Quantitative Fundus Autofluorescence in Systemic Chloroquine/Hydroxychloroquine Therapy: One Year Follow-Up

Abstract

Purpose: Systemic chloroquine/hydroxychloroquine (CQ/HCQ) can cause severe ocular side effects including bull's eye maculopathy (BEM). Recently, we reported higher quantitative autofluorescence (QAF) levels in patients with CQ/HCQ intake. Here, QAF in patients taking CQ/HCQ in a 1-year follow-up is reported.

Methods: Fifty-eight patients currently or previously treated with CQ/HCQ (cumulative doses 94-2435 g) and 32 age- and sex-matched healthy subjects underwent multimodal retinal imaging (infrared, red free, fundus autofluorescence [FAF], QAF [488 nm], and spectral-domain optical coherence tomography (SD-OCT). For analysis, custom written FIJI plugins were used for image processing, multimodal image stacks assembling, and QAF calculation.

Results: Thirty patients (28 without BEM and 2 with BEM, age range = 25–69 years) were followed up (370 ± 63 days). QAF values in patients taking CQ/HCQ showed a significant increase between baseline and follow-up examination: 282.0 ± 67.9 to 297.7 ± 70.0 (QAF a.u.), P = 0.002. An increase up to 10% was observed in the superior macular hemisphere. Eight individuals (including 1 patient with BEM) had a pronounced QAF increase of up to 25%. Compared to healthy controls, QAF levels in patients taking CQ/HCQ were significantly increased (P = 0.04).

Conclusions: Our study confirms our previous finding of increased QAF in patients taking CQ/HCQ with a further significant QAF increase from baseline to follow-up. Whether pronounced QAF increase might predispose for rapid progression toward structural changes and BEM development is currently investigated in ongoing studies.

Translational Relevance: In addition to standard screening tools during systemic CQ/HCQ treatment, QAF imaging might be useful in CQ/HCQ monitoring and could serve as a screening tool in the future.

Introduction

Chloroquine (CQ) and hydroxychloroquine (HCQ) are potent drugs clinically used in the treatment of rheumatologic and dermatologic disorders, such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA). Although the short-term side-effect profile indicates good tolerability for both drugs, long-term use bears the risk of developing CQ/HCQ-related bull's eye maculopathy (BEM)^1,2 in advanced stages, characterized by the irreversible attenuation and degradation of photoreceptors and retinal pigment epithelium (RPE) and subsequent irreversible loss of visual acuity and visual field defects.^3,4 

Main risk factors for developing retinal toxicity include a body weight related high daily dose (>2.3 mg/kg real weight [CQ], >5 mg/kg real weight [HCQ]), concomitant renal disease and/or tamoxifen therapy.¹ Melles and Marmor reported a prevalence for CQ/HCQ maculopathy of less than 2% in the first 10 years, but up to 20% after 20 years of CQ/HCQ intake.¹ Therefore, the long-term systemic use of these drugs in the aforementioned chronic diseases might lead to an increasing number of these patients in the practices and clinics in the future. 

Due to the irreversible nature of CQ/HCQ maculopathy, it is mandatory to recognize retinal changes early in order to reconsider continuation of medication and to prevent further progression and potential impairment of vision. As recommended by the American Academy of Ophthalmology (AAO), CQ/HCQ screening includes functional (visual field testing) and structural (spectral-domain optical coherence tomography [SD-OCT] imaging) methods, at baseline and then annually after 5 years of treatment. In addition, screening can be extended by further objective tools, such as multifocal electroretinogram (mfERG) or fundus autofluorescence (FAF).^3,5,6 

Blue FAF imaging is a noninvasive, safe, and easy to perform imaging technique that excites fluorophores in the photoreceptors and mainly in the RPE – and, therefore, serves as an indicator for outer retinal health. FAF has its intensity maximum around 10 to 15 degrees from the fovea and increases with age.⁷ 

Altered FAF patterns have been reported for several retinal diseases, including but not limited to age-related macular degeneration, hereditary retinal diseases, para- and postinfection, tumors, and others.^8–10 Characteristically, in CQ/HCQ maculopathy, FAF shows hyper- and hypo-autofluorescence at the parafovea in early and intermediate stages, but hypo-autofluorescence at late stage, indicative of altered and loss of photoreceptor and/or RPE cells. However, FAF is limited because it only enables qualitative assessments. 

