We used a prospective cohort approach to address our main goal. The study was approved by institutional review boards of René Rachou Research Center, Oswaldo Cruz Foundation (CAAE no. 37614314.7.3001.5091), and Federal University of Minas Gerais (CAAE no. 37614314.7.3001.5149). All patients provided written informed consent.

We defined a case as illness in a person in the city of Gouveia with a history of fever, headache, lymphadenopathy, asthenia, or myalgia during February 12–May 18, 2015. From 5,276 local health charts, a task force comprised of municipal, state, and federal health authorities identified 201 persons suspected of meeting case definition criteria. Before confirmation of toxoplasmosis, differential diagnosis was made with consultation of an infectious disease specialist and serologic tests for dengue fever, visceral leishmaniasis, and leptospirosis.

We contacted the 201 persons with suspected toxoplasmosis. We were able to reach 151 (75.1%) persons whom we subsequently interviewed and tested for toxoplasmosis. We defined confirmed cases of acute toxoplasmosis as persons having T. gondii IgM and low avidity IgG on enzyme-linked fluorescence assay by using Vidas Toxo IgM, IgGII, and IgG avidity assays (bioMérieux, https://www.biomerieux.com).

Among the 151 suspected cases interviewed and tested, 52 (34.4%) had serologic evidence of T. gondii IgM and low avidity IgG, indicating acute toxoplasmosis. We performed a complete ophthalmic examination on each of the 52 case-patients, including assessment of best-corrected visual acuity (VA), applanation tonometry, slit-lamp examination (SLE; biomicroscopy), and indirect ophthalmoscopy. All case-patients also underwent multimodal imaging evaluation, including fundus photography, reflectances, fundus autofluorescence, and spectral-domain optical coherence tomography (SD-OCT). For patients with confirmed ocular involvement, we also performed fluorescein angiography. We determined prevalence and incidence of ocular changes and clinical characteristics on the basis of clinical and multimodal imaging findings. All patients with ocular involvement received standard therapy for 35–45 days, which consisted of sulfadiazine (1 g 4×/d), pyrimethamine (25–50 mg/d), folinic acid (7.5 mg/d), and prednisone (40–60 mg/d). One patient was allergic to sulfa and was switched from sulfadiazine to clindamycin 300 mg 4×/d. Complete blood counts were monitored at baseline and every 2 weeks during treatment.

For 3 years, we conducted follow-up examinations on case-patients at 5–8-month intervals and conducted the same ophthalmic examination protocol periodically. We reexamined case-patients with active primary or recurrent retinochoroiditis 3–6 weeks after therapy, or more often, if needed. We defined severe ocular involvement as binocular or macular involvement, or extensive necrotizing retinochoroiditis of >3 disk diameter (DD).

We prospectively collected and stored in an electronic database clinical and ophthalmological data, including symptoms, best-corrected VA, applanation tonometry, SLE, indirect ophthalmoscopy, and multimodal imaging. We assessed best-corrected VA by using an early treatment of diabetic retinopathy study chart and reported results in logarithm of the minimum angle of resolution (logMAR) scale. Previous studies suggested increased prevalence of ocular involvement among patients >30 years of age and among patients >50 years of age (9,10). We also analyzed age >40 years at time of infection as a potential risk factor for occurrence and severity of ocular involvement.

We defined the time of primary infection for each case-patient as the time of ocular or systemic symptom onset. To calculate time intervals among persons with new active retinochoroidal lesions during follow-up, we assumed eye disease occurred when patients first noted ocular symptoms. For case-patients without ocular symptoms, we assumed eye disease occurred when consistent retinochoroidal lesions were identified. Among case-patients who displayed new retinochoroidal scars during follow-up exams, we assumed eye disease occurred when ocular symptoms first were perceived; if the case-patient did not notice any ocular symptoms, we assumed eye disease occurred in the time between the prior ocular examination and identification of the scar. We also noted the first instance of retinochoroidal recurrences in either eye.

We performed statistical analyses by using R version 3.5.2 (21) by nonparametric methods and considered p<0.05 statistically significant. We used Mann-Whitney-Wilcoxon test to compare continuous variables, including age, length of follow-up, and VA at baseline. We used the mid-p exact test to compare proportions between subgroups, such as age >40 years, sex, and presence of underlying conditions and symptoms. We reported continuous variables as median (interquartile range [IQR]) and proportions as no. (%). We estimated the 95% CI of relative risks by using a maximum-likelihood estimator and described the follow-up by using the rate of recurrence per person-year (22), rate of recurrence per person-month, and Kaplan-Meier survival plot (23). We estimated survival probability and cumulative risk for ocular involvement and ocular recurrence by using the Kaplan-Meier method and compared results by using a log-rank test (24,25).

We investigated ocular involvement of 52 patients with serologically confirmed acute toxoplasmosis acquired in a presumed waterborne outbreak. We described clinical and multimodal imaging findings and determined the prevalence of retinochoroiditis, the incidence of recurrences and complications, and analyzed risk factors for ocular involvement.

In addition to the standard pattern of necrotizing retinochoroiditis, multimodal imaging revealed a distinct pattern of punctate retinal infiltrates (Figure 3), which might be overlooked if the retina is not examined carefully. This pattern also has been reported in neonates with congenital toxoplasmosis (5) and might represent the result of the primary parasite insult to the retina before a more robust immune response develops.

