Fundus Autofluorescence and mfERG for Early Detection of Retinal Alterations in Patients Using Chloroquine/Hydroxychloroquine

Abstract

purpose. To evaluate and compare the value of fundus autofluorescence (FAF) imaging and multifocal electroretinography (mfERG) in early detection of retinal alterations in patients using chloroquine/hydroxychloroquine (CQ/HCQ).

methods. FAF imaging was performed in a consecutive series of 25 patients with long-term CQ or HCQ treatment (duration, >1 year), with or without visual disturbances. In addition, mfERG was performed in accordance with ISCEV (International Society for Clinical Electrophysiology of Vision) guidelines in 23/25 patients.

results. In 10/25 patients alterations of FAF were observed. Mild changes were limited to a pericentral ring of increased FAF. More advanced stages presented as pericentral mottled loss of FAF with increased FAF in the adjacent retina and later on a complete loss of pericentral FAF. In one case, a pericentral ring was observed when ophthalmoscopy and fluorescein angiography were normal. Marked progression of FAF abnormalities was observed during a 3-year follow-up in two of three patients. With the mfERG, pericentral, central, or generalized amplitude reductions were detected in all patients with FAF abnormalities and in an additional four patients with normal FAF.

conclusions. FAF imaging can be reliably used to detect early retinal pigment epithelial alterations in CQ/HCQ retinopathy. Ophthalmoscopy and fluorescein angiography appear to be less sensitive. With the mfERG, more retinal abnormalities were detected compared with FAF imaging.

Chloroquine (CQ) and hydroxychloroquine (HCQ) are widely used for treatment of autoimmune disorders in many countries. After the introduction of the drugs, irreversible retinal toxicity has been described ^{1 2} ; however, the reduction of the daily dose and a shift from the use of CQ to the more expensive but less toxic HCQ has reduced the frequency of this severe side effect. Recommendations have been made for prophylactic screening, although there is still discussion regarding the frequency and the examination methods. ^{3 4 5 6 7 8 9} In contrast, patients with CQ/HCQ retinal toxicity are detected regularly in retinal service centers. ^{3 4 10 11 12 13 14} Some of these patients have taken daily doses higher than recommended, some have a long duration of treatment; however, in some patients the daily intake as well as the cumulative dose is within recommended safety margins. 

Recently, several studies have been conducted to examine treated patients by multifocal electroretinogram (mfERG). ^{12 15 16 17 18 19} The findings in brief were that the mfERG can reliably detect retinal functional loss associated with CQ/HCQ retinopathy. In some patients the mfERG showed reduced response amplitudes when other functional tests (e.g., visual acuity, color vision, central visual field) or morphologic examinations (e.g., ophthalmoscopy, fluorescein angiography) were normal. ¹⁷ In addition, follow-up studies demonstrated a decline of retinal function when using HCQ ¹⁹ and improvement of retinal function after discontinuation of HCQ treatment, ^{17 19} indicating that the mfERG is a very sensitive test for detection of early retinal abnormalities under CQ/HCQ treatment. 

Histologic examinations and animal experiments have shown that the degenerative process begins in the ganglion cells; however, the photoreceptors are most severely affected. ^{20 21 22} The major pathogenic effect of CQ is the induction of lysosomal dysfunction in photoreceptors and retinal pigment epithelium (RPE) cells. ²³ Fundus autofluorescence (FAF) imaging is a recent method that has been shown to detect early RPE alterations in retinal disorders. ^{24 25 26 27 28} The present study was undertaken to evaluate the possible use of FAF in CQ/HCQ intoxication and to compare mfERG and FAF findings. 

Methods

Patients were seen either at the Department of Ophthalmology at the Charité, Campus Benjamin Franklin, Berlin, Germany (n = 17), or the AugenZentrum Siegburg, Germany (n = 8), between May 2000 and December 2005. Clinical examinations were conducted after explanation of the procedures and receipt of informed consent. The research adhered to the tenets of the Declaration of Helsinki, and Investigational Review Board approval was obtained. 

Included were consecutive patients with long duration (minimum, 1 year) of regular CQ/HCQ intake, with or without visual disturbances suggestive of retinal dysfunction. Patients with signs of other retinal diseases (e.g., age-related macular degeneration or diabetic retinopathy) were excluded. 

