Candida Endophthalmitis: Chorioretinal Lesions Predict Worse Vision
Key Takeaways
- Among immunocompetent patients followed for 12 months after this infusion-related Candida albicans endogenous endophthalmitis outbreak, chorioretinal involvement was linked to more baseline structural macular damage and worse 12-month visual acuity than intraretinal or preretinal involvement.
- Secondary MNV was reported in all evaluable chorioretinal eyes and in none of the intraretinal or preretinal eyes, a clear reported between-group difference in delayed neovascular complications.
- Postoperative complications were reported far more often in eyes with chorioretinal involvement than in eyes with intraretinal or preretinal involvement.
- In the chorioretinal subgroup, RPE lesion area expanded early and then contracted by month 12 even as BCVA improved.
Investigators at Ningde Municipal Hospital reported a Candida endogenous endophthalmitis longitudinal cohort, a retrospective, single-center, longitudinal study with 12 months of follow-up after an infusion-related Candida albicans outbreak in immunocompetent patients. The cohort comprised 26 eyes overall, with 18 initially classified as chorioretinal and 8 as intraretinal/preretinal; 13 chorioretinal and 8 intraretinal/preretinal eyes were included in the 12-month visual and complication analyses, and 10 chorioretinal eyes were evaluable for longitudinal retinal pigment epithelium (RPE) and macular neovascularization (MNV) imaging. Lesion-pattern classification combined perioperative findings with postoperative day-1 optical coherence tomography (OCT) because dense media opacity often limited reliable preoperative imaging. All eyes underwent pars plana vitrectomy (PPV), vitreous sampling, intravitreal voriconazole, and standardized systemic voriconazole, with follow-up at baseline and months 3, 6, 9, and 12. Outcomes centered on best-corrected visual acuity (BCVA), complications, optical coherence tomography angiography (OCTA)-defined MNV, longitudinal RPE lesion-area change, and generalized estimating equation (GEE) modeling of visual trajectories.
Baseline structural macular disruption was concentrated in the chorioretinal group and was absent in the intraretinal/preretinal group. Best-corrected visual acuity improved over time in both groups, but 12-month median BCVA remained worse in chorioretinal eyes at 0.80 versus 0.20 LogMAR (P = 0.005). Across visits, GEE modeling also favored the intraretinal/preretinal group (beta = -0.880; 95% CI, -1.437 to -0.323; P = 0.002). The magnitude of BCVA improvement itself did not differ significantly between groups.
Postoperative structural sequelae also separated the groups. In the OCTA-defined secondary MNV outcomes analysis, at least one complication occurred in 92.3% of chorioretinal eyes versus 25% of intraretinal/preretinal eyes (P < 0.001), and secondary MNV developed in 10/10 evaluable chorioretinal eyes (100%; 95% CI, 69.2%-100%) versus 0/8 intraretinal/preretinal eyes (0%; 95% CI, 0%-36.9%; P < 0.001). That MNV signal came from a small evaluable chorioretinal subset. MNV was the most frequent vision-threatening complication; active lesions were treated with conbercept, inactive networks were observed, and treated active lesions stabilized without further retreatment during follow-up. In the chorioretinal subgroup, RPE lesion area increased through month 3 and then decreased by month 12 while BCVA continued to improve, and early RPE area did not show a clear monotonic relationship with later vision.
The findings come from a small retrospective single-center outbreak cohort, so they describe association rather than causation and may not generalize beyond this immunocompetent, infusion-related setting in China. Because classification partly relied on postoperative day-1 OCT, the authors note that some outer-retinal or RPE disruption used to define chorioretinal involvement could theoretically have been influenced by surgical manipulation, and excluding eyes with severe baseline destruction, retinal detachment, central retinal artery occlusion, or loss to follow-up may have biased longitudinal estimates. Longitudinal retinal and choroidal thickness as well as macular edema volume were not quantified because acute-phase segmentation was unreliable. The universal MNV pattern was observed in a small evaluable subset and carried wide confidence bounds. The authors interpreted early RPE expansion despite improving vision as tissue remodeling, lesion demarcation, scar contraction, and clearer visualization of damage rather than proof of persistent infection.
The authors concluded that initial lesion morphology functioned as an anatomic marker of 12-month prognosis in this outbreak cohort. Chorioretinal involvement tracked with poorer final vision, heavier postoperative complication burden, broader RPE remodeling, and markedly greater secondary MNV risk. Overall, deeper initial structural involvement marked a less favorable long-term course after treatment.
Clinician Questions
How were chorioretinal and intraretinal/preretinal lesions distinguished in Candida endogenous endophthalmitis?
In this Candida endogenous endophthalmitis cohort, chorioretinal involvement was defined by OCT evidence of outer retinal, RPE, or choroidal disruption, subretinal extension, and/or chorioretinal scarring, whereas intraretinal/preretinal involvement was limited to the inner retina or vitreoretinal interface with preserved outer retinal and RPE structures after surgical removal of fungal material. Classification combined perioperative findings with postoperative day-1 OCT because dense media opacity often prevented reliable preoperative OCT.
Why were only 10 chorioretinal eyes included in the longitudinal RPE and MNV analysis?
Thirteen chorioretinal eyes completed the 12-month visual and complication follow-up, but one ungradable peripheral lesion, one severe retinal detachment, and one central retinal artery occlusion prevented longitudinal RPE and MNV assessment, leaving 10 evaluable chorioretinal eyes. That smaller denominator is the basis for the reported MNV estimate in the chorioretinal subgroup.
How was secondary MNV managed after Candida endogenous endophthalmitis in this cohort?
Secondary MNV was identified with OCTA, and lesions were treated only when structural OCT showed active exudative features. Active MNV received intravitreal conbercept 0.5 mg/0.05 mL as 3 monthly loading injections followed by PRN retreatment, whereas inactive MNV without exudation was monitored without intervention. During the 12-month follow-up, treated active lesions did not require additional PRN injections.
Does early enlargement of RPE loss after Candida endogenous endophthalmitis imply persistent infection?
The authors interpreted early enlargement of RPE lesion area despite improving BCVA as tissue remodeling, lesion demarcation, scar contraction, and clearer visualization of damage as the optical media cleared rather than proof of ongoing infection. This was presented as an interpretation of the imaging course, not as an established mechanism.