Purpose
To determine the incidence of, and risk factors for, ocular involvement among people known to have postnatally acquired Toxoplasma gondii infection in a region of southern Brazil where there is a high prevalence of endemic disease.
Design
Retrospective longitudinal cohort study.
Methods
Records of 302 patients with serologic evidence of recent T gondii infection (a positive anti– T gondii IgM antibody test) from Erechim, Rio Grande do Sul state, Brazil (1974–2002) were analyzed. The incidence of ocular involvement was calculated in terms of person-years (PY) of follow-up. Risk factors for ocular involvement were analyzed using log-rank and Fisher exact tests.
Results
At initial ocular examination (baseline), 30 patients (9.9%) had intraocular inflammation only (anterior chamber cells and flare, vitreous inflammatory reactions, retinal whitening), without clinically apparent necrotizing retinochoroiditis. At baseline, men were more likely to have ocular involvement ( P = .043) and antiparasitic treatment was associated with less ocular involvement ( P = .015). Follow-up examinations were performed on 255 patients (median follow-up, 13.7 months [range 0.4–261.9 months]). Among those without ocular involvement at baseline, the incidence of necrotizing retinochoroiditis was 6.4/100 PY. Patients >40 years of age at first IgM test had a greater risk of incident necrotizing retinochoroiditis (hazard ratio = 4.47, 95% CI = 1.67–11.93, P = .003) than younger patients. The incidence of recurrent necrotizing retinochoroiditis was 10.5/100 PY.
Conclusion
Isolated intraocular inflammatory reactions can be an initial manifestation of T gondii infection, with necrotizing retinochoroiditis occurring months or years later. Male sex and older age are risk factors for toxoplasmic retinochoroiditis. Antitoxoplasmic treatment may protect against early ocular involvement.
Current evidence suggests that postnatally acquired Toxoplasma gondii infection is responsible for the majority of ocular toxoplasmosis cases. Sources of infection include ingestion of tissue cysts in raw or undercooked meat and ingestion of oocysts in soil, water, or food contaminated with feces of infected cats. The incidence and characteristics of ocular lesions related to recent postnatally acquired toxoplasmosis have been described during outbreaks, but information about early ocular involvement, course of disease, and risk factors associated with ocular toxoplasmosis from sporadic, postnatally acquired T gondii infection is limited because of low prevalence in most regions and the fact that the time of initial T gondii infection typically is not known. Nonocular T gondii infection is often asymptomatic, and clinically apparent retinal lesions may first develop years after systemic infection; these factors preclude the determination of when most people with endemic ocular toxoplasmosis first became infected.
In Erechim, a city located in an agricultural region of Rio Grande do Sul, the southernmost state of Brazil, up to 85% of the population is infected with T gondii and 17.7% of infected individuals have ophthalmic findings consistent with ocular toxoplasmosis. In this study, we took advantage of the high prevalence of T gondii infection and associated ocular toxoplasmosis in Erechim to study the ophthalmic features of postnatally acquired T gondii infection.
Methods
In this retrospective longitudinal cohort study, we used a dataset created by 1 author (C.S.) in 2002; it includes all patients older than 1 year of age, with at least 1 positive test for anti– T gondii IgM antibodies, who were examined between 1974 and 2002 by 1 ophthalmologist (author C.S.) at Clínica Silveira in Erechim, Rio Grande do Sul, Brazil. Patients either had been referred to Clinica Silveira because of a positive IgM antibody test ordered by a non-ophthalmologist for reasons other than eye disease, or had undergone IgM antibody testing as evaluation of presumed toxoplasmic retinochoroiditis without evidence of prior infection that was identified at Clinica Silveira. The specific reason that IgM testing was ordered for those patients referred to Clinica Silveira with positive tests was not recorded. A known reason for such testing is screening of women during pregnancy, especially for those who have lymphadenopathy and other constitutional signs and symptoms, because of high prevalence of T gondii infection in the general population. Individuals with positive anti– T gondii IgM tests are routinely referred by internists or obstetricians for ocular examination, whether or not they have visual symptoms, because of the high prevalence of ocular involvement among people with T gondii infection in that area. Not included in the dataset were individuals with acquired immunodeficiency syndrome (AIDS) or other immune system diseases and those receiving immunosuppressive drugs (other than corticosteroids for ocular toxoplasmosis). This study was approved by the Institutional Review Board (IRB) of the Universidade Federal de São Paulo (Comitê de Ética em Pesquisa da Universidade Federal de São Paulo–UNIFESP/EPM) prior to retrospective data collection, and analysis of previously collected, de-identified data was approved by the IRBs at the University of California, Los Angeles and the United States Centers for Disease Control and Prevention.
