Ophthalmologic Evaluation and Clinical Outcome in Candidemia: A 7-Year Retrospective Multicenter Cohort Study

Abstract

Background

Ocular candidiasis (OC) is a notorious complication of candidemia, but guidelines on fundoscopic screening are inconsistent. We aimed to determine the frequency of OC in patients with candidemia who underwent fundoscopy, regardless of visual symptoms, and to evaluate its impact on antifungal treatment selection and duration.

Methods

A retrospective cohort study was performed at 2 tertiary care hospitals in the Netherlands in adult patients with at least 1 positive blood culture with Candida species between January 2018 and December 2024. Primary outcomes were the incidence of OC (defined as proven or probable chorioretinitis or endophthalmitis), persistent vision loss, and changes in the choice and/or duration of antifungal therapy in patients with OC.

Results

Of 402 patients, 307 (76.4%) underwent fundoscopy. Ocular candidiasis was diagnosed in 15 of 307 patients (4.9%), including 12 with probable chorioretinitis (3.9%) and 3 with probable endophthalmitis (1.0%). Nine of the 15 patients (60%) were asymptomatic (n = 6) or unable to report symptoms (n = 3). In all 15 patients, the fundoscopic findings resulted in a change in therapy. This consisted of therapy prolongation (n = 15), addition of (n = 7) or switch to (n = 4) an azole, and intravitreal antifungal injections (n = 2). Persistent visual impairment of any degree occurred in 3 of 15 patients (20%), all of whom were initially symptomatic.

Conclusions

Ocular candidiasis is a relatively rare but clinically significant complication of candidemia. Fundoscopic findings are used to guide treatment decisions. Persistent vision loss was uncommon.

Ocular candidiasis occurred in 4.9% of candidemia patients undergoing routine fundoscopy, often without visual symptoms. Fundoscopic findings consistently led to treatment modifications, including therapy prolongation and azole addition. Despite its rarity, ocular candidiasis remains clinically relevant, with persistent vision loss observed in few symptomatic patients.

Candidemia, a bloodstream infection caused by Candida species, is a common invasive fungal disease associated with high morbidity and mortality. Predisposing factors include compromised gut wall integrity from abdominal surgery, mucositis, or graft-versus-host disease (GVHD), central venous catheters (CVCs), and immunodeficiency [1, 2].

Candida bloodstream infections can result in dissemination to any organ. Involvement of the eye can present as chorioretinitis or endophthalmitis. Chorioretinitis typically presents as focal, white, infiltrative retinal lesions, often raised or mound-like. Extension into the vitreous manifests as vitritis, visible on fundoscopy as vitreous haze, fluffy white aggregates (“snowballs”), or linear formations resembling a “string of pearls” [3, 4]. Reported incidences of ocular involvement in candidemia have varied widely, from up to 37% before 1994 to 7.7–20% in more recent studies [5–9]. Concordant endophthalmitis specifically occurred less frequently in recent systematic reviews, as Breazzano et al. reported 0.8% and Phongkhun et al. 1.8% [10, 11]. These differences likely reflect evolving diagnostic criteria and shifts in antifungal therapy, with echinocandins largely replacing azoles and amphotericin B over the past 2 decades.

Guidelines for ophthalmologic evaluation in candidemia differ. In 2012, the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) recommended routine fundoscopy, especially in immunocompromised patients or those with persistent candidemia [12]. In 2016, the Infectious Diseases Society of America (IDSA) advised a dilated retinal exam within the first week after diagnosis, preferably by an ophthalmologist [13]. In 2017, the Dutch Working Party on Antibiotic Policy (SWAB) recommended to consider fundoscopy in patients with candidemia, with strong emphasis on those with visual symptoms or communication barriers [14]. The global guideline from the European Confederation of Medical Mycology (ECMM), the International Society for Human and Animal Mycology (ISHAM), and the American Society for Microbiology (ASM) published in 2025 states that routine screening of all patients with candidemia by indirect funduscopic examination is not formally endorsed due to a lack of definitive outcome and cost-benefit data. However, in the absence of reliable predictors for ocular involvement, the guideline recommends “erring on the side of caution” by screening all patients when feasible and at minimum performing fundoscopy in those with ocular symptoms or signs, those unable to verbalize complaints, immunocompromised individuals, and patients with persistent candidemia lasting 48 hours or more [15]. In contrast, the American Academy of Ophthalmology discouraged routine consultation from 2022 onward [16], citing the low incidence of ocular candidiasis (OC) as found in the recent systematic review mentioned earlier [10] and the lack of evidence for improved outcomes following ophthalmologic evaluation.

