Posterior segment findings by spectral-domain optical coherence tomography and clinical associations in active toxoplasmic retinochoroiditis

Toxoplasmic retinochoroiditis is a common, potentially blinding parasitic infection. We sought to define the spectrum and frequency of signs of active toxoplasmic retinochoroiditis by spectral domain optical coherence tomography (SD-OCT), and to identify clinical associations. Ninety eyes of 90 individuals presenting consecutively to a tertiary referral uveitis service with active toxoplasmic retinochoroiditis and gradable SD-OCT scans were evaluated prospectively. SD-OCT features were collated, and associations with lesion location, primary versus recurrent episode, serological status, human immunodeficiency virus infection and best-corrected Snellen visual acuity were explored. Active toxoplasmic retinochoroiditis presented with thickened (65%) and hyperreflective (61%) retina, choroidal thickening (55%) and hyporeflectivity (61%), hyperreflective vitreous dots (80%) and deposits (36%), and posterior hyaloid thickening (35%) on SD-OCT. Most signs occurred with similar frequency across clinical groups. Retinal hyporeflectivity (17%) was significantly associated with a visual acuity of 20/200 or worse at resolution. Our observations demonstrate that active toxoplasmic retinochoroiditis has diverse SD-OCT signs and that none are universally present. Retinal hyporeflectivity—suggesting liquefactive necrosis—predicts poor visual outcome.

Toxoplasmic retinochoroiditis (TRC) is the most common disease caused by an infection with the parasite, Toxoplasma gondii 1 . There is striking variation in global prevalence of this disease. In the United States, TRC occurs in an estimated 21,000 individuals annually 2 . In South America, the infection is at least 10-times more common than in the United States 3 . Evaluation of donor eyes in eye banks across Brazil has identified TRC in 15% 4 . The diagnosis of active TRC is straightforward in cases with a "satellite lesion" of retinochoroiditis adjacent to a pigmented chorioretinal scar 5 . When there is no scar, however, the diagnosis is less obvious. In a study of 154 individuals with TRC, 28% presented with their first ocular infection and had no scars in either eye 6 . Serious visual sequelae result from TRC: the same study documented blindness in at least one eye in 24% of participants 6 .
Spectral domain optical coherence tomography (SD-OCT) is a well-established modality for imaging the structure of the posterior eye, with resolution of approximately 5 µm. A limited number of studies have used SD-OCT to identify signs of TRC. In the largest cohort study to date, 24 eyes of 24 patients with active TRC were examined, and common SD-OCT signs included posterior hyaloid thickening and retinal hyperreflectivity, thickening and disorganisation 7 . Smaller studies have included both active and healed lesions, and TRCassociated macular oedema and choroidal thickening [8][9][10][11][12][13] . Spectral domain OCT signs that indicate prognosis have not been reported.
We sought to identify signs of TRC on SD-OCT, to define frequency and determine differences between clinical subgroups (i.e. posterior pole vs. peripheral lesions, or primary vs. recurrent, or based on T. gondii and HIV serology results) and to identify signs that predicted visual outcome. We used SD-OCT to examine patients www.nature.com/scientificreports/ with TRC who presented consecutively over a 34-month period to a uveitis practice in an area of Brazil with high prevalence of toxoplasmosis.

Methods
Patient selection. From 16 . Zone 1 lesions extended to within 3000 µm of the fovea and/or 1500 µm from the optic disc margin. Zone 2 lesions were located between zone 1 and the equator. Zone 3 lesions were positioned anterior to the equator, and were excluded as SD-OCT was not possible.
