Hydroxychloroquine-induced retinopathy in Puerto Ricans with systemic lupus erythematosus: Associated factors and early-onset risk

Abstract

Objective

To determine factors associated with hydroxychloroquine (HCQ)-induced retinopathy in Puerto Ricans with systemic lupus erythematosus (SLE) and to evaluate determinants of early HCQ-induced retinopathy.

Methods

We conducted a retrospective study of adult Puerto Ricans with SLE. Patients with ophthalmologist-confirmed HCQ-induced retinopathy were compared with those without retinopathy. A subgroup analysis comparing patients with early-onset (≤5 years) versus late-onset (>5 years) retinopathy was also performed. Demographic factors, clinical manifestations, and pharmacologic treatments were analyzed using bivariate methods and logistic regression models adjusted for SLE duration.

Results

A total of 279 patients were included; 41 had HCQ-induced retinopathy, while 238 did not. Among patients with HCQ-induced retinopathy, the mean age (SD) was 45.2 (13.2) years, and 90.2% were women. The mean (SD) SLE duration and HCQ treatment duration were 10.3 (7.7) years and 9.1 (5.8) years, respectively. Patients with HCQ-induced retinopathy were more likely to have higher daily and weight-adjusted HCQ doses and greater damage accrual, but were less likely to have proteinuria, lymphopenia, anti-dsDNA positivity, low complement levels or exposure to mycophenolate mofetil. Among patients with HCQ-induced retinopathy, 13 (31.7%) discontinued HCQ within 5 years of treatment initiation. In multivariable analysis, early-onset toxicity was associated with corticosteroid exposure.

Conclusions

In this cohort of Puerto Rican Hispanics with SLE, HCQ-induced retinopathy occurred in 14.7% of patients, and nearly 32% of those affected developed early-onset retinopathy. Retinopathy was primarily associated with higher daily and weight-adjusted HCQ doses, while early-onset retinopathy was associated with corticosteroid exposure. These findings provide population-specific data that may help optimize HCQ dosing and underscore the importance of regular ophthalmologic screening, even during the early years of therapy.

Keywords

systemic lupus erythematosus hydroxychloroquine-induced retinopathy early-onset retinopathy associated factors clinical outcomes

Introduction

Hydroxychloroquine (HCQ) is an antimalarial drug with known immunomodulatory effects that is widely regarded as one of the most crucial therapeutic agents in the treatment of systemic lupus erythematosus (SLE).¹ Despite its efficacy, the prolonged HCQ use is associated with a risk of retinopathy, a potentially irreversible condition that can lead to permanent vision loss.² The American Academy of Ophthalmology (AAO) both 2016 and 2025 guidelines recommend a maximum daily HCQ dose of ≤5.0 mg/kg of actual body weight to reduce the risk of developing HCQ-induced retinopathy.^3,4 In addition, the AAO advises a baseline fundus examination to exclude preexisting maculopathy and annual retinopathy screening starting after 5 years of treatment in patients without significant risk factors and on appropriate dosages.³ Although these guidelines recommend performing the first ophthalmologic evaluation 5 years after starting HCQ, some studies have shown that HCQ-induced retinopathy can occur within the first 5 years of therapy.^5,6

HCQ-induced retinopathy in SLE patients shows ethnic variability, yet Hispanic patients remain underrepresented in research.⁷ Also, limited data exist on risk factors and on early-onset retinopathy. Therefore, we sought to determine the association of demographic factors, lupus manifestations, comorbidities, and medication exposure in Hispanics from Puerto Rico with HCQ-induced retinopathy, and identify factors linked to early-onset retinopathy.

Methods

Study design and population

A retrospective observational study was conducted in a cohort of Puerto Ricans with SLE followed at the lupus clinics of the University of Puerto Rico Medical Sciences Campus (UPR-MSC). The study period spanned from December 1987 to July 2024. Eligible participants were adults (≥21 years old), of Puerto Rican ethnicity (self and all four grandparents), and diagnosed with SLE according to the 1997 revised American College of Rheumatology (ACR) classification criteria.⁸ Patients attended routine clinic visits every 3–4 months, with additional visits scheduled as needed based on disease activity or complications. At each routine visit, a structured questionnaire was completed to collect information on demographic characteristics, clinical manifestations, laboratory tests (complete blood count, comprehensive metabolic panel, urinalysis, erythrocyte sedimentation rate, C-reactive protein, lipid panel, anti-double-stranded DNA [anti-dsDNA] antibodies, and C3 and C4 complement levels), disease activity, disease damage, and pharmacologic treatment. A lupus autoantibody panel was performed for all patients at the time of SLE diagnosis, including antinuclear antibodies, anti-dsDNA antibodies, anti-Ro antibodies, anti-La antibodies, anti-Smith antibodies, anti-ribonucleoprotein (RNP) antibodies, anticardiolipin antibodies, lupus anticoagulant test, and C3 and C4 complement levels. Autoantibody testing was subsequently repeated as clinically indicated. This study was approved by the Institutional Review Board of the UPR-MSC (protocol# 2290033247).