This study examined current ophthalmologic consultation practices for candidemia in 2 large tertiary care hospitals in the Netherlands. We assessed the frequency and findings of fundoscopy, their therapeutic impact, and the clinical outcome of OC.

METHODS

Study Design and Patient Population

We performed a multicenter retrospective observational cohort study in line with STROBE criteria [17]. Adult patients (≥18 years) with a positive blood culture for any Candida species or formerly named Candida species were included at the Erasmus University Medical Center (EMC, Rotterdam) and Amsterdam University Medical Center (AUMC, Amsterdam) between January 2018 and December 2024. Both institutions serve as referral centers for solid organ and stem cell transplantation, as well as complex gastrointestinal and hepatopancreatic surgery. Patients were excluded if they were lost to follow-up or opted out of clinical data sharing for research purposes. The study was reviewed by the Medical Research Ethics Committee of Rotterdam, which determined it did not fall under the Dutch Medical Research Involving Human Subjects Act (WMO). The requirement for written informed consent was waived.

Data Collection

Clinical data were retrospectively extracted from electronic medical records and microbiological data were obtained from laboratory systems. Variables included demographics (age, sex, underlying conditions), candidemia characteristics (species, duration), and risk factors at diagnosis: prolonged neutropenia (<0.5 × 10⁹/L for >10 days), immunosuppression from corticosteroids, T- or B-cell targeted drugs, chemotherapy or stem cell transplantation, CVC presence, ICU admission, and impaired gut wall integrity due to surgery, perforations, mucositis, typhlitis, or GVHD. Management variables included antifungal prophylaxis, initial and step-down therapy, and treatment duration. Ophthalmologic evaluations were documented regarding whether they were performed, along with their timing and findings. Clinical outcomes comprised persistent vision loss, persistent abnormalities on repeat fundoscopy, and 90-day mortality.

Persistent candidemia was defined as ≥1 positive culture with the same Candida species ≥96 h after the first. Cultures positive beyond 28 days or with another species were classified as an independent episode and were not included in the analysis. The most likely focus of infection was determined from clinical suspicion, imaging, and supportive cultures; if no clear focus was identified, or 2 were equally suspected by the treating and/or consulting specialists, the origin was recorded as unknown. For patients discharged to hospice with an unknown date of death, the last known follow-up date was used. Cause of death and relation to candidemia were assessed, using the treating physicians' correspondence.

Classification of Ocular Candidiasis

Ocular candidiasis was defined as retinal, choroidal, and/or vitreous lesions likely caused by Candida dissemination, as assessed by an ophthalmologist. Classification of OC was initially carried out by the evaluating ophthalmologist and subsequently reviewed by a second independent ophthalmologist, based on fundoscopy images and, when available, optical coherence tomography. We used these established diagnostic criteria [6, 9, 18]:

• Proven OC: ocular lesions with positive histology and/or positive culture of a vitreous aspirate.

• Probable Candida endophthalmitis: vitritis or fluffy lesions extending into the vitreous.

• Probable Candida chorioretinitis: deep, white focal infiltrates in the retina. Hemorrhages, Roth spots, or nerve fiber layer infarctions (cotton wool spots) were also classified as probable Candida chorioretinitis if no alternative cause, such as thrombocytopenia, diabetes mellitus, or hypertension, was identified.

• No OC: no abnormalities or lesions attributed to other systemic diseases.

Statistical Analysis

Continuous nonparametric data were summarized as medians with interquartile ranges (IQR). Comparisons used Mann–Whitney U or Kruskal–Wallis tests for continuous variables and χ² or Fisher's exact tests for categorical variables. A 2-sided P < .05 was considered significant. Analyses were performed using IBM SPSS Statistics version 28.0.1.0.

