Analysis of Inner and Outer Retinal Thickness in Patients Using Hydroxychloroquine Prior to Development of Retinopathy

Importance  Retinopathy is a known risk of long-term use of hydroxychloroquine sulfate. However, whether the inner as well as outer retina are involved before retinopathy develops and whether changes in the retina might signal impending toxic effects during screening remain unknown.

Objective  To determine the degree of inner and outer retinal involvement in short- and long-term use of hydroxychloroquine before the development of retinopathy.

Design, Setting, and Participants  This retrospective medical record review of spectral-domain optical coherence tomography (SD-OCT) findings was performed at an academic medical center. Thirty-two patients without retinopathy and with high-quality SD-OCT images were studied. Twenty-seven patients were age matched (49-65 years old) for comparison of retinal layers among patients who used the drug less than 5 years (n = 12) or longer than 15 years (n = 15) at the initial visit. Populations were also defined (without age limitation) for comparison of change during 25 to 52 months of follow-up among patients with initial use of less than 5 years (n = 7) or longer than 15 years (n = 8). Data were collected from 2010 to 2015.

Main Outcomes and Measures  Measurements of inner and outer retinal thickness in SD-OCT images using commercial software and a Stanford pixel-by-pixel segmentation software that also provided topographic maps of thickness dimensions and change.

Results  Thirty-two patients (5 men and 27 women) were included in the analysis. Measurements of inner retinal thickness between short- and long-term hydroxychloroquine users (n = 27) in different retinal regions, and during a median 39 months of follow-up (n = 15), showed no statistically significant differences or change. Similarly, no significant differences or changes were identified in outer retinal thickness except for the final visit of 1 patient who developed focal parafoveal thinning, a toxic effect of hydroxychloroquine use. Cirrus ganglion cell analysis measurements were inaccurate in the presence of outer retinal damage.

Conclusions and Relevance  The inner retina appears not to be involved in hydroxychloroquine-induced retinopathy to any clinically relevant degree within the limitations of our sample size. No clinically apparent warning of outer retinal damage was seen in the SD-OCT images of long-term hydroxychloroquine users until the actual appearance of focal retinopathy. Early detection of hydroxychloroquine-induced retinopathy is known to prevent visual acuity loss and serious progression after the therapy is stopped, and these data suggest that screening should seek distinct new areas of retinopathy (shown by topographic thickness maps) rather than long-term progressive thinning.

The predominant clinical finding in human retinopathy caused by hydroxychloroquine sulfate use is thinning of outer retina and eventual damage to the retinal pigment epithelium.¹ However, long-term hydroxychloroquine exposure in primates causes generalized neuronal damage that also injures the inner retina.² This finding has led to concern that the inner retina might be affected by long-term exposure to hydroxychloroquine in humans, and subsequent spectral-domain optical coherence tomography (SD-OCT) studies have suggested that inner retinal changes could be a marker for early toxic effects.^3-5 However, the magnitude of change in these studies was very small, and retinal segmentation was performed with the Cirrus SD-OCT (Carl Zeiss Meditec, Inc) ganglion cell analysis (CGCA) that can be unreliable in the presence of outer retinal disease.^6-8

We reported recently that inner retinal thickness was identical between mild and severe stages of hydroxychloroquine toxic effects, and it was stable after stopping the therapy.⁶ That previous study used a Stanford pixel-by-pixel segmentation paradigm that adjusts for outer retinal disease. However, important questions remain about whether any damage to the inner or outer retina might begin during hydroxychloroquine use before retinopathy develops and, more critically, whether the outer retina shows any gradual changes that would be warning signs of impending retinopathy. This understanding would influence the practice of clinical screening. In the present report, we investigate SD-OCT findings in populations of short- and long-term hydroxychloroquine users without toxic effects and whether the inner or outer retinal thicknesses change during 25 to 52 months of follow-up.