Assessment of HCQ-induced retinopathy

SLE patients were evaluated at baseline and annually by an ophthalmologist or retinologist using fundus examination, spectral-domain optical coherence tomography (SD-OCT), and automated visual field testing. Hydroxychloroquine (HCQ) retinopathy was diagnosed if fundoscopy revealed parafoveal depigmentation or classic bull’s-eye maculopathy, or if optical coherence tomography (OCT) demonstrated thinning of the photoreceptor layer or disruption of the outer retina, particularly in the parafoveal region.³ Patients were classified as cases if they had received HCQ and subsequently discontinued it due to HCQ-induced retinopathy confirmed by an ophthalmologist. Patients who continued HCQ therapy without developing HCQ-induced retinopathy were included as controls. Patients with HCQ-induced retinopathy were further classified as having early-onset (≤5 years) or late-onset (>5 years) retinopathy. Participants were excluded from the study if HCQ was discontinued for reasons other than retinopathy (e.g., non-adherence, allergic reactions, or other adverse effects), if there was no documented history of HCQ use, or if medical records were incomplete and precluded accurate assessment.

Variables

Variables from demographic, clinical, and pharmacologic domains were ascertained. Demographic factors included age, sex, and SLE duration. Cumulative (at any time during follow-up) clinical manifestations according to the American College of Rheumatology (ACR) classification criteria and serologic markers were determined.⁸ Disease activity and damage at last study visit were assessed using the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) and the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SDI).^9,10 Comorbidities included tobacco use, overweight/obesity (body mass index ≥25 kg/m²), dyslipidemia, arterial hypertension, diabetes mellitus, coronary artery disease, congestive heart failure, stroke, chronic kidney disease, cataracts, diabetic retinopathy, and hypothyroidism. Current (at the last study visit) and cumulative pharmacologic treatments were evaluated. These included SLE therapies (HCQ, corticosteroids, mycophenolate mofetil (MMF), azathioprine, tacrolimus, methotrexate, cyclophosphamide, belimumab, and rituximab), as well as nonsteroidal anti-inflammatory drugs (NSAIDs), statins, angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARBs), low-dose aspirin, and tamoxifen. For HCQ, current and cumulative doses and weight-adjusted dosing (mg/kg/day) were also calculated.

Statistical analyses

First, demographic characteristics, SLE clinical manifestations, comorbidities, disease activity, disease damage, and pharmacologic treatments were compared between patients with and without HCQ-induced retinopathy using bivariate analyses. For patients with HCQ-induced retinopathy, the last study visit (current visit) corresponded to the date on which HCQ was discontinued. Student’s t tests were used for continuous variables and chi-square tests for categorical variables. Variables demonstrating statistical significance in bivariate analyses were subsequently evaluated in logistic regression models adjusted for SLE duration, with odds ratios (ORs) and 95% confidence intervals (CIs) reported.

A subgroup analysis was then performed among patients with HCQ-induced retinopathy to compare early (≤5 years) versus late (>5 years) onset of toxicity. Predictors showing statistical significance in the initial subgroup comparisons were entered into penalized logistic regression models adjusted for SLE duration. Firth’s penalized likelihood method was applied to address small-sample bias and separation. Odds ratios with 95% confidence intervals were reported.

In multivariable analyses across all study groups, models were adjusted for disease duration to account for potential differences in follow-up time between groups and to mitigate bias arising from longer exposure periods among patients who remained on therapy without developing HCQ-induced retinopathy compared with those who experienced earlier events.

A two-sided p-value <0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 13 (IBM Corp., Armonk, NY, USA), except for logistic regression analyses, which were conducted using StataNow/SE version 19.5 (StataCorp LLC, College Station, TX, USA).

Results

From the total cohort of 414 SLE patients, 135 did not meet the eligibility criteria and were excluded. Of the remaining 279 patients, 41 developed HCQ-induced retinopathy, while 238 continued HCQ therapy without evidence of retinopathy. Among patients with HCQ-induced retinopathy, the mean age (SD) was 45.2 (13.2) years, and 90.2% were women. The mean (SD) SLE duration was 10.3 (7.7) years, and the mean (SD) HCQ treatment duration was 9.1 (5.8) years. Figure 1 shows the distribution of HCQ treatment duration prior to discontinuation due to HCQ-induced retinopathy. HCQ exposure ranged from 1 to 23 years. The most frequent time to HCQ discontinuation occurred at 5 years (6 patients), followed by 6 years (5 patients). Overall, 13 patients (31.7%) discontinued HCQ within 5 years of treatment initiation. The mean HCQ dose based on actual body weight was 5.4 mg/kg/day. Of the 41 patients, 26 had a mean HCQ dose greater than 5 mg/kg/day, while 15 had a mean dose ≤5 mg/kg/day. 15 patients were diagnosed by fundus examination and 26 by OCT. It is important to note that some patients in this cohort initiated HCQ therapy prior to the 2016 revision of the American AAO guidelines, which recommended a maximum daily dose of ≤6.5 mg/kg based on ideal body weight.

Figure 1.

Distribution of hydroxychloroquine (HCQ) treatment duration among systemic lupus erythematous patients who developed HCQ-induced retinopathy.