In this study, 12 (23%) patients displayed toxoplasmic retinochoroiditis at baseline examination; 14 (27%) had ocular involvement at some point during the study (Table 3), a rate of 16% per person-year. Variable rates of ocular involvement have been reported in toxoplasmosis outbreaks. For instance, risk for ocular involvement was 31% during the first 10.5 months after the Santa Isabel do Ivaí outbreak in Paraná, Brazil during 2001–2002 (15), and 21% after a mean follow-up of 114 weeks in an outbreak in Victoria, British Columbia, Canada in 1995 (13,14).

As expected, eye symptoms, particularly self-reported recently decreased VA and floaters, were associated with ocular involvement, suggesting that clinicians should inquire about symptoms routinely during and after an outbreak. Nevertheless, 2/12 (16.6%) patients had retinochoroiditis at baseline in the absence of symptoms, reinforcing the importance of examining the eyes of every patient with confirmed acute T. gondii infection.

Older age was frequently associated with a higher risk for ocular involvement at baseline and patients >40 years of age at the time of infection had a 5 times greater risk for retinochoroiditis than younger patients. Patients in the subgroup with ocular involvement were much older than patients in the subgroup without ocular involvement, consistent reports in other studies. In a study from the Netherlands, most patients with serologic evidence of recently acquired ocular toxoplasmosis were older, with a mean age of 50.6 years (26). A study from Brazil found age was a major risk factor for ocular involvement, with higher prevalence in patients >50 years of age, and »50% of patients >60 years of age had ocular involvement (10). Another study from Brazil found that persons >30 years of age with recently acquired T. gondii infection were more likely to have ocular involvement by the time of study enrollment (9).

At baseline examination, 75% (9/12) of patients with toxoplasmic retinochoroiditis in our study had severe ocular involvement, defined by large (>3 DD) necrotizing, macular, or bilateral retinochoroiditis (Figure 2). The median age of the 9 patients with severe ocular involvement (43 years of age) was higher than the median age of the 3 patients with less severe disease (31 years of age), but the differences were not statistically significant, probably because of the small number of patients in each subgroup. These findings also are in line with the literature (8), agreeing with 2 previous studies focusing on ocular toxoplasmosis in patients >50 years of age (median 67.5 years) in which 22/34 (64.7%) patients had severe disease, defined as either multiple active lesions, large lesions (>3 DD), or prolonged disease (duration >8 weeks) (27,28). Another possible explanation for the higher rate and severity of ocular involvement in our cohort is involvement of a virulent atypical T. gondii strain. However, this possibility remains elusive because parasite isolation for subsequent genotyping was not successful in the presumably contaminated water source, or in blood samples of patients with serologically confirmed infection (20).

Recurrence of retinochoroiditis was associated with bilateral ocular involvement at initial examination (Figure 4). One possible explanation is that patients with bilateral lesions at initial examination might have had a higher parasite load systemically and in the retina, leading to an increased risk for local reactivation. This finding also might have been associated with some degree of subclinical immune dysfunction in these cases.

The most common complication was epiretinal membrane development in 25% of eyes with retinochoroiditis, a finding consistent with a transversal study including 248 patients with acquired toxoplasmosis in India, which also reported development of epiretinal membranes at 25.1% (12). At the last follow-up examination 3/12 (25%) patients with retinochoroiditis had logMAR VA >0.3, one of them in both eyes. Complications underlying this visual impairment included unilateral macular involvement at baseline in 2 patients (VA 1.9) and other late complications, including 1 eye with an epiretinal membrane (VA 0.4), and the other with a rhegmatogenous retinal detachment (VA 0.8 after pars plana vitrectomy).

In addition to the 4 patients who developed recurrences of retinochoroiditis at 2, 9, 11, 12, or even at 22 months of follow-up, 2 patients without retinochoroiditis had new primary retinal lesions detected >34 months of follow-up (Figure 5). This finding highlights the need for long-term ophthalmologic follow-up in patients with postnatally acquired toxoplasmosis, regardless of previous ocular involvement.

This study has some limitations. We only considered symptomatic patients as having suspected cases of acute toxoplasmosis, and that might represent only a small portion of infected persons; several asymptomatic persons probably were not included, and this selection bias might justify, at least in part, the high prevalence of ocular changes. However, selecting for symptomatic cases has been the rule in most investigations of outbreaks of toxoplasmosis because serologic survey of all persons in affected areas is difficult. Finally, incidence calculations were based on few events, especially within subgroups of cases with ocular involvement, and some statistical tests were underpowered, such as the comparison of age in relation to severity of ocular involvement.

Despite the limitations, our study provides objective documentation of ocular changes in the context of a toxoplasmosis outbreak. By using multimodal imaging, we were able to characterize more subtle lesions, such as punctate retinochoroiditis among 33% of cases with ocular involvement. By comparison, previous reports of ophthalmic assessment of patients during outbreaks mostly were based on indirect ophthalmoscopy alone. Among 52 patients with confirmed infection, 47 (90%) had >6 months of follow-up; the median length for follow-up was 36 months, which enabled us to estimate the incidence of new lesions, recurrences, and complications during a relatively long period. We noted a substantial prevalence of early ocular involvement, with recurrences and new lesions occurring up to 39 months after infection. Thus, patients diagnosed with toxoplasmosis should receive long-term ophthalmic follow-up, regardless of initial ocular involvement. In addition, older patients had higher risk for ocular involvement, possibly reflecting age-related changes in the immune system, which could predispose persons >40 years of age to more severe disease.