Figure 1. Digitally composited fundal photographs from the same patient. A, Multifocal and confluent areas of outer retinal necrosis on initial evaluation. B, After 2 weeks of intravitreal ganciclovir injections, with subsequent scarring (final visual acuity, 6/9).
Figure 2. Digitally composited fundal photographs from the same patient. A, On initial evaluation, extensive posterior pole retinal scarring, arteriole nonperfusion, and well-delineated full-thickness retinal necrosis were evident. B, Two months later, no signs of active retinitis were seen (final visual acuity, hand movements).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Gore DM, Gore SK, Visser L. Progressive Outer Retinal Necrosis: Outcomes in the Intravitreal Era. Arch Ophthalmol. 2012;130(6):700–706. doi:10.1001/archophthalmol.2011.2622
Author Affiliations: Department of Ophthalmology, Nelson R Mandela School of Medicine, University of KwaZulu-Natal, Durban, South Africa.
Objective To describe the functional and anatomic outcomes of progressive outer retinal necrosis treated with intravitreal ganciclovir sodium injections.
Methods A retrospective, interventional case series of all patients fitting established clinical diagnostic criteria for progressive outer retinal necrosis was conducted at a single institution in South Africa. Eyes with salvageable vision were treated with repeated intravitreal ganciclovir injections until regression was achieved or the eye lost light perception. Pars plana vitrectomy was performed when retinal detachments failed to resolve spontaneously. The main outcome measures were visual acuity (VA) and response to intravitreal ganciclovir.
Results Thirty-nine patients (67 eyes), all of whom were HIV-positive (median CD4+ T-lymphocyte count, 30/μL), were included; 12 patients (31%) died during the study period. Twenty-eight of 36 patients (78%) had a recent history of cutaneous varicella zoster virus infection. At the initial evaluation, the mean VA was 6/120, with 12 eyes (18%) having already lost perception of light. Intravitreal ganciclovir injections were started immediately in all salvageable eyes (n = 50). Improvements in visual outcomes trended toward significance in eyes responding early (≤21 days), achieving a median final VA of 6/36 (P = .046). Retinal detachment occurred in 34 eyes (51%), predicating a significantly worse visual outcome (P < .001). Excluding eyes with no light perception at the start of the study period, median final VA was hand movements (range, 6/4 to no light perception); 9 eyes lost perception of light despite treatment.
Conclusions Progressive outer retinal necrosis remains a devastating condition, often with acute and profound loss of vision. Intravitreal ganciclovir may offer a more targeted approach and, compared with earlier reports using systemic therapy alone, may result in better visual outcomes.
To date, the largest and most comprehensive descriptive series of progressive outer retinal necrosis (PORN, also known as varicella zoster virus [VZV] retinitis or rapidly progressive herpetic retinal necrosis) remains that published by Engstrom and colleagues in 1994,1 crystallizing the devastating nature of the condition in which disease in two-thirds of eyes progressed to no light perception (NLP) within 4 weeks of onset. Occurring in profoundly immunocompromised patients, the cardinal features are multifocal lesions characterized by deep retinal opacification without granular borders, sometimes including areas of confluent opacification; lesion location in the peripheral retina, with or without macular involvement; extremely rapid progression; absence of vascular inflammation; and minimal or no intraocular inflammation.1,2 Although almost exclusively found in persons with AIDS, other mechanisms of immunosuppression have been described, principally iatrogenic (ie, chemotherapy).3
Early reports, including that by Engstrom et al,1 documented treatment only with systemic antiviral therapy. More recently, better outcomes appear to have been achieved with local therapy, including intravitreal antiviral injections and/or implants.4-11 These publications have for the most part been single case reports or small case series. We reviewed our large collection of patients with PORN treated with intravitreal antiviral injections on a background of the South African human immunodeficiency virus (HIV) epidemic.
This study received ethics and institutional board approval from the Biomedical Research Ethics Administration of the University of KwaZulu-Natal and Inkosi Albert Luthuli Central Hospital. The research adhered to the tenets of the Declaration of Helsinki.
