Ocular Immunology and Inflammation

ISSN: 0927-3948 (Print) 1744-5078 (Online) Journal homepage: https://www.tandfonline.com/loi/ioii20

Viral Acute Anterior Uveitis: Clinical Signs Useful

for Differential Diagnosis

Massimo Accorinti, Luigi Petitti, Aurelia Gaeta, Daniela Giannini & Daniele
De Geronimo

To cite this article: Massimo Accorinti, Luigi Petitti, Aurelia Gaeta, Daniela Giannini & Daniele
De Geronimo (2020): Viral Acute Anterior Uveitis: Clinical Signs Useful for Differential Diagnosis,
Ocular Immunology and Inflammation, DOI: 10.1080/09273948.2020.1751213

To link to this article: https://doi.org/10.1080/09273948.2020.1751213

Published online: 14 May 2020.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ioii20
OCULAR IMMUNOLOGY AND INFLAMMATION
https://doi.org/10.1080/09273948.2020.1751213

ORIGINAL ARTICLE

Viral Acute Anterior Uveitis: Clinical Signs Useful for Differential Diagnosis
Massimo Accorinti, MD, PhD a, Luigi Petitti, MDa, Aurelia Gaeta, MDb, Daniela Giannini, MEng, PhDc,
and Daniele De Geronimo, MDc
a
Department of Ophthalmology, Sapienza University of Rome, Rome, Italy; bDepartment of Public Health and Infectious Disease, Sapienza
University of Rome, Rome, Italy; cIRCCS- Fondazione Bietti, Rome, Italy

ABSTRACT ARTICLE HISTORY

Purpose: To assess the frequency of clinical signs in patients with viral acute anterior uveitis (AAU), and Received 16 December 2019
their ability to differentiate viral versus non-viral AAU. Revised 26 March 2020
Methods: 168 patients with AAU, including 84 with presumed viral etiology, were evaluated. Sensitivity, Accepted 31 March 2020
specificity, area under the curve (AUC), positive and negative predictive value were calculated for each KEYWORDS
clinical sign. The model built with these parameters was tested on a validation group comprising 66 Acute anterior uveitis; viral
patients with AAU. etiology; clinical signs;
Results: The most useful clinical signs were unilaterality (sensitivity: 98.8%, specificity: 57.1%), intraocular sensitivity; specificity
pressure (IOP) ≥24 mmHg (sensitivity: 68.7%, specificity: 91.7%), and the association between the two
(sensitivity: 68.7%, specificity: 95.2%). In the validation group, the model built with these parameters
presented AUC of 0.939. Adding iris atrophy AUC increased to 0.97. Considering these signs, it was
possible to diagnose viral uveitis in 93.9% of the patients.
Conclusion: Unilaterality, IOP≥24 mmHg and iris atrophy are significant predictors of possible viral
etiology in AAU.

Uveitis is a relatively rare disease, with an incidence estimated The aims of this study were to assess the frequency of the
between 17 and 52 per 100,000 people annually, and clinical features of viral uveitis in a large cohort of AAU
a prevalence of 38–714 cases per 100,000 in the developed patients presenting at a tertiary eye care center, and to esti-
world.1 Anterior uveitis is by far the most common form of mate the specificity and sensitivity of the clinical signs able to
uveitis encountered in the general population, accounting for differentiate presumed viral AAU from that of other origins.
approximately 75% of cases, with an annual incidence of Furthermore, we assessed the consistency of our results in
around 12 per 100,000 population.2 It accounts for 90% of a simple clinical model to help general ophthalmologist to
all cases of uveitis in primary care settings and 50–60% in differentiate between viral and non-viral AAU.
tertiary referral centers.3,4 Among patients with anterior uvei-
tis, 75% have acute anterior uveitis (AAU), and for 25% the
Patients and Methods
condition is chronic.5,6 Up to 50% of patients with AAU may
be HLA-B27 positive.7 Another important cause of AAU is We performed a retrospective chart review of patients with
viral infection, which has been proposed as the cause of presumed viral AAU consecutively admitted to the Uveitis
anterior uveitis in approximately 10% of cases.8 Service of the Sapienza University of Rome, Italy, from 2006
More recently, with the modern widespread use of aqueous to 2012.
analysis by polymerase chain reaction (PCR), some cases of We additionally selected an equal number of patients
idiopathic anterior uveitis can be diagnosed as viral.9 examined in the same period with a diagnosis of non-viral
Nevertheless, this technology is not used worldwide, and for AAU.
ophthalmologists who do not have access to such a laboratory A total of 168 patients were included in the study, com-
examination, it might be difficult to suspect viral etiology in prising 90 males (53.6%) and 78 females (46.4%).
AAU. However, some clinical features have been proposed as Exclusion criteria were chronic uveitis, as defined by the
more common in patients with viral uveitis than in other standardization of uveitis nomenclature (SUN)14; patients
types. These features include unilateral involvement, previous who had no laboratory investigation to investigate the etiology
corneal diseases or the presence of corneal scars, increased of their uveitis; patients who refused laboratory investigation;
intraocular pressure (IOP) at the time of uveitis onset or and patients with no clear clinical data reported in the chart
relapse, sectorial or diffuse iris atrophy and pupillary and/or no follow-up to show the clinical characteristics of
dilation.10-13 Early etiological diagnosis based on characteris- their uveitis.
tics detected by clinical examination might be important to The diagnosis of presumed viral uveitis was made at the
establish adequate therapy as soon as possible, limiting the time of presentation to our department or during follow-up,
complications of prolonged inflammation. on clinical features. The clinical features considered helpful

CONTACT Massimo Accorinti [email protected] Sapienza University of Rome, Viale Del Policlinico 155, 00167 Rome, Italy
© 2020 Taylor & Francis Group, LLC
2 M. ACCORINTI ET AL.