Anti– T gondii IgM antibody tests were performed at the Laboratório Fleury (São Paulo, Brazil), using either an indirect immunofluorescence assay or a microparticle enzyme immunoassay (Abbot AxSYM; Abbott Laboratories, Abbott Park, Illinois, USA). For the immunofluorescence assay, a titer ≥1:16 was considered positive. For the immunoenzymatic assay, values ≤0.499 were considered negative; values between 0.500 and 0.599 were considered indeterminant; and values ≥0.600 were considered positive. For the immunoenzymatic assay, the reported test sensitivity is 96.3% and specificity is 99.8% (package insert for Abbot AxSYM anti– T gondii IgM antibody assay), although sensitivity and specificity of the test has not been determined specifically for the population studied. The aforementioned values pertain to published test standards at the time of data collection. Although test kits may have changed with different cut-off values during the 28-year period of data collection, recording of positive results was always in reference to contemporary cut-off values.
Data Collection
The following demographic and medical data were collected: age at time of first anti– T gondii IgM antibody test; age at first identification of ocular involvement related to T gondii infection (intraocular inflammatory reactions, necrotizing retinochoroiditis, retinochoroidal scars); age at diagnosis of recurrent toxoplasmic retinochoroiditis; sex; presence of signs or symptoms of nonocular toxoplasmosis at time of IgM testing; and use of systemic anti– T gondii treatment at or before baseline. The following information was collected for each involved eye at each examination: presence or absence of intraocular inflammatory reactions (anterior chamber cells and flare, vitreous inflammatory reactions, retinal vascular sheathing, or focal retinal whitening without clinical evidence of retinal necrosis); presence or absence of necrotizing retinochoroiditis; and presence or absence of retinochoroidal scars. In a previous publication, the phenomenon of isolated focal retinal whitening in people with serologic evidence of recent T gondii infection has been shown to resolve without clinically apparent scar formation. We have hypothesized that lesions represent foci of retinal infiltration where T gondii tissue cysts have colonized the retina. In contrast to a study of epidemic disease, retinal whitening was not categorized separately from other intraocular inflammatory reactions. For eyes with necrotizing retinochoroiditis, the size of the largest lesion (<1 optic disc area [da] vs ≥1 da) and the presence or absence of macular and foveal involvement were determined. Data on intraocular pressure and visual acuity were not analyzed, as these were not relevant to the purpose of this study.
Conventions and Definitions
Baseline was defined as the date of the first eye examination that was performed either within 3 months before or at any time after the first positive anti– T gondii IgM antibody test. Ocular involvement that occurred within the window of 3 months before or 3 months after the first positive anti– T gondii IgM antibody test was considered to be “immediate” for purposes of the study. We used the same definitions of terms and study conventions that were used in our study of epidemic T gondii infection. On the basis of these definitions, we categorized ophthalmic findings further using 1 or more of the following terms: necrotizing retinochoroiditis (denoting active disease); retinochoroidal scars consistent with healed T gondii retinal infection; isolated intraocular inflammatory reactions; initial necrotizing retinochoroiditis; recurrent necrotizing retinochoroiditis; primary lesions; satellite lesions; and first incident ocular disease.
Data Analysis and Statistical Techniques
Incidence of ocular involvement was calculated in terms of events per 100 person-years (PY) of follow-up. For comparison of outcomes between those with and those without intraocular inflammatory reactions at baseline, Time 0 was defined as the baseline date. For other longitudinal analyses, Time 0 was defined as the date of first positive anti– T gondii IgM antibody test. Primary necrotizing retinochoroiditis lesions that occurred 6 months or longer after first positive anti– T gondii IgM antibody test were considered incident lesions, whether or not a prior eye examination had been performed. If primary necrotizing retinochoroiditis lesions were identified on a baseline examination that was performed between 3 and 6 months after the first positive anti– T gondii IgM antibody test, we felt that they could not be categorized reliably as being immediate or incident, and they were excluded from some analyses. For individuals with new retinochoroidal scars during follow-up, ocular disease was assumed to have occurred when the scar was identified, for purposes of calculating intervals.