Table 1 summarizes the demographic, clinical characteristics, and pharmacological therapies of patients with and without HCQ-induced retinopathy. Patients without HCQ-induced retinopathy had a longer mean (SD) SLE duration (14.0 [8.9] years vs 10.3 [7.7] years, p = 0.012) and longer HCQ treatment duration (14.4 [15.2] years vs 9.1 [5.8] years, p = 0.030) compared with patients with HCQ-induced retinopathy. Also, a higher frequency of proteinuria, leukopenia, anti-dsDNA antibodies, and lower C3 and C4 complement levels was observed among patients without HCQ-induced retinopathy. In contrast, patients with HCQ-induced retinopathy had a higher mean (SD) current HCQ dose (380.5 [60.1] mg/day vs 316.8 [91.7] mg/day, p < 0.001) and higher mean (SD) weight-adjusted HCQ dosing (5.4 [1.4] mg/kg/day vs 4.4 [1.2] mg/kg/day, p < 0.001). Patients with HCQ-induced retinopathy also had higher mean SDI score (2.3 [1.4] vs 1.2 [1.4], p < 0.001). However, this difference was no longer significant after removing the SDI point corresponding to retinopathy. No differences were observed in SLEDAI scores or comorbidities between the groups. In terms of pharmacological therapies, no differences were observed in immunosuppressive therapies or non–lupus-related medications, except that patients without HCQ-induced retinopathy had greater exposure to mycophenolate mofetil (current and cumulative) and angiotensin II receptor blockers compared with those with retinopathy. After adjustment for SLE duration (Table 2), variables from Table 1 that showed statistical significance in the bivariate analyses were evaluated in logistic regression models. Higher HCQ dose (OR 1.011, 95% CI 1.005–1.017, p < 0.001) and higher weight-adjusted HCQ dosing (OR 1.682, 95% CI 1.304–2.170, p < 0.001) were independently associated with increased odds of HCQ-induced retinopathy, whereas HCQ treatment duration and cumulative dose were not. Higher Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SDI) scores were also associated with increased odds of retinopathy (OR 1.817, 95% CI 1.439–2.293, p < 0.001). In contrast, proteinuria, lymphopenia, anti-dsDNA positivity, and low complement levels (C3 and C4) were associated with lower odds of HCQ-induced retinopathy.

Table 1.

Demographic factors, HCQ dosages, lupus manifestations, comorbidities, disease activity, disease damage and current (last visit) and cumulative (at any time) pharmacologic therapy in SLE patients with and without HCQ-induced retinopathy.

Features	Patients with retinopathy n = 41	Patients without retinopathy n = 238	p-value
Age, mean (SD) years	45.18 (13.2)	43.18 (13.3)	0.381
Sex, % female	90.2	94.1	0.351
SLE duration, mean (SD) years	10.3 (7.7)	14.0 (8.9)	0.012
Hydroxychloroquine dosage
Treatment duration, mean (SD) years	9.12 (5.8)	14.4 (15.2)	0.030
Current dose, mean (SD) mg/day	380.5 (60.1)	316.8 (91.7)	<0.001
Dose per weight, mean (SD) mg/kg/day	5.4 (1.4)	4.4 (1.2)	<0.001
Cumulative dose, mean (SD) grams	1277.8 (871.0)	1639.4 (1912.0)	0.236
Cumulative SLE manifestations, %
Malar rash	65.0	63.0	0.811
Discoid rash	12.5	13.9	0.816
Oral ulcers	45.0	39.1	0.479
Photosensitivity	77.5	72.3	0.490
Arthritis	65.0	66.4	0.864
Pericarditis	2.5	9.2	0.152
Pleuritis	2.5	12.2	0.068
Cellular casts	10.0	8.4	0.739
Proteinuria	30.0	47.1	0.045
Seizures	12.5	9.7	0.581
Psychosis	0.0	2.9	0.272
Hemolytic anemia	7.5	12.2	0.390
Leukopenia (<4.0 × 10⁹/L)	42.5	42.9	0.735
Lymphopenia (<1.5 × 10⁹/L)	52.5	73.9	0.006
Thrombocytopenia (<100 × 10⁹/L)	15.0	18.1	0.638
Antinuclear antibodies	90.0	92.3	0.627
Anti-dsDNA antibodies	38.5	64.9	0.002
Anti-Smith antibodies	20.8	27.6	0.484
Anticardiolipin antibodies	43.8	24.7	0.167
Lupus anticoagulant test	21.4	21.0	0.969
Low C3	55.0	60.3	<0.001
Low C4	32.5	57.4	0.003
Anti-Ro antibodies	38.5	39.4	0.924
Anti-La antibodies	21.7	15.7	0.458
Comorbidities, %
Tobacco use	4.9	12.2	0.169
Overweight/obesity	65.9	64.3	0.846
Dyslipidemia	39.0	35.7	0.684
Arterial hypertension	61.0	56.7	0.611
Diabetes mellitus	9.7	10.1	0.942
Coronary artery disease	4.9	1.7	0.192
Congestive heart failure	4.9	1.7	0.192
Stroke	9.8	4.6	0.178
Chronic kidney disease	7.3	6.7	0.889
Hypothyroidism	31.7	24.5	0.327
SLEDAI, mean (SD) score	3.1 (3.7)	2.6 (3.3)	0.458
SDI, mean (SD) score
Not including retinopathy item in cases^a	1.3 (1.4)	1.2 (1.4)	0.402
Including retinopathy item in cases	2.3 (1.4)	1.2 (1.4)	<0.001
Pharmacological therapy, %
Corticosteroids
Current	61.0	56.3	0.577
Cumulative	87.8	89.5	0.747
IV Pulse corticosteroids
Current	0.0	0.0	--
Cumulative	12.2	21.0	0.190
Mycophenolate mofetil
Current	17.1	6.6	0.015
Cumulative	26.8	44.1	0.038
Azathioprine
Current	9.8	9.7	0.985
Cumulative	19.5	28.6	0.229
Tacrolimus
Current	2.4	6.7	0.287
Cumulative	2.4	8.0	0.204
Methotrexate
Current	4.9	2.5	0.404
Cumulative	4.9	8.4	0.439
Cyclophosphamide
Current	0.0	0.8	0.556
Cumulative	19.5	19.7	0.972
Belimumab
Current	0.0	0.0	--
Cumulative	0.0	0.0	--
Rituximab
Current	0.0	0.8	0.556
Cumulative	2.4	4.6	0.525
NSAIDs
Current	12.2	10.1	0.682
Cumulative	30.0	29.0	0.897
ACE inhibitors
Current	24.4	16.4	0.214
Cumulative	39.0	30.3	0.264
ARBs
Current	15.0	31.9	0.030
Cumulative	24.4	38.7	0.080
Statins
Current	22.0	22.3	0.964
Cumulative	26.8	29.8	0.697
Low-dose aspirin
Current	22.0	23.5	0.825
Cumulative	29.3	32.8	0.657
Tamoxifen
Current	0.0	0.0	--
Cumulative	0.0	0.0	--