for the diagnosis of presumed viral uveitis, based on the (CI). Positive predictive value (PPV), negative predictive value
literature10-13, were unilateral involvement, history of kerati- (NPV), positive (PLR) and negative likelihood ratio (NLR), and
tis, corneal edema, reduced corneal sensation, increased IOP area under the curve (AUC) analyses of receiver operating
of ≥24 mmHg, medium to large keratic precipitates and iris characteristic (ROC) curve were also calculated for each clinical
atrophy. Additionally, even if presenting clinical features sug- finding.
gestive for viral uveitis, the patients must also show: antiviral To assess the consistency of our results, we used a binary
antibodies (either IgG or IgM) in the serum; negative labora- logistic regression model including two and three clinically
tory investigations for other etiology of uveitis (including tests significant parameters as predictive variables able to differ-
for syphilis, QuantiFERON-TB Gold, chest x-ray, urinalysis, entiate between viral and non-viral AAU. A threshold of 0.7
kidney, and liver function) and a positive response to antiviral was chosen.
therapy, in combination with corticosteroids. Furthermore, we selected a validation group of new
Eighty-four patients with a clinical chart reporting the patients with a diagnosis of AAU observed after 2012 to
diagnosis of presumed viral AAU were included in Group 1. validate the use of a combination of clinical parameters to
Twenty-six of those patients also received aqueous tap and assess the viral or non-viral etiology of uveitis. The validation
real-time PCR analysis to detect viral DNA (herpes simplex group comprised 66 patients, 32 males and 34 females, with
virus 1 [HSV1], herpes simplex virus 2 [HSV2], varicella- a mean age at uveitis onset of 39.6 ± 18.6 years. Thirty-three
zoster virus [VZV], cytomegalovirus [CMV], and Epstein- patients, 17 males and 16 females, were diagnosed with viral
Barr virus [EBV]), either for diagnostic purposes in the uveitis. Of these, 14 tested positive on aqueous analysis, six
acute phase of their uveitis, or when they were operated on showed uveitis shortly after a herpes zoster ophthalmicus
for cataract or glaucoma during follow-up. Patients with posi- infection with no serologic evidence of other possible uveitic
tive PCR in aqueous humor and those presenting uveitis etiology, one presented AAU in combination with a systemic
concomitantly or immediately after a herpes zoster ophthal- CMV infection (serum IgM positive, IgG negative) and 12
micus (HZO) on the same side were considered patients with showed clinical features suggesting viral etiology without
proven diagnosis of viral uveitis. serologic evidence for other possible etiology. They also
Group 2 included the first consecutive 84 patients with responded adequately to antiviral therapy, systemic (acyclo-
a clinical diagnosis of non-viral AAU. The patients were vir 2400–4000 mg/day in the acute phase) and/or local (gan-
classified as having non-viral AAU if their history was nega- ciclovir gel 5 times/day), in combination with topical
tive for a recent viral disease, serology was negative for corticosteroid treatment. The remaining 33 patients were
a recent viral infection at uveitis presentation, uveitis was affected by other forms of AAU: 16 (48.5%) were HLA-B27
associated with a systemic disease (e.g. ankylosing spondylitis, +, 11 idiopathic (33.3%), 3 suffered from ankylosing spondy-
psoriasis, inflammatory bowel disease, sarcoidosis, etc.), HLA litis (9.1%) 2 from psoriasis (6%) and 1 from fibromial-
typing was positive for B27 antigen, or if all laboratory inves- gia (3%).
tigations resulted negative (including tests for syphilis, All the patients admitted to our service have to sign an
QuantiFERON-TB Gold, chest x-ray, urinalysis, kidney and informed consent allowing the use of data drawn from their
liver function), and the uveitis could not be defined etiologi- clinical charts, for scientific purpose, in an anonymous form.
cally (idiopathic uveitis). The study was performed in accordance with the tenets of the
The following parameters, either available from the patient Declaration of Helsinki and institutional review board
clinical reports or found during our examination, were eval- approval was obtained.
uated in patients of both groups: age at uveitis onset, visual Sample size calculation was performed considering the
acuity at presentation, unilateral or bilateral involvement formula based on the work of Peduzzi et al.15 for logistic
(either concomitant or not), type of onset, course of uveitis, regression: N = 10 k/p, where p was the smallest of the
ciliary injection at onset or during relapse, reduced corneal proportions of negative or positive cases in the population
sensation, corneal edema, history of keratitis, keratic precipi- and k the number of covariates (the number of independent
tates, grade of inflammation in anterior chamber according to variables). In our case, for a model with three covariates, the
SUN classification14, presence of posterior synechiae, hetero- minimum number of cases to include was approximately 60.
chromia, iris atrophy, iris nodules, endotheliitis, IOP Moreover, the ROC analysis required a sample size of around
≥24 mmHg at onset or during relapse without corticosteroid 100 observations for meaningful qualitative conclusions16,
therapy, onset and type of cataract during follow-up, vitreous with a minimum of 50 cases in each of the two groups.
cells, retinal scars, and macular edema. Due to the retrospec- Statistical evaluation was performed using SPSS (IBM SPSS
tive nature of the evaluation, the reported prevalence of the Statistics version 25). Continuous variables, including age, and
considered parameters does not always refer to the total age of onset, were expressed as mean ± standard deviations,
number of the patients in both groups, but to those whose while categorical variables were expressed as frequencies.
clinical chart reported the specific parameter. The normal data distribution was tested using the one-
Forty-five out of 84 (53.6%) of the patients in Group 1 and sample Kolmogorov–Smirnov test. The independent sample
50 out of 84 patients (59.5%) of patients in Group 2 had t-test and the Mann–Whitney test were used to compare the
a follow-up at ≥12 months. parameter values between the two groups. A chi-square test or
Sensitivity and specificity were evaluated for each parameter Fisher exact test, two sides, as appropriate, was performed to
and for the association between parameters that were considered investigate the relationship between the groups and the clin-
clinically significant, reported with 95% confidence intervals ical categorical variables.
 OCULAR IMMUNOLOGY AND INFLAMMATION 3