All 25 patients underwent a complete eye examination, including best-corrected visual acuity, slit lamp, and ophthalmoscopy. Fluorescein angiography was performed in seven patients. The in vivo measurement of FAF was performed with a confocal scanning laser ophthalmoscope (Heidelberg Retina Angiograph 1 or 2; Heidelberg Engineering, Heidelberg, Germany), as described previously. ²⁸ Argon laser light (488 nm) was used to excite RPE autofluorescence. A wide band-pass filter with a cutoff at 500 nm was inserted in front of the detector. A 30° field-of-view mode was used. The image resolution was 512 × 512 (HRA1)/768 × 768 (HRA2) pixels. The maximum illumination of a 10° × 10° field-of-view was approximately 2 mW/cm². Six images per second were recorded, and approximately eight single images were averaged depending on the fixation of the patient. 

Recording of the mfERG was performed according to the International Society for Clinical Electrophysiology of Vision (ISCEV) guideline. ²⁹ The recording protocols have been described in detail elsewhere. ²⁸ mfERGs were recorded and analyzed with the VERIS system (EDI, San Mateo, CA) or the RetiScan system (Roland Consult, Brandenburg, Germany). Recording was performed unilaterally with maximum dilated pupils using a contact lens electrode (Jet; LKC Technologies, Gaithersburg, MD). Refractive errors were corrected. For stimulation, a black-and-white pattern of 61 hexagons was presented on a monitor (200 cd/m² for white, 99.3% contrast). The duration of data acquisition was 4 to 5 minutes, divided into eight sessions. Data analysis (first order kernel) was performed with the software of the respective system. The response elicited by the central hexagon (ring 1) and summated responses elicited by concentric rings of hexagons surrounding the center (rings 2–5) were evaluated. Based on manually controlled cursor placement, amplitudes and implicit times were determined for the first positive component (P1) of each trace. Amplitudes were expressed relative to their respective area (nV/deg²). The normal ranges for these amplitudes and implicit times were defined by calculation of the median values and the 95% confidence intervals in one eye of 50 age-similar probands with each system. Because of variations in the recording techniques between both recording systems a direct comparison of response parameters is not feasible, because in normal control subjects amplitudes are slightly higher and implicit times are longer with the RetiScan system. ³⁰ Therefore, the percentage of P1 amplitude loss and implicit time delay in comparison to the respective normal values for each system was calculated. mfERG stimuli location and anatomic areas correspond roughly as follows: ring 1 to the fovea, ring 2 to the parafovea, ring 3 to the perifovea, ring 4 to the near periphery, and ring 5 to the central part of the middle periphery. The area of the rings was comparable between both systems. In an unpublished series of 15 patients with different macular disorders, we found no difference in the distribution of retinal abnormalities when recording with both systems. 

Results

An overview of all 25 patients including details of medication, clinical findings and a summary of FAF imaging and mfERG results is displayed in Table 1 . All except patient 5188 were women, and their ages ranged between 29 and 73 years. The treatment was CQ in 19 patients and HCQ in 6 patients, with a duration ranging between 1.3 and 20 years. As the treatment schedule varied between patients and application of drug was partly interrupted in some, the duration of treatment as well as the calculated total intake of CQ/HCQ are displayed. The daily dose of CQ was 250 mg except for patient 1843 who used 500 mg/d by her own decision for 5.3 years. The daily dose of HCQ was 400 mg except for the initial stage of treatment in which 600 mg was used in some patients. Visual acuity ranged between 0.3 and 1.0, except for one eye with a severe visual loss due to traumatic optic atrophy. 

Fundus Autofluorescence

FAF imaging was performed in both eyes of all 25 patients. Results of FAF were similar in stage on both eyes of each patient with slight variations of the total area affected. In 15 patients, FAF imaging was normal on both eyes (Fig. 1A) . 