Optical coherence tomographic imaging and interpretation. The posterior pole and lesion were imaged by color photography and SD-OCT, as permitted by media clarity. Spectral domain OCT scans were acquired using the SPECTRALIS HRA + OCT (Heidelberg Engineering, Heidelberg, Germany) and analyzed with Heidelberg Eye Explorer software version 6.9.4.0. Automatic real-time tracking mode was selected, with 24 frames per location, using volume scans of the macula and lesion (20 × 15 degree area formed by 19 B-scans, each separated by 242 µm), single slice acquisition, and enhanced-depth imaging (EDI) in some cases. An SD-OCT scan was considered gradable if the media were sufficiently clear to allow visualization of the posterior segment, as judged on resolution at the vitreoretinal interface. Patients whose presentation scans were not gradable were excluded from the study. Vitreous, posterior hyaloid, neural retina, retinal pigment epithelium and choroid were individually assessed, and findings were reported only for those scans in which the specific tissue or layer was well visualized. The SD-OCT segmentation was corrected manually. Measurements included: macular central subfield volume (1000 µm diameter), maximum lesion thickness (measured in the center of the lesion, from internal limiting membrane to base of retinal pigment epithelium, less subretinal fluid if present), and choroidal thickness under the maximum lesion thickness. Choroidal thickness was measured with the calliper tool of the SD-OCT software, and was considered thickened if it measured at least 300 µm. One eye per patient was analyzed. If both eyes had active lesions, the eye with the most posterior lesion was selected for analysis. If bilaterally active disease was symmetrical, the right eye was selected for analysis. Treatment and follow-up schedule. Patients with active TRC were offered treatment. Antimicrobial drug regimens included a course (6-7 weeks) of oral trimethoprim and sulfamethoxazole (160/800 mg twice daily) or oral sulfadiazine (1 g 4-times daily) and pyrimethamine (25 mg daily). Prednisone was prescribed on a case-by-case basis depending on the presence and intensity of posterior segment inflammation, and even in immunocompromised patients, at an initial dose of 20-60 mg/day with a 4-week taper. Intravitreal and intravenous anti-parasitic drugs or corticosteroids were not used. Unless the clinical situation required earlier review, patients were re-evaluated at 2 and 6-8 weeks after presentation. In selected cases, SD-OCT of the lesion and/or macula was repeated during follow-up. www.nature.com/scientificreports/

Results
Of a total of 262 subjects (344 eyes) with TRC, 90 individuals (90 eyes) qualified for inclusion in this study ( Table 1, Supplementary Fig. S1). Mean age at presentation was 37 years (median, 33; range, 12 to 84), and 44 participants (49%) were male. A total of 8 of 90 participants (9%) had bilateral active TRC. For 90% of persons (n = 81), timing of infection was unknown. Recurrent TRC was diagnosed in 55 participants (61%), and 35 (39%) had primary ocular disease. Lesions were located in zone 1 in 30 eyes (33%) and zone 2 in 60 eyes (67%). Fourteen people (16%) were T. gondii IgM-positive, and all these subjects were also T. gondii IgG-positive. Seven participants (8%) had evidence of HIV infection; the CD4-positive T-cell count was known for 6 of these 7 individuals, with range from 45 to 1253 cells/µl (mean, 447 cell/µl), and just one having a CD4-positive T-cell count below 200 cells/µl. None of these individuals had positive syphilis serology. Most participants were treated with trimethoprim and sulfamethoxazole (n = 63, 70%), or sulfadiazine and pyrimethamine (n = 12, 13%), plus oral prednisolone (n = 77, 85%). The mean duration of follow-up was 23.9 months (range, 1-30 months). At presentation, the majority of SD-OCT signs involved the retina ( Supplementary Fig. S2), which was thickened at the site of TRC relative to adjacent uninvolved retina in two-thirds of cases (n = 46/71, 65%) ( Table 2, Fig. 1). This was more common in zone 1 lesions, which were on average 201 µm thicker than zone 2 lesions (p < 0.01). Retinal hyperreflectivity (with the hyporeflective inner and outer nuclear layers becoming hyperreflective, resulting in full-thickness hyperreflectivity) was a presenting feature in 43 of 70 lesions (61%), and more likely in zone 1 lesions or primary disease (p < 0.05). Disorganization of the retinal layers was evident in 16 of 71 lesions (23%), and noted more often in zone 2 or recurrent disease (p < 0.05). Intraretinal hyperreflective dots were seen in 7 of 70 lesions (10%). Large outer retinal cystic spaces were present in 7 of 73 lesions (10%).