For the parameters showing a statistically significant asso- uveitis onset, visual acuity >0.3 was found in 82.9% of the
ciation between the two groups, the AUC from a ROC curve patients, while IOP ≥24 mmHg at uveitis onset and/or during
analysis was calculated to measure the diagnostic accuracy. relapse without corticosteroid therapy was diagnosed in 57
The AUC has a value ranging from 0.5 to 1.0, where 1.0 out of 83 patients (68.7%). Ciliary injection was present in
represents the perfect ability to discriminate viral from non- 72.2% of patients, reduced corneal sensation in 42.5%, iris
viral AAU, and 0.5 represents discrimination resulting from atrophy in 42.3%, fine keratic precipitates in 44.1%, corneal
pure chance. An AUC greater than 0.9 is considered excellent, edema in 33.3%, history of keratitis in 22%, and cells in the
between 0.8 and 0.9 very good, greater than 0.7 up to 0.8 anterior chamber >2+ in 14.9% (Table 2).
good, between 0.6 and 0.7 average, and greater than 0.5 up to In the 84 patients with non-viral AAU (Group 2), the mean
0.6 poor.17 A binary logistic regression model18 was applied age at uveitis onset was 36.27 ± 13.98 years (range:
using the variables, with the highest AUC scores as explana- 9–66 years), younger than in Group 1 (t = 3.17, p = .002).
tory independent variables (i.e. predictors) to classify viral and Patients included in this group have had the following
non-viral cases (i.e. the categorical dependent variables). The diagnosis: idiopathic uveitis in 26 cases (30.9%), undifferen-
variance inflation factor (VIF), which assesses how much the tiated spondyloarthritis in 6 (7.1%), ankylosing spondylitis in
variance of an estimated regression coefficient increases if the 5 (5.9%), psoriasis in 2 (2.4%), Crohn disease in 2 (2.4%),
predictors are correlated, was used to assess multicollinearity. systemic lupus erythematosus in 2 (2.4%), tubulo-interstitial
A VIF between 5 and 10 indicates a moderate correlation nephritis and uveitis syndrome in 2 (2.4%), undifferentiated
among predictors. In this study, the maximum threshold for connective tissue disease in 2 (2.4%), ulcerative colitis in 1
VIF outcome was set at 2. (1.2%). All these patients were HLAB27 negative, and those
The efficacy of the proposed model in predicting labeled as idiopathic uveitis did not present infectious causes,
a diagnosis of viral uveitis was measured in terms of sensitiv- association with systemic diseases or clinical features of viral
ity and specificity, PPV, NPV, PLR, and NLR. PLR values in uveitis as previously described. HLA B27 + uveitis, with or
the range of <2, 2–5, 5–10 and >10 are considered not mean- without associated systemic disease, was diagnosed in addi-
ingful, small, moderate, and large increases of probability, tional 36 subjects (42.8%).
respectively. NLR values in the range of >0.5, 0.2–0.5, The clinical features of patients in Group 2 are also
0.1–0.2, and <0.1 represent not meaningful, small, moderate, reported in Table 2.
and large decreases of probability, respectively.19 Comparing the clinical features detected in Groups 1 and
Statistically significant differences were set at p < .05 for all 2, we found that unilateral involvement (98.8% versus 42.6%,
the tests performed. p < .001), reduced corneal sensation (42.5% versus 6.3%,
p = .009), corneal edema (33.3% versus 10.8%, p = .001),
history of keratitis (22% versus 0%, p < .001), iris atrophy
Results
(42.3% versus 4.9%, p < .001), and IOP ≥24 mmHg (68.7%
In the 84 patients with a diagnosis of presumed viral uveitis, versus 8.3%, p < .001) were all statistically associated with
the mean age at uveitis onset was 45.4 ± 22.0 years (range: presumed viral uveitis, while ciliary injection (72.2% versus
2–87 years). 93.8%, p < .001), fine keratic precipitates (76.5% versus 44.1%,
Herpes Simplex Virus (HSV) uveitis was diagnosed, either p < .001) and anterior chamber cells >2+ (14.9% versus 31.1%,
clinically or by PCR in aqueous humor, in 49 cases (58.3%), p = .044) were statically associated with non-viral uveitis
Varicella Zoster Virus (VZV) uveitis in 15 (17.8%), (Table 2)
Cytomegalovirus (CMV) uveitis in 19 (22.6%), and Epstein- Therefore, specificity, sensitivity, PPV, NPV, PLR, NLR, and
Barr virus (EBV) uveitis in 1 case (1.2%). Twenty-six of these AUC of the aforementioned clinical findings have been calcu-
subjects received PCR analysis of the aqueous humor for lated for the whole group 1 and 2, and are reported in Table 3.
HSV1, HSV2, VZV, CMV and EBV, with 19 testing posi- Between the parameters described, unilaterality and IOP
tive (73.1%). ≥24 mmHg were the two clinical signs with the highest overall
A comparison among the frequency of clinical features values considering both sensitivity (98.8% and 68.7%, respec-
found in 27 patients with “proven” viral uveitis (19 with tively) and specificity (57.1% and 91.7%, respectively). Iris
positive PCR in aqueous humor, and 8 presenting uveitis atrophy and corneal edema presented high values of specifi-
concomitantly or immediately after a HZO on the same city (95.1% and 89.2%, respectively) but low values of sensi-
side) and those detected in 57 “clinically diagnosed” viral tivity (42.3% and 33.3%, respectively).
uveitis was made and is reported in Table 1. Excluding The association between unilaterality and IOP ≥24 mmHg
a higher frequency of patients with a history of herpetic presented the highest overall value considering both sensitiv-
keratitis in subjects carrying a “clinical diagnosis” of viral ity (68.7%) and specificity (95.2%), and had the highest AUC
uveitis (p = .008), no other differences were found. value (0.82). All the other associations between two, three or
Therefore, providing that no differences in terms of clinical four parameters (unilaterality, IOP ≥24 mmHg, iris atrophy,
features can be found in patients with “proven” or “clinically and corneal edema) presented values of specificity equal or
diagnosed” acute viral uveitis, we have defined all these sub- near 100%, but low values of sensitivity.
jects as patients with viral uveitis and included them in Iris atrophy and IOP ≥24 mmHg were the parameters with
Group 1. the highest PLR values (8.67 and 8.24, respectively).
In this group, unilateral uveitis was diagnosed in 83 cases Figure 1 shows the ROC curves of the most frequently
(98.8%), with sudden onset in 73 out of 83 patients (88%). At encountered clinical signs and of some of their associations.
4 M. ACCORINTI ET AL.