Box Section Ref ID

The study was based on a retrospective review of medical records and SD-OCT images. All patients were under the care of one of us (M.F.M.) for evaluation or screening. Patients were included who underwent evaluation from 2010 to 2015 with good-quality SD-OCT imaging studies (ie, recordings not distorted by movement, artifacts, or low OCT signal, so that an instrument software was able to produce thickness measurements) and who showed no signs of retinal abnormalities at the beginning of the study. Patients were divided into those seen initially less than 5 years of starting hydroxychloroquine therapy (short-term use) and those seen initially after more than 15 years of exposure (long-term use). A mean value was calculated from data from both eyes for each patient. One scan from 1 eye was excluded owing to low OCT quality, in which case only the data from the other eye were considered. This study was approved by the institutional review board of Stanford University School of Medicine, who waived the need for informed consent.

All SD-OCT data were acquired using a Cirrus instrument (Carl Zeiss Meditec, Inc) in the form of scan volumes with a topographic dimension of 6 × 6 mm and scan depth of 2 mm. The scanning pattern had a different resolution in each direction so that the individual voxels were 12 µm horizontally, 47 µm vertically, and 2 µm axially, for a total of 512, 128, and 1024 voxels in the respective directions. The raw data were imported into the OCT proprietary software (Cirrus Research Browser, version 6.2.0.3; Carl Zeiss Meditec, Inc) and analyzed using Cirrus full-thickness and CGCA measurements (Cirrus analysis) and the previously developed pixel-by-pixel analysis tool⁶ (Stanford analysis).

Full-thickness values were read from the Cirrus Early Treatment Diabetic Retinopathy Study (ETDRS) cube diagrams, and measurements of inner retinal ganglion cell layer thickness were read from the CGCA diagram. The full-thickness ETDRS cube values show the distance between the inner limiting membrane and the retinal pigment epithelium, with the mean measurements calculated within the following 8 nonfoveal topographic regions: parafovea and peripheral fovea in a superior, nasal, inferior, or temporal quadrant. The parafovea is defined as a ring 0.5 to 1.5 mm from the foveal center; the peripheral fovea, as a ring 1.5 to 3.0 mm from the foveal center.

The CGCA records the distance between the outer boundary of the retinal nerve fiber layer and the inner boundary of the inner nuclear layer. The mean values are calculated from 0.5 to 0.6 mm from the center to 2.0 to 2.4 mm from the center (long axis horizontally) within the superior, superonasal, inferonasal, inferior, inferotemporal, and superotemporal nonfoveal sectors. We reviewed the Cirrus segmentation cross-sectional image to confirm the segmentation lines for every case, and demarcation of the ganglion cell layers was correct, except for the final visit of 1 patient who developed signs of retinopathy (described in the Results section).

The collected SD-OCT volumes were exported to files describing the reflectivity measured at each voxel location with 8-bit precision (limit imposed by Carl Zeiss Meditec, Inc, proprietary software). These data were processed with the Stanford analysis software, which we have developed and validated.⁶ This software can accurately demarcate 10 intraretinal boundaries across the entire central macula, whether or not the outer retina has focal damage, and allows topographic and linear analysis of thickness and changes over time. This method has been verified previously,^6,9 but cross sections from every patient were also checked to confirm accurate layer separation.

As in previous work,⁶ thickness measurements were generated in a pixel-by-pixel basis between the inner limiting membrane and inner nuclear layer–outer plexiform layer junction for the inner retina and between the inner nuclear layer–outer plexiform layer junction and the outer border of the retinal pigment epithelium for the outer retina. We also generated topographic maps that include the parafovea and peripheral fovea, with data from the right and left eyes combined in proper nasal and temporal orientation.

When analyzing longitudinal data (multiple visits from the same patient), we also computed the rate of thickness change by determining the slope of a linear fit over time of the recorded thickness values at each pixel location. To reduce variability in these progression maps owing to noise in the scans and possible misalignments in the longitudinal images, the thickness maps were aligned with the center of the image corresponding to the center fovea pit location and passed through a median filter with a circular kernel of a 0.115-mm radius, in a manner similar to what we have reported previously.⁶

A total of 32 patients were studied (5 men and 27 women). All patients in this report were judged to be free of hydroxychloroquine toxic effects at their initial visit based on results of at least 2 screening tests, including 10-2 visual fields, SD-OCT, fundus autofluorescence, and multifocal electroretinography (mfERG). We stratified the patients into 2 groups with short-term (<5 years) and long-term (>15 years) hydroxychloroquine exposure at the initial visit.