HCQ: hydroxychloroquine; SLE: systemic lupus erythematosus; SD: standard deviation; dsDNA: double stranded DNA; SLEDAI: systemic lupus erythematosus disease activity index; SDI: systemic lupus international collaborating clinics/ACR damage index; IV: intravenous; NSAIDs; nonsteroidal anti-inflammatory drugs; ACE: angiotensin converting enzyme; ARBs: angiotensin II receptor blockers.

^aThe score corresponding to the retinopathy item was excluded from the SDI.

Table 2.

Logistic regression analysis of factors associated with HCQ-induced retinopathy in SLE patients.

Features	OR	95% Confidence interval	p-value
HCQ treatment duration, years	0.948	0.868–1.034	0.225
HCQ current dose, mg/day	1.011	1.005–1.017	<0.001
HCQ dose per weight, mg/kg/day	1.682	1.304–2.170	<0.001
Proteinuria	0.459	0.222–0.951	0.036
Lymphopenia (<1.5 × 10⁹/L)	0.377	0.190–0.749	0.005
Anti-dsDNA antibodies	0.348	0.171–0.707	0.004
Low C3	0.238	0.113–0.502	<0.001
Low C4	0.355	0.174–0.724	0.004
SDI	1.817	1.439–2.293	<0.001
Current MMF treatment	0.339	0.143–0.802	0.014
Cumulative MMF treatment	0.436	0.207–0.919	0.029
Current ARBs treatment	0.401	0.160–1.003	0.051

HCQ: hydroxychloroquine; SLE: systemic lupus erythematosus; dsDNA: double stranded DNA; SDI: systemic lupus international collaborating clinics/ACR damage index; MMF: mycophenolate mofetil; ARB: angiotensin II receptor blockers.

Adjusted for SLE duration in the multivariable model.

Among SLE patients with HCQ-induced retinopathy, a subgroup analysis was performed comparing those with early-onset (≤5 years) and late-onset (>5 years) retinopathy (Table 3). No significant differences were observed between the groups in daily HCQ dose or weight-adjusted HCQ dosing. Patients with late-onset retinopathy had a longer mean (SD) SLE duration (12.6 [7.0] years vs 5.2 [6.8] years, p = 0.003). As expected, patients with late-onset retinopathy also had significantly longer HCQ treatment duration and higher cumulative HCQ doses. Patients with early-onset retinopathy had a significantly higher frequency of hemolytic anemia and thrombocytopenia, whereas patients with late-onset retinopathy had a higher frequency of anticardiolipin antibodies. No differences were observed in comorbidities, disease activity (SLEDAI), or disease damage (SDI) between the groups. A significantly higher proportion of patients with early-onset retinopathy were receiving corticosteroids at the time of evaluation. No other differences were observed. In the logistic regression model adjusted for SLE duration (Table 4), shorter HCQ exposure (OR 0.023, 95% CI 0.018–0.030) and lower cumulative HCQ dose (OR 0.994, 95% CI 0.991–0.999) were associated with early-onset retinopathy. In addition, current corticosteroid therapy was independently associated with early toxicity (OR 7.67, 95% CI 1.23–47.71). Hemolytic anemia, thrombocytopenia, and anticardiolipin antibodies were not statistically significant in the multivariable analysis.

Table 3.

Demographic factors, HCQ dosages, lupus manifestations, comorbidities, disease activity, and disease damage and current (last visit) and cumulative (at any time) pharmacologic therapy in SLE patients with early-onset (≤5 years) and late-onset (>5 years) HCQ-induced retinopathy.

