNA; aPL = antiphospholipid antibodies; LDL = low-density lipoprotein; ACE = angiotensin converting enzyme. † All ethnicities combined. ‡ According to US Federal government guidelines, adjusted for the number of persons in the household. § IgG and/or IgM aPL and/or the lupus anticoagulant. ¶ Prednisone dose.
Age, years	32.9 ± 10.7	32.4 ± 12.5	0.7934
Female, %	90.5	91.3	0.8665
Ethnicity, %			0.2805†
Texan-Hispanic (n = 54)	35.7	24.2
Puerto Rican-Hispanic (n = 26)	7.1	14.3
African American (n = 100)	42.9	50.9
Caucasian (n = 23)	14.3	10.6
Education level, years	12.5 ± 3.1	11.8 ± 3.0	0.1941
Have medical insurance, %	72.8	66.7	0.4350
Below the poverty line, %‡	39.9	47.7	0.3996
Smoking, %	10.0	21.4	0.0453
SLE clinical manifestations, %
Arthritis	87.6	64.3	0.0004
Photosensitivity	60.9	38.1	0.0081
Malar rash	62.7	38.1	0.0040
Serositis	65.2	78.6	0.0985
WHO class IV glomerulonephritis	9.9	33.3	0.0003
Hematologic disease	88.2	88.1	0.9852
Neurologic disease	14.9	16.7	0.7777
Non-SLE clinical manifestations, %
Diabetes mellitus	6.8	7.1	0.9423
Hypertension	34.8	28.6	0.4474
SLAM-R	9.0 ± 5.8	10.8 ± 6.8	0.0839
SDI	0.50 ± 0.90	0.68 ± 1.31	0.3219
Anti-dsDNA antibodies, %	83.2	76.2	0.2926
aPL antibodies, %§	39.8	40.5	0.9319
Abnormal serum creatinine, %	20.5	16.7	0.5784
HLA–DRB1*08, %	20.9	31.7	0.1430
HLA–DRB1*1503, %	21.5	9.8	0.0877
LDL cholesterol	110.4 ± 48.0	138.8 ± 95.8	0.0155
Illness Behavior Questionnaire score	19.4 ± 6.7	18.5 ± 6.5	0.4729
Hospitalizations due to SLE, %	64.7	62.2	0.7792
Emergency room visits due to SLE, %	55.6	54.1	0.8638
Medications
ACE inhibitors, receptor blockers, %	20.5	12.5	0.4577
Statins, %	7.3	12.5	0.4698
Azathioprine use, %	34.2	19.1	0.0594
Glucocorticoid dose, mg¶	11.3 ± 12.0	16.8 ± 20.5	0.0247
Cyclophosphamide use, %	34.8	28.6	0.4475
Mycophenolate mofetil use, %	8.5	7.7	0.9275
Death during followup, %	13.0	42.9	< 0.0001

Multivariable analyses.

Hydroxychloroquine was associated with a longer time to the occurrence of renal damage either in a full (HR 0.12, 95% CI 0.02–0.97) or a reduced (HR 0.29, 95% CI 0.13–0.68) model, after adjusting for confounders (Table 1). Other significant variables in the final model include Texan-Hispanic ethnicity (HR 2.86, 95% CI 1.51–5.40), SLAM-R (HR 1.09, 95% CI 1.04–1.14), and LDL cholesterol (HR 1.00, 95% CI 1.00–1.01) (Table 2). In the alternative model in which proteinuria was omitted from the end point, hydroxychloroquine was also associated with a longer time to the occurrence of the event, reaching significance in the reduced model (HR 0.38, 95% CI 0.16–0.86) (Table 3).

Table 2. Protective effect of hydroxychloroquine in renal damage development among patients with SLE and renal involvement from the LUMINA cohort by multivariable Cox regression analyses*