Because only women were screened routinely for anti– T gondii IgM antibodies during pregnancy, there was the possibility of ascertainment bias, in which more women without disease were examined than men without disease. To address this issue and to investigate whether it was likely to influence relationships between sex or age and ocular disease, we compared men and women for the following factors: age at baseline; percent with ocular involvement at baseline; percent with systemic disease, but no ocular involvement at baseline; and treatment at baseline. Because of the possibility that age at presentation could vary by indication for examination, we also compared the percent on treatment at baseline between younger (≤40 years of age) and older (>40 years of age) patients.
Statistical analysis was performed using SAS software version 9.3 (SAS, Inc, Cary, North Carolina, USA). Cumulative risk of ocular involvement was estimated using the Kaplan-Meier method and compared using the log-rank test. Relative risks were expressed as hazard ratios (HR), estimated from Cox proportional hazards regression models. The Fisher exact test was performed to evaluate relationships between categorical variables, and the student t test was used in the analysis of continuous variables. A P value of <.05 was considered to be statistically significant.
Results
A total of 302 patients met inclusion criteria. At baseline, 5 patients had inactive retinochoroidal scars, suggesting remote healed disease; these patients were excluded from analysis. The majority of patients (n = 239, 79.1%) had been referred for ocular evaluation because of a positive anti– T gondii IgM antibody test; for the other 63 patients (20.9%), the IgM test had been ordered by the examining ophthalmologist during investigation of initial primary necrotizing retinochoroiditis lesions. Ocular examination was performed within 3 months of, or at the same time as, the first positive anti– T gondii IgM antibody test in 258 patients (85.4%). Follow-up examinations after first positive anti– T gondii IgM antibody test were performed on 255 patients (median follow-up: 13.7 months [range 0.4–261.9 months]). During follow-up, persistence of anti– T gondii IgM antibodies for 1 year or longer on repeat testing was known to have occurred in 25 patients; they did not differ in demographic or clinical characteristics from the rest of the study population (data not shown).
Table 1 lists demographic, medical, and ophthalmic characteristics for the studied population. Isolated intraocular inflammation was present in 30 patients (9.9%) at baseline; inflammation was unilateral in 27 patients. In all but 2 patients, intraocular inflammation was observed within 3 months of the first positive IgM test (5 months in 1 case; 8 months in another). Women were more likely than men not to have clinically apparent systemic disease or ocular involvement at baseline (25.7% [38 of 148 patients] vs 5.0% [6 of 120 patients], respectively, P < .001), as might be expected because of serologic screening during pregnancy. There were no significant differences, however, between men and women for the following factors at baseline: percent with ocular involvement; percent with systemic disease, but no ocular involvement; and treatment ( Supplemental Table , available at www.ajo.com ). With regard to age at baseline, men were slightly younger than women overall (mean, 18.3 ± 15.2 years vs 24.5 ± 13.3 years, respectively, P = .0002), but the difference was unlikely to be clinically important in terms of disease characteristics. There was no significant difference between the proportion of men and women in the older age group at baseline ( Supplemental Table , available at www.ajo.com ). There was no significant difference between younger and older patients with respect to proportion treated at baseline (55.2% [144 of 261 patients] vs 48.4% [15 of 31 patients], respectively, P = .568).
| Characteristics | Value |
|---|---|
| Sex, n (%) | |
| Male | 129 (42.7%) |
| Female | 173 (57.3%) |
| Age at baseline (y) | |
| Mean ± SD | 21.9 ± 14.4 |
| Median (range) | 21 (1–63) |
| Clinically apparent nonocular toxoplasmosis at baseline, a n (%) | 183 (68.3%) (total n = 268) b |
| Ophthalmic findings at baseline, n (%) | |
| Active disease | 134 (44.4%) |
| Isolated intraocular inflammation c | 30 (9.9%) |
| Initial, primary necrotizing retinochoroiditis | 104 (34.4%) |
| Retinochoroidal scar only | 5 (1.7%) |
| Antitoxoplasmic treatment at or before baseline, n (%) | 159 (54.5%) (total n = 292) b |
| Follow-up data available after infection (n = 297), d n (%) | 255 (85.9%) |
| Duration of follow-up after infection (mo) | |
| Mean ± SD | 33.7 ± 49.8 |
| Median (range) | 13.7 (0.4–261.9) |
a Arthralgia, fatigue, fever, malaise, lymphadenopathy, sore throat, or a combination of these disorders at the time of positive anti– T gondii IgM antibody testing.
b Number of individuals for whom values were known, if different than 302.
c Anterior chamber cells, retinal vitreous humor cells or haze, retinal vascular sheathing, retinal infiltrates without retinal necrosis, or a combination of these findings in the absence of necrotizing retinochoroiditis.