Table 1. Comparison of clinical features in patients with “proven” or “clinically diagnosed” acute anterior viral uveitis.
“Proven” acute
anterior viral Chi-square/
uveitis Fisher’s
n°/total (%) “Clinically diagnosed” acute anterior viral uveitis n°/total (%) exact test p
Unilaterality 27/27 (100) 56/57 (98.2) - 1.000
Sudden Onset 25/27 (92.6) 48/56 (85.7) - 0.487
Visual acuity >0.3 21/24 (87.5) 42/52 (80.8) - 0.744
Limited Duration 18/26/(69.2) 46/55 (83.6) 2.209 0.137
Recurrent course 9/27 (33.3) 27/55 (49.1) 1.826 0.177
Ciliary injection 19/25 (76.0) 38/54 (70.4) 0.270 0.604
Reduced corneal sensitivity 4/12 (33.3) 13/28 (46.4) - 0.505
Corneal Edema 5/25 (20.0) 21/53 (39.6) - 0.123
Previous keratitis 1/25 (3.8) 17/56 (30.4) - 0.008
Keratic precipitates 27/27 (100) 53/55 (96.4) - 1.000
Fine keratic precipitates 9/25 (36.0) 21/43 (48.8) 1.057 0.304
Endotheliitis 1/25 (4.0) 2/53 (3.8) - 1.000
Anterior chamber flare (0-2+) 13/15 (86.7) 28/32 (87.5) - 1.000
Anterior chamber cells (0-2+) 12/15 (80.0) 28/32 (87.5) - 0.664
Iris Atrophy 9/24 (37.5) 24/54 (44.4) 0.328 0.567
Sectorial iris Atrophy 3/9 (33.3) 6/23 (26.1) - 0.685
Posterior synechiae 6/26 (23.1) 13/53 (24.1) 0.020 0.887
Heterochromia 0/27 (0.0) 4/56 (7.1) - 0.299
Iris nodules 0/27 (0.0) 1/57 (1.8) - 1.000
IOP (≥ 24 mmHg) 10/26 (38.5) 16/57 (28.1) 0.896 0.344
Cataract 7/27 (25.9) 11/57 (19.3) 0.478 0.489
Subcapsular cataract 4/7 (57.1) 2/11 (18.2) - 0.141
Vitreous cells > 2+ 1/26 (3.8) 0/54 (0.0) - 0.325
Cystoid macular edema 1/26 (3.8) 1/55 (1.8) - 0.542
Retinal scars 1/26 (3.8) 2/55 (3.6) - 1.000

Table 2. Clinical signs in viral acute anterior uveitis (Group 1) and non-viral acute anterior uveitis (Group 2).
Group 1 Group 2 Chi-square/
n°/total (%) n°/total (%) Fisher’s exact test p
Unilaterality 83/84 (98.8) 36/84 (42.6) - <0.001
Sudden onset 73/83 (88) 77/83 (92.8) 1.107 0.293
Visual acuity >0.3 63/76 (82.9) 75/80 (93.8) 4.499 0.034
Limited duration 64/81 (79) 81/83 (97.6) - <0.001
Recurrent course 36/82 (43.9) 57/82 (69.5) 10.953 0.001
Ciliary injection 57/79 (72.2) 75/80 (93.8) 13.152 <0.001
Reduced corneal sensitivity 17/40 (42.5) 1/16 (6.3) - 0.011
Corneal edema 26/78 (33.3) 9/83 (10.8) 11.955 0.001
Previous keratitis 18/82 (22) 0/83 (0) - <0.001
Keratic precipitates 80/82 (97.6) 72/77 (93.5) - 0.265
Fine keratic precipitates 30/68 (44.1) 52/68 (76.5) 14.865 <0.001
Endotheliitis 3/75 (3.8) 0/82 (0) - 0.114
Anterior chamber flare (0-2+) 41/47 (87.2) 54/74 (73) 3.465 0.063
Anterior chamber cells (0-2+) 40/47 (85.1) 51/74 (68.9) 4.039 0.044
Iris atrophy 33/78 (42.3) 4/82 (4.9) - <0.001
Sectorial iris atrophy 19/33 (57.6) 0/4 (0) - 0.029
Posterior synechiae 19/79 (24.1) 26/84 (31) 0.970 0.325
Heterochromia 4/83 (4.8) 1/84 (1.2) - 0.210
Iris nodules 1/84 (1.2) 2/84 (2.4) - 1.000
IOP (≥ 24 mmHg) 57/83 (68.7) 7/84 (8.3) 64.311 <0.001
Cataract 18/84 (21.4) 16/80 (20) 0.051 0.822
Subcapsular cataract 10/18 (55.6) 7/16 (43.8) 0.472 0.492
Vitreous cells > 2+ 1/80 (1.3) 2/83 (2.4) - 1.000
Cystoid macular edema 2/81 (2.5) 1/84 (1.2) - 0.616
Retinal scars 3/81 (3.7) 1/84 (1.2) - 0.361

