Purpose
To report rapid onset of retinal toxicity in a series of patients followed on high-dose (1000 mg daily) hydroxychloroquine during an oncologic clinical trial studying hydroxychloroquine with erlotinib for non–small cell lung cancer.
Design
Retrospective observational case series.
Methods
Ophthalmic surveillance was performed on patients in a multicenter clinical trial testing high-dose (1000 mg daily) hydroxychloroquine for advanced non–small cell lung cancer. The US Food & Drug Administration–recommended screening protocol included only visual acuity testing, dilated fundus examination, Amsler grid testing, and color vision testing. In patients seen at Stanford, additional sensitive screening procedures were added at the discretion of the retinal physician: high-resolution spectral-domain optical coherence tomography (OCT), fundus autofluorescence (FAF) imaging, Humphrey visual field (HVF) testing, and multifocal electroretinography (mfERG).
Results
Out of the 7 patients having exposure of at least 6 months, 2 developed retinal toxicity (at 11 and 17 months of exposure). Damage was identified by OCT imaging, mfERG testing, and, in 1 case, visual field testing. Fundus autofluorescence imaging remained normal. Neither patient had symptomatic visual acuity loss.
Conclusions
These cases show that high doses of hydroxychloroquine can initiate the development of retinal toxicity within 1–2 years. Although synergy with erlotinib is theoretically possible, there are no prior reports of erlotinib-associated retinal toxicity despite over a decade of use in oncology. These results also suggest that sensitive retinal screening tests should be added to ongoing and future clinical trials involving high-dose hydroxychloroquine to improve safety monitoring and preservation of vision.
Hydroxychloroquine has been a mainstay of treatment for rheumatologic conditions such as systemic lupus erythematosus and rheumatoid arthritis, and there are extensive risk and toxicity data for the typical range of doses. The risk of hydroxychloroquine-mediated retinal toxicity is relatively low within the first 5–10 years of therapy when used at daily doses that do not exceed 5 mg/kg, typically 200–400 mg daily. Risk is clearly a balance between daily dose and duration of use, but there are little data available on the degree to which highly elevated doses can accelerate toxicity.
Loading doses of up to 1200 mg hydroxychloroquine for starting therapy of rheumatologic conditions have been described, but have typically been given only for brief durations of less than 6 weeks. Hydroxychloroquine has recently been proposed as an adjunct chemotherapeutic agent for a number of types of cancer, after in vivo and in vitro studies showed potential efficacy in various tumor types. However, relatively high blood concentrations are required to achieve tumor inhibition in preclinical studies, with dose escalation studies showing that an excess of 600 mg daily may be necessary. These trials might in theory provide knowledge about the risk of high-dose hydroxychloroquine; however, US Food & Drug Administration–mandated ocular screening is usually minimal by ophthalmologic standards, and patients are rarely followed for extended periods.
Stanford was a study site for a phase II open-label, randomized multisite clinical trial investigating the efficacy of high-dose hydroxychloroquine (1000 mg daily) in addition to erlotinib vs erlotinib alone for advanced lung cancer (Neal JW, et al. J Clin Oncol 2014;32:ASCO Abstract 8088). All patients in the hydroxychloroquine arm were referred for ongoing ophthalmic screening. The trial protocol only required visual acuity, color testing, Amsler grid testing, and fundus examination (which are no longer considered sufficient for screening according to current recommendations). We added sensitive imaging modalities and functional testing to the follow-up of a number of these patients. We present ophthalmologic findings on those patients who had specialized retinal testing and maintained high-dose hydroxychloroquine for at least 6 months.
Methods
The phase II trial evaluated EGFR-mutant advanced stage non–small cell lung cancer ( NCT00977470 ) and was initiated at the Massachusetts General Hospital; Stanford participated as a subsite with approval by the Institutional Review Board of both institutions. Erlotinib was administered at standard oncologic doses, and hydroxychloroquine was started in patients randomized to the combination arm at 1000 mg daily, with dose reductions to 800 mg, 600 mg, 400 mg, or complete discontinuation allowed for toxicity. The protocol required examination before treatment and every 3 months for those patients in the hydroxychloroquine arm, but only with dilated fundus examination, visual acuity, color vision testing, and Amsler grid testing. Our purpose was not to study every patient, but to follow as many as possible of those who continued on hydroxychloroquine beyond 6 months, adding (as clinically indicated) automated 10-2 fields (Humphrey Field Analyzer, Zeiss-Meditec, Dublin, CA, USA), spectral-domain optical coherence tomography (SD OCT) (Cirrus Model 4000; Zeiss-Meditec, Dublin, CA, USA), fundus autofluorescence (FAF) (Heidelberg Engineering, Heidelberg, Germany), and multifocal electroretinography (mfERG; Diagnosys LLC, Lowell, MA, USA). All specialized imaging and diagnostic studies were reviewed by 2 retinal specialists (L.B.L. and M.F.M.).