	Full model			Reduced model
	HR	95% CI	P	HR	95% CI	P
* Renal damage was determined by the Systemic Lupus Collaborating Clinics/American College of Rheumatology Damage Index (estimated or measured <50% glomerular filtration rate, proteinuria ≥3.5 gm, or end-stage renal disease). HR = hazard ratio; 95% CI = 95% confidence interval. See Table 1 for additional definitions. † Diagnosis-recruitment or baseline visit. ‡ Without renal manifestations.
Age	0.92	0.82–1.02	0.1060	0.97	0.95–1.00	0.0547
Female sex	1.18	0.12–11.29	0.8839
Ethnicity
Texan-Hispanic	0.60	0.05–7.28	0.6883	2.86	1.51–5.40	0.0012
Puerto Rican-Hispanic	0.06	0.00–1.12	0.0598
African American	0.18	0.01–2.40	0.1954
Caucasian (reference group)
Time (TD - T0)†	0.90	0.52–1.56	0.7116
Smoking	1.40	0.05–37.72	0.8405
Malar rash	0.55	0.10–2.91	0.4781	0.59	0.31–1.11	0.0989
Photosensitivity	0.65	0.11–3.70	0.6225
Arthritis	1.02	0.10–10.17	0.9851
Serositis	0.88	0.15–5.13	0.8878
WHO class IV glomerulonephritis	2.00	0.28–14.62	0.4913
SLAM-R‡	1.03	0.92–1.15	0.6537	1.09	1.04–1.14	0.0001
LDL cholesterol	1.03	1.00–1.05	0.0137	1.00	1.00–1.01	0.0005
HLA–DRB1*1503	8.78	0.69–112.06	0.0949	2.00	0.98–4.06	0.0567
ACE inhibitors and receptor blockers	1.27	0.15–11.02	0.8288
Hydroxychloroquine	0.12	0.02–0.97	0.0464	0.29	0.13–0.68	0.0043
Azathioprine	0.58	0.07–4.78	0.6139

Table 3. Protective effect of hydroxychloroquine in diminished glomerular filtration rate and/or end-stage renal disease among SLE patients with renal involvement from the LUMINA cohort by multivariable Cox regression analyses

	Full model			Reduced model
	HR	95% CI	P	HR	95% CI	P
*Renal damage was determined by the Systemic Lupus Collaborating Clinics/American College of Rheumatology Damage Index (estimated or measured <50% glomerular filtration rate or end-stage renal disease). HR = hazard ratio; 95% CI = 95% confidence interval. See Table 1 for additional definitions. † Without renal manifestations.
Age	0.97	0.94–1.01	0.1159
Female sex	1.25	0.26–6.04	0.7826
Ethnicity
Texan-Hispanic	2.68	0.49–14.52	0.2533	2.99	1.44–1.22	< 0.0001
Puerto Rican-Hispanic	0.97	0.07–13.02	0.9827
African American	0.76	0.13–4.40	0.0954
Caucasian (reference group)
Malar rash	0.28	0.11–0.75	0.0111	0.46	0.23–0.96	0.0371
Photosensitivity	0.75	0.31–1.86	0.5373	0.47	0.23–0.97	0.0398
Arthritis	2.81	0.58–13.74	0.2019
Serositis	1.64	0.65–4.14	0.2932
WHO class IV glomerulonephritis	1.22	0.49–3.04	0.6746
SLAM-R†	1.14	1.08–1.21	< 0.0001
LDL cholesterol	1.00	1.00–1.00	0.4520
HLA–DRB1*1503	3.03	1.19–7.72	0.0201	2.13	0.96–4.70	0.0615
Hydroxychloroquine	0.38	0.13–1.06	0.0647	0.38	0.16–0.86	0.0206
Azathioprine	0.69	0.28–1.62	0.3788

Survival analysis.

The cumulative probabilities of developing renal damage at 5 and 10 years for those patients who were taking hydroxychloroquine were 20% and 38% compared with 47% and 70% for those who were not (P ≤ 0.0001) (Figure 2A). The corresponding figures for those patients with WHO class IV glomerulonephritis were 65% and 85% compared with 30% and 43% for those without it (P < 0.0001) (Figure 2B).

Figure 2. Cumulative probability of developing renal damage in LUMINA (LUpus in MInorities, NAture versus nurture) patients with lupus nephritis by Kaplan-Meier survival analyses. A, number of patients at risk at each time point; as a function of hydroxychloroquine intake. B, number of patients at risk at each time point; as a function of the presence of World Health Organization (WHO) Class IV glomerulonephritis.