The selection of predictors for the binary logistic regres- was 0.015 (95% CI 0.002–0.119) and for IOP ≥ 24 mmHg
sion analysis was based on the AUC analysis results. 15.674 (95% CI: 5.435–45.200).
Unilaterality and IOP ≥24 mmHg, and these variables in By adding iris atrophy to unilaterality and IOP ≥24 mmHg,
association with iris atrophy, were selected as predictors for the model improved its performance. The three predictors were
building two different models. significant (p < .05), explaining 72.3% of the variability in
Using a two-parameter model binary logistic regression diagnosis for uveitis. The OR for unilaterality was 0.011 (95%
suggested that unilaterality and IOP ≥24 mmHg were signifi- CI: 0.001–0.129), for IOP ≥24 mmHg 13.435 (95% CI: 4.176–-
cant predictors of the viral nature of uveitis (Wald = 15.541, 43.219) and for iris atrophy 0.031 (95% CI: 0.003–0.288).
p < .001; Wald = 25.937, p < .001; respectively). All the The performance of logistic regression models in model
predictors account for 65.5% (Nagelkerke R2) of the variability development cohort (Table 4) showed very good diagnostic
in diagnosis for uveitis. The odds ratio (OR) for unilaterality accuracy, with AUC values of 0.820 (sensitivity: 68.7%,
Table 3. ROC analysis for clinical signs of uveitis and their association.
Between Group 1 (viral uveitis) and Group 2 (non viral uveitis) patients
Sensitivity [%] Specificity [%] PLR NLR PPV [%] NPV [%]
Variable AUC P value 95% CI 95% CI 95% CI 95% CI 95% CI 95% CI
Unilaterality 0.782 <0.0001 98.8 [93.5–100.0] 57.1 [45.9–67.9] 2.31 [1.8–3.0] 0.021 [0.003–0.1] 69.7 [64.3–74.7] 98.0 [87.1–99.7]
Visual acuity>0.3 0.554 0.0344 17.1 [9.4–27.5] 93.7 [86.0–97.9] 2.74 [1.0–7.3] 0.88 [0.8–1.0] 72.2 [49.3–87.4] 54.3 [51.4–57.2]
Limited duration 0.593 0.0001 21.0 [12.7–31.5] 97.6 [91.6–99.7] 8.71 [2.1–36.5] 0.81 [0.7–0.9] 89.5 [67.0–97.3] 55.9 [53.0–58.7]
Recurrent course 0.640 0.0002 56.1 [44.7–67.0] 69.5 [58.4–79.2] 1.84 [1.3–2.7] 0.63 [0.5–0.8] 64.8 [55.7–72.9] 61.3 [54.4–67.8]
Ciliary injection 0.608 0.0002 27.8 [18.3–39.1] 93.7 [86.0–97.9] 4.46 [1.8–11.2] 0.77 [0.7–0.9] 81.5 [63.7–91.7] 56.8 [53.2–60.4]
Reduced corneal sensitivity 0.681 0.0003 42.5 [27.0–59.1] 93.7 [69.8–99.8] 6.80 [1.0–46.9] 0.61 [0.5–0.8] 94.4 [71.1–99.2] 39.5 [32.7–46.7]
Corneal edema 0.612 0.0004 33.3 [23.1–44.9] 89.2 [80.4–94.9] 3.07 [1.5–6.1] 0.75 [0.6–0.9] 74.3 [59.1–85.2] 58.7 [54.5–62.9]
Previous keratitis 0.610 <0.000 21.9 [13.6–32.5] 100.0 [95.7–100.0] - 0.78 [0.7–0.9] 100.0 56.5 [53.6–59.3]
Fine keratic precipitates 0.658 0.0001 55.9 [43.3–67.9] 76.5 [64.6–85.9] 2.37 [1.5–3.8] 0.58 [0.4–0.8] 70.4 [59.6–79.3] 63.4 [56.3–70.0]
Anterior chamber cells (0-2+) 0.579 0.0392 85.1 [71.7–93.8] 31.1 [20.8–42.9] 1.23 [1.0–1.5] 0.48 [0.2–1.0] 44.0 [39.2–48.8] 76.7 [60.5–87.6]
Iris Atrophy 0.687 <0.0001 42.3 [31.2–54.0] 95.1 [88.0–98.7] 8.67 [3.2–23.3] 0.61 [0.5–0.7] 89.2 [75.4–95.7] 63.4 [58.8–67.8]
IOP (≥ 24 mmHg) 0.802 <0.0001 68.7 [57.6–78.4] 91.7 [83.6–96.6] 8.24 [4.0–17.0] 0.34 [0.2–0.5] 89.1 [79.8–94.4] 74.8 [68.2–80.4]
Iris atrophy+ Unilaterality 0.712 <0.0001 42.3 [31.2–54.0] 100.0 [95.7–100.0] - 0.58 [0.5–0.7] 100.0 [89.4–100.0] 64.8 [55.9–73.1]
Iris atrophy + IOP (≥ 24 mmHg) 0.652 <0.0001 30.4 [20.5–41.8] 100.0 [95.7–100.0] - 0.70 [0.6–0.8] 100.0 [85.8–100.0] 60.1 [51.5–68.4]
Iris atrophy + corneal edema 0.567 <0.0001 13.4 [6.9–22.7] 100.0 [95.7–100.0] - 0.87 [0.8–0.9] 100.0 [71.5–100.0] 53.9 [45.7–61.9]
Unilaterality + IOP (≥ 24 mmHg) 0.820 <0.0001 68.7 [57.6–78.4] 95.2 [88.3–98.7] 14.42 [5.5–37.9] 0.33 [0.2–0.5] 93.4 [84.1–98.2] 75.5 [66.1–83.3]
Unilaterality + corneal edema 0.631 <0.0001 33.3 [23.1–44.9] 92.9 [85.1–97.3] 4.67 [2.0–10.7] 0.72 [0.6–0.8] 81.2 [63.6–92.8] 60.0 [51.0–68.5]
IOP (≥ 24 mmHg) + corneal edema 0.608 <0.0001 21.5 [13.1–32.2] 100.0 [95.7–100.0] - 0.78 [0.7–0.9] 100.0 [80.5–100.0] 57.5 [49.1–65.7]
Iris atrophy + IOP (≥ 24 mmHg) + unilaterality 0.652 <0.0001 30.4 [20.5–41.8] 100.0 [95.7–100.0] - 0.70 [0.6–0.8] 100.0 [85.8–100.0 60.4 [51.8–68.6]
Iris atrophy + corneal edema + unilaterality 0.567 0.0004 13.4 [6.9–22.7] 100.0 [95.7–100.0] - 0.87 [0.8–0.9] 100.0 [71.5–100.0] 54.2 [46.0–62.2]
Unilaterality + IOP (≥ 24 mmHg) + corneal edema 0.608 <0.0001 21.5 [13.1–32.2] 100.0 [95.7–100.0] - 0.78 [0.7–0.9] 100.0 [80.5–100.0] 57.5 [49.1–65.7]
Iris atrophy + IOP (> 24 mmHg) + corneal edema 0.548 0.0031 9.5 [4.2–17.9] 100.0 [95.7–100.0] 0.90 [0.8–1.0] 100.0 [63.1–100.0] 52.5 [44.5–60.4]
Iris atrophy+ unilaterality + 0.548 0.0031 9.5 [4.2–17.9] 100.0 [95.7–100.0] - 0.90 [0.8–1.0] 100.0 [63.1–100.0] 52.5 [44.5–60.4]
IOP (≥ 24 mmHg)+corneal edema
AUC: Area under the curve; CI: confidence interval; PLR: positive likelihood ratio; NLR: negative likelihood ratio; PPV: positive predictive value; NPV: negative predictive value
OCULAR IMMUNOLOGY AND INFLAMMATION
5
6 M. ACCORINTI ET AL.