RESULTS

Study Population

Between 2018 and 2024, 417 patients had at least 1 positive blood culture for Candida species. Eleven were lost to follow-up and 4 declined data use, leaving 402 evaluable patients. The median age was 63 years (IQR 50–69), and 63.2% of patients were male. Frequent underlying conditions were solid malignancies (108/402, 26.9%), hematological malignancies (89/402, 22.1%), and cardiovascular disease (44/402, 10.9%), such as aortic dissections, endocarditis, and vascular bypass infections. At diagnosis, 300/402 patients (74.6%) had a CVC and 148/402 (36.8%) were in the ICU. Other risk factors included recent abdominal surgery (107/402, 26.6%) and mucositis or typhlitis (72/402, 17.9%). Seventy-one of 402 patients (17.7%) received systemic antifungal prophylaxis or therapy at the time of the first positive blood culture. For all baseline characteristics, see Table 1.

Table 1.

Baseline Characteristics of All Patients

	All Patients (N = 402) Mdn (IQR) or n (%)
Age, y	63 (50–69)
Sex, male	254 (63.2)
Underlying disease
Solid malignancy	108 (26.9)
Hematological malignancy	89 (22.1)
Cardiovascular disease	44 (10.9)
Solid organ transplantation	26 (6.5)
Liver/biliary tract disease	24 (6.0)
Autoimmune disease	16 (4.0)
Trauma	14 (3.5)
Pulmonary disease	13 (3.2)
Other	68 (16.9)
Presence of central venous cathetera	300 (74.6)
Admission to the intensive care unita	148 (36.8)
Immunosuppression
Prolonged neutropeniab	76 (18.9)
Recent chemotherapyc	59 (14.7)
Corticosteroids	28 (7.0)
Recent hematopoietic stem cell transplantd	8 (2.0)
T- or B-cell immunosuppressants	2 (0.5)
Combination of immunosuppressants	66 (16.4)
No immunosuppressants	232 (57.7)
Other risk factors
Recent abdominal surgery and/or intestinal perforations	133 (33.1)
Mucositis or typhlitis	72 (17.9)
Major non-abdominal surgery	40 (10.0)
Parenteral nutrition	18 (4.5)
Severe pancreatitis	12 (3.0)
Other	66 (16.4)
None	61 (15.1)
Receipt of antifungal prophylaxis or therapya	71 (17.7)
Fluconazole	62 (15.4)
Mold-active azole	4 (1.0)
Echinocandin	2 (0.5)
Azole and echinocandin	3 (0.8)

^aAt time of first positive blood culture.

^b<0.5 × 10⁹/L for more than 10 days temporally related to candidemia.

^cWithin 2 months before candidemia.

^dWithin 3 months before candidemia.

Characteristics of Candidemia

Candida albicans was the most commonly isolated species (in 165/402 cases, 41.0%) followed by Nakaseomyces glabratus (81/402, 20.1%), formerly known as Candida glabrata, and Pichia kudriavzevii (45/402, 11.2%), formerly known as Candida krusei. Mixed infections occurred in 19/402 patients (4.7%). The most frequently reported sources were intestinal translocation (88/402, 21.9%) and CVC infections (86/402, 21.4%), while 125/402 (31.1%) had no clear or multiple potential sources. Median duration of candidemia was 1 day (IQR 1–3), which was shorter for C. albicans candidemia (median 1, IQR 1–2) than non-albicans species (median 1, IQR 1–4; P < .001). Persistent candidemia (≥96 h) occurred in 95/402 (23.6%).

Ophthalmologic Evaluation and Ocular Candidiasis

Fundoscopy was recommended by the consulting microbiologist or infectious diseases specialist in 324/402 (80.6%) and performed in 307/402 (76.4%). Nonrecommendation or noncompliance with the recommendation was mainly due to early death or palliative care (74/95, 77.9%).

Among those undergoing fundoscopy, probable OC was diagnosed in 15/307 patients (4.9%): 12 with chorioretinitis (3.9%) and 3 with endophthalmitis (1.0%). No proven OC occurred, as invasive sampling was never performed. Six of 15 patients (40.0%) had new visual complaints, 6 (40.0%) had no complaints, and in 3 (20.0%) patients, symptom status could not be evaluated. Overall, OC occurred in 6/29 patients (20.7%) that reported new visual symptoms, 6/187 (3.2%) without symptoms, and 3/91 (3.3%) in whom the information on visual symptoms was lacking (P = .003).