Results
The Stanford site enrolled a total of 15 patients to the hydroxychloroquine plus erlotinib arm, of which 7 had exposure beyond 6 months and underwent serial specialized diagnostic studies (mean exposure 16.5 months, range 7.5–25 months). Patient characteristics are shown in the Table . The last follow-up eye examination with retinal imaging studies ranged from 9 to 28 months after the start of therapy. Two patients developed definitive signs of toxicity without symptoms and are described below. Among those patients who did not develop signs of toxicity, 1 patient developed a retinal pigment epithelial detachment after discontinuing hydroxychloroquine, and an additional patient developed both a retinal pigment epithelial and serous neurosensory retinal detachment in 1 eye over the course of therapy, both of which resolved without recurrence. Neither event was believed to be associated with hydroxychloroquine.
| Patient a | Age (y) | Sex | Race | Weight (kg) | Dose/Kg | HCQ Exposure b (mo) | Duration of Follow-up (mo) | Cumulative Dose (g) | Ophthalmic Adverse Events |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 68 | M | W | 81 | 12 | 7.5 | 21.5 | 229 | Retinal pigment epithelial detachment after discontinuation of hydroxychloroquine |
| 2 | 56 | M | A | 76 | 13 | 10 | 10 | 298 | |
| 3 | 71 | M | A | 59 | 17/14 c | 19 | 19 | 539 | |
| 4 | 60 | M | A | 58 | 17 | 9 | 9 | 273 | |
| 5 | 65 | F | A | 78 | 13 | 25 | 28 | 755 | Retinal toxicity, trichomegaly |
| 6 | 77 | F | A | 58 | 17 | 18 | 23 | 532 | Neurosensory retinal detachment left eye |
| 7 | 66 | F | W | 50 | 20 | 13 | 15 | 403 | Retinal toxicity |
a No patient had renal or liver disease.
b Exposure at the last eye examination.
Case 1
A 66-year old white woman was newly diagnosed with advanced lung adenocarcinoma. The patient had no other medical diseases, and renal function remained normal throughout her therapy. The patient was 1.59 m tall (ideal body weight: 53 kg ) and weighed about 50 kg (daily hydroxychloroquine dose: 20 mg/kg). Baseline examination showed visual acuity of 20/20 OD and 20/25 OS. The anterior and posterior segment examination was unremarkable, except for a subtle yellow spot in the fovea of the left eye associated with mild vitreomacular traction. SD OCT ( Figure 1 ) was otherwise normal with good visualization of the ellipsoid zone (EZ) and interdigitation zone (IZ). FAF was also normal. Baseline Humphrey 10-2 visual fields were unremarkable for any loss of central macular sensitivity (not shown).
Follow-up examinations with SD OCT and FAF remained unchanged until the 11-month visit, although the SD OCT showed very subtle intensity loss in the parafoveal EZ. There was no change in thickness on the subfield cube map, and the EZ alteration was not felt to be definitive enough to discontinue potentially effective anticancer therapy.
The patient stopped hydroxychloroquine therapy at 13 months owing to symptoms of dry mouth. When next examined at 15 months ( Figure 1 ), she showed clear dropout of the parafoveal EZ (corresponding to the region of intensity loss at 11 months). There was also thinning of the overlying outer nuclear layer that averaged 30 μm in the parafoveal areas of the thickness map. FAF and ophthalmoscopic fundus examination were normal, but an mfERG showed widespread signal depression across the macula, with the greatest loss in the parafoveal region ( Figure 2 ). Follow-up visual field testing was performed elsewhere and was not available. The patient did not report any symptoms of visual impairment.
Case 2
A 65-year-old Asian woman with history of diabetes, hypertension, and hyperlipidemia was diagnosed with adenocarcinoma of the lung metastatic to the spine and brain. The patient was also 1.59 m tall (ideal body weight 53 kg) and weighed 78 kg (daily dose: 12.8 mg/kg). Baseline examination showed visual acuity of 20/30 OU and no fundus abnormalities aside from minimal nonproliferative diabetic retinopathy. Visual fields (10-2) were unremarkable, as was the SD OCT ( Figure 2 ). SD OCT, fundus autofluorescence imaging, and 10-2 visual fields were performed quarterly over the next 15 months without change.
After 17 months, some patchiness of the inferior parafoveal IZ was noted, more prominently in the left eye, and by 19 months there was zone of diminished EZ brightness. But there was no thinning on the SD OCT subfield maps, nor was there any diagnostic abnormality in FAF or visual field at this time or on subsequent follow-up 3 months later. These changes were felt to be borderline and the decision was made to continue therapy.
Evaluation at 25 months revealed clear loss of the inferior parafoveal EZ and IZ with visible thinning of the outer nuclear layer between about 2–6 degrees ( Figure 3 ). These were the same areas showing borderline change at 17 months. The thickness map still remained within normal limits for age and FAF was unremarkable. An mfERG showed diffusely depressed signals throughout the macula, with the greatest loss in the parafoveal region ( Figure 4 ). Erlotinib was continued, but hydroxychloroquine was stopped because of the observed retinal toxicity.