Features	Early retinopathy n = 13	Late retinopathy n = 28	p-value
Age, mean (SD) years	41.0 (16.8)	47.2 (11.0)	0.169
Sex, % female	92.4	89.3	0.762
SLE duration, mean (SD) years	5.2 (6.8)	12.6 (7.0)	0.003
Hydroxychloroquine dosages
Treatment duration, mean (SD) years	3.2 (2.0)	11.9 (5.0)	<0.001
Current dose, mean (SD) mg/day	384.6 (55.5)	378.6 (63.0)	0.769
Dose per weight, mean (SD) mg/kg/day	5.5 (1.6)	5.3 (1.4)	0.827
Cumulative dose, mean (SD) grams	442.6 (273.8)	1665.5 (773.8)	<0.001
Cumulative SLE manifestations, %
Malar rash	76.9	60.7	0.308
Discoid rash	23.1	7.1	0.142
Oral ulcers	23.1	53.6	0.067
Photosensitivity	69.2	82.1	0.353
Arthritis	46.1	75.0	0.096
Pericarditis	0.0	3.6	0.491
Pleuritis	0.0	3.6	0.491
Cellular casts	0.0	14.3	0.151
Proteinuria	23.1	32.1	0.553
Seizures	15.4	10.7	0.667
Psychosis	0.0	0.0	---
Hemolytic anemia	23.1	0.0	0.008
Leukopenia (<4.0 × 10⁹/L)	30.8	42.9	0.599
Lymphopenia (<1.5 × 10⁹/L)	38.5	57.1	0.265
Thrombocytopenia (<100 × 10⁹/L)	30.8	7.1	0.046
Antinuclear antibodies	76.9	96.4	0.050
Anti-dsDNA antibodies	38.5	35.7	0.931
Anti-Smith antibodies	0.0	17.9	0.146
Anticardiolipin antibodies	0.0	25.0	0.017
Lupus anticoagulant test	0.0	10.7	0.188
Low C3	23.1	28.6	0.716
Low C4	23.1	35.7	0.416
Anti-Ro antibodies	7.7	32.1	0.158
Anti-La antibodies	7.7	14.3	0.869
Comorbidities, %
Tobacco use	0.0	7.1	0.314
Overweight/obesity	69.2	64.3	0.756
Dyslipidemia	23.1	46.4	0.182
Arterial hypertension	61.5	60.7	0.960
Diabetes mellitus	23.1	3.6	0.056
Coronary artery disease	7.7	3.6	0.585
Congestive heart failure	0.0	7.1	0.314
Stroke	7.7	10.7	0.737
Chronic kidney disease	0.0	10.7	0.211
Cataracts	30.8	32.1	0.930
Diabetic retinopathy	7.7	3.6	0.585
Hypothyroidism	46.2	25.0	0.113
SLEDAI, mean (SD) score	3.2 (4.5)	3.0 (3.3)	0.925
SDI, mean (SD) score	2.4 (1.4)	2.3 (1.4)	0.837
Pharmacological therapy, %
Corticosteroids
Current	84.6	51.9	0.034
Cumulative	92.3	85.7	0.548
IV Pulse corticosteroids
Current	0.0	0.0	--
Cumulative	7.7	14.3	0.809
Mycophenolate mofetil
Current	15.4	17.9	0.809
Cumulative	23.1	28.6	0.663
Azathioprine
Current	7.7	10.7	0.762
Cumulative	7.7	25.0	0.193
Tacrolimus
Current	0.0	3.6	0.482
Cumulative	0.0	3.6	0.482
Methotrexate
Current	7.7	3.6	0.586
Cumulative	7.7	3.6	0.585
Cyclophosphamide
Current	0.0	0.0	--
Cumulative	15.4	21.4	0.612
Belimumab
Current	0.0	0.0	--
Cumulative	0.0	0.0	--
Rituximab
Current	0.0	0.0	--
Cumulative	7.7	0.0	0.143
NSAIDs
Current	23.1	7.1	0.157
Cumulative	38.5	25.9	0.453
ACE inhibitors
Current	15.4	28.6	0.326
Cumulative	30.8	42.9	0.399
ARBs
Current	23.1	11.1	0.343
Cumulative	23.1	25.0	0.847
Statins
Current	15.4	25.0	0.452
Cumulative	23.1	28.6	0.663
Low-dose aspirin
Current	30.8	17.9	0.353
Cumulative	30.8	28.6	0.886
Tamoxifen
Current	0.0	0.0	--
Cumulative	0.0	0.0	--

Table 4.

Logistic regression analysis of factors associated with early-onset HCQ-induced retinopathy in SLE patients.

Features	OR	95% Confidence interval	p-value
HCQ treatment duration	0.023	0.018–0.030	<0.001
HCQ cumulative dose	0.994	0.991–0.999	0.009
Hemolytic anemia	8.402	0.378–186.5	0.178
Thrombocytopenia	9.981	0.737–135.2	0.084
Anticardiolipin antibodies	0.028	0.000–2.777	0.127
Current corticosteroid treatment	7.671	1.233–47.71	0.029

HCQ: hydroxychloroquine; SLE: systemic lupus erythematosus.

Adjusted for SLE duration in the multivariable model.

Discussion

This study provides important insights into HCQ–induced retinopathy in a Puerto Rican population of patients with SLE, demonstrating both a relatively high overall prevalence and a notable frequency of early-onset HCQ-induced retinopathy. In our cohort of 279 patients treated with HCQ, nearly 15% developed HCQ-associated retinopathy. The prevalence observed in our population is higher than that reported in several previous studies. For example, in a Canadian case–control study by Mukwikwi et al., which included 326 patients exposed to antimalarial drugs, 5.5% developed HCQ-induced retinopathy over a mean follow-up of 12.8 years.¹¹ This study included patients treated with both HCQ and chloroquine (CQ), although approximately 90% were receiving HCQ. Similarly, in a large cohort study conducted within the Kaiser Permanente Northern California system by Melles et al., HCQ retinopathy was identified in 2.4% of 3325 patients.¹² However, several methodological differences may partly explain the higher prevalence observed in our cohort. In the Kaiser Permanente study, retinopathy may have been underestimated among patients who initiated HCQ before the implementation of screening with OCT, a more sensitive modality for detecting early HCQ-induced retinopathy than fundoscopy.³ In addition, that cohort included patients with a variety of rheumatologic diseases rather than exclusively patients with SLE, which may further contribute to differences in the observed prevalence rates.