Hospital records were reviewed for cases of PORN between 2004 and 2011. All diagnoses were made by a single consultant retinal surgeon (L.V.) with, to date, more than 15 years of experience managing retinitis related to HIV/AIDS with intravitreal therapy. Diagnoses were based on funduscopic findings in a confirmed HIV-positive patient: well-demarcated, multifocal, coalescing, and deep or full-thickness areas of predominantly posterior retinal necrosis. Vitritis and vasculitis were not considered exclusion criteria because of frequently advanced disease in our patients when first seen. If HIV was not already diagnosed, patients were counseled for testing and, where available, highly active antiretroviral therapy (HAART) was started. Information on a recent episode of VZV reactivation (shingles) or primary varicella (chicken pox) was sought. Laboratory confirmation by polymerase chain reaction was unfortunately not available because of financial constraints. Estimation of the extent of retinal involvement (ie, percentage) was not possible because of inadequate documentation; instead, we recorded the location according to a standard zone grading system: zone 1 encompasses an area extending 3000 μm from the center of the fovea or 1500 μm from the optic nerve, zone 2 extends from zone 1 to the vortex veins (equator), and zone 3 extends anterior to the equator.12 Fundus photographs were taken where possible.
Patients with clinical features consistent with acute retinal necrosis and cytomegaloviral retinitis (CMVR) were excluded. Regarding acute retinal necrosis, diagnostic criteria from the American Uveitis Society were used: peripheral, well-delineated foci of retinal necrosis with occlusive vasculopathy, circumferential spread and prominent vitritis, and anterior chamber inflammation.13 Diagnostic features of CMVR included the following subtypes: fulminant (posterior retinal hemorrhage with necrotic retina), granular (peripheral retinitis, with little or no edema, hemorrhage, or vasculitis), and frosted-branch angiitis (retinal perivasculitis).
Any eye with active retinitis that had potential for vision (ie, light perception [LP] or better with good projection) was treated with repeated intravitreal ganciclovir injections (2 mg in 0.08 mL of normal saline). Injections were administered twice weekly (Tuesdays and Fridays) for 2 weeks (induction phase) and then weekly, with activity and response to treatment monitored as standard at each visit. When no signs of active disease were visible, treatment was tapered to injections every 2 weeks (maintenance phase), continued until HAART-mediated immune reconstitution was confirmed by improved CD4+ T-lymphocyte counts above 100/μL. Before the availability of HAART, maintenance injections were continued weekly until the patient died. Patients with bilateral disease received simultaneous injections. Activity was defined as retinal opacification consistent with necrosis. Response to treatment was qualified as the first signs of localized disease control, specifically, retinal scarring. Because serial fundal photographs were not available for most patients, no analysis of the extent of scarring was made; instead, the interval before a response to treatment was observed. Pars plana vitrectomy was performed when retinal detachment (RD) failed to resolve spontaneously.
For purposes of statistical analysis, Snellen visual acuity (VA) values were converted to logMAR equivalents (−log decimal acuity) before being reconverted to Snellen acuity for presentation in this article. LogMAR equivalents for counting fingers, hand movements, LP, and NLP were estimated at 1.85, 2.30, 2.70, and 3.00, respectively.14 Data were analyzed with commercial software (Excel for Mac, 2011; Microsoft Corp).
In total, 67 eyes of 39 patients (38 black, 1 Indian) with a clinical diagnosis of PORN were included. Twelve patients (31%) died during the study period (Table 1) and 18 patients (46%) were lost to follow-up, including 6 who later returned. All patients were HIV positive, including 6 new diagnoses at their initial visit to the clinic. Of the 33 patients with previously diagnosed HIV, only 12 (36%) were already receiving HAART on their first visit to the clinic, including 2 who had been treated for less than 1 month.
The median CD4+ T-lymphocyte count was 30/μL. Of 36 patients for whom a history was documented, cutaneous VZV infection occurred in 28 (78%) (24 shingles and 4 primary varicella); 5 patients (14%) also had a concurrent disciform keratitis (4 ipsilaterally), presumed VZV in origin. Another patient was initially treated elsewhere for presumed VZV-meningitis with intravenous acyclovir before being referred to our department with a decrease in vision on day 3 of admission. Twenty-three of 28 patients (82%) had either active or previous pulmonary tuberculosis.
