To determine classification criteria for acute retinal necrosis (ARN).
Machine learning of cases with ARN and 4 other infectious posterior uveitides / panuveitides.
Cases of infectious posterior uveitides / panuveitides were collected in an informatics-designed preliminary database, and a final database was constructed of cases achieving supermajority agreement on diagnosis, using formal consensus techniques. Cases were split into a training set and a validation set. Machine learning using multinomial logistic regression was used on the training set to determine a parsimonious set of criteria that minimized the misclassification rate among the infectious posterior uveitides / panuveitides. The resulting criteria were evaluated on the validation set.
Eight hundred three cases of infectious posterior uveitides / panuveitides, including 186 cases of ARN, were evaluated by machine learning. Key criteria for ARN included (1) peripheral necrotizing retinitis and either (2) polymerase chain reaction assay of an intraocular fluid specimen positive for either herpes simplex virus or varicella zoster virus or (3) a characteristic clinical appearance with circumferential or confluent retinitis, retinal vascular sheathing and/or occlusion, and more than minimal vitritis. Overall accuracy for infectious posterior uveitides / panuveitides was 92.1% in the training set and 93.3% (95% confidence interval 88.2, 96.3) in the validation set. The misclassification rates for ARN were 15% in the training set and 11.5% in the validation set.
The criteria for ARN had a reasonably low misclassification rate and seemed to perform sufficiently well for use in clinical and translational research.
T he first published use of the term acute retinal necrosis (ARN) was by Young and Bird in 1978, to describe the clinical findings of 4 patients with the sudden onset of a bilateral, symmetrical, and rapidly progressive retinal necrosis. In the course of their disease all 4 patients also were described as having aqueous and vitreous inflammatory cells and retinal vascular occlusion. Young and Bird noted similarities between their 4 patients and patients with herpes simplex virus (HSV) and varicella zoster virus (VZV) retinitis, as well as with 2 prior cases of necrotizing retinitis and retinal vascular occlusion of unknown cause described by Willerson and associates. There also was a striking clinical similarity between the patients described by Young and Bird and 6 patients described in the Japanese literature by Urayama and associates and a single case report of a patient with herpes zoster ophthalmicus complicated by panuveitis and retinal arteritis described by Brown and Mendis. Despite its originally being described as a bilateral disease, it subsequently became evident that the majority of ARN cases were unilateral.
Although herpesviruses were suspected causes of the ARN syndrome since its original descriptions, the first immunolocalization of a herpesvirus antigen in the eyes of patients with the ARN syndrome was reported by Peyman and associates in 1984. In 1986 Culbertson and associates provided ultrastructural, immunohistochemical, and viral culture evidence of VZV as a cause of the ARN syndrome. Numerous subsequent reports support HSV type 1 (HSV-1), HSV type 2 (HSV-2) and VZV as pathologic causes of ARN. Although some authors suggested that ocular infections with cytomegalovirus (CMV) and Epstein-Barr virus also may produce the characteristic findings of the ARN syndrome, the data to support these assertions have not been robust. Cases series have consistently reported that the viral cause of ARN segregates by age. HSV-2 is present in the youngest group (mean age early third decade of life); HSV-1 next youngest (mean age late third decade to early fourth decade); and VZV the oldest (mean age sixth decade of life).
The ARN syndrome is rare. In a prospective population-based surveillance study in the United Kingdom, 45 confirmed cases of ARN were reported in a 14-month study period, resulting in an estimated incidence of 0.63 cases per million population per year, an estimate that was very similar to results previously published on the incidence of ARN in the United Kingdom 5 years earlier (0.50 cases per million population).
Of note, 8 patients captured in the United Kingdom 2012 surveillance study were recognized as having either a preceding or concurrent central nervous system herpetic disease. Several case series and case reports similarly have reported an association between ARN and prior, subsequent, and/or concurrent viral meningitis or encephalitis, particularly among younger patients with HSV-1 or HSV-2 ARN. , , Given the low incidence of ARN, the high prevalence of latent HSV and VZV infection, and the data linking herpesvirus encephalitis or meningitis with selective defects in viral immunity, it is possible that there may be genetic risk factors in the immune response for the ARN syndrome, although there are as yet limited data to support this inference.
Because ARN is a rare disease, the data on treatment are of limited quality, largely coming from case series. Nevertheless, because of the viral etiology, rapid progression, and poor prognosis, there is a consensus that prompt treatment with antiviral agents is needed. In early reports patients were treated with intravenous acyclovir at a dose of 500 mg/m 2 every 8 hours (∼850 mg every 8 hours in an average adult). Subsequent studies have used oral valacyclovir, typically at a dose of 2 g 3-4 times per day, and 1 case series suggested similar results between oral valacyclovir plus intravitreal foscarnet and intravenous acyclovir. Given that systemically administered drugs need 5 half-lives to achieve steady- state and the rapid progression of the disease, initial therapy with intravitreal antiviral agents (eg, foscarnet) in order to achieve high intraocular drug levels seems appropriate. , Systemic therapy seems to decrease the risk of second-eye involvement. Although the optimal duration of oral therapy is uncertain, the high-risk period for second-eye involvement is the first 14 weeks after presentation, and many experts suggest treating with lower-dose maintenance therapy for at least 6 months. Retinal detachment is a frequent sequela of ARN, occurring in up to 85% of eyes. , Its frequency does not seem to be decreased by systemic antiviral therapy, perhaps owing to the extensive amount of retinal necrosis at presentation, but 1 retrospective case series suggested that it may be decreased by the early use of adjunctive intravitreal foscarnet. , , Because of the poor visual outcomes in eyes with ARN (∼50% 20/200 or worse at 6 months after presentation), prevention of second-eye involvement is an important goal of therapy.