A milestone in quantification of FAF was Delori and colleagues’ introduction of quantitative FAF (QAF) imaging using an internal reference in their device setup.⁷ QAF enables the comparison of FAF between different individuals as well as between sessions of the same individual. 

Recently, we were able to show that QAF is significantly increased across the posterior pole in patients taking CQ/HCQ, as compared to healthy age and sex matched controls,¹¹ later confirmed by others.^12,13 Interestingly, increased QAF was not associated with structural (SD-OCT) or functional (visual field and mfERG) changes in our study cohort, and independent from the CQ/HCQ cumulative dose. It is, however, currently unclear whether an increased QAF level could serve as a biomarker for early detection of CQ/HCQ-related maculopathy development. Therefore, longitudinal studies will help to further clarify QAF's potential role in CQ/HCQ maculopathy screening. 

The goal of the current study was to follow-up on our recent CQ/HCQ patient group and to analyze QAF 1 year after baseline. Our data confirm our initial findings on CQ/HCQ therapy related FAF and adds further knowledge on QAF as a possible imaging tool for CQ/HCQ maculopathy screening. 

Methods

Study Population

In this follow-up cohort study, 30 patients with a history of CQ/HCQ treatment were re-examined 1 year (mean ± STD =: 370 ± 63 days) after their baseline (BL) visit.¹¹ Study participant characteristics are shown in Table 1. The control group of healthy age- and sex-matched subjects is a subgroup of the study collective reported in our previous study.^11,14 All subjects were recruited from the Department of Ophthalmology, University Hospital Würzburg, Germany, between September 2017 and October 2020. All study procedures were in compliance with the tenets of the Declaration of Helsinki, and the study had been approved by the University of Würzburg's ethics Committee (#69/17). All study participants gave written consent after explanation of the study's purpose, risk, and consequences. 

Table 1.

 

Characteristics of Patients at Baseline, Follow-Up, and Control Group

 

Inclusion criteria have been reported previously.¹¹ Briefly, clear optic media, age between 18 and 79 years, and current or past CQ/HCQ intake were required. Subjects were excluded from enrollment if they had spherical equivalent >5 Diopters, had a history or showed retinal pathologies other than CQ/HCQ related maculopathy or BEM, or had a history of vitreoretinal surgery. 

BEM has been diagnosed by retina specialists at the Department of Ophthalmology, University Hospital Würzburg, based on the history of CQ and/or HCQ intake, clinical presentation, retina imaging (SD-OCT and FAF), abnormal multi-focal electroretinography, and/or paracentral defects in the central /paracentral visual fields. 

Screening Modalities

At the 1 year follow-up visit, all patients taking CQ/HCQ underwent examinations as recommended by the AAO³ including functional testing (12 degrees white-on-white automated threshold static perimetry [Octopus, Haag Streit, Bern, Switzerland]; mfERG with 61 segments, 27 degrees angle view [RETIscan, version 6.16.3.8; Roland Consult, Brandenburg an der Havel, Germany]; best visual acuity testing [using early treatment of diabetic retinopathy {ETDRS} charts]; color vision testing using an anomaloscope [HMC anomaloscope; Oculus, Wetzlar, Germany] and Farnsworth-Panel D’15 test [saturated and unsaturated]), as well as multimodal retinal imaging (see below). The follow-up visits also included an ophthalmologic examination with biomicroscopy of the anterior segment, funduscopy (after pupillary dilatation [5% tropicamide and 2.5% phenylephrine]), and applanation tonometry for intraocular pressure. 