A history of deterioration and/or loss of vision was obtained at diagnosis in all patients, who had symptoms for a median of 3 weeks (range, 1-12 weeks) (Table 2). Eleven patients (28%) had unilateral disease; of the remaining bilaterally affected patients, all but one had signs in the fellow eye by the time of initial presentation. Median VA was 6/120 (range, 6/6 to NLP), including 11 patients (28%) already experiencing NLP (1 bilaterally). Fifty-nine (88%) of 67 eyes exhibited active disease; of the remaining 8 eyes, 6 had widespread retinal atrophy (the retinas in 5 of these had extensive detachments) and 2 eyes (with VAs of 6/5 and 6/6) had midperipheral retinal scarring without retinal detachment. Multifocal lesions were observed in 23 of 53 eyes (43%) with areas of confluence evident in 52 of 55 eyes (95%) (Figure 1). Among 65 eyes, less than or equal to 1+ anterior chamber and vitreous cells were observed in 52 (80%) and 37(57%), respectively, with 14 of 48 eyes (29%) exhibiting vasculitis. Immune status, when subdivided into patients with CD4+ T-lymphocyte counts of 0 to 24, 25 to 49, 50 to 74, and more than 75/μL, correlated positively with both the degree of vitritis (P = .03, χ2 test) and presence of vasculitis (P = .05, χ2 test).
In 50 active eyes with salvageable vision, intravitreal ganciclovir injections were started immediately (Table 3). Six treated eyes were excluded from further analysis because of disciform keratitis precluding useful retinal view (n = 2) and follow-up of 1 week or less (n = 4). A median total of 12 injections (range, 2-36) were given. Localized scarring (ie, response to treatment) (Figure 2) was observed at a mean interval of 21 days (range, 7-42). Improvements in visual outcomes trended toward significance in eyes responding early (≤21 days), achieving a median final VA of 6/36 (P = .046, paired t test). This compares with eyes responding late (>21 days), finishing with a median Snellen equivalent of approximately counting fingers at 1 m (2.08 logMAR). Excluding 9 other patients in whom direct macular necrosis or atrophy was evident at their first visit (ie, prejudicing therapy-mediated visual improvement), the benefits of an earlier response to treatment were more evident with a median final VA of 6/21 (P = .003, paired t test). No patient developed bacterial endophthalmitis during the study period (547 total injections).
Of 34 eyes (51%) with RD, 30 retinas (88%) detached within 7 weeks (median interval, 7 days; range, 0-14.2 months), including 14 that were detached when the patient was first evaluated (Table 4). The remaining 4 retinas detached at 26, 35, 42, and 61 weeks. Retinas in patients receiving HAART at their first visit were significantly less likely to detach, with an odds ratio of 0.07 (95% CI, 0.02-0.26; P < .001). One patient who developed a dense cataract following uncontrolled immune recovery uveitis underwent vitrectomy for a presumed RD detected on B-scan ultrasonography, although no RD was found during the operation. For the most part, eyes with RD not undergoing surgery were not considered salvageable; 1 patient declined surgery and another with VA of 6/18 was lost to follow-up after the previous week's visit. The retinas of 3 patients with small peripheral detachments overlying or adjacent to thinned atrophic (ie, inactive) retina spontaneously reattached: a reduction in fluid was documented after 2 weeks in 2 patients and by 3 weeks in the third patient. Resorption of fluid was complete by 6, 8, and 16 weeks, respectively. Seven patients underwent vitrectomy, with successful anatomic reattachment maintained in 6 at the most recent follow-up. The development of an RD was significantly associated with poorer visual outcome (P < .001, paired t test).
Excluding eyes with NLP (n = 12) at the start of the study period, the median final VA was counting fingers (range, 6/4 to NLP) (Table 5). Nine eyes lost LP despite treatment at a median of 12 days after diagnosis. Visual acuity at the patient's first visit significantly correlated with the final VA (P < .001, paired t test), with better outcomes evident in patients with good vision at the start of treatment.
