Purpose
Endogenous fungal endophthalmitis (EFE) is a rare, sight-threatening intraocular infection with heterogeneous clinical presentations. Candida and Aspergillus species are the most common etiologic agents and may manifest with retinochoroiditis lesions that can be challenging to differentiate clinically. Accurate identification of the causative organism is essential for targeted antifungal therapy, anticipating disease progression, and optimizing outcomes. This study aimed to differentiate the etiology of EFE-associated retinochoroiditis using detailed clinical and optical coherence tomography (OCT) analysis.
Design
International, multicenter, retrospective comparative case series.
Methods
Demographic information, clinical characteristics, fundus photography, and OCT scans of patients with culture-proven Candida or Aspergillus EFE presenting with retinochoroiditis lesions were collected from eight tertiary-care centers worldwide. Lesion morphology (size, location, multifocality, and associated features) and OCT patterns (vitreal changes, vitreoretinal interface abnormalities, inner/outer retinal infiltration, and choroidal involvement) were compared between groups. Student’s t-tests were used for continuous variables and Fisher’s exact tests for categorical variables. Multivariable logistic regression and a random forest classifier were applied to identify the features most predictive of fungal species.
Results
Thirty-eight eyes of 30 patients (mean age: 64.7 ± 15 years) were included: 28 with Candida and 10 with Aspergillus EFE. Foveal involvement occurred only in Candida cases (28.6% eyes). Compared with Aspergillus, Candida EFE showed significantly more multifocal lesions ( P =.008), mid-peripheral/peripheral involvement ( P =.004), satellite lesions ( P =.001), and “string-of-pearls” vitreous exudates ( P =.05). Aspergillus eyes had larger lesions (2.4 vs. 1.2 disc-diameters; P =.001), more pre-/subretinal hemorrhage ( P =.03), and higher rates of occlusive vasculitis ( P =.008). On OCT, Candida eyes demonstrated more vitreous condensations/rain-cloud sign ( P =.03), preretinal aggregates ( P =.02), and intraretinal fluid (IRF) ( P =.04). Aspergillus infections more commonly exhibited full-thickness involvement with dense shadowing ( P =.001) and choriocapillaris alteration ( P =.008). Multivariable regression and random forest analysis identified lesion size, multifocality, satellite lesions, hemorrhage/vasculitis, IRF, and choriocapillaris alteration as the most discriminative features. These features allowed species differentiation with ∼85% accuracy using machine-learning classification.
Conclusions
Candida and Aspergillus EFE presenting with retinochoroiditis exhibit distinct features that allow reliable differentiation between these fungal etiologies. These OCT-based biomarkers may inform early organism-specific management while awaiting microbiological confirmation.
Introduction
E ndogenous fungal endophthalmitis (EFE) is a relatively uncommon but sight-threatening ocular condition. EFE occurs due to the spread of infection to the retinochoroid via the bloodstream. ,, EFE is a potentially blinding disease with generally poor visual outcomes. ,,, The predisposing risk factors in patients with EFE include immunosuppression, diabetes mellitus, abdominal surgery, indwelling catheters, renal disorders, liver disorders, cancers, intravenous drug usage, organ transplantation, hyperalimentation, neutropenia, among others. ,
Candida and Aspergillus species are the most common causative fungi for EFE. ,,, In the initial stages of infection, especially Candida EFE, the disease can be missed until progress to a sight-threatening condition, particularly when it starts from the peripheral retina and the choroid. EFE secondary to Aspergillus species is rare and typically results in poorer visual outcomes. , Both organisms can produce retinochoroiditis lesions characterized by vitreous inflammation and variably sized creamy white fluffy infiltrates that lead to necrosis of the retina, retinal pigment epithelium (RPE)-Bruch’s membrane complex, and possibly the choroid. The pattern of ocular involvement, the intraocular spread, biological behavior, and simultaneous systemic features may indicate the causative fungal agent. ,,, The sensitivity of the anti-fungal drugs varies according to the species involved, so early diagnosis and initiation of the most appropriate treatment are crucial for the prognosis of this sight-threatening disease. ,,
Fungal cultures from intraocular fluid samples remain the gold standard for establishing a definitive diagnosis. However, results often take several weeks, and isolation and culture of fungal organisms from ocular specimens, such as vitreous fluid, are challenging. Therefore, it is desirable to make a presumptive diagnosis with non-invasive imaging modalities so that early treatment with appropriate antifungal agents and prognostication can be initiated. Recently, optical coherence tomography (OCT) has emerged as a diagnostic (and in vivo histological) tool to differentiate various infective entities and obtain clues regarding the causative agent in conditions such as toxoplasmosis, viral retinitis and ocular tuberculosis. ,,,,,, OCT may also provide clues regarding the specific etiology of the EFE, guiding initial management until culture results are available.
In this multicenter study, we present the differentiating features of Candida and Aspergillus species based on detailed patterns observed in fundus imaging, especially OCT, with the aim of identifying distinctive characteristics that may aid in differential diagnosis.