Multimodal Imaging

Imaging of the fundus was obtained after sufficient mydriasis of at least 6 mm pupil diameter. Modalities included structural high resolution SD-OCT examination of the macula (fovea-centered, 20 degrees × 20 degrees field, 35 ART frames, and 49 B-scans using the Spectralis, Heidelberg Engineering, Heidelberg, Germany), infrared and red free images (30 degrees × 30 degrees), FAF (excitation = 488 nm and 787 nm), and QAF (excitation = 488 nm, emission = 00–750 nm, and image size = 30 degrees × 30 degrees, and 768 × 768 pixels) using an HRA (Heidelberg Engineering, Heidelberg, Germany), as previously reported.^11,14 

For SD-OCT imaging, the baseline SD-OCT scans served as reference, and the Spectralis internal software (TruTrack Active Eye Tracking) enabled to do the follow-up SD-OCT at exactly the same positions. QAF imaging was performed using a modified, experimental HRA2 camera (Heidelberg Engineering, Heidelberg, Germany), as previously reported.¹¹ The HRA2 device contains an internal reference simultaneously excited and imaged during QAF imaging to enable normalization across subjects and follow-up examinations. Prior to QAF imaging, pigments of the photoreceptors were bleached for at least 20 seconds,¹⁵ followed by registration of 12 single QAF frames. At least nine of these frames were used to plot an average QAF image using the manufacturer's software (criteria for inclusion: homogeneous illumination of the posterior pole, and image centered on the fovea). Subjects were excluded from analysis if fewer than nine frames were useable. The variability between two QAF measurements (2 measurements on 1 day with complete reposition of patient and camera) was tested in a subsequent study with five subjects from each age decade randomly picked for test–retest reliability. The mean difference between two QAF measurements was 7.9% (median = 6.4%, range = 0.05%–22.3%). Further, the device was regularly calibrated to ensure consistent QAF measurements. 

The QAF imaging procedure was then repeated (after new positioning of the subject and camera), and the QAF image with the highest quality was chosen for final analysis. In addition, to further assure high QAF quality, we calculated intersession repeatability on 30 patients (30 eyes) with QAF images being taken on the same day. For that, the difference (%) of QAF97 between the final images of the two recording series was calculated for each patient. 

To determine proper image sizes during analysis, data from previously measured c-curves¹¹ (IOL Master 500; Zeiss, Oberkochen, Germany) of all patients were transferred into the Spectralis software. 

Image PostProcessing

For QAF image processing, custom written FIJI plugins were used, as reported.^11,14 Initially, the image gray scale was adjusted to the reference calibration factor of the modified HRA and to the age-correlated lens transmission of the study participants. The resulting QAF images (gray scale of 0 to 511) were then color coded to an 8-bit image (scale of 0 to 255). 

During the SD-OCT scanning process, an enface IR image was recorded simultaneously. All other images of the different modalities were then aligned to this image by manually selecting anatomic landmarks, such as vessel bifurcations resulting in an image stack with only minimal rotational and translational error of the different images. 

The fovea center point (FCP) was determined using the OCT scan with the deepest foveal depression and the rise of the external limiting membrane.^11,14,16 Using FCP, the same coordinate system could then be generated for all images of the image stack, allowing correct positioning of the QAF analysis grid. 

QAF Analysis

For QAF analysis, the ETDRS and the previously described QAF97 grid were used.¹¹ Both grids are centered at the FCP. Although the ETDRS grid consists of 3 rings with defined diameters of 1, 3, and 6 mm, the maximum diameter of the QAF97 grid is determined by the distance between the edge of the optic disc and the fovea and may, therefore, differ between individuals. Basically, QAF97 consists of 6 concentric rings (with 16 radial segments in each ring) plus foveal and parafoveal regions (see Supplementary Fig. S1). 

For each segment of the two grids, the mean QAF intensities ± standard deviation, the minimum and maximum QAF values, and the number of pixels for each segment were calculated. The reported QAF97 values are the averaged QAF intensities of all segments excluding the fovea. 

Because the control group was imaged at BL only, there are no data to directly determine QAF development over 1 year in this group. Therefore, using recently published QAF data from >150 subjects (5-85 years),¹⁴ the expected increase of QAF a.u./year was calculated by applying a linear regression to the data and using the result to extrapolate the change in QAF after 1 year. 

Retinal Thickness Analysis

All retinal layers in the SD-OCT B-scans were automatically segmented using the HEYEX2 (Heidelberg Engineering, Heidelberg, Germany) software, each scan double-checked for accuracy and manually corrected, if necessary, by two trained and experienced readers. A retinal thickness map was generated using the ETDRS grid in the Spectralis Viewing Module (Heidelberg Eye Explorer; Heidelberg Engineering, Heidelberg, Germany). 