We confirmed that higher current daily and weight-adjusted HCQ doses are associated with an increased risk of HCQ-induced retinopathy.² However, in our cohort, 36.5% of retinopathy cases occurred in patients receiving an HCQ dose of ≤5 mg/kg/day based on actual body weight. This finding is particularly noteworthy, also considering that the mean daily dose in our population was 5.4 mg/kg/day, which closely approximates the 2016 recommendations of the American Academy of Ophthalmology (AAO) that advise a maximum daily dose of ≤5.0 mg/kg/day based on actual body weight to reduce the risk of retinopathy.³ It is important to note that some patients in our cohort initiated HCQ therapy under earlier AAO guidelines that recommended dosing up to 6.5 mg/kg/day based on ideal body weight, which may have contributed to higher cumulative exposure over time. In a cohort study by Melles et al., a progressively higher risk of retinopathy was observed across increasing hydroxychloroquine dose categories after 15 years of therapy, ranging from approximately 3% among patients receiving ≤5 mg/kg/day, to 11% among those receiving >5–6 mg/kg/day, and up to 22% among those receiving >6 mg/kg/day.¹² Similarly, in the case-control study by Mukwikwi et al., 12 of the 18 patients who developed retinopathy had average daily doses exceeding recommended limits.¹¹ In an Indian cohort study by Manoj et al., which included patients with various rheumatologic diseases including SLE, the mean daily HCQ dose among those who developed retinopathy was 5.7 ± 0.9 mg/kg/day based on actual body weight.¹³ In contrast, the substantial proportion of retinopathy cases occurring within guideline-recommended dosing in our cohort suggests that dosing considered acceptable by current standards may not fully mitigate retinopathy risk in this population. This finding highlights the potential contribution of additional patient-specific or disease-related factors influencing susceptibility to HCQ-induced retinopathy.

The potential for confounding by indication may explain our finding of a higher prevalence of MMF use among patients without HCQ-induced retinopathy. This association likely reflects a greater burden of lupus nephritis within this subgroup, as MMF is a cornerstone therapy for renal involvement in SLE.¹⁴ This interpretation is supported by the increased frequency of proteinuria, hypocomplementemia, and anti-dsDNA positivity, hallmarks of active nephritis.¹ These patterns suggest differences in disease phenotype and management rather than a protective effect of inflammatory activity itself. Patients with renal involvement typically undergo intensified immunosuppressive therapy, closer clinical monitoring, and more frequent HCQ dose adjustments, which may limit sustained high-dose exposure and enhance early toxicity detection.¹⁵ In contrast, patients without major organ involvement may remain on higher HCQ doses for longer periods, increasing cumulative retinal exposure. Together, these findings suggest that treatment strategies and disease severity, particularly in the context of lupus nephritis, may indirectly influence the risk of HCQ-induced retinopathy rather than any pharmacologic interaction between MMF and HCQ. Consequently, the observed difference in MMF use should not be interpreted as evidence of a protective effect against HCQ-induced retinal toxicity, but rather as a reflection of underlying differences in disease manifestations and treatment indications between groups. Finally, the higher proportion of lymphopenia observed in our patients who did not develop HCQ-induced retinopathy probably reflects the bidirectional relationship between hematologic disease activity and immunosuppressive management in SLE. Lymphopenia is a common manifestation of active immune-mediated disease and often serves as an indication for intensified immunosuppressive therapy, particularly in patients with concurrent lupus nephritis and serologic activity.¹⁶

Our study found that 31.7% of HCQ-induced retinopathy cases occurred in patients with a hydroxychloroquine exposure duration of ≤5 years. This higher proportion is notably greater than that reported in most previous studies. One possible explanation for this difference is the more intensive ophthalmologic surveillance in our cohort, as patients underwent annual ophthalmologic screening beginning immediately after initiation of HCQ therapy. This approach may have increased the detection of early or subclinical retinal abnormalities that could have been missed in cohorts following less frequent screening protocols. Another potential explanation is the higher cumulative exposure to HCQ doses in earlier treatment eras, as a significant proportion of our patients were treated under previous AAO guidelines that recommended dosing up to 6.5 mg/kg/day and different baseline and follow-up screening practices, in contrast to current recommendations. Furthermore, the earliest retinopathy in our population occurred less than a year after starting HCQ therapy. In contrast, the Canadian study by Mukwikwi et al. had no cases detected in the first 5 years of therapy, and the minimum number of years of exposure to HCQ before retinopathy developed was 8 years.¹¹ A shorter duration of treatment before toxicity was detected in the Systemic Lupus International Collaborating Clinics (SLICC) inception cohort by Almeida-Brasil et al., with the earliest case being detected after 4 years of HCQ initiation.¹⁷ Our findings challenge the AAO guideline recommending annual screening for retinopathy to begin after 5 years of HCQ use for patients on acceptable doses and without significant risk factors, like renal disease. However, in our study, the prevalence of chronic kidney disease did not differ significantly between patients with early-onset and those with late-onset HCQ-induced retinopathy. Our data suggest that a significant proportion of retinopathy can manifest earlier, underscoring the potential for early-onset toxicity in this population. The contrasts between our results and those seen in other studies highlight the need to consider ethnic variations, particularly the possibility that genetic predispositions or unique metabolic profiles in Puerto Ricans may contribute to the earlier onset of HCQ toxicity.

A higher proportion of patients with early HCQ-induced retinopathy had greater current corticosteroid exposure at the time of HCQ discontinuation. This observation was further supported by adjusted analyses, in which corticosteroid use was identified as an independent factor associated with early-onset retinopathy. This association may reflect the higher disease activity and greater corticosteroid requirements typically observed during the early years of SLE, rather than a direct effect of corticosteroid therapy itself.¹⁸ Although a higher proportion of hemolytic anemia and thrombocytopenia was observed among patients with early HCQ-induced retinopathy in the bivariate analyses, these associations did not reach statistical significance in the regression models.