In four patients, a pericentral ring of increased FAF without reduced FAF in other areas was observed (Fig. 1B) . The retina was ophthalmoscopically normal in the patient with the smallest ring (patient 1709). In the other three patients, mild pigment epithelial abnormalities were detected during ophthalmoscopy. In two more advanced cases, the pericentral area showed a mottled loss of FAF with increased FAF in the adjacent peripheral area (Fig. 1C) . Three more advanced cases presented with a pericentral ring with total loss of FAF with increased FAF in the adjacent peripheral area (Fig. 1D) . The patient with the most severe functional dysfunction showed mottled loss of FAF at the posterior pole with increased FAF in the adjacent peripheral area (Fig. 1E) . In these latter six patients with advanced abnormalities on FAF imaging, pigment epithelial abnormalities were observed on ophthalmoscopy; however, the area affected was usually larger in FAF imaging than expected from ophthalmoscopy. 

These proposed stages are indicated by comparison of FAF images of different patients. A progression through all stages in one patient was not observed, as CQ/HCQ treatment was discontinued after detection of any RPE alterations on FAF. 

Follow-up examinations were available in six patients between 1 and 3 years (Figs. 2 3) . In two patients normal FAF remained unchanged after 1 year under continuous treatment. In the other four patients treatment was discontinued. In one case with mild CQ retinopathy there were only limited changes within 3 years (Fig. 3) . In three patients with severe changes at the first examination progression of FAF abnormalities was observed after 1 to 3 years. Progression presented as additional loss of FAF in the previously affected area as well as in an increase of affected area (Fig. 2) . 

Fluorescein angiography was performed in seven patients. In three patients with normal FAF imaging, the angiography was unremarkable as well. In three patients with severe morphologic changes on FAF imaging, the fluorescein angiography was abnormal, delineating similar areas of pigment epithelial alterations as FAF imaging. In one patient with mild abnormalities on FAF, fluorescein angiography was normal (Fig. 3) . 

Comparison of mfERG and Autofluorescence

In 23 of 25 patients with FAF imaging a multifocal ERG could be recorded. Of these, 15 patients had normal FAF and 8 patients had FAF abnormalities. Alterations of the mfERG responses were detected in 13 of 23 patients, including all patients with FAF abnormalities. Amplitude reduction in ring 2 was the most frequent finding followed by amplitude reduction in rings 3, 4, and 1 (Table 2) . Delayed implicit times were less frequent. Changes in the mfERG could be differentiated into the following patterns, which were similar in both eyes (Figs. 4 5) : pericentral amplitude reduction, predominantly in rings 2 and 3 (six patients), central amplitude reduction in rings 1 to 3 (three patients), generalized amplitude reduction in rings 1 to 5 (three patients), and no measurable mfERG responses in at all (one patient). The degree of amplitude reduction in the mfERG corresponded to the severity morphologic alterations in FAF imaging; however, a direct association between mfERG and FAF patterns did not exist. In patient 1503, loss of function appeared to precede FAF alterations. Initially, the mfERG showed central amplitude reduction, whereas FAF imaging showed an incomplete ring of RPE loss. Three years later, the mfERG was nearly unchanged, but FAF alterations had markedly progressed (Figs. 2C 2D 2E 2F) . In 4 of 15 patients with normal FAF, a pericentral amplitude reduction in the mfERG was detected, three of them treated with CQ and one with HCQ. 

A full-field ERG (FF-ERG) was recorded in six patients and was normal in one of them. In one patient with FAF alterations, in whom mfERG recording was not possible due to a technical defect, the cone response in the FF-ERG was abnormal. In one patient with central amplitude reduction in the mfERG, the FF-ERG cone amplitudes were mildly reduced in one eye. Two patients with generalized amplitude reduction in the mfERG presented with either moderately or markedly reduced FF-ERG responses at all stimulus conditions. In the patient with no measurable mfERG responses, the full-field ERG was not detectable as well, indicating severe CQ retinopathy. 

Discussion

FAF imaging is a recent method that has the potential to detect abnormalities of the retinal pigment epithelium. ^{24 25 26 27 28} Results of FAF imaging correspond to retinal pigment epithelium lipofuscin characteristics. ²⁴ An increased FAF indicates accumulation of lipofuscin due to, for example, abnormal metabolism with increased phagocytosis of photoreceptor outer segments or an inherited or acquired defect of the phagocytotic processes of the retinal pigment epithelial cells. ^{31 32} Absence of FAF indicates photoreceptor or pigment epithelial cell loss. ²⁵  