Thehickenss of all inner retinal layers (i.e. the internal limiting membrane [ILM] to the outer nuclear layer [ONL]) and all outer retinal layers (external limiting membrane [ELM] to Bruch membrane), as well as RPE/Bruch membrane's thickness were determined for each subfield of the ETDRS grid.¹¹ Retinal layer thicknesses were correlated to the corresponding QAF values of the ETDRS grid. 

Quantification of the Atrophic Zone in Patients With BEM

In patients with BEM, the atrophic zones in the QAF image were quantified using the following algorithm (see Supplementary Fig. S2): from the 8-bit QAF image of patients with BEM, the foveal and parafoveal area including the atrophic zone was cropped. The size was chosen manually according to the extent of the atrophic area. Blood vessels were excluded because these appear hypo-autofluorescent in QAF imaging and could mimic atrophic areas. In addition, due to blue light blocking properties of macular pigment, the area of its highest density (FCP to approximately 1 degree eccentricity)^17,18 was also excluded from analysis if not affected by CQ/HCQ related atrophy. We then arbitrarily defined a brightness threshold below 120 (QAF a.u.) as atrophy related decrease in QAF. All pixels below this threshold were counted and displayed. Pixels above the threshold were colored in white. With the help of the individual's c-curves, the pixel count was converted to an area (mm²) covering the atrophic zone. 

Statistical Analysis

Data collection, organization, and analysis were performed using the SPSS statistics software package (IBM SPSS 26.0; IBM Corporation, Armonk, NY, USA) and Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA). Categorical variables are presented as numbers and percentages, and continuous variables are expressed as means ± standard deviation. Statistical significance was set at P < 0.05. Differences between mean values were tested using Student's t-test (independent samples for comparison to control group and paired for development from baseline to follow-up). The theoretical increase of QAF for the healthy control group was determined using linear regression and extrapolation. Correlations between retinal thickness and QAF97 were analyzed with Spearman's rank correlation coefficient (rs) (to check for a monotonic relation) and Pearson product-moment correlation coefficient (to check for a linear relation). 

Results

From the initial 58 study participants at BL,¹¹ 42 were examined in the 1-year follow-up. Twelve participants had to be excluded from the study due to insufficient image quality or follow-up examination after the study's cutoff date, resulting in 30 subjects included for statistical analysis. Detailed information about the 30 individuals is summarized in Table 2. The mean ± STD follow-up period was 370 ± 63 days. 25 out of 30 subjects were still on systemic HCQ. Five patients (including 1 patient with BEM) had taken both CQ and HCQ in the past, all others had used only HCQ. All subjects were phakic. All included patients with BEM showed signs of advanced stage CQ/HCQ related maculopathy (parafoveal reduced retinal layer thickness in SD-OCT, reduced FAF, reduced amplitudes in multifocal electroretinography, and affected visual fields). 

Table 2.

 

Characteristics of Study Participants

 

QAF97 Change From Baseline to Follow-Up

Mean QAF97 intensities increased significantly during the course of one year. For the 28 patients without BEM, mean QAF97 intensity increased significantly from 283.0 ± 70.2 (QAF a.u.) to 297.2 ± 71.7 (QAF a.u.) (P = 0.006), whereas for the 2 patients with BEM QAF97 increased from 259.8 to 276.9 and from 277.2 to 332.6 (QAF a.u.). There were only two patients who had taken a significant amount of CQ (>1000 g). One of them (CQ0002) had developed BEM before study start, the other showed the highest QAF97 value of 481.8 (QAF a.u.) of the whole cohort. 

In healthy age- and sex-matched controls, mean QAF97 was significantly lower with 244.2 ± 72.9 (QAF a.u.), P = 0.04, compared to QAF97 of the patients taking CQ/HCQ at their BL examination. Figure 1 shows QAF97 intensities versus age of all study subjects taking CQ/HCQ. Here, the slope corresponds to the yearly increase of QAF97 which equals 2.84 (QAF a.u.)/year, calculated and interpolated from our previous report. However, the difference of the two mean values at BL and follow-up yields QAF_FU - QAF_BL = 297.2 to 283.0 = 14.2 (QAF a.u.), which exceeds the expected age-related effect (see Figs. 1, 2). 