Figure 1. ROC curve of the clinical signs most frequently associated with viral anterior uveitis and some of their associations.

Table 4. Performance of logistic regression model in model development cohort.

AUC PPV NPV
[%] P value Sensitivity (%) Specificity (%) PLR NLR [%] [%]
Unilaterality + IOP ≥ 24 mmHg 0.820 <0.0001 68.7 95.2 14.42 0.33 93.4 75.5
Iris atrophy + unilaterality + IOP ≥ 24 mmHg 0.872 <0.0001 79.2 95.1 16.24 0.22 93.8 83.0
AUC: Area Under the Curve; PLR: positive likelihood ratio; NLR: negative likelihood ratio; PPV: positive predictive value; NPV: negative predictive value

Table 5. Performance of logistic regression model in validation cohort.

AUC [%] P value Sensitivity (%) Specificity (%) PLR NLR PPV [%] NPV [%]
Unilaterality + IOP ≥ 24 mmHg 0.939 <0.0001 87.9 100.0 - 0.12 100.0 89.2
Iris atrophy + Unilaterality + IOP ≥ 24 mmHg 0.970 <0.0001 93.9 100.0 - 0.061 100.0 94.3
AUC: Area under the curve; PLR: positive likelihood ratio; NLR: negative likelihood ratio; PPV: positive predictive value; NPV: negative predictive value

specificity: 95.2%) for two parameters (unilaterality and IOP AUC values of 0.950 (sensitivity: 90%, specificity: 100%) for
≥24 mmHg) and 0.872 (sensitivity: 79.2%, specificity: 95.1%) two parameters (unilaterality and IOP ≥24 mmHg) and 0.975
for three parameters (unilaterality, IOP ≥24 mmHg, and iris (sensitivity: 95%, specificity: 100%) for three parameters (uni-
atrophy). The two-parameter model allowed correct classifica- laterality, IOP ≥24 mmHg, and iris atrophy). The two-
tion of 57 out of 83 (68.7%) cases of viral uveitis and 80 out of parameter model allowed correct classification of 18 out of
84 (95.2%) of non-viral uveitis. The three-parameter model 20 (90%) cases of viral uveitis and 33 out of 33 (100%) of non-
allowed correct classification of 61 out of 77 (79.2%) cases of viral uveitis. The three-parameter model allowed correct clas-
viral uveitis and 78 out of 82 (95.1%) of non-viral uveitis. sification of 19 out of 20 (95%) cases of viral uveitis and 33
The performance of logistic regression models in the 66 out of 33 (100%) of non-viral uveitis.
patients included in the validation group (Table 5) confirmed
excellent diagnostic accuracy, with AUC values of 0.939 (sen-
Discussion
sitivity: 87.9%, specificity: 100%) for two parameters (unila-
terality and IOP ≥24 mmHg) and 0.970 (sensitivity: 93.9%, Anterior uveitis is the most frequent intraocular inflammation
specificity: 100%) for three parameters (unilaterality, IOP found in the general population, and AAU is usually mana-
≥24 mmHg, and iris atrophy). ged, at least at the onset, by general ophthalmologists. Some
The two-parameter model allowed correct classification of clinical signs can suggest infection by the commonest viruses
29 out of 33 (87.9%) cases of viral uveitis and 33 out of 33 responsible, such as HSV, VZV, or CMV. These features are
(100%) of non-viral uveitis. The three-parameter model unilateral involvement, previous or concomitant corneal
allowed correct classification of 31 out of 33 (93.9%) cases lesions, increased IOP at uveitis onset or during relapse,
of viral uveitis and 33 out of 33 (100%) of non-viral uveitis. diffuse, or sectorial atrophic changes of the iris, and perma-
Furthermore, we also evaluated the performance of logistic nent dilation of the pupil following uveitis onset.5,6,12,20,21
regression models in the validation group, considering only In this setting, some differences can help to differentiate
a subgroup of 20 patients with proven viral uveitis, as pre- between the various etiologic agents. HSV uveitis typically has
viously described, and 33 patients with non-viral uveitis. This an acute unilateral course, often appears in older patients, is
confirms an excellent diagnostic accuracy of the model, with characterized by conjunctival redness, and may show
 OCULAR IMMUNOLOGY AND INFLAMMATION 7