The accumulation of significant numbers of patients with PORN is a reflection of the HIV/AIDS epidemic in southern Africa. As of 2008, South Africa had the largest burden of any nation, with an estimated prevalence of 10.6% (approximately 5.2 million people). Our hospital is located in the province of KwaZulu-Natal, which has the highest overall prevalence (15.8%). Peak prevalence occurs in women aged 25 to 29 years (32.7%) and men aged 30 to 34 years (25.8%).15 The availability of HAART in South Africa, long contentious by its absence, received governmental support in March 2007 after initiation of the National Strategic Plan for HIV/AIDS aiming to ensure that, by 2011, at least 80% of individuals eligible for antiretroviral treatment have access to it.
Although Engstrom and colleagues’1 original article set the benchmark for both diagnostic criteria and visual outcomes for PORN, the clinical entity was first described and so named in 1990 by Forster et al,2 and evidence for VZV involvement was reported by Margolis et al16 the following year. Historically, initial treatments involved 1 or more systemic antiviral agents, with combination therapy appearing to yield better outcomes: Spaide and colleagues17 demonstrated preservation of 20/100 VA or better in 6 patients treated with combination systemic antivirals; patients in 2 further studies18,19 maintained 20/80 VA or better in 10% and 22% of eyes, respectively.In contrast, 5 of 12 eyes (42%) treated with both intravenous and intravitreal antivirals achieved the same final VA of 20/80 or better,6 while a more recent series of 5 eyes receiving intravitreal therapy attained a final VA of 20/50 or better.7 Several other publications,4,8,10,11,20-24 all but 2 of them single case reports, have reported mixed outcomes, with limited follow-up. Our article, representing, to our knowledge, the largest published series of patients with PORN treated with intravitreal antiviral therapy, found an improvement in final VA, with only 9 (13%) eyes losing perception of light compared with 60% in the original series of Engstrom et al1 (excluding those with NLP on their first evaluation).
In addition to outcomes, there were differences in features at the initial examination in our series compared with those described by Engstrom et al. First, fewer of our patients exhibited absent or minimal inflammation when first seen, with 80% and 57% of patients having 1+ anterior chamber or less and vitreous cells, respectively, compared with 85% and 95% in the study by Engstrom et al. Second, at the start of the study period, 58% of our patients had retinal involvement in all 3 zones compared with just 28% among the patients of Engstrom et al. These differences may be linked: delays in seeking care, as in this series, would allow for greater spread of the disease throughout the retina in terms of confluence (95% vs 40%) and in progression from outer to inner retinal layers, both of which might be expected to elicit a greater vitreous response. Absence of a significant inflammatory response is generally a useful means of distinguishing presumed herpetic necrotizing retinopathies (ie, PORN vs acute retinal necrosis); however, our data suggest that, in patients with AIDS and PORN, vitritis can occur, with severity increasing in line with the CD4+ T-lymphocyte count.
Analysis of outcomes with and without RD revealed, unsurprisingly, a strongly significant trend to worse VA following detachment despite successful reattachment in 6 of 7 cases. Our results also suggest that the retinas in patients receiving HAART when first evaluated were significantly less likely to detach. However, we caution regarding interpretation of this, since we were unable to confirm in most patients the duration of treatment and whether their HIV was resistant to current therapy. Both of these factors may explain the low initial CD4+ T-lymphocyte counts, even in patients receiving HAART.
The timing of detachment appears to fall into 2 distinct groups, with the vast majority of retinas detaching within the first few weeks (ie, during the active necrotic phase) and a few detaching many months later (ie, during the inactive phase). This bimodal trend is also seen in patients with CMVR-related RD (unpublished data). The early group can be explained by full-thickness necrosis, leading to holes though both the inner and outer retina, as a prelude to rhegmatogenous RD. These holes are usually large enough to be identified on funduscopy but in a few cases are presumably too small. We propose that these latter examples are associated with one or more sieve-like holes that can occasionally self-seal with the ensuing scarring process, allowing a functioning retinal pigment epithelium to pump the detachment flat. This may explain 3 RDs in this series that spontaneously reattached.
An explanation for the later, inactive RD group may be found in work by Brar and colleagues,25 who described pathologic vitreoretinal interface changes on optical coherence tomography in patients with healed CMVR scars. They noted several abnormalities, including epiretinal membranes and vitreoretinal gliosis, concluding that the associated traction may explain the higher incidence of retinal elevation, breaks, and detachment in these eyes. We can assume, although not confirm, that such a mechanism applies to this series as well, since both CMVR and PORN involve viral-mediated full-thickness retinitis with resulting retinal atrophy. We postulate that the later (inactive) RD group may be a sequela of posterior vitreous detachment in areas incorporating these pathologic vitreoretinal adhesions.