Candida albicans was the causative agent in 12/15 OC patients (80.0%), including 2 mixed infections (C. albicans with either N. glabratus or C. dubliniensis). The 3 remaining episodes were due to C. dubliniensis, C. parapsilosis, and N. glabratus. Ocular candidiasis was more frequent in patients with C. albicans candidemia (12/128, 9.4%) versus non-albicans candidemia (3/179, 1.7%; P = .003). Persistent candidemia was not associated with OC (4/84, 4.8% vs 11/223, 4.9%; P = .608). The details of all patients with OC are found in Table 2.

Table 2.

Details on All Individual Patients With Ocular Candidiasis

Patient No.	Sex/Age (y)	Underlying Disease	Causative Species (Duration of Candidemia in Days)	Visual Symptoms	Time to Fundoscopy in Daysa	Additional Diagnostics	Classification of OC	Initial Antifungal Therapy	Therapy Changes	Persistent Vision Loss at Last Follow-Up (Duration of Follow-Up in Daysb)	Persistent Abnormalities on Fundoscopy at Last Follow-Up	Status at 90 Days
1	M/58	Cryptogenic liver cirrhosis	Candida albicans (1 d)	Intermittent floater but uncertain whether new complaint	1	No	Probable endophthalmitis OS	Anidulafungin	Prolonged therapy, addition of fluconazole, 3 times IVI with voriconazole	Yes, OD 95% with pinhole, OS 50% with pinhole (13 d)	OS: multiple white lesions	Deceased
2	F/28	Liver transplant for liver cirrhosis due to PSC and AIH	Candida albicans (2 d)	Yes, floaters for 1 wk	1	Yes, OCT	Probable endophthalmitis ODS	Micafungin	Prolonged therapy, switch to fluconazole	No, OD: 1.2-, OS 1.2 (225 d)	OD: old retinal lesion macula	Alive
3	F/68	Distal cholangiocarcinoma	Candida albicans (3 d)	No	0	No	Probable chorioretinitis OS	Anidulafungin	Prolonged therapy, addition of fluconazole	Unknown (7 d)	OS: pigmented lesion, no signs of active infection	Deceased
4	F/51	Pancreatic carcinoma	Candida albicans (1 d)	Yes, blurry vision and floaters for 4 d	4	No	Probable chorioretinitis OD	Fluconazole	Prolonged therapy, patient refused IVI with amphotericin B	No follow-up due to comfort care	No follow-up due to comfort care	Deceased
5	F/63	Type A dissection	Candida dubliniensis (3 d)	Unknown due to sedation	3	No	Probable chorioretinitis ODS	Anidulafungin	Prolonged therapy, addition of fluconazole	Unknown (42 d)	ODS: preretinal bleeds, improving Roth spots, stable and possibly improving Candida lesions	Deceased
6	F/78	Rectal carcinoma	Candida albicans (2 d)	No	1	No	Probable chorioretinitis OS	Anidulafungin	Prolonged therapy, addition of anidulafungin to fluconazole	Unknown (15 d)	OD: 1 Roth spot, OS: 2 possible Candida lesions, ODS: (preexistent) atrophic lesions macula	Alive
7	M/22	Necrotizing ischemic colitis	Candida albicans (1 d)	Yes, blurry vision OD	4	No	Probable chorioretinitis OS	Anidulafungin	Prolonged therapy, addition of fluconazole	No, OD 1.0, OS 1.0 (13 d)	OS: atrophic lesions, no signs of active infection	Alive
8	M/65	Traumatic brain injury	Candida parapsilosis (1 d)	Unknown, seen by ophthalmologist in other hospital	4	Unknown	Probable chorioretinitis, localization unknown	Anidulafungin	Prolonged therapy, switch to fluconazole	Unknown	Unknown	Alive
9	F/63	Short bowel syndrome with TPN dependency	Candida albicans (1 d)	No	1	No	Probable chorioretinitis OS	Anidulafungin	Prolonged therapy, addition of fluconazole	No, “no visual complaints” (31 d)	OS: preexistent atrophic lesion, no signs of active infection	Alive
10	M/66	Cecal carcinoma	Candida albicans (11 d)	Yes, floaters	13	No	Probable chorioretinitis ODS	Anidulafungin	Prolonged therapy, addition of fluconazole	Yes, OD 80%, OS 67% (23 d)	ODS: white yellow parafoveal lesions	Deceased
11	M/63	Sigmoid carcinoma	Candida albicans (4 d)	Yes, floaters since a few days	5	Yes, OCT	Probable chorioretinitis ODS	Micafungin	Prolonged therapy, switch to fluconazole	No, OD 1.0, OS 1.2 (34 d)	OS: 2 hypopigmented, well-circumscribed lesions, no signs of active infection	Alive
12	M/63	Non-small cell lung carcinoma	Candida albicans and Nakaseomyces glabratus (2 d)	No	6	No	Probable endophthalmitis ODS	Micafungin	Prolonged therapy with echinocandin due to N. glabrata	No, ODS 99% (11 d)	ODS: no lesions, no signs of active infection	Alive
13	M/48	Type A dissection	Candida albicans (1 d)	No	2	No	Probable chorioretinitis ODS	Micafungin	Prolonged therapy with fluconazole	No, OD 1.0, OS 1.2 (83 d)	OD: hypopigmentation or scar, no signs of active infection	Alive
14	M/53	Major trauma after road traffic accident	Nakaseomyces glabratus (7 d)	No	6	Yes, OCT	Probable chorioretinitis ODS	Micafungin	Prolonged therapy, switch to voriconazole due to N. glabrata	No, OD 1.2, OS 1.0 (59 d)	OS: atrophic lesion, no signs of active infection	Alive
15	F/39	Necrotizing post-ERCP pancreatitis	Candida albicans and C. dubliniensis (8 d)	Yes, “loss of vision”	5	No	Probable chorioretinitis ODS	Micafungin	Prolonged therapy, addition of fluconazole, 1 time IVI OS with voriconazole	Yes, OD 0.7, OS 0.7++ (278 d)	OS: central scar, no signs of active infection	Alive

