Outcomes of Combination Systemic and Intravitreal Antiviral Therapy for Acute Retinal Necrosis

Abstract

Objective:

Determine the efficacy of combination intravitreal and systemic antiviral therapy for the treatment of acute retinal necrosis (ARN) and risk factors impacting visual acuity (VA) and retinal detachment (RD) outcomes

Design:

Single center retrospective case-series

Participants:

Patients with an ARN diagnosis based on clinical features and polymerase chain reaction confirmation who were treated at a tertiary referral, university-based academic practice

Methods:

Patient records were reviewed for demographic information including age and sex. Snellen VA, disease findings including RD outcomes, optic nerve involvement and treatments were recorded. Incidence rates of major VA and RD outcomes were calculated based on the number of events and exposure time. Cox proportional hazards regression modeling and survival analyses were used to identify factors related to VA and RD outcomes over time.

Main Outcome Measures:

Logarithm of the minimal angle of resolution (logMAR) VA, ≥ 2-line VA gain, severe vision loss ≤ 20/200 (SVL), RD development, and fellow eye involvement

Results:

Twenty-three eyes of 21 patients (11 male, 10 female) were reviewed. Thirteen cases (62%) were due to herpes simplex virus (HSV) and 8 cases (38%) were due to varicella zoster virus (VZV). The event rate for ≥ 2-line VA gain was 0.49/eye-year (EY) (95% CI 0.26 – 0.86) while the rate of SVL was 0.61/EY (95% CI 0.34 – 1.02). RD development was observed at a rate of 0.59/EY (0.33–1.00). Thirteen of 23 eyes (57%) developed RD with a mean time of 120 days following ARN diagnosis. With each additional quadrant of retina involved, a greater risk of RD development over time was observed (Hazard ratio 2.21, 95% CI 1.12 – 4.35). Nine percent of eyes progressed with additional quadrantic involvement despite combination systemic and intravitreal antiviral therapy; however, none of the 19 patients presenting with unilateral ARN showed fellow eye involvement following initiation of therapy.

Conclusions:

Combination intravitreal and systemic antiviral therapy for ARN can be effective in improving VA and limiting retinitis progression. Each additional quadrant of retina involved was associated with a 2.2-fold greater risk of RD, which may impact monitoring, timing of intervention, and patient counseling.

Introduction

Acute retinal necrosis (ARN) is a viral-induced full thickness retinitis syndrome that rapidly leads to permanent vision loss and retinal detachment (RD) if not promptly treated. The first descriptions of the clinical findings of ARN were reported in the 1970s¹ with the definitive identification of a viral etiology by electron microscopy.² The American Uveitis Society (AUS) published clinical criteria for a diagnosis of ARN,³ although polymerase chain reaction (PCR)-based diagnostics have been increasingly utilized to identify the precise viral etiology. Varicella zoster virus (VZV), herpes simplex virus-1 and −2 (HSV-1/HSV-2) are the most common etiologies, although cytomegalovirus and Epstein Barr virus may also lead to ARN, albeit less frequently.^4,5 The treatment of ARN with intravenous acyclovir was initially described with the goal of halting disease progression in the affected eye and preventing fellow eye involvement.^6–7 Adjunctive interventions to antiviral medication include aspirin, corticosteroids, laser retinopexy, and vitrectomy with variable results in the improvement of outcomes.^8–10

More recently, emerging reports have described the benefits of intravitreal antiviral agents.^8–15 Meghpara et al followed 25 eyes of 22 patients and found that intravitreal injections improved vision in patients with moderate extent of retinitis, defined as 25–50% of the retina affected.¹¹ However, patients with greater than 50% of the retina affected did not appear to derive this benefit.¹¹ Wong et al showed a trend towards a reduction of RD among patients with VZV-ARN treated with combination systemic antiviral and intravitreal treatment when compared to systemic antiviral alone.¹² Yeh et al previously showed VA improvement and reduction in RD in patients who received combination systemic and intravitreal antiviral therapy compared to systemic therapy alone.¹³ More recently, Butler et al demonstrated that nearly 60% of eyes develop RD and eyes with retinitis involving ≥ 25% of the retina develop RD at nearly 12 times the rate of eyes with < 25% retinal involvement. Those eyes that develop RD have a less than 5% chance of achieving ≥ 20/40 visual acuity (VA).¹⁴ Despite these advances in therapy, patients with ARN remain at-risk for vision-threatening complications including RD, epiretinal membrane, and cystoid macular edema that may contribute to vision loss.

