A nationwide survey of hydroxychloroquine retinopathy presenting to the hospital eye service in the United Kingdom

Frequency of hydroxychloroquine retinopathy presenting to the hospital eye service in long-term users

Over a 1-year study period, 66 cases of hydroxychloroquine toxicity were reported through BOSU or by personal communication to the investigators (Fig. 2). One case was excluded due to duplication of reporting. Personal communication was received regarding 2 unreturned questionnaires where the diagnosis had changed. Sixty-three study questionnaires were sent to reporting consultants, of which 46 were returned (73% response rate). After exclusion of a further case of duplication, a case of quinine toxicity, a case of erroneous reporting and cases which received their diagnosis outside the study period of 1 July 2018 to 1 July 2019 (n = 20), 24 incident cases of hydroxychloroquine retinopathy were identified during the study period (Fig. 2). Adjusting for the under-ascertainment (76% BOSU report card return rate) and under-reporting (73% response rate), the potential number of cases of hydroxychloroquine retinopathy presenting to the hospital eye service during the study year was between 24 and 43 cases.

Fig. 2: Flow of case recruitment and exclusion.figure 2

BOSU British Ophthalmological Surveillance Unit.

Twenty-three of the 24 reported patients (96%) had taken hydroxychloroquine for more than 5 years. Using a denominator value of 71,144–77,170 long-term hydroxychloroquine users in the UK, these data suggest that approximately 1 in 1655–3215 (0.03-0.06%) long-term hydroxychloroquine users were diagnosed with retinopathy by the hospital eye service in the UK in the year of study.

Demographic characteristics of reported cases

Complete datasets were obtained for 19 patients and partial datasets for 5 patients. Demographic information, clinical parameters and dosing characteristics of reported patients are summarised in Table 1. In the cohort, there were 23 females and 2 males. The most common indication for hydroxychloroquine use was SLE (n = 12), followed by RA (n = 6), Sjögren’s syndrome (n = 5), mixed connective tissue disease (n = 1) and malaria (n = 1). 75% of individuals were White British (n = 17), with Black Caribbean (n = 3), Other White (n = 2), Other Black (n = 1), and Irish (n = 1).

Table 1 Summary of patient characteristics.

The age distribution at diagnosis of patients reported in this study are presented in Fig. 3.

Fig. 3: Age distribution of patients with hydroxychloroquine retinopathy at diagnosis.figure 3

The mean age at diagnosis was 56.3 years (standard deviation of 17.5 years).

Hydroxychloroquine dosing characteristics

Data relating to daily dose and duration of hydroxychloroquine use were not normally distributed within the cohort. The median daily hydroxychloroquine dose was 262 mg (IQR: 200–400 mg). Although most patients were taking 200 mg or 400 mg per day, some patients were on variable dosing regimens. The median duration of hydroxychloroquine therapy was 19 years (IQR: 14.5–23 years).

Clinical characteristics

Data relating to logMAR visual acuity and visual field parameters were not normally distributed within the cohort. Median visual acuity at diagnosis was logMAR 0.0 for the right and 0.02 for the left eyes (IQR right 0–0.2, left -0.02-0.2). Median 10-2 visual field sensitivity at diagnosis was −3.73 dB in right eyes and -2.77 dB in left eyes (IQR right: −8.52 dB to −3.21 dB, left: −7.55 dB to −2.05 dB). There was no significant difference in visual acuities (p = 0.91) or visual field sensitivities (p = 0.1; Wilcoxon matched-pairs signed rank test for both tests), between the right and left eyes. For subsequent analyses we used visual acuity and visual field data from the worse-seeing eye for each patient. 14 patients underwent automated 10–2 visual field testing, and 3 patients, automated 24-2 visual field testing. Due to the small number of patients with 24–2 field data, these were excluded from statistical analysis of visual field data.