Abbreviations: AIH: autoimmune hepatitis; ERCP: endoscopic retrograde cholangiopancreatography; F: female; IVI: intravitreal injection; M: male; OC: ocular candidiasis; OCT: optical coherence tomography; OD: oculus dexter; ODS: oculus dexter et sinister; OS: oculus sinister; PSC: primary sclerosing cholangitis; TPN: total parenteral nutrition.

^aTime between the start of antifungal therapy and fundoscopy in days.

^bTime between the first and last fundoscopy in days.

Timing of Fundoscopy and Follow-Up

The median time from antifungal treatment initiation to fundoscopy was 4 days (IQR 1–8). Of 307 patients examined, 196 (63.8%) underwent fundoscopy within 6 days, 96 (31.3%) between days 7 and 13, and 15 (4.9%) on day 14 or later. Ocular candidiasis was found in 14/196 patients (7.1%) examined within 6 days versus 1/111 (0.9%) examined on or after day 7 (P = .013). Timing of fundoscopy differed between centers: AUMC median 1 day (IQR 1–4) versus EMC 7 days (IQR 4–9; P < .001), consistent with protocol differences, as the EMC protocol recommends ophthalmology consultation after 7 days in asymptomatic patients, whereas this is not mentioned in the AUMC protocol.

At least 1 follow-up fundoscopy was performed in 70/307 patients (22.8%), with a median follow-up duration of 13.5 days (IQR 8–25, range 1–278) from the first to last fundoscopy. Thirteen of 15 OC patients (86.7%) received follow-up; 2 were lost to follow-up due to transfer to another hospital or death. Optical coherence tomography was performed in 3/15 patients (20.0%). Persistent vision loss occurred in 3/15 patients (20.0%): 2 with probable chorioretinitis and 1 with probable endophthalmitis, all symptomatic at baseline. Visual acuity at the last ophthalmologic follow-up was unrecorded in 5/15 (33.3%).

Systemic Dissemination

Additional imaging or cultures apart from fundoscopy were performed in 284/402 patients (70.6%). Dissemination beyond the eye occurred in 34/284 patients (12.0%), typically confirmed with culture or PCR, and was further suspected in 22/284 (7.7%). Among these 56 patients, common sites of dissemination were the lungs (12/56, 21.4%), liver/spleen (7/56, 12.5%), muscle/bone (7/56, 12.5%), and implanted devices (6/56, 10.7%). Multiple sites were involved in 9/56 (16.1%).