A recent report by the American Academy of Ophthalmology (AAO) in 2017 on the diagnosis and treatment of ARN recommended additional studies to evaluate the efficacy of combination therapy with systemic and intravitreal antivirals.⁸ In this study, we evaluated the treatment of ARN at a single tertiary referral academic center and evaluate factors associated with VA outcomes and RD in patients treated with combined systemic and intravitreal antiviral therapy.

Methods

A retrospective case series of patients with ARN was performed at the Emory Eye Center over a 5-year period from 2010–2015. This study was approved by the Institutional Review Board of Emory University including a waiver of informed consent for all retrospectively obtained and anonymized data set. This study adhered to the Declaration of Helsinki and the Health Insurance Portability and Accountability Act of 1996. Patients met the diagnostic criteria for ARN set forth by the AUS and underwent anterior chamber paracentesis or vitreous tap for PCR testing.

Demographic and clinical information assessed

Demographic variables collected included the patient’s sex and age at diagnosis. History of human immunodeficiency virus (HIV) or medications leading to systemic immunosuppression were also reviewed. The clinical features of patients identified included time from symptom onset to presentation, number of quadrants of retinitis based on fundus photography or clinical medical record documentation, Snellen VA at the initial visit, and follow-up visits (i.e. 1, 3, 6, 9, 12 months, and final follow-up visit), number and type of intravitreal injections, optic nerve involvement, presence or absence of RD during follow-up, and time to RD from initial presentation. To quantify the amount of retinal area affected, the number of quadrants with retinitis was measured at the time of diagnosis and again at the time of maximal disease activity. The quadrants of the retina were divided with the optic nerve being the dividing reference point and grouped by superotemporal, superonasal, inferotemporal and inferonasal.

Treatments administered

All patients received systemic antiviral medication in combination with intravitreal antiviral injections. Systemic antiviral medication consisted of valacyclovir 1 gram orally two to three times daily or acyclovir, initially dosed intravenously at 10 mg/kg/dose three times daily. If the patient received intravenous dosing of acyclovir initially, this was followed by oral acyclovir or valacyclovir. Patients were treated with valacyclovir 1 gram three times daily until disease inactivity was achieved. The dosage was subsequently reduced to 1 gram twice daily and tapered to a long-term prophylactic dosage of 1 gram daily at the discretion of the treating physician. Antiviral medication dosing was adjusted for patients with evidence of renal insufficiency. Intravitreal antiviral medications administered included either foscarnet 2.4 mg/0.1 cc or ganciclovir 2.0 mg/0.1 cc. Patients received injections twice weekly as induction treatment until the retinitis stabilized, at which time the frequency was decreased to once weekly until retinitis was considered inactive. In one case, the patient received a ganciclovir implant (Vitrasert, Bausch and Lomb) while the device was still commercially available. Laser retinopexy and oral corticosteroids were not routinely administered but were administered if deemed clinically indicated by the individual provider.

Snellen VA data was converted to logarithm of the minimal angle of resolution (logMAR) using the technique described by Holladay.¹⁶ For patients who were unable to read the Snellen VA chart, VA were assigned as follows: Counting fingers – logMAR VA 2.0; Hand motion – logMAR VA 3.0. Light perception and no light perception were noted but were excluded from VA calculation.

Descriptive and Inferential Statistical analysis

Microsoft Excel and IBM SPSS software were used for descriptive and inferential statistical analysis. Demographic information and treatment information were summarized as frequencies. Incidence rates of the major outcomes assessed were expressed as event rates per eye-year (EY), defined as 365.25 days of follow-up time of a single eye, to account for differential follow-up times. These outcomes include ≥ 2-line VA gain, severe vision loss to ≤ 20/200 (SVL), and RD.

To ascertain whether the outcome had been achieved, patients were required to maintain the level of VA gain or loss for 2 consecutive visits. Visual acuity at initial, 1-, 3-, 6-, 12-month and final time points were compared using a two-tailed paired T-test.