Figure 1.

 

QAF development of all patients taking CQ/HCQ for baseline (blue) and at 1-year follow-up (orange). The dotted lines represent the trend lines for each group (orange = follow-up, blue = baseline, green = control group, and light green = 95% confidence interval for control group). For all groups, an increase in QAF with age can be seen.

Figure 2.

 

Mean values of QAF97 intensities for different subject groups at baseline and at follow-up. For the healthy control group (yellow bar), interpolated QAF97 intensity after 1 year according to age-related increase of 2.84 (QAF a.u.) per year (dotted yellow bar) are shown, (subgroup analysis using data from Kleefeldt et al.¹⁴). QAF97 intensities of patients taking CQ/HCQ without BEM (blue bar) and with BEM (orange bar) at baseline and follow-up visit are plotted.

Repeatability and QAF97 Changes at Follow-Up

In order to evaluate QAF changes from BL to follow-up, we determined the repeatability of QAF measurements. For that purpose, we compared two measurements taken on the same day from the same subject. The between-session repeatability yielded a mean difference between two measurements of 1.2 ± 5.4% (range = −8.9% to +10.5%), which is in line with previously reported variability.^15,19,20 Therefore, we considered a variability of ± 10% within the limits of measurement accuracy for the following analysis. The Bland-Altman method showed that the difference between two measurements is within the statistical limits of agreements (no more than 2 standard deviations from the average) indicating good repeatability (data in Supplementary Fig. S3). 

Our analysis of BL to follow-up changes yielded a QAF97 change between −10% and +10% for the majority of patients (67%, 20 participants), which is within the range of measurement inaccuracy. All non-BEM patients who had stopped taking CQ/HCQ fell into that category. Two study participants showed a decreased QAF97 at the follow-up visit of more than −10%, which can be attributed to a slightly lower image quality at follow-up. Significant development of lens opacity between BL and follow-up was ruled out. Eight subjects (including 1 patient with BEM who had stopped taking CQ/HCQ) had a pronounced QAF97 increase of more than 10% and up to 25%. Neither cumulative CQ/HCQ dosage nor age correlated to the QAF change. 

QAF in Individual Segments and Hemispheres

The autofluorescence intensity distribution across the posterior pole is similar in patients taking CQ/HCQ and in healthy controls with a maximum temporal-superior and a minimum nasal-inferior, as previously reported.^7,11 Similarly, looking at all patients taking CQ/HCQ , including our 2 patients with BEM, the highest change from BL to follow-up is at the superior region with a maximum of approximately 10% (QAF a.u.; Fig. 3). 

Figure 3.

 

Change in QAF intensity (in %) for all QAF97 segments. Change of QAF intensities (follow-up/baseline) were converted to gray scale. Highest QAF change of approximately +10% (black) can be found at the superior region, whereas the lowest QAF change of approximately +1% (light gray) is present inferior segments.

To quantify the difference between QAF intensities in segments of the superior and inferior hemisphere, the mean QAF of all segments above and below the horizontal meridian was calculated and plotted in Figure 4. In all subjects (patients taking CQ/HCQ and control group), the superior segments showed significantly higher QAF values than the inferior ones. Of note, this effect was detectable in both patients taking CQ/HCQ and healthy controls. For the control group, mean QAF of the superior segments was 259.0 ± 76.1 (a.u.) and mean QAF of the inferior segments was 234.1 ± 71.9, P < 0.001; for the patients taking CQ/HCQ, 316.7 ± 79.4 (a.u.) superior and 277.6 ± 65.3 (a.u.) inferior, P < 0.001; for the patients with BEM, 346.1 ± 70.59 (a.u.) superior and 263.35 ± 8.2 (a.u.) inferior, P = 0.31 (n = 2), respectively. 

Figure 4.

 

Average QAF values of segments in the superior and segments in the inferior hemisphere are plotted for patients taking CQ/HCQ in the follow-up group without BEM (blue points), CQ/HCQ patients with BEM (red points), and the healthy control group (yellow points). The dots above the slope represent segments in the superior hemisphere that have higher QAF than in inferior segments in an individual patient.