In this small series, patients with long-term CQ/HCQ intake developed pigment epithelial abnormalities beginning with a fine pericentral ring of increased FAF. In more advanced cases, the ring appeared to broaden, before first mottled and later on general loss of pigment epithelium was indicated by an absence of FAF. Similar to histologic findings ^{20 21} and retinal functional loss in the mfERG, the abnormalities were limited in the beginning to the pericentral area of the posterior pole. In severe cases, however, the fovea might become involved as well with complete loss of RPE and more peripheral areas showing an increase in FAF. In two of three patients, marked progression was observed within 3 years after discontinuation of the drug. It is not possible to decide whether this progressive loss of pigment epithelium and loss of function is due to fatal damage to these cells at the time of discontinuation of the drug or whether other factors were responsible (e.g., predisposing genetic factors ³³ or storage of CQ/HCQ in the RPE). Individual susceptibility is indicated by the fact that retinal toxicity can be observed even in some cases with a relatively low cumulative dose (patient 1709), whereas a much longer treatment did not result in functional loss (patient 2702). Age may be an additional factor, at least in this small series of patients with normal FAF and mfERG had a mean age of 45.8 years, whereas those with any abnormality detected with one or both methods had a mean age of 59.9 years. 

In one patient, RPE abnormalities were detected by FAF imaging when ophthalmoscopy as well as fluorescein angiography were normal. It is of interest that, in this patient, the FAF pattern did not show marked progression after discontinuation of CQ in a 3-year follow-up compared with the more advanced cases. In this series, FAF imaging was more sensitive in detecting RPE abnormalities than was ophthalmoscopy. Although a comparison with fluorescein angiography was available in only a limited number of patients, FAF imaging appears to be at least as sensitive as fluorescein angiography, may allow earlier detection of RPE abnormalities in some cases, and is a helpful, fast, and noninvasive tool for monitoring of CQ/HCQ retinal toxicity. 

Recent studies have shown that the mfERG is a very sensitive method of detecting early loss of retinal function in CQ/HCQ retinopathy. ^{12 17 19} In a series of 19 patients, mfERG alterations were detectable earlier than changes of visual acuity, Amsler grid testing, or color vision tests. ¹⁷ We had a similar experience in a previous series of 20 patients without FAF imaging in whom the mfERG was more sensitive in detecting retinal abnormalities than was ophthalmoscopy, fluorescein angiography, color vision, and static perimetry (Missner S, Kellner U, manuscript submitted). Four different patterns of mfERG abnormalities have been defined previously ¹⁷ : pericentral, central, peripheral, and generalized functional loss. In this series, these patterns were confirmed except for the peripheral loss. As an extension of the generalized loss, one patient had an unmeasurable mfERG. 

In the present study, functional abnormalities in the mfERG were more frequent than were morphologic abnormalities in the FAF imaging. All patients with abnormal FAF patterns presented with reduced amplitudes in the mfERG when tested. To date, mfERG recording appears to be the most sensitive test for detection of CQ/HCQ-induced retinal functional loss. ^{17 19} Similar to FAF imaging, in the mfERG, the pericentral region is first affected with a progression to central and generalized functional loss and completely unmeasurable responses in the most severe case of CQ retinopathy. 

Although demonstrated in histology ^{20 21} it is still unclear, why the pericentral region is more vulnerable for CQ/HCQ toxic damage. Because the first abnormality detected is the loss of function in pericentral cones, the answer to this puzzle is most likely the difference between foveal and pericentral cones. The region of RPE damage as demonstrated by FAF imaging most likely mirrors the area of cone damage. Long-term follow-up in age-related macular degeneration presenting as geographic atrophy has demonstrated that areas of increased FAF indicate areas of subsequent loss of RPE. ²⁵ Although a follow-up in an individual patient is not feasible in toxic retinopathies, the comparison of severity indicates that, similar to geographic atrophy, increased FAF precedes loss of RPE. In severe cases, progressive loss of RPE may occur, even after discontinuation of the drug. 