The relatively elevated frequency of HCQ-induced retinopathy, including early-onset retinopathy, in our population could reflect underlying pharmacogenomic and pharmacokinetic variability. HCQ-induced retinopathy is characterized by progressive damage to the outer retina, particularly the photoreceptors and retinal pigment epithelium (RPE).⁴ The pathophysiology involves the accumulation of HCQ in the RPE, likely due to its binding to melanin, which may concentrate the drug and prolong its toxic effects.⁴ Pharmacogenomic and pharmacokinetic genetic variants, particularly in CYP2D6, CYP2C8, and CYP3A4/5, can influence HCQ metabolism and systemic exposure, potentially affecting toxicity risk.¹⁹ Poor metabolizer phenotypes may result in higher drug concentrations and increased risk of retinopathy, while ultra-rapid metabolizers may have lower exposure and reduced efficacy.²⁰ Ethnic differences in the frequency of these alleles are well documented. For example, CYP2D6*10 is more common in East Asians, while CYP2D6*4 is prevalent in Europeans. Latin American populations, including Puerto Ricans, are highly admixed, with variable frequencies of CYP2D6, CYP2C8, and CYP3A4/5 alleles depending on the proportions of European, African, and Amerindian ancestry.^21,22 CYP2D6*31 is a nonfunctional allelic variant that results in a poor metabolizer phenotype for CYP2D6 substrates, including HCQ. In Puerto Rican populations, the CYP2D6*31 allele is present at a frequency of approximately 2%, which is higher than in other major ancestral groups such as North American Caucasians and African Americans, where it is absent.²³ This allele has been directly associated with poor metabolism of CYP2D6 substrates in vivo, as demonstrated by increased parent drug concentrations and reduced metabolite formation.²³ On the other hand, the CYP3A5*3 allele (rs776746), which results in a nonfunctional enzyme due to aberrant splicing, has a high minor allele frequency (MAF ≈76.4%) in the Puerto Rican population, which was similar to frequencies in the general Hispanic group of undetermined ethnicities.²⁴ Comprehensive, population-specific data on the Puerto Rican population’s pharmacogenomic and pharmacokinetic genetic variants remain scarce. Puerto Ricans’ diverse genetic ancestry significantly affects allele frequencies and drug responses; neglecting this may lead to misestimation of pharmacogenetic effects.

This study has some limitations that should be acknowledged. First, its retrospective observational design introduces inherent constraints. Second, the relatively small sample size and limited number of outcome events in the subgroup analyses comparing patients with early versus late HCQ-induced retinopathy may have increased the risk of overfitting in the multivariable models, potentially affecting the stability, precision, and generalizability of the observed associations. Consequently, these findings should be interpreted with caution and warrant confirmation in larger, adequately powered studies. Third, we assumed adherence to prescribed HCQ dosing; however, medication compliance could not be objectively verified, which may have affected estimates of cumulative exposure. Fourth, genetic susceptibility and pharmacogenomic variability were not assessed and may contribute to individual differences in the risk of HCQ-induced retinal toxicity. Finally, because the study period spanned from December 1987 to July 2024, our study may have introduced selection bias, as patients with HCQ-induced retinopathy during the earlier years of the study were diagnosed primarily by fundus examination. Therefore, changes and advances in screening modalities over time may have influenced detection rates. For example, subclinical or early retinal changes in patients who did not discontinue treatment may not have been detected. In fact, 15 patients were diagnosed by fundus examination and 26 by OCT. Although stratification by treatment era would have been valuable, the number of patients within each subgroup was too small to allow meaningful comparisons or adequately powered statistical analyses. Despite this limitation, standardized AAO criteria were consistently applied, and ophthalmologic follow-up was even more rigorous than current AAO recommendations, as all patients underwent baseline examination and annual follow-up after initiating HCQ therapy, rather than baseline examination followed by the first screening at 5 years, as recommended by the AAO guidelines.

In conclusion, the findings of this study have important implications for clinical practice, particularly regarding the monitoring of HCQ-induced retinopathy in the Puerto Rican population. Nearly 30% of HCQ-induced retinopathy cases occurred in patients treated for ≤5 years, suggesting that earlier and more frequent retinal screening, potentially beginning at the initiation of therapy, may be warranted. Additionally, 36.5% of patients who developed retinopathy were receiving HCQ at or below the AAO-recommended maximum daily dose of ≤5.0 mg/kg/day, indicating that even guideline-recommended dosing may not fully lessen the risk of HCQ-induced retinopathy in this population. Although higher daily and weight-adjusted doses were associated with increased risk, a substantial proportion of cases occurred within AAO guideline-recommended dosing ranges. This observation should be interpreted cautiously, as it likely reflects the interplay of multiple factors, including cumulative exposure, disease characteristics, and individual susceptibility, rather than limitations of current recommendations alone. Further studies are needed to better define risk stratification across diverse populations and to optimize screening strategies within existing guideline frameworks.