Table 2 lists the characteristics of retinal involvement at baseline and during follow-up. Lesions that developed during follow-up presented characteristics similar to those seen at baseline. Table 3 shows the comparison of selected characteristics between patients with and those without ocular involvement at baseline. There were significantly more men than women with ocular involvement at baseline ( P = .043). The use of antitoxoplasmic medications before or at baseline was associated with less ocular involvement at initial examination ( P = .015); however, we could not confirm that such treatment prevented incident necrotizing retinochoroiditis lesions during follow-up (HR: 0.71, 95% confidence interval [CI] = 0.34–1.50, P = .37).
| Lesion Characteristic | Necrotizing Retinochoroiditis | ||
|---|---|---|---|
| Immediate Lesions a (N = 97) d | Incident Primary Lesions b (N = 18) d | Recurrent Disease c (N = 31) d | |
| Laterality, n (%) e | |||
| Unilateral | 87 (89.7%) | 15 (83.3%) | 30 (96.8%) |
| Bilateral | 10 (10.3%) | 3 (16.7%) | 1 (3.2%) |
| Location in either eye, n (%) e | N = 15 f | ||
| Extramacular only | 49 (49.5%) | 7 (46.7%) | 15 (48.4%) |
| Macular | 48 (50.5%) | 8 (53.3%) | 16 (51.6%) |
| Foveal involvement g | 32 (68.1%) | NA | NA |
| Size of largest lesion, either eye, n (%) e | N = 95 f | N = 15 f | |
| ≥1 DA | 43 (45.3%) | 4 (26.7%) | 13 (41.9%) |
| <1 DA | 52 (54.7%) | 11 (73.3%) | 18 (58.1%) |
| Number of lesions per eye, h n (%) e | N = 96 f | N = 15 f | N = 30 f |
| Unifocal | 81 (84.8%) | 13 (86.7%) | 28 (93.3%) |
| Multifocal | 15 (15.6%) | 2 (13.3%) | 2 (6.7%) |
a Primary necrotizing retinochoroiditis on initial eye examination, if the initial eye examination was prior to or at the same time of the first positive anti– T gondii IgM test (n = 87), or on initial eye examination performed within 3 months after the first positive anti– T gondii IgM antibody test (n = 10). Not considered as immediate lesions in this analysis were 7 of 104 initial primary necrotizing retinochoroiditis lesions at baseline examinations, for the following reasons. Patients with initial primary necrotizing retinochoroiditis lesions on baseline examinations performed 3–6 months after the first positive anti– T gondii IgM antibody test (n = 2) were excluded because these lesions could not be classified as either immediate or incident lesions. Patients with initial primary necrotizing retinochoroiditis lesions on baseline examinations performed more than 6 months after the first positive anti– T gondii IgM+ antibody test (n = 5) were considered incident lesions, as described in the text.
b Necrotizing retinochoroiditis on examination after the first positive anti– T gondii IgM antibody test, if additional, previous ophthalmic examinations had been performed since the first positive anti– T gondii IgM antibody test and revealed no evidence of retinochoroidal involvement (n = 13; 10 had no ocular involvement at baseline and 3 had intraocular inflammatory reactions at baseline); or an initial primary necrotizing retinochoroiditis lesion that was identified more than 6 months after the first positive anti– T gondii IgM antibody test (n = 5), whether or not there had been prior ophthalmic examinations.
c Recurrent disease was defined as necrotizing retinochoroiditis in the presence of old retinochoroidal scars in either eye. Data presented refers to the first recurrence.
d Number of individuals with necrotizing retinochoroiditis lesions.
e Number of individuals with characteristic among those with necrotizing retinochoroiditis in each column.
f Denominator, if less than all individuals with necrotizing retinochoroiditis for a given column.