Systemic Antifungal Treatment and Clinical Outcome

Systemic antifungals were given to 371/402 patients (92.3%). Of the 31 untreated patients, 27 (87.1%) died before therapy could be initiated, 3 (9.7%) received palliative care, and 1 (3.2%) improved spontaneously. Among treated patients, 359/371 (96.8%) initially received an echinocandin, and step-down to fluconazole was performed in 120/359 (33.4%) and to another azole in 12/359 (3.3%). Excluding 112 patients who stopped early due to death or comfort care, median treatment duration was 19 days (IQR 15–41). Therapy was longer in the 46 patients with ophthalmologic and/or systemic dissemination (median 52.5 days, IQR 32.8–96.5) versus 17 days (IQR 15–23) in the 211 patients without dissemination (P < .001).

All 15 OC patients had treatment adjustments. Intravitreal injections with voriconazole were added to systemic antifungal therapy in one patient with probable Candida endophthalmitis and in 1 patient with probable Candida chorioretinitis. One patient was offered an intravitreal injection with amphotericin B, but this was refused in the context of a short life expectancy. In 7 patients (46.7%), an azole was added, while in 4 (26.7%), the therapy was switched to azole monotherapy. In all 15 OC patients, the treatment duration was prolonged.

The overall 90-day mortality was 40.0% (161/402 patients). Candidemia was the suspected cause in 42/161 cases (26.1%), contributory in 53/161 (32.9%), and the cause of death was unknown in 5/161 (3.1%). Mortality did not differ between patients with or without systemic dissemination (24/70, 34.3% vs 88/281, 31.3%; P = .633), with or without ocular dissemination (5/15, 33.3% vs 74/292, 25.3%; P = .546), or with versus without dissemination beyond the eye (20/56, 35.7% vs 76/228, 33.3%; P = .736).

DISCUSSION

This multicenter retrospective cohort study evaluated the use and yield of routine fundoscopy in 402 patients with candidemia at 2 Dutch tertiary care centers. Fundoscopy was performed in 76.4% (307/402 patients), and probable OC was diagnosed in 4.9% (15/307), comprising 3.9% chorioretinitis and 1.0% endophthalmitis. These rates confirm that anno 2020 OC is relatively rare, consistent with prior reports of overall OC between 1.3% and 7.6% [19–21] and endophthalmitis between 0.8% and 1.8% [9–11]. Notably, 40% of OC patients were asymptomatic, and 20% were unable to report symptoms, highlighting that chorioretinitis can be asymptomatic and that fundoscopy may detect cases otherwise missed [6, 21, 22].

Timing and Impact of Fundoscopy

In the entire cohort, fundoscopy was performed early with a median of 4 days after antifungal initiation. The timing of screening differed across the 2 centers (Amsterdam UMC after median 1 day [IQR 1–4] and Erasmus MC after 7 days [IQR 4–9; P < .001]), consistent with the ongoing debate on optimal timing of fundoscopy. While some studies suggest higher sensitivity after day 7, as lesions may develop later after an initially negative exam [9, 21, 23, 24], we found a higher incidence in patients examined within 6 days. However, as patients with visual symptoms will undergo fundoscopy earlier, this result may be biased. Also, we cannot exclude that some of these cases diagnosed very early could have been resolved had the fundoscopy performed later.

Findings prompted changes in antifungal therapy in all OC cases: treatment was prolonged in 100%, azoles were added or substituted in 73.4%, and intravitreal injections were administered in 13.3%. These adjustments align with guidelines recommending azoles for improved ocular penetration. However, the clinical benefit of these changes in therapy remains uncertain, as persistent vision loss to any degree was relatively rare (20% of OC patients vs 1.0% of all examined patients) and did not lead to blindness, with the lowest visual acuity reported at 50% in one eye. None of the patients without visual symptoms at the time of OC diagnosis developed persistent vision loss during follow-up. This supports the previously suggested idea [24] that OC represents a spectrum of disease, ranging from more prevalent and relatively mild chorioretinitis (when lesions are not near the macula) to severe endophthalmitis with risk of persistent vision loss. This finding suggests, but does not prove, that the early detection and the resulting changes in antifungal therapy may have prevented progression of OC.