Footnotes

ORCID iDs

Valeria Valentín-Rosa

Luis M. Vilá

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Siegel

Sammaritano

. Systemic lupus erythematosus: a review. JAMA 2024; 331(17): 1480–1491. https://doi.org/10.1001/jama.2024.2315

Jorge

Melles

Marmor

, et al. Risk factors for hydroxychloroquine retinopathy and its subtypes. JAMA Netw Open 2024; 7(5): e2410677. https://doi.org/10.1001/jamanetworkopen.2024.10677

Marmor

Kellner

Lai

TYY

, et al. Recommendations on screening for chloroquine and hydroxychloroquine retinopathy (2016 revision). Ophthalmology 2016; 123(6): 1386–1394. https://doi.org/10.1016/j.ophtha.2016.01.058

Marmor

Ahn

Ehlers

, et al. Special AAO report: recommendations on screening for hydroxychloroquine retinopathy (2025 revision). Ophthalmology 2026; 133(4): 439–450. https://doi.org/10.1016/j.ophtha.2025.11.001

Melles

Marmor

. The risk of toxic retinopathy in patients on long-term hydroxychloroquine therapy. JAMA Ophthalmol 2014; 132(12): 1453–1460. https://doi.org/10.1001/jamaophthalmol.2014.3459

Petri

Elkhalifa

, et al. Hydroxychloroquine blood levels predict hydroxychloroquine retinopathy. Arthritis Rheumatol 2020; 72(3): 448–453. https://doi.org/10.1002/art.41121

Daftarian

Yue

Levasseur

, et al. Risk of retinopathy associated with long-term use of hydroxychloroquine in patients with rheumatic diseases: a systematic review and meta-analysis. Arthritis Care Res 2026. Epub ahead of print. https://doi.org/10.1002/acr.80033

Tan

Cohen

Fries

, et al. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982; 25(11): 1271–1277. https://doi.org/10.1002/art.1780251101

Mikdashi

Nived

. Measuring disease activity in adults with systemic lupus erythematosus: the challenges of administrative burden and responsiveness to patient concerns in clinical research. Arthritis Res Ther 2015; 17(1): 183. https://doi.org/10.1186/s13075-015-0702-6

10.

Gladman

Goldsmith

Urowitz

, et al. The systemic lupus international collaborating clinics/American college of rheumatology (SLICC/ACR) damage index for systemic lupus erythematosus international comparison. J Rheumatol 2000; 27(2): 373–376.

11.

Mukwikwi

Pineau

Vinet

, et al. Retinal complications in patients with systemic lupus erythematosus treated with antimalarial drugs. J Rheumatol 2020; 47(4): 553–556. https://doi.org/10.3899/jrheum.181102

12.

Melles

Jorge

Marmor

, et al. Hydroxychloroquine dose and risk for incident retinopathy: a cohort study. Ann Intern Med 2023; 176(2): 166–173. https://doi.org/10.7326/M22-2453

13.

Manoj

Sahoo

Singh

, et al. Prevalence of hydroxychloroquine retinopathy with long-term use in a cohort of Indian patients with rheumatic diseases. Rheumatol Int 2021; 41(5): 929–937. https://doi.org/10.1007/s00296-021-04831-5

14.

Sammaritano

Askanase

Bermas

, et al. 2024 American college of rheumatology (ACR) guideline for the screening, treatment, and management of lupus nephritis. Arthritis Rheumatol 2025; 77(9): 1115–1135. https://doi.org/10.1002/art.43212

15.

Kidney Disease: Improving Global Outcomes (KDIGO) Lupus Nephritis Work Group Ayoub

Chan

, et al. KDIGO 2024 clinical practice guideline for the management of lupus nephritis. Kidney Int 2024; 105(1S): S1–S69. https://doi.org/10.1016/j.kint.2023.09.002

16.

Jung

Kwak

Choe

, et al. The effect of mycophenolate mofetil on non-renal manifestations in systemic lupus erythematosus: results from Korean lupus network registry. J Kor Med Sci 2019; 34(27): e185. https://doi.org/10.3346/jkms.2019.34.e185

17.

Almeida-Brasil

Hanly

Urowitz

, et al. Retinal toxicity in a multinational inception cohort of patients with systemic lupus on hydroxychloroquine. Lupus Sci Med 2022; 9(1): e000789. https://doi.org/10.1136/lupus-2022-000789

18.

Wen

Chen

Jin

, et al. Clinical and laboratorial outcome of different age-onset systemic lupus erythematosus patients in Jiangsu, China: a multicentre retrospective study. Sci Rep 2022; 12(1): 10683. https://doi.org/10.1038/s41598-022-14840-4

19.

Paludetto

Kurkela

Kahma

, et al. Hydroxychloroquine is metabolized by cytochrome P450 2D6, 3A4, and 2C8, and inhibits cytochrome P450 2D6, while its metabolites also inhibit cytochrome P450 3A in vitro. Drug Metab Dispos 2023; 51(3): 293–305. https://doi.org/10.1124/dmd.122.001018

20.

Paniri

Hosseini

Rasoulinejad

, et al. Molecular effects and retinopathy induced by hydroxychloroquine during SARS-CoV-2 therapy: role of CYP450 isoforms and epigenetic modulations. Eur J Pharmacol 2020; 886: 173454. https://doi.org/10.1016/j.ejphar.2020.173454

21.

Biswas

Sukasem

. Pharmacogenomics of chloroquine and hydroxychloroquine: current evidence and future implications. Pharmacogenomics 2023; 24(15): 831–840. https://doi.org/10.2217/pgs-2023-0124

22.

Zhou

Lauschke

. Population pharmacogenomics: an update on ethnogeographic differences and opportunities for precision public health. Hum Genet 2022; 141(6): 1113–1136. https://doi.org/10.1007/s00439-021-02385-x

23.

Gaedigk

Isidoro-García

Pearce

, et al. Discovery of the nonfunctional CYP2D6 31 allele in Spanish, Puerto Rican, and US Hispanic populations. Eur J Clin Pharmacol 2010; 66(9): 859–864. https://doi.org/10.1007/s00228-010-0831-4

24.

Claudio-Campos

Duconge

Cadilla

, et al. Pharmacogenetics of drug-metabolizing enzymes in US hispanics. Drug Metab Pers Ther 2015; 30(2): 87–105. https://doi.org/10.1515/dmdi-2014-0023