The Standardization of Uveitis Nomenclature (SUN) Working Group is an international collaboration that has developed classification criteria for 25 of the most common uveitides using a formal approach to development and classification. Among the diseases studied was ARN.
The SUN Developing Classification Criteria for the Uveitides project proceeded in 4 phases, as previously described: (1) informatics, (2) case collection, (3) case selection, and (4) machine learning.
As previously described, the consensus-based informatics phase permitted the development of a standardized vocabulary and the development of a standardized, menu-driven hierarchical case collection instrument.
Case Collection and Case Selection
De-identified information was entered into the SUN preliminary database by the 76 contributing investigators for each disease, as previously described. , Cases in the preliminary database were reviewed by committees of 9 investigators for selection into the final database, using formal consensus techniques described in the accompanying article. , Because the goal was to develop classification criteria, only cases with a supermajority agreement (>75%) that the case was the disease in question were retained in the final database (ie, were “selected”). ,
The final database then was randomly separated into a training set (∼85% of the cases) and a validation set (∼15% of the cases) for each disease, as described in the accompanying article. Machine learning was used on the training set to determine criteria that minimized misclassification. The criteria then were tested on the validation set; for both the training set and the validation set, the misclassification rate was calculated for each disease. The misclassification rate was the proportion of cases classified incorrectly by the machine learning algorithm when compared to the consensus diagnosis. For infectious posterior uveitides and panuveitides, the diseases against which ARN was evaluated were CMV retinitis, syphilitic uveitis, tubercular uveitis, and toxoplasmic retinitis.
Two hundred fifty-two cases of ARN were collected and 186 (74%) achieved supermajority agreement on the diagnosis during the “selection” phase and were used in the machine learning phase. These cases of ARN were compared to cases of infectious posterior uveitides / panuveitides, including 211 cases of CMV retinitis, 174 cases of toxoplasmic retinitis, 35 cases of syphilitic posterior uveitis, and 197 cases of tubercular uveitis. The details of the machine learning results for these diseases are outlined in the accompanying article. The characteristics of cases with acute retinal necrosis are listed in Table 1 , and the classification criteria developed after machine learning are listed in Table 2 . In all of the cases the retinitis involved the peripheral retina, though in 18% it had extended into the posterior pole. Key features of the criteria include a peripheral necrotizing retinitis and either confirmation of HSV or VZV infection on polymerase chain reaction (PCR) of an intraocular fluid specimen or the characteristic clinical picture ( Figure 1 ). The characteristic clinical picture includes (1) either circumferential or confluent retinitis and (2) retinal vascular inflammation (sheathing, leakage, and/or occlusion), and (3) greater than minimal vitritis, unless the patient is immunocompromised. The overall accuracy for infectious posterior uveitides / panuveitides was 92.1% in the training set and 93.3% (95% confidence interval 88.2, 96.3%) in the validation set. The misclassification rate for acute retinal necrosis was 15% in the training set and 11.5% in the validation set. In both the training set and the validation set, the diseases with which it was most often confused were CMV retinitis and ocular toxoplasmosis.
|Number of cases||186|
|Age, median, years (25th, 75th percentile)||50 (33, 63)|
|Asian, Pacific Islander||15|
|Uveitis course (%)|
|Keratic precipitates (%)|
|Anterior chamber cells, grade (%)|
|Anterior chamber flare, grade (%)|
|Iris atrophy (sectoral, patchy, or diffuse)||0|
|IOP, involved eyes|
|Median, mm Hg (25th, 75th percentile)||15 (12, 19)|
|Proportion of patients with IOP > 24 mm Hg either eye (%)||6|
|Vitreous cells, grade (%)|
|Vitreous haze, grade (%)|
|Number of lesions (%)|
|Paucifocal (2 to 4)||25|
|Lesion shape (%)|
|Round or ovoid||13|
|Lesion character (%)|
|Lesion location (%)|
|Posterior pole and periphery involved||18|
|Midperiphery and/or periphery only||82|
|Lesion size (%)|
|Other features (%)|
|Retinal vascular sheathing or leakage or occlusion||54|
|Immunocompromised patients (%)||15|
|Human immunodeficiency virus infection||3|
|Chemotherapy or other immunosuppression||12|
|Laboratory data (%)|
|Aqueous or vitreous specimen PCR positive for HSV a||29|
|Aqueous or vitreous specimen PCR positive for VZV a||53|