Clinical Relevance and Protocol Adherence

Routine fundoscopy provided clinically relevant information, particularly in asymptomatic patients. However, the low incidence of vision-threatening OC suggests that a more selective approach may be justified. The AAO recommends fundoscopy only in symptomatic patients based on the low incidence of 0.8% Candida endophthalmitis found by Breazzano et al [10, 16], whereas the IDSA supports routine screening to prevent severe complications based on the more recent meta-analysis of Phongkhun et al, with 1.8% Candida endophthalmitis [11, 13]. As also mentioned in a recent review [24], the incidence differences may stem from inclusion of non-English papers, but also from true regional variations, which will have to impact clinical decision-making. In another recent retrospective study from the UK and USA, the rates of Candida endophthalmitis were as high as 9.47% (9/95) in patients examined but much lower at 2.5% (9/359) in all included patients with candidemia [25]. The broad incidence range across studies is probably one of the reasons that recommendations differ across recent guidelines as well. With several more recent studies now becoming available, future updated guidelines should be able to come to more concordant recommendations. Amidst these controversies, our findings suggest that fundoscopy is most warranted in patients with visual symptoms and those unable to communicate symptoms.

The proportion of patients with persistent candidemia (≥96 h) in our cohort (23.6%) was higher than expected, possibly due to the relatively high prevalence of P. kudriavzevii and N. glabratus as causative agents, the high proportion (22.1%) of patients with an underlying hematological disease and a methodological aspect in which positive blood cultures with the same species within 28 days would be considered ongoing infection instead of a recurrent episode. Nevertheless, our proportion of persistent candidemia was comparable to the 20.2% of patients with positive follow-up blood cultures (median duration of 6 days) in a recently published subgroup analysis of the Candida III study [26]. Although not observed in our study, one could argue that patients with prolonged candidemia could have an increased risk for ocular dissemination, consistent with the stance of the Global ECMM/ISHAM/ASM guideline [15].

Protocol adherence was high: fundoscopy was performed in 76.4% of patients (while non-adherence was typically due to early death), echinocandin therapy was administered in 96.8%, azoles were switched to or added in 73.4% of OC patients, and treatment duration was significantly longer in patients with dissemination, reflecting appropriate clinical response to systemic involvement. In our cohort, there was a relatively high rate of possible dissemination to the lungs and/or pleural cavity: 12/284 (4.2%) patients who underwent additional imaging displayed abnormalities such as pulmonary micronodules, of whom 5 had pleural fluid cultures positive for Candida species. An autopsy study by Kontoyiannis et al. in cancer patients found that 36/676 (5.3%) had histologically proven pulmonary candidiasis [27], although it was not clear whether these patients had had prior candidemia. Further studies on the exact prevalence of pulmonary dissemination following candidemia are limited, as the specific location of invasive candidiasis is often not reported in recent clinical trials [28].

Strengths and Limitations

Strengths include a large sample size, the multicenter design, and the review of fundoscopic findings by independent ophthalmologists, enhancing generalizability and diagnostic reliability. Limitations include the retrospective design and absence of histopathological confirmation. In addition, not all fundoscopic signs are specific; for example, cotton wools spots could be classified as probable Candida chorioretinitis in absence of alternative explanations such as hypertension or diabetes mellitus, potentially leading to overdiagnosis. Conversely, as bedside fundoscopy in critically ill or noncooperative patients can be challenging, underdiagnosis can also not be fully excluded.

Recommendations for Clinical Practice and Future Research

Given that almost half of OC cases were asymptomatic and fundoscopy impacted treatment, we think that routine fundoscopic screening is valuable in selected high-risk populations. We recommend a single fundoscopy in the first 5–10 days following the onset of candidemia, with earlier evaluation in symptomatic or patients unable to communicate visual symptoms. Prospective studies should try to identify predictors of OC. Ideally, a randomized trial should assess the (cost-)effectiveness of routine versus selective fundoscopy (eg, in symptomatic or unconscious patients), but it is unlikely that such a study will ever be performed.

CONCLUSION

Ocular candidiasis is a relatively rare but clinically significant complication of candidemia. In a setting of systematic fundoscopic screening, persistent vision loss was uncommon and in all cases preceded by visual symptoms. When OC was diagnosed, antifungal therapy was prolonged and the diagnosis prompted a switch to or the addition of an azole in most cases. A targeted screening approach may balance diagnostic yield and resource use.