Univariate and multivariate linear regression were utilized to assess factors contributing to logMAR VA outcomes. Univariate and multivariate logistic regression were performed to assess factors associated with dichotomous variables including the major outcomes of the study, which included the following: 2-line or greater VA improvement during follow-up, SVL, and RD. To determine variables for multivariate linear or multivariate regression, a forward stepwise (conditional) selection algorithm was used. “Better eye” and “worse eye” analyses were also performed to assess the contribution of factors to these outcomes given the bilateral presentation of ARN in two patients and potential for within eye correlation. Variables were assessed for multivariate regression modeling if p < 0.10 but included in regression models only if p < 0.05. Cox proportional hazards regression modeling was utilized to identify factors associated with RD outcomes based on variables of interest. Kaplan-Meier survival analysis was performed to describe RD occurrence over time. Because viral subtypes (HSV compared to VZV) are a known disease consideration that may affect outcomes, their contribution to the major VA and RD outcome was independently assessed. An alpha of 0.05 was considered significant for all analyses.

Results

Baseline demographic, clinical features, and therapies administered

We identified 23 eyes of 21 patients with ARN based on AUS criteria. Baseline clinical and demographic characteristics are summarized in Table 1. There were 11 male (52%) and 10 female patients (48%). Two of the 21 patients (9.5%) were immunosuppressed (i.e. one patient status post recent chemotherapy and another patient who was HIV-positive). Anterior chamber paracentesis was performed in all patients, and the viral etiology was confirmed by PCR in 20 of 21 cases (95%).

Table 1.

Baseline demographic, clinical characteristics and treatment regimens of ARN patients.

Demographic or Clinical Feature	Number
Total patients (Eyes)	21 (23)
Male	11 (52%)
Female	10 (48%)
Median Age ± Standard Deviation	44 ± 24.6
Mean Initial VA logMAR ± Standard Deviation (Snellen VA)	1.37 ± 1.09 (20/470)
Mean Final VA logMAR ± Standard Deviation (Snellen VA)	0.85 ± 1.10 (20/140)
Mean Follow-Up Time in Months (Range)	44 (3–52)
Quadrants of Retinitis
Eyes with 1 Quadrant	5 (22%)
Eyes with 2 Quadrants	7 (30%)
Eyes with 3 Quadrants	3 (13%)
Eyes with 4 Quadrants	5 (22%)
Viral etiology by patient
Herpes simplex virus, type 1	7 (33%)
Herpes simplex virus, type 2	6 (29%)
Varicella zoster virus	8 (38%)
Patients receiving the following treatments
Intravenous Acyclovir	1 (4.8%)
Oral Acyclovir or Valacyclovir	21 (100%)
Intravitreal Foscarnet	21 (100%)
Median number of foscarnet injections per eye (Range)	6 (1 – 10)
Total number of foscarnet injection for all patients	126
Intravitreal Ganciclovir	2 (9.5%)
Median number of ganciclovir injections per eye (Range)	1 (0 – 1)
Total number of ganciclovir injections for all patients	5
Intravitreal Ganciclovir implant	1

All patients received systemic antiviral therapy including valacyclovir or acyclovir, while one patient was hospitalized for intravenous acyclovir given her VA was poorer than 20/200 in both eyes. All patients received intravitreal foscarnet and a median of 6 injections were administered per eye (range 1–10 injections). A total of 126 foscarnet injections were recorded during the study period. Three patients received intravitreal ganciclovir, most often in eyes with particularly severe disease that prompted combination therapy with foscarnet and ganciclovir.

Visual acuity outcomes: LogMAR VA improvement, 2-line VA gain and SVL impairment

The mean initial logMAR VA was 1.37 ± 1.09 (Snellen VA 20/470). The mean final logMAR VA was 0.85 ± 1.10 (20/140) with a mean follow-up time of 44 months (range 3–52 months). Significant VA improvement was observed between initial presentation and 1-month follow-up where logMAR VA improved to 1.13 ± 1.04 (Snellen VA 20/270, p=0.002). However, after the 1-month follow-up visit, no further improvement was observed at the 3-, 6-, 12-month or final follow-up visit (p > 0.05 for comparison of logMAR VA at these time points to baseline).

The overall incidence rate for a 2-line or greater VA improvement was 0.49 events/EY (95% CI 0.26 – 0.86, Table 2). Given these findings, we assessed factors contributing to final logMAR VA and a 2-line or greater gain in VA. Final logMAR VA was significantly associated with initial logMAR VA (p=0.022) and worsened significantly with RD occurrence (p=0.001) in univariate analyses (Table 3). However, in the multivariate analyses, only RD was significantly associated with a worse final logMAR VA. In addition, univariate logistic regression analysis showed that increased retinal area involved by quadrants affected (p=0.038) and RD occurrence (p=0.011) were associated with a significantly decreased odds of a 2-line or greater VA improvement. Multivariate logistic regression analyses showed that only RD was the predominant contributor to VA loss.