96% of patients (23/24) had at least 2 abnormal tests of the following testing types: fundoscopy, Humphrey visual field, fundus autofluorescence and optical coherence tomography: 76% had at least 3 abnormal tests (16 out of 21). 6 patients went on to have an ERG of which 4 were abnormal; 5 had a multifocal ERG, all of which were reported as abnormal. 57% of patients had disease identifiable on fundoscopy (13/23), and 100% of patients had abnormal visual fields. 92% had abnormal SD-OCT imaging (22/24) and 77% had an abnormal fundus autofluorescence imaging test result (17/22). 63% of patients had ELM disruption at diagnosis (n = 15). The ratio of paracentral: pericentral disease: mixed disease was 18:2:3.

Symptomatic versus asymptomatic patients

The ratio of asymptomatic to symptomatic patients at diagnosis was 14:9 (data missing for one male patient). Of the symptomatic patients, 56% complained of worse visual acuity (one patient had both scotoma and reduced visual acuity and has been counted in both groups, and one further patient had both reduced visual acuity and micropsia and has been counted in both groups) (n = 5), 22% of scotoma (n = 2), 22% of a deficit in colour vision (n = 2), and 33% had ‘other’ symptoms including glare and photopsia (n = 3). We found no statistically significant differences in patient age (p = 0.79; Welch’s unpaired t-test, parametric, Fig. 4A), duration of hydroxychloroquine usage (p = 0.29; Mann–Whitney unpaired t-test, non-parametric, Fig. 4B) visual acuity (p = 0.51; Welch’s unpaired t-test, parametric, Fig. 4C), visual field sensitivity (p = 0.23; Mann–Whitney unpaired t-test, non-parametric, Fig. 4D), hydroxychloroquine dose (p = 0.52; Mann–Whitney unpaired t-test, non-parametric), or the number of patients with ELM disruption (p = 0.13; Fischer’s exact contingency test for dichotomous variables of symptom and ELM disruption status) between the symptomatic and asymptomatic patient groups.

Fig. 4: Comparison of asymptomatic and symptomatic hydroxychloroquine retinopathy patients.figure 4

A Age (median of 59.5 years versus 52 years; p = 0.79, Welch’s t-test). B Duration of therapy (median of 20 years versus 15 years; p = 0.29, Mann–Whitney test). C Visual acuity (median logMAR 0.00 versus 0.06; p = 0.51; Welch’s t-test). D Visual field (median mean deviation −3.5 dB versus −7.26 dB; p = 0.23, Mann–Whitney test). All box plots present median values, interquartile ranges and minimum and maximum values for each chart. The Tukey method was used to exclude outliers which are indicated as points on the relevant box plots.

External limiting membrane (ELM) status on optical coherence tomography imaging

We performed a subgroup analysis of patients with ELM disruption on OCT imaging compared to those with a preserved ELM at diagnosis (Supplementary Fig. 2). There was no statistically significant difference in age (p = 0.30; Welch’s unpaired t-test, parametric), duration of hydroxychloroquine usage (p = 0.82, Mann–Whitney unpaired t-test, non-parametric) hydroxychloroquine dose (p = 0.94, Mann–Whitney unpaired t-test, non-parametric), or presence or absence of symptoms (p = 0.23, Mann–Whitney unpaired t-test, non-parametric) between patients based on ELM status. However, patients with ELM preservation at diagnosis had significantly better visual acuity (median logMAR of 0.0; IQR 0.1; non-parametric) than patients with ELM disruption (median logMAR of 0.14; IQR 0.28; non-parametric) (p = 0.002; Mann–Whitney unpaired t-test, non-parametric). Although there was no statistically significant difference in visual field sensitivity between patients with ELM disruption or ELM preservation at diagnosis (p = 0.18; Mann–Whitney unpaired t-test, non-parametric), there appeared to be a trend towards better visual field status in patients with a preserved ELM (median MD value: −3.51 dB; IQR −2.1 dB) (median MD value: −13.55 dB; IQR −22.75 dB) (Supplementary Fig. 2).

Management

Eighty-three percent (n = 20) of patients were managed with cessation of hydroxychloroquine at diagnosis. 13% (n = 3) were continued at the same dose pending rheumatological review (data unavailable for one case).

留言 (0)

沒有登入
gif