The findings in CQ/HCQ retinopathy are in contrast to a similar ring of increased FAF observed in patients with retinitis pigmentosa, ²⁴ in which the ring of increased FAF demarcates the border between the central retina with preserved retinal function and the peripheral retina with loss of function. The retinal sensitivity over the area with increased FAF is better than in the adjacent peripheral area with more normal FAF. ²⁷ This indicates that, although different pathologic mechanisms may result in increased FAF, these morphologic findings do not allow conclusions with regard to photoreceptor function in the respective area. As most of our patients were older than 50 years, it is important to note that the FAF pattern in CQ/HCQ retinopathy differs from patterns described in early age-related macular disease. ³⁴  

When early changes of retinal abnormalities are detectable with the mfERG and FAF imaging, we recommend discontinuation of CQ/HCQ treatment. In patients with mfERG abnormalities only, follow-up examinations may be feasible if there is no better treatment option for the underlying autoimmune disease. FAF imaging will be especially worthwhile in patients in whom mfERG recording is difficult to obtain, as less cooperation is needed for FAF imaging. It remains to be discussed whether present recommendations for retinal toxicity screening of treated patients should be revised now that more powerful examination methods are at hand. 

 Supported in part by a grant from the Deutsche Forschungsgemeinschaft (Ke442/11-1,2) and an institutional grant (UK) from Charité Campus Benjamin Franklin (previously Freie Universität Berlin).

 Submitted for publication September 28, 2005; revised January 28, 2006; accepted May 24, 2006.

 Disclosure: U. Kellner, None; A.B. Renner, None; H. Tillack, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Patient Data

ID	Age (y)	CQ HCQ	Cum Dose (g)	Duration (y)	VA (OD OS)	FAF	FAF Follow-up	Angiography	mfERG Amplitudes	mfERG Follow-up
2596	48	CQ	180	2.5	1.0 1.0	OU normal	—	—	OU normal	1 y unchanged
1709	70	CQ	180	2	1.0 0.8	OS > OD pericentral ring of increased FAF	3 y only small changes	OU normal	OU central reduction, FF-ERG OS mildly reduced	1 y mild progression
2558	68	CQ	225	3	0.3 0.9	OU normal	—	—	OU pericentral reduction	—
422	58	CQ	230	4.5	0.8 0.8	OD > OS pericentral ring of increased FAF	—	—	OS > OD generalized reduction	6 months unchanged
1609	53	CQ	260	2.7	1.0 1.0	OU normal	1 y unchanged	OU normal	OU pericentral reduction	1 y unchanged
5153	62	CQ	275	5	0.8 0.8	OU normal	—	—	OU normal	—
5169	41	CQ	275	3	1.0 1.0	OU normal	—	—	OU normal	—
2572	48	CQ	370	12	1.0 1.0	OU normal	—	—	OU pericentral reduction	—
2713	73	CQ	410	4.5	0.4 0.1	OU pericentral loss of FAF and adjacent increased FAF	—	—	OU generalized reduction, FF-ERG moderately reduced	—
1690	64	CQ	460	5.0	0.7 0.05	OS > OD broad pericentral ring with increased FAF	3 y progression with pericentral loss of FAF	OS > OD pericentral RPE window defects	OU central reduction	1 y mild progression
1904	53	CQ	495	5.5	0.6 0.9	OU mottled pericentral loss of FAF and adjacent increased FAF	—	—	OU generalized reduction, FF-ERG markedly reduced	—
2561	63	CQ	637	7	0.9 0.9	OU pericentral ring of increased FAF	—	—	OU pericentral reduction	—
5188	53	CQ	709	7.8	1.0 1.0	OU normal	—	—	OU normal	—
1503	44	CQ	1000	11	0.7 0.8	OU mottled pericentral loss of FAF and adjacent increased FAF	3 y progression to pericentral ring of reduced FAF	OU pericentral RPE window defects	OU central reduction	3 y minimal changes
2688	58	CQ	1095	12	1.0 1.0	OU normal	—	—	OU pericentral reduction	—
2702	29	CQ	1278	14	1.0 1.0	OU normal	—	—	OU normal	—
1843	64	CQ	1500	11	0.5 0.3	OU mottled loss of FAF and adjacent increased FAF	1 y progression, further loss of FAF	—	OU not measurable FF-ERG not measurable	—
2203	62	CQ	1825	20	0.3 0.3	OU pericentral loss of FAF and adjacent increased FAF	—	—	OU not done	—
2201	53	CQ	1825	20	0.7 0.7	OU pericentral loss of FAF and adjacent increased FAF	—	OU pericentral RPE window defects	OU not done FF-ERG reduced	—
2200	49	HCQ	195	1.3	1.0 0.9	OU normal	—	—	OD normal OS not done	—
1735	41	HCQ	220	3	1.0 1.0	OU normal	—	OU normal	OU normal	—
1813	62	HCQ	310	2.8	0.9 1.0	OU normal	—	—	OU normal	1 y unchanged
1804	31	HCQ	350	8	1.0 1.0	OU normal	—	—	OU normal	—
1726	42	HCQ	660	9	0.6 0.6	OU normal	—	OU normal	OU normal	2 y unchanged
2658	67	HCQ	1390	19	1.0 1.0	OU normal	1 y unchanged		OU pericentral reduction	1 y progression of amplitude reduction