endotheliitis, medium to large keratic precipitates and accuracy (AUC: 0.872, sensitivity: 79.2%, specificity: 95.1%).
increased IOP.8 VZV uveitis is often associated with keratitis, The chance to diagnose viral uveitis using the two-parameter
may have a chronic recurrent course, and may lead to sector- model (unilaterality and IOP increase, both signs that might
ial iris atrophy, pupillary distortion, and increased IOP.22 be found even in the first episode of uveitis) was 68.7%, and to
CMV anterior uveitis can manifest with either an acute diagnose non-viral uveitis 95.2%. The three-parameter model
onset presenting as recurrent hypertensive anterior uveitis allowed correct classification of 79.2% of viral uveitis and
with few granulomatous keratic precipitates (resembling 95.1% of non-viral uveitis. Nevertheless, as the presence of
Posner-Schlossman syndrome)10,20,23 or with a chronic course iris atrophy is more common in the subsequent course of the
resembling Fuchs heterochromic iridocyclitis or idiopathic uveitis, this might be the correct scenario to use the three-
chronic anterior uveitis.10,24 parameter model.
Nevertheless, the real and statistical value of the clinical Subsequently, we validated the data using a validation
signs associated with a viral agent as the cause of AAU has not group of 66 consecutive cases, 33 showing viral uveitis.
yet been established. To address this problem, we compared Interestingly, we confirmed excellent diagnostic accuracy of
the clinical features of patients diagnosed and treated for viral the two-parameter model (AUC: 0.939, sensitivity 87.9%, spe-
AAU with those of patients with non-viral AAU. cificity 100%), with a correct classification of 87.9% of viral
Our study has demonstrated that patients with acute viral and 100% of non-viral AAU. The three-parameter model
anterior uveitis were older than those with other types of acute showed the same results (AUC: 0.97, sensitivity 93.9%, speci-
anterior uveitis (p = .002). Unilateral uveitis (p < .001), reduced ficity 100%), with a correct classification of viral and non-viral
corneal sensation (p = .021), corneal edema (p < .001), history of AAU in 93.9% and 100% of cases, respectively.
keratitis (p < .001), iris atrophy (p < .001) and IOP ≥24 mmHg A possible bias in our analysis might be found in the
(p < .001) were all statistically related to a diagnosis of viral diagnosis of viral acute uveitis made predominantly on clin-
uveitis. In contrast, ciliary injection (p < .001), fine keratic ical features, which was done in both the study group and the
precipitates (p = .002) and cells in the anterior chamber greater validation group. To reduce this bias we have demonstrated
than 2+ (p = .044) were statistically more frequently encoun- that in a referral center for uveitis, no statistical differences in
tered in non-viral AAU (Table 2). terms of frequency of clinical features can be found in patients
Not all the clinical signs statistically associated with viral with proven viral uveitis (i.e. those with a PCR positive in
etiology of uveitis showed a useful combination of sensitivity aqueous humor for viruses or with an uveitis occurred con-
and specificity. Most, such as corneal edema, reduced corneal comitantly or shortly after and HZO on the same side), and in
sensation, history of keratitis and iris atrophy, presented good subjects with clinically diagnosed viral uveitis (Table 1), sug-
specificity but low sensitivity, with a limited value in the gesting that the clinical features alone are good enough to
differential diagnosis. make the diagnosis of viral acute anterior uveitis.
Unilaterality and IOP ≥24 mmHg were the two clinical To corroborate our results, we have also tested the perfor-
signs with the highest overall values considering both sensi- mance of the logistic regression models in a subgroup of 20
tivity (98.8% and 68.7%, respectively) and specificity (57.1% patients of the validation group with proven viral uveitis.
and 91.7%, respectively) and, therefore, should always be Those patients had a diagnosis of viral uveitis due to either
considered in the diagnostic process of AAU. Corneal PCR detection of viral DNA in the aqueous humor, or onset
edema might present in the acute onset of viral uveitis: it of uveitis concomitantly or shortly after an episode of herpes
has a high specificity (89.2%) but a low sensitivity (33.3%). zoster ophthalmicus, without laboratory evidence of any other
Iris atrophy, which might be encountered more frequently as possible cause of AAU.
a sign of recurrent uveitis, is also highly specific to viral Even in this subgroup of patients the AUC was 0.950
etiology (95.1%), but with low sensitivity (42.3%). (sensitivity: 90%, specificity: 100%) for two parameters (uni-
To provide some useful suggestions for clinical diagnosis of laterality and IOP ≥24 mmHg) and 0.975 (sensitivity: 95%,
viral uveitis without any invasive procedure, we statistically specificity: 100%) for three parameters (unilaterality, IOP
tested some associations between clinical signs. ≥24 mmHg, and iris atrophy). The two-parameter model
We found that the association between unilaterality and allowed correct classification of 18 out of 20 (90%) cases of
IOP ≥24 mmHg presented the highest overall value consider- viral uveitis and 33 out of 33 (100%) of non-viral uveitis. The
ing both sensitivity (68.7%) and specificity (95.2%), and had three-parameter model allowed correct classification of 19 out
the highest AUC value (0.82). All other associations between of 20 (95%) cases of viral uveitis and 33 out of 33 (100%) of
the most frequently encountered clinical signs (unilaterality, non-viral uveitis.
IOP ≥24 mmHg, iris atrophy and corneal edema; Table 3) In conclusion, our study has confirmed that there are
were highly specific for a diagnosis of viral uveitis, but with clinical signs suggestive of viral etiology in patients with
low values of sensitivity, suggesting a possibly high number of AAU. Some, such as corneal edema, abnormal corneal sensa-
false negatives. tion, previous history of keratitis and iris atrophy, if detected
In order to give the accuracy of the provided data, we individually, are highly specific to viral disease but not suffi-
developed two logistic regression models: one with two para- ciently sensitive to be able to suggest a diagnosis of viral
meters only (unilaterality and IOP ≥24 mmHg) and the other uveitis. Unilateral involvement and IOP ≥24 mmHg during
with three parameters (adding iris atrophy). The first model the acute phase of the disease and in the absence of any
provided very good diagnostic accuracy (AUC: 0.820, sensi- corticosteroid therapy showed a good combination of specifi-
tivity: 68.7%, specificity: 95.2%), and the second even greater city and sensitivity, suggesting a role in the clinical diagnostic
8 M. ACCORINTI ET AL.