g Among those with macular lesions; presence or absence of foveal involvement known for 47 of 48 macular lesions, among those with immediate necrotizing retinochoroiditis.
h Based on eye with largest number of lesions; 1 eye per patient.
| Immediate Eye Disease a (N = 125) | No Immediate Eye Disease (N = 169) | P Value b | |
|---|---|---|---|
| Age at baseline (y) | .19 c | ||
| Mean ± SD | 23.1 ± 15.1 | 20.9 ± 13.8 | |
| Median (range) | 22 (1–63) | 19 (1–60) | |
| Age groups | .12 | ||
| 1–40 years | 108 (86.4%) | 156 (92.3%) | |
| >40 years | 17 (13.6%) | 13 (7.7%) | |
| Male sex, n (%) d | 63 (50.4%) | 64 (37.9%) | .043 |
| Clinically apparent nonocular toxoplasmosis, e n (%) d (N = 261) | 80 (70.8%) (n = 113) f | 101 (68.2%) (n = 148) f | .69 |
| Antitoxoplasmic treatment at baseline, n (%) d | 53 (45.3%) (n = 117) f | 101 (60.5%) (n = 167) f | .015 |
| Follow-up data available, n (%) d | 114 (91.2%) | 122 (72.2%) | <.001 |
| Follow-up (mo) | .40 g | ||
| Mean ± SD | 42.9 ± 60.8 | 22.4 ± 30.4 | |
| Median (range) | 14.1 (0.6–261.9) | 10.9 (0.5–157.8) |
a Primary necrotizing retinochoroiditis or isolated intraocular inflammation on initial eye examination, if the initial eye examination was within a 3-month window prior to or 3 months after the first positive anti– T gondii IgM antibody test (n = 97 necrotizing retinochoroiditis and 28 isolated intraocular inflammation).
b Fisher exact test, except when noted otherwise.
d Number of individuals with characteristic among all individuals for each column.
e Arthralgia, fatigue, fever, malaise, lymphadenopathy, sore throat, or a combination of these disorders at the time of positive anti– T gondii IgM antibody testing.
f Denominator if less than all individuals for each column.
Necrotizing retinochoroiditis after first positive IgM test developed in 29 individuals (either primary lesions at baseline or incident lesions during follow-up), with an incidence 10.0/100 PY. The cumulative risk of necrotizing retinochoroiditis at longest follow-up after T gondii infection for those at risk (214.9 months) was 64.5% ( Figure 1 ), but the 95% CI was broad (43.9%–84.4%). Based on Kaplan-Meier analysis, the period of greatest risk for necrotizing retinochoroiditis was during the interval 0–48 months after T gondii infection; cumulative risk of necrotizing retinochoroiditis at the end of this interval was 49.9% (95% CI, 41.2%–59.3%). To allow comparison to a study of epidemic disease, we also calculated the cumulative risk of necrotizing retinochoroiditis at 10.5 months after T gondii infection in this study (38.5%; 95% CI, 32.9%–44.7%).
The incidence of necrotizing retinochoroiditis after baseline for all individuals without such lesions at baseline was 6.8/100 PY. The incidence for those with isolated intraocular inflammation at baseline (9.0/100 PY; 3 patients with incident lesions) was not statistically different than the incidence for those without any previous ocular involvement (6.4/100 PY; 10 patients with incident lesions; HR = 1.63, 95% CI, 0.44–6.05, P = .46; Figure 2 ). The incidence of first recurrences among patients with necrotizing retinochoroiditis at baseline or during follow-up was 10.5/100 PY ( Figure 3 ). Among 31 patients with recurrences, 19 (61%) had satellite lesions and 12 (39%) had primary recurrent lesions.
Patients without retinal lesions at the time of first positive IgM test who were older than 40 years were at higher risk of developing retinal lesions during follow-up than were younger age groups (HR = 4.47, 95% CI = 1.67–11.93, P = .003; Figure 4 ). Table 4 shows the comparison of selected host factors and retinal lesion characteristics. Older patients (>40 years) had larger retinal lesions ( P = .044), but no other associations were identified. In 3-way comparisons (ages 1–20 years; 21–40 years; >40 years), statistical differences were not identified when considering either risk for development of retinal lesions or the size of lesions; findings were similar for the 2 youngest age groups in each assessment (data not shown).