QAF in Patients With BEM: Tool for Atrophy Progression in Patients With BEM

Generally, atrophic retinal areas are hypo-autofluorescent due to changes in the RPE´s subcellular content of autofluorescent granules, or complete loss of RPE. In the two patients with BEM in our follow-up study, the sizes of areas with autofluorescence loss within the QAF images were used to determine atrophy growth in the 1-year follow-up (Fig. 5). 

Figure 5.

 

Quantification of the atrophic zone after 1-year follow-up. For both patients, the corresponding 8-bit-QAF images including the atrophic zone at baseline and follow-up are shown. Atrophy was defined as QAF values below 120 (QAF a.u.). For the entire atrophic zone, pixels count of the atrophic zone and corresponding retinal area are depicted. (A) The subject is 35 years old; cumulative dose = 887.25 g CQ; who stopped using HCQ 2 months before baseline examination. (B) The subject is 56 years old; cumulative dose of HCQ = 146 g + CQ = 1567 g; stopped using CQ/HCQ 121 months before baseline examination.

The atrophic area was more extensive for the patient who stopped CQ/HCQ treatment over 10 years ago than for the patient who recently stopped her treatment. Furthermore, the atrophic zone is more pronounced inferior to the fovea for both subjects. For both patients, there is a slight increase in atrophy over the course of 1 year. 

Correlation of Retinal Thicknesses With QAF

In order to investigate a potential correlation between QAF and retinal changes, we determined the thickness of several retinal layers, including the RPE. Overall, there were no significant changes of the retinal layers detectable in the 1-year follow-up. 

No correlation was found between the thickness of different retinal layers and the QAF intensities. Figure 6 shows QAF and retinal thickness in a vertical cross section through the FCP. Although there is a rather homogenous loss of retinal thickness in patients with BEM, their QAF intensities decrease most prominently in segment 4 (i.e. inferior parafovea), whereas QAF values in the periphery tend to increase (segments 6 and 8). 

Figure 6.

 

QAF and retinal thickness changes in a 1-year follow-up. (A) A vertical cross section through the ETDRS grid (denoted by the yellow dotted line) was used for analysis of QAF and retinal thickness. (B)Retinal thicknesses of the whole retina and QAF of BL and follow-up and patients with BEM. There is a sharp decrease in QAF of patients with BEM in segment 4 which corresponds to the inferior parafoveal region.

Discussion

In this study, we report the 1-year follow-up of QAF in patients with current or previous systemic CQ/HCQ treatment. Our study confirms our earlier finding of increased QAF in patients taking CQ/HCQ as compared to normal healthy subjects.¹¹ In addition, at 1 year follow-up, the QAF increase significantly continued, most markedly in the superior hemisphere. QAF, therefore, might be a suitable candidate for CQ/HCQ therapy control. 

The intrinsic autofluorescence (AF) signal of the human fundus at 488 nm excitation arises mainly from the outer retina, specifically from the bisretinoids at the photoreceptor and the RPE level.^21,22 Bisretinoids are products/by-products of the visual cycle and some might be deposited as lipofuscin or melanolipofuscin in the RPE.²³ Lipofuscin granules accumulate with age,²⁴ but get lost at higher age and in age-related macular degeneration.^25,26 Therefore, it is believed that FAF reflects the distribution of lipofuscin/melanolipofuscin within the RPE. 

With the development of QAF, the limitation in standard fundus AF imaging of only qualitative interpretation has been surpassed. An internal fluorescent reference in a FAF camera and simultaneous capturing fundus and probe AF enables comparison of AF intensities in follow-up studies and among individuals. Since its introduction in 2011,¹⁵ many clinical studies have investigated QAF signals for healthy eyes as well as in several retinal diseases.^8–10 In addition, different analysis grids have been evaluated and the QAF97 grid has been found most useful for our patients taking CQ/HCQ because it not only covers the parafoveal region but also enables sub-segmental analysis.¹⁴ QAF imaging depends on operator qualification and experience, and cooperation of the subject. In addition, QAF image quality depends on clear optic media, particularly the lens. This was taken into account by both our inclusion criteria and the fact that most patients were below 60 years of age. Higher age could significantly alter QAF and most QAF studies set their cut off at 60 years.⁷ 