MATERIALS AND METHODS
Study Subjects
The clinical records and OCT imaging of patients with culture-proven EFE with retinochoroidal lesions (blood culture or culture from intraocular fluids) were retrospectively reviewed. Deidentified data of patients from eight tertiary-care centers across different regions of the world (Supplementary Table A) were analyzed. The records of patients were reviewed from January 2018 to September 2025. Institutional ethics approval was obtained from the Institutional Review Boards of all participating centers, including the Cleveland Clinic Abu Dhabi Research Ethics Committee (REC) (#A-2022-050). The study adhered to the principles of the Declaration of Helsinki and complied with the regulations of the Health Insurance Portability and Accountability Act (HIPAA) of 1996. As this was a retrospective study, a waiver of informed consent was granted.
EFE was diagnosed as an infective intraocular inflammation affecting the inner coats of the eye due to the hematogenous spread of fungal organisms from distant foci. We included patients diagnosed with either Candida or Aspergillus EFE based on the isolation of organisms from culture of the blood or vitreous fluid. Only patients who presented with clinically visible retinochoroiditis lesions amenable to SD-OCT imaging were eligible. Patients of either gender (male/female), between the ages of 18-85 years were included in the study. The records were assessed for the clinical presentation, predisposing factors, SD-OCT findings, organisms isolated on cultures, specific anti-fungal therapies, and visual and functional outcomes. The exclusion criteria of the study were exogenous endophthalmitis or endogenous endophthalmitis due to unknown organisms or bacteria, patients with structural retinal abnormalities like retinal detachments, media opacities like vitreous hemorrhage and in whom regular follow-up was not available. Patients with fungal EFE lacking discernible posterior segment lesions, such as those with isolated anterior uveitis, panuveitis without retinal lesions, or endophthalmitis with no view of the posterior segment, were also excluded.
A detailed history taken from the medical records of all recruited patients, including visual acuity, any history suggestive of other ophthalmological diseases or pathology was noted. The visual acuity (VA) of the patients was noted using Snellen’s visual acuity chart at baseline and at follow-up visits. The visual acuity was converted to LogMAR for statistical analysis.
Image Analyses
Color fundus photography was used as a surrogate for clinical fundus examination in all included patients. Retinochoroiditis lesions associated with fungal EFE were evaluated for their morphological characteristics, including size (measured in disc diameters), location, laterality, appearance, multifocality, and the presence of satellite lesions. Lesion size was recorded based solely on the principal lesion, excluding any satellite foci. Additional associated findings, such as retinal hemorrhages, retinal vasculitis, vitreous inflammation, and optic nerve involvement, were also systematically assessed.
The primary objective of the study was to differentiate between Candida and Aspergillus fungal etiologies based on SD-OCT findings in the acute stage of the disease . The secondary objectives of the study were to assess the anatomical and functional outcomes of EFE. Therefore, OCT images (without the enhanced-depth imaging; EDI) were used to analyze the retinal findings, and EDI-OCT was used for determining the choroidal findings in EFE. The OCT scans passing through active lesions were analyzed. Given the retrospective and multicentric nature of the study, the pattern of the OCT and the size of the scan could vary among cases. The scans chosen were either 30 or 55 degrees and of sufficient quality to permit image analysis. The OCT images were analyzed by two independent experienced uveitis specialists (AA and NKM). Any discrepancies between their assessments were resolved by a third independent reviewer (AI).
Statistical Analysis
For statistical analysis, data were entered into Microsoft Excel and analyzed using Stata version 18 (StataCorp LLC). Continuous variables are presented as mean ± SD, and categorical variables as frequencies and percentages. Comparisons between the Candida and Aspergillus groups were performed using the Student’s t-test for continuous variables and the Fisher’s exact test for categorical variables, as appropriate. VA changes within each group (pre- vs post-treatment) were evaluated using the Wilcoxon signed-rank test. Comparisons of VA between groups were performed using the Mann-Whitney U test (non-parametric independent samples test). To identify independent predictors differentiating the two fungal species, multivariable logistic regression analysis was performed including variables with statistical significance in univariate analysis. Adjusted odds ratios with 95% CIs (CI) were reported. Additionally, a random forest classifier was applied to evaluate the relative importance of imaging features in distinguishing Candida from Aspergillus EFE. Given the limited sample size inherent to this rare condition, the random forest model was used primarily for feature importance ranking rather than for predictive deployment. Model performance was assessed by out-of-bag error rate and overall accuracy. A p -value <.05 was considered statistically significant.
RESULTS
Thirty-eight eyes of thirty patients (twelve women) fulfilled the inclusion criteria and were included in the study. Twenty-eight eyes had an infection secondary to Candida species, while ten eyes had Aspergillus infection. Bilateral retinochoroiditis occurred in 8 patients (26.7%), including 1 patient with Aspergillus EFE ( P =.37). Among the included patients, there were 19 Caucasians, 7 from the Middle East, and 4 from Asian Indian ethnicity. The mean age of all the patients was 64.7 ± 15 years (range: 30-84 years). The visual acuity at presentation was 1.2 ± 0.6 LogMAR units (range: 0.2-2.3 LogMAR units, corresponding approximately to Snellen 20/32-hand motions. The systemic predisposing factors for EFE, demographic features, and clinical findings are listed in Table 1 . To identify the most discriminative imaging features, we applied both multivariable regression and machine-learning classification.