Because retinal thickness may change with age,^10,11 we restricted the comparison of short- vs long-term users to those aged 49 to 65 years. Table 1 shows the patient characteristics of these populations, including the daily hydroxychloroquine dosage in milligrams per kilogram of real weight (a more reliable predictor of risk than ideal weight).¹² The short- (n = 12; mean [SD] age, 55.5 [5.5] years) and long-term (n = 15; mean [SD] age, 56.9 [4.9] years) populations were similar in all clinical characteristics other than duration of exposure (P > .19, paired t test).

Table 2 shows the characteristics of short-term (n = 7) and long-term (n = 8) hydroxychloroquine users who were followed up for at least 24 (median, 39; maximum, 52) months after their initial visit. These groups were not age matched because progression of retinopathy is independent of age, and only partial overlap with the Table 1 data occurred. The only significant difference between these groups was in the mean initial daily hydroxychloroquine sulfate dosage per kilogram, which was 5.6 mg for the short-term users and 1.2 mg less for the long-term users (P = .04). The lower average dose in the long-term group came mostly from 3 individuals and may have contributed to their treatment longevity, but the long-term users were still at much greater risk to show toxic effect than patients in the short-term group.¹² The difference in doses was not statistically significant during the follow-up period because some high initial hydroxychloroquine doses were lowered (Table 2).

Quiz Ref IDFigure 1 compares the inner and outer retinal thickness between the age-matched short- (<5 years) and long-term (>15 years) exposure groups. Measurements were obtained at the initial visit. Figure 1A shows the Cirrus analysis data for full retinal thickness in each parafoveal and peripheral sector as well as CGCA thickness in each sector. Mean differences between exposure groups were less than 5 μm (within scan reproducibility error) for full-thickness or CGCA sectors, and none of the differences were statistically significant (P > .26, paired t test). Thinning was not localized to the inferior and temporal parafoveal regions, which are typical sites of initial retinopathy.^1,6 The CGCA measurements were slightly lower in all sectors of the group with use longer than 15 years, but the SDs were also greater.

Figure 1B shows Stanford analysis with data based on pixel-by-pixel retinal segmentation of the inner and outer retina. The topographic plots show the mean inner and outer retinal thickness for the short- and long-term users in gray scale and differences between the short- and long-term groups in color maps. The difference plots indicated no major thickness differences between short- and long-term users in the inner or outer retina. The scattergrams depict similar mean inner and outer retinal thickness values for the individual patients between the short- and long-term users, with no major outliers.

Quiz Ref IDBecause any 2 populations may have slightly different mean thickness values, we also investigated whether inner or outer retinal thickness changed over time in short- or long-term users. Figure 2A shows the Cirrus analysis measurements. The plots show the difference in thickness from baseline for every patient to give an overview of measurement variability, to show trends of change, and to identify outliers. The full-thickness ETDRS cube measurements (Figure 2A, left) show a fair amount of individual variability, but no consistent gain or loss in any region, for the short- or the long-term hydroxychloroquine users. One outlying temporal parafoveal point was found in the long-term exposure group at the final visit of patient 136, who showed evidence of retinopathy at this visit. The CGCA measurements (Figure 2A, right) showed extremely stable values over time in all regions, except for 1 outlying point at the final visit of patient 136. However, this value is actually an error resulting from incorrect retinal segmentation by the CGCA software (below).

Figure 2B shows the rates of change over time (micrometers per year) between baseline and the final visit for inner and outer retinal thickness using the Stanford analysis. The color topographic plots show no consistent change among short- or long-term hydroxychloroquine users. The small focal lesion in patient 136 was not large enough to alter the mean rate of change substantially from that of all patients combined. The scattergrams (Figure 2B, right) display the annual rates of thickness change in individual patients to show consistency and outliers. The rate of inner retinal thickness change was essentially minimal during the follow-up interval for short-and long-term users, including patient 136. Outer retinal thickness change was also minimal, save for 1 outlier among short-term users with increased thickness and 1 negative outlying point among long-term users for patient 136 (see below).