Table 2.

Incidence rate for visual acuity and retinal detachment outcomes

Outcome/Event	Event rate, eye-year (95% CI)
2-line or greater visual acuity improvement	0.49 (0.26 – 0.86)
Severe visual loss 20/200 or worse	0.61 (0.34 – 1.02)
Retinal detachment	0.59 (0.33 – 1.00)

Table 3.

Factors contributing to visual acuity and retinal detachment outcomes.

Factors contributing to final visual acuity outcome
Factor	Coefficient	SE	T statistic	P-value
Initial VA	0.69	0.27	2.54	0.022*
Age	0.015	0.013	1.20	0.251
Sex	0.18	0.055	0.33	0.748
Symptom duration (Days)	0.042	0.037	1.14	0.273
Retinal Detachment	1.63	0.37	4.41	0.001**
Retinal Area (Quadrants)	0.48	0.24		0.063
Optic nerve involvement	0.40	0.69	0.57	0.584
Virus type (HSV or VSV)	−0.74	0.54	−1.35	0.198
Factors contributing to 2-line or greater visual acuity improvement
Factor	Coefficient	SE	Odds Ratio (95% CI)	P-value
Initial VA	−0.59	0.46	0.557 (0.23–1.36)	0.20
Age	0.003	0.018	1.00 (0.97 – 1.03)	0.13
Sex	1.41	0.93	4.08 (0.66 – 25.38)	0.88
Symptom duration (Days)	−0.11	0.06	0.89 (0.79 – 1.01)	0.06
Retinal Detachment	−3.18	1.25	0.04 (0.004 – 0.485)	0.011**
Retinal Area (Quadrants)	−1.00	0.48	0.37 (0.14 – 0.95)	0.038*
Optic nerve involvement	−0.81	1.38	0.44 (0.02 – 6.7)	0.56
Virus type (HSV or VSV)	0.15	0.70	1.17 (0.20 – 6.8)	0.86
Factors contributing to severe visual loss 20/200 or worse
Factor	Coefficient	SE	Odds Ratio (95% CI)	P-value
Initial VA	1.58	0.68	4.8 (1.28 – 18.2)	0.02*
Age	0.07	0.032	1.08 (1.01 – 1.14)	0.02*
Sex	−0.69	0.88	0.50 (0.09 – 2.80)	0.43
Symptom duration (Days)	0.07	0.055	1.07 (0.96 – 1.19)	0.23
Retinal Detachment	3.09	1.10	22.0 (2.5 – 191)	0.005**
Retinal Area (Quadrants)	0.89	0.44	2.44 (1.0 – 5.7)	0.04*
Optic nerve involvement	1.39	1.38	4.0 (0.27–60.3)	0.32
Virus type (HSV or VZV)	−2.23	1.20	0.11 (0.01 – 1.1)	0.06
Factors associated with retinal detachment
Factor	Coefficient	SE	Odds Ratio (95% CI)	P-value
Initial VA	1.972 ± 0.841	0.84	7.2 (1.38 – 37.3)	0.019*,†
Age	0.039 ± 0.022	0.022	1.04 (1.00 – 1.09)	0.08
Sex	−1.099 ± 0.890	0.89	0.33 (0.058 – 1.9)	0.22
Symptom duration (Days)	0.102 ± 0.061	0.061	1.11 (0.98 – 1.25)	0.096
Retinal Area (Quadrants)	1.716 ± 0.663	0.66	5.6 (1.5 – 20.4)	0.01**,†
Optic nerve involvement	0.811 ± 1.384	1.38	2.3 (0.15 – 33.9)	0.56
Virus type (HSV or VZV)	−1.10	0.98	0.33 (0.05 – 2.23)	0.26

^*P < 0.05 in Univariate regression models;

^**P < 0.05 in multivariate regression model

^†P = 0.07 for initial VA as a factor associated with retinal detachment; P = 0.036 in multivariate regression model for retinal area as a factor contributing to retinal detachment

Abbreviations - VA Visual acuity, HSV Herpes simplex virus, VZV Varicella zoster virus, SE Standard Error, CI Confidence Interval, RD Retinal detachment

Note – For data analysis and interpretation, for Sex (1=Female, 0=Male), Optic nerve involvement (1=Involved, 0=Not involved), Virus type (1=HSV, 0=VZV). No changes in statistical significance for multivariate analyses were observed for “better-eye” analyses. For “worse-eye” analyses, RD was not an independent predictor for 2-line visual acuity gain or severe visual acuity loss.