SD-OCT feature at presentation
Total/gradable scans www.nature.com/scientificreports/ www.nature.com/scientificreports/ was observed in 10 of 86 eyes (12%), and vitreomacular traction was an uncommon presenting feature (n = 3/86, 3%). Presenting SD-OCT findings in the vitreous included hyperreflective dots (n = 64/80 eyes, 80%), although these were not associated with lesion location, disease type, T. gondii IgM positivity and serologic evidence of HIV infection ( Table 2, Supplementary Table S1). The posterior hyaloid was attached at the lesion in 70 of 80 eyes (88%), and this was more common in primary TRC (p < 0.05). A partial posterior vitreous detachment was noted in 13 of 80 eyes (16%), and occurred more often in zone 2 TRC (p < 0.05). Posterior hyaloid thickening and large hyperreflective deposits at the vitreoretinal interface were present in approximately one-third of eyes (n = 28/80 and 29/80, respectively) at presentation, and there was an epiretinal membrane across 4 of 80 lesions (5%).
Two types of retinal destruction were visualized by SD-OCT. The most common form of necrosis presented with full-thickness retinal hyperreflectivity and thickening, and a well-demarcated border to adjacent normalappearing retina, with underlying choroidal hyporeflectivity (presumed to represent shadowing in the majority of cases) and hyperreflective dots in the vitreous (Fig. 2D-F, interpreted as coagulative retinal necrosis). Over time, the lesion progressively thinned, and became disorganized and less hyperreflective. Less commonly, retinal destruction presented as hyporeflective spaces or signal voids within the neural retina, and preservation of the retinal pigment epithelium and internal limiting membrane (Fig. 2G-L, interpreted as liquefactive retinal necrosis). There were relatively fewer hyperreflective dots in the vitreous, and hyporeflective retinal spaces were slowly replaced by disorganized retinal tissue in follow-up.

Discussion
In this large cohort of patients presenting with TRC, SD-OCT of the lesion frequently demonstrated thickened hyperreflective retina, choroidal thickening and hyporeflectivity, hyperreflective dots in the vitreous and posterior hyaloid thickening. Adjacent disorganization of the retinal layers was often seen. Over time, the retina thinned and became less hyperreflective, but did not return to a normal appearance, and there was retinal pigment epithelial thickening with choroidal hyporeflectivity, or epithelial atrophy with choroidal hyperreflectivity.
The necrotizing retinitis that defines active TRC clinically represents tissue destruction related to proliferation of T. gondii and the resultant inflammation 17 . Coagulative necrosis is the most common form of the process, characterized by preservation of tissue architecture. In TRC, this is likely represented as full-thickness retinal hyperreflectivity on SD-OCT. For 5% of eyes, the presentation SD-OCT showed hyporeflective retinal spaces or Table 3. Macular findings on SD-OCT in active TRC at presentation for 86 eyes (86 patients). Comparison was made between TRC lesions located in zones 1 and 2. The p values were calculated using Fisher's exact test or Student t-test (* p < 0.05, ** p < 0.01, *** p < 0.001). SD-OCT = spectral domain optical coherence tomography; TRC = toxoplasmic retinochoroiditis; RPE = retinal pigment epithelial/epithelium. www.nature.com/scientificreports/ signal voids with an intact internal limiting membrane. A similar finding was described in the case report of TRC imaged by SD-OCT, and was assumed to represent liquefactive necrosis 18 . While there are obvious difficulties in providing a retinal biopsy to correlate SD-OCT and clinical findings in TRC, the technique of OCT-mounted core biopsy in oncology has been used to confirm that liquefactive necrosis appears as hyporeflective tissue spaces 19 . Liquefactive necrosis probably occurs where there is heavy parasite proliferation with massive cell lysis, and profound tissue destruction. We observed hyporeflective retinal spaces in 17% of eyes over the course of TRC, and found this to be predictive of poor visual outcome. Subretinal or intraretinal fluid, usually adjacent to the lesion, was observed in approximately 15% of eyes presenting with active TRC, and was more common in primary disease. Published SD-OCT studies have identified subretinal fluid in as many as 43% of patients 20 . Two studies from the same group documented intraretinal www.nature.com/scientificreports/ and subretinal fluid in SD-OCT scans of TRC, although the authors did not explore clinical associations 10,11 . An unusual presentation of outer retinal oedema-"huge outer retinal cystic spaces" 11 -was identified in two of eyes in our series. This form of retinal oedema usually occurs at the macula, affects the outer plexiform, outer nuclear and photoreceptor layers, and is accompanied by subretinal fluid. Macular oedema was a presenting SD-OCT sign in approximately 20% of eyes; this was equally frequent in zone 1 and zone 2 lesions, highlighting that zone 2 TRC is not a benign process. Macular epiretinal membrane was noted during follow-up in 40% of eyes in our series. Vitritis is a common clinical feature of TRC, represented as hyperreflective dots in the vitreous cavity on SD-OCT. Large hyperreflective deposits (over 25 µm) were seen in the vitreous of one-third of eyes at presentation. This feature has been noted in both viral and parasitic retinal necrosis, and absorption of these deposits into the retina has been documented with both time domain OCT and SD-OCT 8,13,21,22 . The deposits may represent clumps of inflammatory cells and/or parasite cysts 23 . In one report of an individual with severe TRC, the authors found clusters of macrophages attached to the posterior hyaloid and internal limiting membrane in the enucleated globe 24 . Encysted T. gondii has been isolated from the vitreous of patients with TRC who required diagnostic vitrectomy 23,[25][26][27] . Cysts contain multiple dormant parasites, and measure 20 to 40 µm in diameter 26 .