DISCUSSION

In this study we have shown, after adjusting for confounding variables associated with its intake, that hydroxychloroquine retards the development of renal damage in lupus nephritis patients, whether proteinuria is omitted from the end point or not. These data are relevant to the care of patients with SLE given that renal damage is one of the most important causes of morbidity and mortality in these patients (34, 35), and it imposes a burden on the patient and society as well as creates a significant portion of SLE-related health care costs (9). Although the occurrence of clinically evident renal involvement in SLE can be as high as 75% over the course of the disease (36), renal damage has been reported in up to 20% of the patients in some studies (37–39). We realize that although renal involvement and damage represent a continuum, and that some of the histopathologic changes observed in lupus nephritis indeed indicate “tissue damage,” these changes are not recorded in the damage index. Furthermore, these histopathologic changes are reversible with treatment, albeit that does not occur in all patients. Therefore, for the purpose of this study, and according to the definitions used, these 2 constructs, renal involvement and renal damage, can be distinguished. By definition, renal damage according to the SDI is a composite end point. Nevertheless, excluding proteinuria (which can be argued may reverse even after 6 months), the protective effect of hydroxychloroquine was still evident. Several investigators have attempted to determine possible predictors of renal damage (33, 40, 41) and significant work is taking place to develop better treatment strategies than those currently available. There is, however, limited information about medications that may either retard the occurrence of lupus nephritis (42) or delay the onset of renal failure (16). Our results reinforce those from Kasitanon et al indicating that when hydroxychloroquine is taken by patients with membranous lupus nephritis being treated with mycophenolate mofetil, remission is more likely to occur (15), and the recent data from Sisó et al suggesting that exposure to antimalarials prior to the onset of lupus nephritis may prevent the occurrence of renal failure (16). Although propensity score analyses were not performed to adjust for confounding by indication, all the variables that differed between those who took hydroxychloroquine and those who did not were included in the analyses; as noted by several investigators, regression models provide comparable results to those obtained by propensity score analyses, and therefore inferences derived using such methods should be regarded as being entirely adequate (43, 44).

Our findings can be explained by the variety of effects possessed by antimalarials. First, antimalarials have immunoregulatory properties; they inhibit intracellular toll-like receptors, and the traffic of nuclear material into the cells, preventing the formation of autoantibodies and the activation of plasmocytoid dendritic cells with the subsequent diminished production of interferon α, a hallmark of active lupus (45–48). Second, antimalarials exert mild anticoagulant properties inhibiting platelet aggregation and adhesion, reducing blood viscosity and thrombus size (26, 49, 50). Third, antimalarial agents have a favorable effect on serum lipid profile and glucose concentrations (51, 52). Therefore, by preventing the formation of autoantibodies and immune-complexes, diminishing inflammation, and favorably acting at the vascular endothelial level, antimalarials may contribute to an adequate therapeutic response in patients with lupus nephritis retarding the onset of renal damage. We were, however, unable to examine the precise dose of hydroxychloroquine, or the length of time needed for it to exert this beneficial effect, since these data are not captured in our study visits. However, although the average daily dose of hydroxychloroquine was comparable for those who had not developed renal damage with those who had, the maximum average dose was higher in the first than in the second group, again supporting its protective role.

Some limitations of our study are worth noting. First, we could not include the prevalent cases of renal damage in our analyses since the exact temporal relationship between renal damage and hydroxychloroquine use could not be inferred from the data collected. Second, hydroxychloroquine exposure was recorded in a nonblinded manner at the time of study visits; however, the hypothesis tested in this study occurred subsequently and therefore a systematic bias is unlikely to have occurred. Third, all autoantibodies were examined only once, and this may significantly impact the possible relationship between renal damage and aPL antibodies, which tend to significantly fluctuate over time. Finally, although ethnicity was entered into the multivariable analyses, and the data presented are likely to apply to lupus patients of similar characteristics to those of our LUMINA patients, we cannot absolutely conclude that they apply to patients of each ethnic group individually since we could not perform ethnic-specific analyses (sample size).

In summary, using multivariable analyses to adjust for confounding by indication, we have shown that hydroxychloroquine retards the onset of renal damage in patients with lupus nephritis. It is possible that not all confounders have been eliminated. However, based on our current knowledge and our previously published data, we strongly believe that residual confounding has been kept to a minimum. Although the best possible way to reduce confounding to a minimum is to conduct a randomized clinical trial, given the many proven beneficial effects of hydroxychloroquine taken by patients with lupus, such a study is unlikely to be conducted. In fact, it may even be considered unethical. The data presented, taken in conjunction with those previously reported by others (15, 16), suggest that renal damage can be prevented with the administration of hydroxychloroquine.

AUTHOR CONTRIBUTIONS

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 submitted for publication. Dr. Alarcón 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. Pons-Estel, Alarcón, McGwin.

Acquisition of data. Pons-Estel, Alarcón, Bastian, Reveille, Vilá.

Analysis and interpretation of data. Pons-Estel, Alarcón, McGwin, Danila, Zhang.

Acknowledgements

The authors would like to acknowledge all LUMINA patients without whom this study would have not been possible, our supporting staff (Jigna M. Liu, MPH, and Ellen Sowell, AA, at the University of Alabama at Birmingham, Carmine Pinilla-Díaz, MT, at the University of Puerto Rico Medical Sciences Campus, and Robert Sandoval, BA, at the University of Texas Health Science Center at Houston) for their efforts in securing our patients' followup and performing other LUMINA-related tasks, and Ms. Maria Tyson, AA, for her most expert assistance in the preparation of this manuscript. We also gratefully acknowledge Drs. William J. Koopman and Vineeta Kumar for their most helpful comments to an earlier version of this manuscript.