Table 1
Baseline Demographic and Clinical Features of Patients with Fungal Endogenous Endophthalmitis Included in the Study
| Variable | All Subjects (n = 30, 38 eyes) | Candida (n = 21, 28 eyes) | Aspergillus (n = 9, 10 eyes) | P value |
|---|---|---|---|---|
| Age (years, SD) | 64.7 ± 15.1 | 65.6 ± 15.9 | 62.2 ± 13.3 | .28 |
| Gender (n, %) | ||||
| Male | 18, 60 | 12, 57.1 | 6, 66.7 | .70 |
| Female | 12, 40 | 9, 42.9 | 3, 33.3 | |
| Ethnicity (n) | ||||
| Caucasian | 18 | 17 | 1 | – |
| Middle Eastern | 8 | 3 | 5 | |
| South Asian/Indian | 4 | 1 | 3 | |
| Country (n) | ||||
| France | 3 | 3 | 0 | – |
| India | 5 | 3 | 2 | |
| Italy | 12 | 11 | 1 | |
| Spain | 2 | 1 | 1 | |
| United Arab Emirates | 8 | 3 | 5 | |
| Systemic Comorbidity (n) | ||||
| Cardiac disease | 10 | 7 | 3 | – |
| Gastrointestinal | 5 | 4 | 1 | |
| Genitourinary | 4 | 3 | 1 | |
| Renal failure | 2 | 2 | 0 | |
| Sepsis/infections | 4 | 2 | 2 | |
| Transplant | 3 | 2 | 1 | |
| Miscellaneous | 2 | 1 | 1 | |
| Diagnosis (n) | ||||
| Blood culture | 8 | 6 | 2 | – |
| Anterior chamber paracentesis | 5 | 5 | 0 | |
| Vitreous fluid | 18 | 11 | 7 | |
| Laterality (n, %) | ||||
| Unilateral | 22, 73.3 | 14, 66.7 | 8, 88.9 | .37 |
| Bilateral | 8, 26.7 | 7, 33.3 | 1, 1.1 | |
| Slit-lamp findings (eyes, %) | ||||
| Anterior chamber cells | 31, 81.6 | 23, 82.1 | 8, 80 | .99 |
| Keratic precipitates | 26, 68.4 | 19, 67.9 | 7, 70 | .99 |
| Hypopyon | 3, 7.9 | 2, 7.1 | 1, 10 | .99 |
| Vitreous cells/haze | 38, 100 | 28, 100 | 10, 100 | – |
| Total follow-up (months ± SD) | 10.3 ± 6.2 | 10.2 ± 6.3 | 10.5 ± 6.4 | .90 |
| Treatment (n, %) | ||||
| Pars plana vitrectomy | 21, 55.3 | 14, 50 | 7, 70 | – |
| Intravitreal antifungal | 17, 44.7 | 14, 50 | 3, 30 | |
| Systemic antifungal therapy | 38, 100 | 28, 100 | 10, 100 | |
| BCVA (LogMAR ± SD) | ||||
| At initial presentation | 1.2 ± 0.6 | 1.1 ± 0.5 | 1.4 ± 0.6 | .38 |
| At final follow-up visit | 1.0 ± 0.6 | 0.9 ± 0.6 | 1.2 ± 0.7 | .34 |
| P value | 0.12 | 0.08 | 0.72 |
Lesion Morphology
All eyes included in this series presented with fungal retinochoroidal lesions consistent with previously described characteristics in the literature. The retinochoroiditis lesions appeared as creamy-white or yellow-white, fluffy, necrotic, elevated plaque-like (or poorly defined) lesions involving the full thickness of the retina and underlying choroid, exhibiting feathery margins and variable vitreous inflammation. Overall, 18 eyes (47.3%) had lesions located in the posterior pole. Foveal involvement was observed in 8 eyes (21.1%) and showed a trend toward association with Candida EFE. Mid-peripheral and peripheral lesions were present in 26 eyes (68.4%), significantly more common in Candida EFE cases ( P =.004). Multifocal disease was significantly more frequent in Candida EFE (17 eyes) compared to Aspergillus EFE (1 eye; P =.008). Lesion size was larger in the Aspergillus group (2.4 ± 0.9 disc-diameters) than in the Candida group (1.2 ± 0.6 disc-diameters; P =.001). Additionally, satellite lesions surrounding the main lesion were seen in 23 eyes (82.1%) with Candida EFE but only in 1 eye (10%) with Aspergillus EFE ( P =.001). These findings indicate significant morphological differences between eyes affected by Candida vs Aspergillus EFE ( Figure 1 and 2 ).