Patient 136 in our series developed signs of hydroxychloroquine retinopathy during the follow-up period, and her findings over time are shown in Figure 3A. This white woman in her late 50s at the initial visit had mixed connective tissue disease and used a daily dose of hydroxychloroquine sulfate, 5.45 mg/kg, for 23 years. She had no renal disease and did not use tamoxifen citrate. Her SD-OCT scans and 10-2 fields were normal at baseline and 1 and 2 years later. The mfERG was normal at baseline and not repeated. At the examination 4 years after baseline (she missed the 3-year visit), she had no visual complaints but her tests showed evidence of early retinopathy (Figure 3). Her 10-2 fields revealed bilateral nasal parafoveal scotomas with some superior extension, and a new mfERG showed relative signal reduction in a ring 5° to 6° from the foveal center. The SD-OCT cross sections and ETDRS cube thicknesses were still normal in the right eye but showed a distinct focal region of temporal thinning in the left eye (Figure 3) with loss of outer segment marker lines. Stanford analysis topographic maps of retinal thickness and thickness change showed no abnormalities or warning signs until the appearance at 4 years of focal outer retinal thinning.

Figure 3B compared Cirrus and Stanford OCT segmentation of the inner vs outer retina at this patient’s final visit. The Cirrus program incorrectly localized the outer edge of the ganglion cell layer because the segmentation line failed to dip into the region of photoreceptor loss. This failure led to the false reporting of inner retinal thinning, which was shown in Figure 2A. The Stanford analysis followed the contour of outer retinal damage and showed no thinning in the inner retina.

Our data show 2 major findings based on SD-OCT data. First, the inner retina was not affected by hydroxychloroquine exposure to any clinically relevant degree. Second, we could not identify any chronic or progressive thinning of the outer retina during hydroxychloroquine use before toxic effects. When retinopathy was identified in 1 patient, it appeared in areas with normal thickness 2 years earlier.

Inner retinal thickness was essentially identical between our short- and long-term users of hydroxychloroquine. The small decrease in users for longer than 15 years with Cirrus data could result from greater variability or represent a small population difference. We also could not identify inner retinal thinning during 25 to 52 months of follow-up of short- or long-term users, including the patient who developed retinopathy. Cirrus data from other studies of the inner retina^3-5,13 may have been influenced by errors in CGCA measurements when the outer retina was abnormal,^6-8 as shown in Figure 3B. One report¹³ found major inner retinal loss in select cases with severe toxic effects, but the investigators also used Cirrus analysis. Our own prior data using the Stanford analysis in patients with severe toxic effects⁶ showed no loss of inner retinal thickness. We checked the segmentation lines for every eye in this study, and the CGCA appeared to be highly reliable in eyes without outer retinal damage. However, CGCA should not be used for screening hydroxychloroquine retinopathy or evaluating other disorders that affect the outer retina.

None of our patients showed measurable outer retinal thinning (or any focal warning signs of toxic effects in the parafovea) while taking hydroxychloroquine before the appearance of toxic effects, despite the fact that the groups with long-term use had a mean duration of use that exceeded 20 years and an approximately 20% risk for toxic effects.¹² The findings in our patient who developed hydroxychloroquine retinopathy during follow-up are instructive. Her imaging studies showed no evidence that toxic effects were developing during the initial 2 years of follow-up, when her retinal thickness measurements and visual fields were normal. When she returned after a 2-year gap, she had partial ring scotomas in both eyes (evident in visual fields and mfERG), and the SD-OCT scan showed a focal zone of rather severe parafoveal thinning in 1 eye, presumably because some threshold of cellular decompensation was reached. The fields and mfERG showed more widespread and bilateral functional loss than was evident in the SD-OCT imaging. This disparity has been observed before,^14,15 and it reinforces our belief that annual screening for hydroxychloroquine toxic effects should ideally include visual fields (which may not be reliable for all patients) and SD-OCT.