We also assessed factors contributing to SVL of 20/200 or poorer. The overall incidence of SVL in this cohort was 0.61 events/EY (95% CI 0.34 −1.02). Univariate logistic regression showed that worse initial VA (i.e. increased logMAR VA, p=0.02), age (p=0.02), RD development (p=0.005), and quadrantic retinal area involved (p=0.04) were associated with SVL. Factors not associated with SVL included sex, symptom duration prior to presentation, optic nerve involvement and virus etiology (p > 0.05 for all analyses). Multivariate logistic regression modeling showed that RD remained the primary driver of SVL (p<0.05).

Retinal detachment outcomes

The overall incidence rate of RD was 0.59 events/EY (95% CI 0.33–1.00). Of the 23 eyes with ARN, 13 (57%) developed a RD during the follow-up period. RD occurred at a mean time of 120 ± 75 days after diagnosis. Nine of thirteen patients (69%) presented with macula-off status at the time of RD. Univariate regression analysis showed that factors associated with development of RD included the number of quadrants of retina involved (p=0.01) and initial VA (0.019). Age and symptom duration were borderline statistically significant (p=0.08 for age and 0.10 for symptom duration), but were not statistically significant with multivariate logistic regression modeling. With multivariate regression analysis, the retinal area was the major factor contributing to RD (p=0.036), whereas initial VA was borderline significant (p=0.07).

To account for the effect of these variables over time, we performed a Cox proportional hazards model analysis to assess these potential contributing factors. Initial VA was a predictor of RD with a hazard ratio of 2.69 (95% CI 1.45 – 5.01, p=0.002) and retinal area also conferred greater RD risk with a hazard ratio of 2.21 (95% CI 1.12 – 4.35, p=0.023). These data suggest that for each quadrant of retina involved, there was roughly a 2.2-fold increased rate of RD (i.e. 2.2x, 4.8x, and 10.6x increased hazard for 2, 3, and 4 quadrants involved, respectively). Survival analyses illustrate the timeline for RD development in Figure 1.

Figure 1.

Survival curve for retinal detachment (RD) outcome show that the majority of RDs developed before 6 months. The overall incidence rate of RD was 0.59 events/eye-year (95% CI 0.33–1.00). Initial visual acuity (VA) showed a hazard ratio of 2.69 (95% CI 1.45 – 5.01, p=0.002) and retinal quadrantic area showed a hazard ratio of 2.21 (95% CI 1.12 – 4.35, p=0.023).

Eleven of 13 eyes (85%) detached within 6 months of diagnosis. The two other RDs occurred at 7 months and 3 years after diagnosis in patients with 1 and 4 quadrants of retinitis. Only 2 of 23 eyes (8.7%) developed additional quadrantic involvement following the initiation of combination therapy, with a one quadrant-increase in each patient.

Varicella zoster virus vs. herpes simplex virus retinitis

Given reports of differential outcomes between VZV and HSV-ARN, we assessed the demographics and major outcomes by viral etiology. Patients with VZV-ARN were older at 63 ± 24.9 years-old compared to HSV-ARN at 33 years ± 21.2 years-old (p=0.046). Females comprised 25% of the VZV-ARN cohort and 62% of the HSV-ARN although these proportions did not differ significantly (p>0.05). The incidence rate for VA gain ≥ 2-lines for HSV, 0.57 events/EY (95% CI 0.25–1.12), was greater than for VZV at 0.40 events/EY (95% CI 0.12 – 0.96, p> 0.05, two-tailed Fisher’s exact test). The incidence rate for SVL for VZV was 2.15 events/EY (95% CI 0.94–4.3) and was significantly greater than the HSV incidence rate of 0.33 events/EY (95% 0.14 – 0.69, p=0.003 for comparison). The RD incidence rate was greater in the VZV cohort at 1.07 events/EY (95% CI 0.44 – 2.2) compared to HSV at 0.43 events/EY (95% CI 0.19 – 0.86, p>0.05). Survival curves illustrate the divergence of curves between HSV and VZV for 2-line VA gain, SVL and RD rates (Figures 2–4).