Only few data exist on the association between FAF and CQ/HCQ intake.^11–13 In our baseline report,¹¹ we detected a significantly increased QAF as compared to healthy subjects and discussed different mechanisms that could lead to an increased QAF signal associated with CQ/HCQ intake. In different animal studies (rats and rhesus monkeys),^27,28 an accumulation of CQ/HCQ and its metabolites were found, especially in the RPE, presumably due to its high binding ability to melanin. Alternatively, other studies suggested that CQ/HCQ related retinal toxicity results from altered RPE lysosomes.²⁹ As a weak base, CQ/HCQ can easily diffuse past the lysosomal membrane where it gets protonated and trapped. As a consequence, the pH inside the lysosomes increases and, in parallel, reduces the degradative capacity of different hydrolases.^30,31 This impairs RPE functionality (e.g. the degradation of the outer segments of photoreceptors) and could lead to an increased accumulation of autofluorescent bisretinoids. In cell culture experiments, CQ has shown stronger enhancement of lipofuscinogenesis compared to HCQ.³⁰ The clinical observation of pronounced BEM using CQ might further support this finding. Out of our two patients with a significant intake of CQ during recent years, one had developed BEM (CQ0002) and the other showed the highest QAF97 intensity of the whole cohort; however, also one of our patients taking HCQ had signs of BEM. Finally, CQ/HCQ or its respective metabolites themselves might exhibit fluorescent properties³² and result in increased QAF throughout the posterior pole. If the latter is true, the storage rather than the effect of CQ/HCQ could be visualized using QAF, although parallel effects cannot be ruled out. 

A question remains: is QAF a suitable candidate for CQ/HCQ screening? Typically, first changes in the CQ/HCQ related affected retina are visible at the parafoveal region and comprise thinning of retinal layers and changes in reflectivity of retinal bands in SD-OCT imaging.^33–35 Loss of AF is a sign of BEM,³⁶ the late stage of CQ/HCQ related maculopathy, and, therefore, common AF is not recommended as diagnostic tool for early retinal changes.³⁷ Because QAF enables quantitative analysis, changes in AF intensities (e.g. pixel by pixel comparison) in individuals might help to detect early changes in the long-term follow-up. Our results from baseline and follow-up might allow prediction of changes in QAF intensity growth at the parafovea (slowdown in QAF increase or QAF decrease before atrophy). FAF, however, might not be available ubiquitous and only offered in a few private practices or clinics. Furthermore, QAF is currently in an experimental stage and not used for regular screening which would further limit current FAF and QAF screening recommendation. 

The overall AF distribution across the posterior pole in our patients taking CQ/HCQ is similar to healthy subjects,^7,11,14,16 with lower AF centrally, increasing AF with increasing distance to the fovea (probably following rods distribution),³⁸ and an AF maximum temporal-superiorly.²⁶ Regional differences in the distribution of lipofuscin can be attributed to different characteristics of the ocular fundus (e.g. density of photoreceptors and the number of RPE cells),³⁹ related to the fovea and which then would result in radial changes, but currently does not explain difference in hemispheres. Our non-BEM patients show higher QAF in the superior hemisphere and higher QAF values have also been reported inferior and nasal and temporal to the fovea in recent studies, for both non-BEM patients and patients with BEM.^12,13 However, we were not able to detect any significant changes at the parafoveal region (i.e. extraordinarily high or low levels of QAF compared to the healthy cohort), except for the patients with BEM. Follow-up studies will clarify whether regional differences exist and if so their impact on CQ/HCQ maculopathy. 