Small round outer nuclear hyperreflective bodies (approximately 80 µm) was an unusual SD-OCT sign that we observed in two T. gondii IgM-positive patients. These bodies might represent encysted parasites, although histopathological reports indicate cysts are most commonly found in the ganglion cell layer. We otherwise found no distinction between SD-OCT features of T. gondii IgM-positive versus -negative TRC. Although differences between IgM-positive and -negative disease have been speculated, a study of post-mortem-enucleated globes with TRC showed IgM serology did not distinguish pathological features, including extension of necrosis and number of cysts 28 . Aggressive TRC has been associated with acquired immunodeficiency syndrome 29 . Our study suggested SD-OCT signs were not impacted by HIV infection status, although retinal pigment epithelial "bumps" and retinal arteriolar hyperreflective were identified in one eye of an HIV-positive patient and no other cases. These RPE bumps were observed in 12 of 13 cases of active TRC cases in a recent retrospective study 30 . Since only one HIV-positive patient had a CD4-positive T cell count below 200 cells/µl, it was not possible to evaluate associations between OCT signs and counts.
One previous study of SD-OCT in TRC specified 71% of gradable scans, which is comparable to our rate of 68% 7 . However, in contrast to this and other studies, we observed a relatively low prevalence of signs. As one example, rates of hyaloid thickening were 90-95% in other studies, compared to just 35% in our study 7,12 . As other examples, we recorded lower rates of large hyperreflective deposits and focal hyaloid attachment than were identified in these studies (i.e. 50-60% vs. 36%, and 50% vs. 16%, respectively) 7,13 . Our patient group was considerably larger than any other studied to date. In addition, we likely had little referral bias in this work; over 75% of the population uses public healthcare in Brazil, and thus the Uveitis Service at Ribeirão Preto General Hospital serves a comprehensive patient population 31 .
Recently, SD-OCT-angiography has been used for evaluation of active TRC, showing hyporeflective signals at the level of superficial and deep capillary retinal plexus, as well as in choriocapillaris and Sattler layers, corresponding to nonperfused or hypoperfused areas 20 . Partial blood flow recovery occurs in most cases following resolution of an active lesion 32 . Intraretinal vascular abnormalities 32,33 , retinal 32 and choroidal neovascularization 20 also are reported SD-OCT-angiographic features of TRC.
We present the largest prospective cohort of individuals with active TRC studied by SD-OCT. Given the challenge of diagnosing primary TRC in particular, one of our goals was to identify diagnostic features. A broad spectrum of SD-OCT signs was identified in patients with active TRC, but no sign was universal. All the posterior segment SD-OCT signs we observed have been reported in other forms of infectious and non-infectious uveitis 13,34,35 . Another goal was to identify prognostic signs: we have showed that hyporeflective retinal spaces -probably representing liquefactive necrosis -is associated with poor visual outcome. Future refinements in the capacity of OCT to resolve tissue architecture may identify additional signs to aid in the management of TRC. Table 4. Features on SD-OCT at any time during clinical course and visual outcomes in active TRC for 47 eyes of 47 patients, who were followed for at least 8 weeks and did not have a recurrence. Comparisons were made between eyes with good visual acuity (20/40 or better) and eyes with poor visual acuity (20/200 or worse). The p-values were calculated using Fisher's exact test (** p < 0.01). SD-OCT = spectral domain optical coherence tomography; TRC = toxoplasmic retinochoroiditis; ERM = epiretinal membrane.