The level of immune recovery at which antiviral therapy in PORN may be safely discontinued has not been established. More evidence exists in the literature26,27 related to the treatment of CMVR, in which most authors consider CD4+ T-lymphocyte counts greater than 100/μL as a safe level to stop maintenance treatment. Our group's experience of CMVR concurs with this, having no documented cases of relapse above this level when patients are adherent to HAART (unpublished data), and this same level was used as a cut-off point in this PORN series.
There are several limitations to this study, principally related to the inherent difficulties in developing-country health care: delays in seeking care (due to lack of transport, lack of money for transport, public sector strikes, and lack of awareness of visual symptoms) and a more general lack of hospital resources. It is in light of these facts that our patients sought care later than would be expected in the developed world, by which time 12 eyes (18%) had already lost LP. Consequently, several eyes had confluent areas of full-thickness retinal necrosis at the patient's first visit, having presumably progressed beyond the early stages of multifocal outer retinitis. Similar reasons, in addition to the 31% mortality rate, are likely to contribute to the large numbers of patients lost to follow-up. This resulted in considerable variations in treatment length (eg, range of number of ganciclovir injections, 2-36) and a high proportion of patients being excluded from the final analysis.
Additionally, the explicit rationing of hospital financial resources meant that we were unable to perform polymerase chain reaction analysis to confirm the etiologic agent. Varicella zoster virus has been confirmed as the causative agent in the overwhelming majority of PORN cases in the literature16,19,28-32; one notable exception identified both herpes simplex virus type 1 and cytomegalovirus but without VZV.33 In the absence of these diagnostic facilities, we sought, where available, a recent history of either VZV reactivation (shingles) or primary varicella (chicken pox). It is therefore possible that several patients in this series did not have PORN; however, we believe that this number is likely to be insignificant.
A criticism regarding treatment, specifically the lack of systemic antiviral therapy, can be made, particularly in light of potentially life-threatening complications of visceral dissemination in immunocompromised patients.34 Three articles19,31,35 describing VZV antigen–positive encephalitis, meningitis, or cerebral vasculitis closely associated with PORN exist; an additional study11 of patients with PORN highlighted 8 of 11 patients (73%) with concurrent central nervous system disease varying from unexplained confusion episodes to severe encephalopathic disease. Despite this, the prohibitive costs of oral valganciclovir (induction, R19 479 [US $2531]; maintenance: R13 914 [US $1808] per month) or inpatient intravenous ganciclovir and/or foscarnet therapy precluded systemic treatment in any of our patients.36 Instead, a single vial of ganciclovir (R358 [US $47]) was used for all intravitreal injections on any given day, mostly for the large numbers of patients with CMVR also in our clinics (approximately 6 new patients per week at the peak; unpublished data).37 Although we were able to confirm that 12 patients (31%) died during the study period, details of the causes of death were unavailable.
In summary, despite some patients retaining good vision, the functional outcomes overall in this series remain dire, and it is a reminder of the often catastrophic loss of vision that accompanies PORN. By far the most encouraging result from these data appears to be improved visual outcomes associated with early response to intravitreal ganciclovir injections. The best outcomes are likely to be seen in patients who begin intravitreal (and systemic) therapy within a few days of symptom onset, before macular involvement is apparent.
Correspondence: Daniel M. Gore, MRCOphth, Department of Ophthalmology, University of KwaZulu-Natal, Private Bag 7, Congella 4013, South Africa (email@example.com).
Submitted for Publication: September 20, 2011; final revision received November 29, 2011; accepted December 4, 2011.
Financial Disclosure: None reported.
Funding/Support: This study was supported in part by subsistence travel grants from the Ellison-Cliffe Travelling Fellowship of The Royal Society of Medicine (Dr D. M. Gore) and the Sir William Lister Travel Award and Dorey Bequest of the Royal College of Ophthalmologists (Drs D. M. Gore and S. K. Gore).