In BEM, the atrophy tends to start inferior to the fovea at the parafovea.⁴⁰ In our patients with BEM, the atrophic area was more pronounced in those who were diagnosed with CQ related retinal toxicity more than 10 years ago, whereas atrophic areas were markedly smaller for those patients who stopped treatment more recently. In this 1-year follow-up, a slight increase of the atrophic area was found for patients with BEM even after cessation of drug intake, suggestive for disease progression, and in line with previous reports.^41–44 A possible reason could be the long half-life period of CQ/HCQ metabolites in retinal tissue.⁴⁵ 

SD-OCT imaging of the macula is a screening test recommended in patients with systemic CQ/HCQ treatment.³ In CQ/HCQ therapy, several studies reported structural changes including thinning of several retinal layers (RNFL,⁴⁶ photoreceptors,⁴⁷ outer nuclear layer)⁴⁸ and disruption and reduced reflectivity of the ellipsoid zone.⁴⁸ All of these SD-OCT biomarkers might be early signs of retinal toxicity; and, in future studies, artificial intelligence and machine learning based algorithms could further help to recognize these early alterations of photoreceptors and/or RPE.^49,50 In this 1-year follow-up study, no clinically significant changes in retinal thicknesses were observed in either ETDRS or QAF97 grid segments, or the horizontal meridian through the fovea. No correlation of retinal thickness to QAF intensities/change was found. Altered retinal layers (i.e. photoreceptor and RPE layer), might, however, impact QAF. A thinning of the photoreceptor layer or loss of the outer segments could result in a lesser amount of light-blocking photoreceptor pigment which in turn might lead to an increase of QAF. Altered RPE which often comes with changes in lipofuscin/melanolipofuscin load then also results in altered AF with hyper- and/or hypoautofluorescence.⁵¹ 

Our study has strengths and limitations. One limitation is the unequal gender distribution of the patient cohort probably due to a higher prevalence of rheumatologic disorders like SLE⁵² and RA⁵³ in women, and as a consequence thereof, leads to more frequent prescriptions of CQ/HCQ. We addressed this problem by using not only an age- but also a sex-matched control group. However, the control group were healthy subjects without rheumatologic disorders and it is unclear whether such diseases can impact the AF measurements per se, currently examined in our ongoing QAF studies. Further, even though we assessed overall retinal thickness and thickness of multiple retinal layers, we did not test for specific thickness changes in ellipsoid zone-RPE and outer nuclear layer-RPE layers. In addition, no follow-up data existed for the control group limiting the comparison to calculations based on assumptions from previous normative databases. Here, a simplified linear model was assumed. Furthermore, cumulative dosages are based on historic data reported by the patients, as well as concomitant diseases or disorders that also might impact QAF and BEM development. Because QAF is still in an experimental stage, currently limited to only a few retina centers, and image acquisition is highly reliant on the photographer´s experience and the patients' cooperation, QAF cannot be considered for regular CQ/HCQ screening at this stage and follow-up, longitudinal studies are warranted to further test this. 

Major strengths of this study include a large patient population, experienced QAF imaging operators, high-resolution retinal imaging with follow-up mode SD-OCT imaging, custom-developed software plugins that enabled accurate alignment of baseline and follow-up, and subsequently accurate QAF measurements. 

In summary, this study presents first longitudinal data of QAF change in patients with systemic CQ/HCQ treatment in a 1-year follow-up. It not only confirms our previous results of increased QAF levels in patients taking CQ/HCQ but also shows that this increase continues and further exceeds the age-related effect in healthy people. Future studies will focus on those patients with the steepest QAF increase and whether those are at higher risk for developing BEM. 

Acknowledgments

Disclosure: V. Radun, None; A. Berlin, None; I.-S. Tarau, None; N. Kleefeldt, None; C. Reichel, None; J. Hillenkamp, None; F.G. Holz, Acucela (C, F), Allergan (F), Apellis (C, F), Bayer (C, F), Boehringer-Ingelheim (C), Bioeq/Formycon (F, C), CenterVue (F), Ellex (F), Roche/Genentech (C, F), Geuder (C, F), Graybug (C), Gyroscope (C), Heidelberg Engineering (C, F), IvericBio (C, F), Kanghong (C, F), LinBioscience (C), NightStarX (F), Novartis (C, F), Optos (F), Oxurion (C), Pixium Vision (C, F), Stealth BioTherapeutics (C), and Zeiss (F, C); K.R. Sloan, None; M. Saßmannshausen, Heidelberg Engineering (Non-financial); Optos (Non-financial), Zeiss (Non-financial), and CenterVue (Non-financial); T. Ach, Roche (C), Novartis (C), Novartis (R), Bayer (C), and Apellis Pharmaceuticals (C)