Figure 2.

Survival curve demonstrates severe vision loss to VA ≤ 20/200 at rates of 2.15 events/eye-years (EY, 95% CI 0.94–4.3) in Varicella zoster virus-acute retinal necrosis (VZV-ARN) and 0.33 events/EY (95% 0.14 – 0.69, p=0.003) in Herpes simplex virus-acute retinal necrosis (HSV-ARN). The incidence rates were defined by EY to account for variable follow-up.

Figure 4.

Survival curve for RD outcomes by virus type show rates of RD at 1.07 events/EY (95% CI 0.44 – 2.2) for VZV-ARN compared to HSV at 0.43 events/EY (95% CI 0.19 – 0.86). These differences were not statistically significant (p>0.05).

Fellow Eye Involvement

Two patients presented with bilateral disease, one each in the HSV and VZV groups. One patient who presented with bilateral HSV-ARN and central retinal artery occlusions following intravenous corticosteroid administered for presumptive bilateral optic neuritis had been described in detail previously.¹⁷ None of the other 19 patients who had unilateral retinitis progressed to bilateral disease after the initiation of combination therapy.

Discussion

The management of ARN has evolved over the last three decades with an increasing use of intravitreal antiviral therapy in its treatment paradigm. Despite aggressive treatment, a subset of patients may develop poor outcomes, particularly if the diagnosis of ARN is delayed from late presentation, often with widespread retinitis at disease presentation. Recent work has showed that combination systemic and intravitreal antiviral therapy for ARN may reduce the risk of SVL, reduce RD risk, and potentially improve VA.^11,13–14 This study was undertaken to evaluate our more recent experience in the treatment of ARN, as well as to identify risk factors that portend better or poorer VA and RD outcomes.

In our cohort of patients with ARN, patients demonstrated an improvement in VA at long-term follow-up, although the major improvement was observed at the 1-month time point. The development of RD was a critical factor associated with SVL and decreased likelihood of meaningful VA improvement, which is consistent with recent literature demonstrating that only 4% of patients who developed RD were able to achieve 20/40 or better VA.¹⁴ A majority of RDs occurred within 6 months of diagnosis suggesting the greatest risk for RD during this time.

The area of retinitis, classified by retinal quadrants involved, was directly related to RD development. Specifically, we observed a 2.2-fold increased rate of RD for each quadrant of retina involved at presentation. These findings are consistent with other reports of poor VA with greater extent of retinal involvement,^{11–14, 18–19} and emphasize the need for aggressive therapy to detect and address these risk factors for RD. It is also notable that following treatment with combination systemic and intravitreal antiviral therapy, less than 10% of patients developed additional quadrantic involvement, which has the potential to decrease the risk of RD.

Enumerating retinal quadrantic involvement involves a practical approach that is quantifiable with landmarks (i.e. optic nerve, fovea separating superior and inferior hemiretina), which can be otherwise difficult to visualize in cases of ARN with severe vitreous haze. Extent of retinitis by quadrant can thus be quantified and followed to identify increasing areas of retina, which are susceptible to necrotic breaks and RD in the setting of vitreous liquefaction and vitreoretinal traction. This methodology differs from percentage of retina involved or zonal involvement, which may be subjective, particularly if these parameters are defined by the investigators’ exam and not by standardized photography. In the AAO Ophthalmic Technology Assessment of interventions for ARN, the standardization of outcome measurements was highlighted as a key area of future research to understand best practices for ARN management.⁸

Interestingly, besides the number of quadrants of retina involved, poorer VA at presentation also appeared to increase the hazard rate for RD development. Whether this is due to more severe disease, greater vitreous inflammation, or other factors related to disease could not be assessed with the limitations of the dataset. Yet, detection of clinical factors associated with poorer VA at follow-up (i.e. posterior pole findings, greater vitreous inflammation) may be important for patient counseling purposes regarding final VA outcome, the need for early aggressive intervention, and potential need for surgical intervention given the increased risk of RD. The burden of viral infection has also been associated with poorer outcomes. Specifically, Calvo et al demonstrated that increased quantitative PCR measures of VZV and HSV were associated with more extensive retinitis, worse visual acuity, and development of RD. While our study did not specifically assess quantitative PCR, this metric is an important factor to consider for disease prognosis and follow-up in patients with ARN.²⁰