process for AAU. Considering that the great majority of 9. Anwar Z, Galor A, Albini TA, Miller D, Perez V, Davis JL. The
patients with AAU are diagnosed and managed by general diagnostic utility of anterior chamber paracentesis with polymer-
ase chain reaction in anterior uveitis. Am J Ophthalmol. 2013
ophthalmologists, at least at their first episode of uveitis, these May;155(5):781–786. doi:10.1016/j.ajo.2012.12.008.
practitioners should be aware of the statistical possibility of 10. The standardization of uveitis nomenclature (SUN) working
viral AAU if the uveitis is unilateral and with IOP ≥24 mmHg. group. Standardization of uveitis nomenclature for reporting clin-
After some days and in recurrent cases, the additional evi- ical data. Results of the first international workshop. Am
dence of iris atrophy strongly supports the diagnosis of viral J Ophthalmol. 2005;140(3):509–516. doi:10.1016/j.ajo.2005.03.057.
11. Chan NS, Chee SP, Caspers L, Bodaghi B. Clinical features of
uveitis.
CMV-associated anterior uveitis. Ocul Immunol Inflamm. 2018;26
Viral uveitis is usually a life-long problem, characterized by (1):107–115. doi:10.1080/09273948.2017.1394471.
frequent recurrences. Considering viral etiology of AAU, from 12. Sakai JI, Usui Y, Suzuki J, Kezuka T, Goto H. Clinical features of
the first appearance of uveitis or immediately after, might anterior uveitis caused by three different herpes viruses.
allow appropriate early antiviral therapy. Further studies are Int Ophthalmol. 2019 May 27;39(12):2785–2795. doi:10.1007/
s10792-019-01125-5. [Epub ahead of print].
needed to evaluate whether such treatment can reduce the 13. Tugal-Tutkun I, Otük-Yasar B, Altinkurt E. Clinical features and
number of recurrences, thus lowering the associated compli- prognosis of herpetic anterior uveitis: a retrospective study of 111
cations, especially cataract and glaucoma development. cases. Int Ophthalmol. 2010;30(5):559–565. doi:10.1007/s10792-
010-9394-8.
14. Goldstein DA, Mis AA, Oh FS, Deschenes JG. Persistent pupillary
Declaration of interest dilation in herpes simplex uveitis. Can J Ophthalmol. 2009 Jun;44
(3):314–316. doi:10.3129/i09-018.
The authors report no conflicts of interest and no financial disclosures.
15. Peduzzi P, Concato J, Kemper E, et al. A simulation study of the
number of events per variable in logistic regression analysis. J Clin
Epidemiol. 1996;49(12):1373–1379. doi:10.1016/S0895-4356(96)
Funding 00236-3.
This research was supported by the Italian Ministry of Health and 16. Metz CE. Basic principles of ROC analysis. Semin Nucl Med.
Fondazione Roma but without any role in study design, data collection 1978;8(4):283–298. WB Saunders. doi:10.1016/S0001-2998(78)
and analysis, decision to publish, or preparation of the manuscript. 80014-2.
17. Ying GS, Maguire M, Quinn G, et al. Vision in preschoolers (VIP)
study group. ROC analysis of the accuracy of noncycloplegic
ORCID retinoscopy, retinomax autorefractor, and suresight vision
screener for preschool vision screening. Invest Ophthalmol Vis
Massimo Accorinti http://orcid.org/0000-0003-0321-1455 Sci. 2011;52:9658–9664. doi:10.1167/iovs.11-8559.
18. Press SJ, Wilson S. Choosing between logistic regression and
discriminant analysis. J Am Stat Assoc. 1978;73(364):699–705.
References doi:10.1080/01621459.1978.10480080.
19. Grimes DA, Schulz KF. Refining clinical diagnosis with likelihood
1. Tsirouki T, Dastiridou A, Symeonidis C, et al. A focus on the ratios. Lancet. 2005;365(9469):1500–1505. doi:10.1016/S0140-
epidemiology of uveitis. Ocul Immunol Inflamm. 2018;26(1):2–16. 6736(05)66422-7.
doi:10.1080/09273948.2016.1196713. 20. Pleyer U. Chee SP: current aspects on the management of viral
2. Islam N, Pavesio C. Uveitis (acute anterior). BMJ Clin Evid. Apr 8, uveitis in immunocompetent patients. Clin Ophthalmol.
2010;2010:pii: 0705. 2015;9:1017–1028. doi:10.2147/OPTH.S60394.
3. Wakefield D, Chang JH. Epidemiology of uveitis. Int Ophthalmol 21. Wensing B, Relvas LM, Caspers LE, et al. Comparison of rubella
Clin. 2005;45:1–13. doi:10.1097/01.iio.0000155938.83083.94. virus- and herpes virus-associated anterior uveitis. Clinical man-
4. Chang JH, Wakefield D. Uveitis: a global perspective. Ocul ifestations and visual prognosis. Ophthalmology. 2011;118
Immunol Inflamm. 2002;10(4):263–279. doi:10.1076/ocii.10. (10):1905–1910. doi:10.1016/j.ophtha.2011.03.033.
4.263.15592. 22. Tugal-Tutkun I, Cimino L, Akova YA. Review for disease of
5. Nussenblatt RB, Whitcup SM. Uveitis. Fundamentals and Clinical the year: varicella zoster virus-induced anterior uveitis. Ocul
Practice. New York, NY: Mosby, Elsevier Inc; 2010. Immunol Inflamm. 2018;26(2):171–177. doi:10.1080/09273948.
6. Foster SC, AT V. Diagnosis and Treatment of Uveitis. New Delhi: 2017.1383447.
Jaypee Med Publ; 2013. 23. Accorinti M, Gilardi M, Pirraglia P, et al. Cytomegalovirus ante-
7. Chang JH, McCluskey PJ, Wakefield D. Acute anterior uveitis and rior uveitis: long-term follow-up of immunocompetent patients.
HLA-B27. Surv Ophthalmol. 2005 Jul-Aug;50(4):364–388. Graefe’s Arch Clin Exp Ophthalmol. 2014;252(11):1817–1824.
doi:10.1016/j.survophthal.2005.04.003. doi:10.1007/s00417-014-2782-4.
8. Wensing B, Mochizuki M, De Boer JH. Clinical characteristics of 24. Chee SP, Jap A. Presumed Fuchs heterochromic iridocyclitis and
herpes simplex virus associated anterior uveitis. Ocul Immunol Posner-Schlossman syndrome: comparison of cytomegalo
Inflamm. 2018;26(3):333–337. doi:10.1080/09273948.2017.142 virus-positive and negative eyes. Am J Ophthalmol. 2008 Dec;146
0806. (6):883–889. doi:10.1016/j.ajo.2008.09.001.