Quiz Ref IDOne limitation of this report is the modest size of our short- and long-term user groups. Although population differences between these groups cannot be entirely ruled out, no differences in retinal thickness were found during follow-up (that would be independent of population norms). The follow-up periods had a median duration of only about 3 years, but few data are available for longer follow-up because SD-OCT is a relatively recent innovation. Finally, we have only 1 patient who developed toxic effects during follow-up, so conclusions about the pattern of initial toxic effects cannot be generalized. However, the critical finding is that no warning signs of toxic effects occurred in any of the cases. We calculated a mean value for data from both eyes, so a very small change in only 1 eye might have been obscured in this presentation, but this process avoided the issue that different eyes would not be independent variables. We were looking for consistent change rather than minimal unilateral changes. Our data show how topographic SD-OCT thickness maps can be an important tool for hydroxychloroquine retinopathy screening because focal damage might be missed by some cross-sectional scans. Focal damage can affect mean thickness in an ETDRS sector, but topographic assessment shows the location and the extent of the damage. Although our single case shows how focal thinning on SD-OCT can be quite dramatic, parafoveal changes in visual fields, mfERG, and fundus autofluorescence are often more diffuse^1,15,16 and may begin outside the parafovea in Asian patients.¹⁷ This study was based on routine SD-OCT scans that do not extend beyond the vascular arcades (although none of our patients who had wider-angle testing with other modalities showed abnormality). We cannot say whether early toxic damage in Asian eyes, which is often outside the arcades, would behave in an identical manner.

Quiz Ref IDEarly detection of hydroxychloroquine retinopathy can prevent visual acuity loss and serious progression after the drug is stopped.¹⁸ The implication of this study relative to hydroxychloroquine retinopathy screening with SD-OCT is that health care professionals need to be alert at each visit for the development of new functional or anatomic loss rather than seek measures of gradual or continual loss in the photoreceptor layers. Nonetheless, the development of confirmable signs of toxic effects seems to require 1 to 2 years, which supports recommendations for annual routine screening and retesting borderline abnormalities before stopping treatment with a useful medication.^12,19 Patient 136 was diagnosed as having retinopathy at a relatively early stage before any retinal pigment epithelium damage and thus is not likely to be at risk for foveal damage or a loss of visual acuity.¹⁸

Our data suggest that the inner retina is not involved in hydroxychloroquine retinopathy to any clinically relevant degree, within the limitations of our sample size. No clinically apparent warning of outer retinal damage in the SD-OCT scans of long-term hydroxychloroquine users was seen until the actual appearance of focal retinopathy. These data may guide ophthalmologists to look for distinct new areas of retinopathy rather than long-term progressive thinning, and they demonstrate the value of topographic thickness maps for this purpose.

Corresponding Author: Michael F. Marmor, MD, Byers Eye Institute at Stanford, Department of Ophthalmology, Stanford University School of Medicine, 2452 Watson Ct, Palo Alto, CA 94303 (marmor@stanford.edu).

Submitted for Publication: September 18, 2015; final revision received January 17, 2016; accepted January 20, 2016.

Published Online: March 17, 2016. doi:10.1001/jamaophthalmol.2016.0155.

Author Contributions: Drs Marmor and de Sisternes had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: All authors.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: de Sisternes, Rubin, Marmor.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: de Sisternes, Hu.

Obtained funding: Rubin, Marmor.

Administrative, technical or material support: Rubin, Marmor.

Study supervision: de Sisternes, Rubin, Marmor.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This work was supported by a Spectrum–Stanford Predictives and Diagnostics Accelerator (SPADA) innovation grant from Stanford University and a research award from Retina Research Foundation and the Retina Society (Dr Marmor). Spectrum-SPADA is part of the Clinical and Translational Science Award program, funded by grant UL1 TR001085 from the National Center for Advancing Translational Sciences at the National Institutes of Health.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Previous Presentation: This work was presented in part at the 48th Annual Meeting of the Retina Society; October 10, 2015; Paris, France.