As several prior studies have observed worse VA outcomes in VZV-ARN compared to HSV-ARN,^12,18 we also assessed whether differences existed in patient demographics and incidence of major VA outcomes. Patients with VZV tended to be older than individuals with HSV-ARN. The rate of SVL was greater in VZV patients when compared to HSV. Whether this relationship is due to the virulence of the organism or other contributing factors is unknown, but reported risk factors for VZV-ARN include increasing age with immunosenescent changes and systemic conditions leading to immunocompromised status.²¹

Valacyclovir induction dosing for ARN has varied in the literature and ranges from 1 gram to 2 grams three times daily to achieve increasing vitreous concentrations of antiviral medication similar to intravenous dosing.¹³ Whether an increased dosage of valacyclovir could potentially decrease the number of intravitreal injections warrants further study. However, acute kidney injury associated with valacyclovir warrants laboratory monitoring, and adjustments in antiviral therapy are needed for patients with chronic renal insufficiency.

Limitations of the study included the retrospective, uncontrolled nature of the study, and sample size which potentially limit the variables collected for data analysis. Moreover, the few patients who received laser retinopexy and oral corticosteroid limited our ability to evaluate these variables and their contribution to VA and RD outcomes in multivariate analysis. Only two of the patients were immunosuppressed and consequently, we were unable to statistically assess the precise role of immunity in ARN outcomes. Lastly, the numbers of patients in the VZV- and HSV-ARN cohorts may have limited our ability to identify statistical differences in outcomes, although our analyses showed a trend towards increased severity in VZV-ARN. Consequently, the generalizability of our outcomes may be limited and additional larger studies are needed.

Nonetheless, our study, which evaluated consecutive patients receiving combination systemic and intravitreal antiviral therapies showed VA improvement overall at long-term follow-up, no evidence of fellow eye involvement with systemic antiviral therapy, and a low rate (i.e. less than 10% rate) of additional quadrantic retinitis progression once therapy was instituted. While we did not specifically compare combination systemic and intravitreal antiviral therapy with systemic antiviral therapy alone as previously assessed in other retrospective, comparative series^12–13, our findings are consistent with these studies that support the efficacy of combination systemic and intravitreal therapy for ARN. In addition, we quantified an approximately 2.2 fold increased risk of RD with each additional quadrant of retina involved. These data support need for early, aggressive therapy to optimize the VA and anatomic outcome of the affected eye, while protecting the fellow eye from ARN. Our finding of quantifiable RD risk with each additional retinal quadrant involved may have prognostic value for counseling in anticipation of potential RD surgery in a substantial proportion of patients.

While recent studies including our series support combination systemic and intravitreal antiviral therapy with PCR confirmation of clinical ARN diagnosis, a number of questions remain that warrant further study. These questions include the optimal length of time that intravitreal antivirals are needed, optimal dosing strategies for intravitreal therapies, and the need for long-term prophylactic therapy. Further prospective studies to assess specific dosing strategies with prespecified VA and anatomic endpoints would help to define these parameters for this challenging and often sight-threatening infectious uveitis syndrome.

Figure 3.

Proportion of VZV-ARN and HSV-ARN with ≥ 2-line VA gain over time. The incidence rate for HSV-ARN, 0.57 events/EY (95% CI 0.25–1.12) was greater than VZV at 0.40 events/EY (95% CI 0.12 – 0.96) but not statistically significant (p > 0.05).

Figure 5.

Representative photos of ARN. A Human immunodeficiency virus-positive patient developed diffuse retinitis and hemorrhage due to VZV-ARN (A). After treatment with multiple intravitreal foscarnet and ganciclovir injections, resolution of the retinitis is observed with diffuse residual retinal pigment change and retinal fibrosis (B). A gentleman with VZV-ARN showed 360 degrees of retinal whitening in the right eye (C) and focal retinitis in the opposite eye (not shown). During treatment with oral valacyclovir and intravitreal foscarnet, a RD ensued prompting successful RD repair with pars plana vitrectomy, endolaser, and silicone oil instillation (D).

Abbreviations:

AAO

American Academy of Ophthalmology

ARN

acute retinal necrosis

AUS

American Uveitis Society

CI

confidence interval

EY

eye-year

HIV

human immunodeficiency virus

HSV

herpes simplex virus

IV

intravenous

logMAR

logarithm of the minimal angle of resolution

PCR

polymerase chain reaction

RD

retinal detachment

VA

visual acuity

VZV

varicella zoster virus