 Cum. Dose, cumulative dose, FF-ERG, full-field ERG, mfERG, multifocal ERG; VA, visual acuity.

Figure 1.

 

Variation of FAF findings in patients using CQ/HCQ. (A) Normal FAF (patient 2658), (B) pericentral ring of increased FAF (patient 2561), (C) pericentral mottled loss of FAF with increased FAF in the adjacent peripheral area (patient 1904), (D) pericentral ring with total loss of FAF with increased FAF in the adjacent peripheral area (patient 2201), (E) mottled loss of FAF at the posterior pole and increased FAF in the adjacent peripheral retina in the most severe case with unmeasurable mfERG and FF-ERG (patient 1843).

Figure 2.

 

Progression of FAF abnormalities over a 3-year-period after discontinuation of CQ intake. Progressive loss of FAF in patient 1690 between 64 years of age, visual acuity 0.7 (A), and 67 years of age, visual acuity 0.5 (B). Progressive loss of FAF and increase of affected area in patient 1503 between 44 years of age, visual acuity 0.8 (C), and 47 years of age, visual acuity 0.5 (D). In the same patient, the mfERG did not change markedly between 44 (E) and 47 (F) years of age.

Figure 3.

 

Comparison between pericentral ring of increased FAF (A) and normal fluorescein angiography (B). The FAF pattern remained nearly unchanged in a 3-year-interval (C) in patient 1709.

Table 2.

 

Multifocal ERG: P1 Amplitude and Implicit Time at First mfERG Recording

Multifocal ERG	Ring 1	Ring 2	Ring 3	Ring 4	Ring 5
P1 amplitude (n = 45 eyes)
Reduced in number of eyes (%)	14 (31.1)	26 (57.8)	23 (51.1)	15 (33.3)	12 (26.7)
Reduction to %*	52.7–0	60.3–0	57.7–0	55.9–0	52.7–0
Mean ± 1 SD (%)*	23.0 ± 15.8	29.6 ± 19.4	33.1 ± 19.7	32.5 ± 22.3	27.7 ± 20.7
P1 implicit time (n = 43 eyes), †
Increased in number of eyes (%)	9 (20.9)	10 (23.3)	9 (20.9)	9 (20.9)	8 (18.6)
Increased to %*	113.3–133.8	112.7–149.8	114.5–127.3	112.7–127.3	113.1–120.8
Mean ± 1 SD (%)*	122.2 ± 8.6	118.9 ± 4.7	119.5 ± 5.6	120.0 ± 6.5	117.3 ± 3.4

*  Values are given in percentage of the mean of the corresponding age-related norm for eyes with reduced amplitudes.

†  In two eyes without measurable amplitude no implicit time could be determined.

Figure 4.

 

Variation of mfERG (VERIS; EDI) findings in patients using CQ/HCQ. (A) Normal mfERG (patient 1804); (B) moderate pericentral amplitude loss, predominantly in rings 2 and 3 (patient 422); (C) central amplitude loss in rings 1 to 3 (patient 1709); and (D) generalized amplitude loss (patient 1904).

Figure 5.

 

Variation of mfERG (RetiScan; Heidelberg Engineering) findings in patients using CQ. (A) Normal mfERG (patient 2702); (B) moderate pericentral amplitude loss predominantly in rings 2 and 3 (patient 2558); (C) generalized amplitude loss predominantly in rings 1 to 3 (patient 2713).

The authors thank the technicians Hannelore Kraus, Elke Cropp, and Silke Weinitz for assistance with mfERG recording and FAF imaging.