Endogenous Aspergillus endophthalmitis: report of 3 cases and review of the literature.

Abstract

Introduction Invasive aspergillosis typically involves the lung, but subsequent hematogenous dissemination to other organs is common in immunosuppressed patients (8,42,49,67). Rarely, disseminated disease can also be acquired following intravenous inoculation. Lesions are typically found in the brain, myocardium, kidneys, and skin with disseminated aspergillosis. An uncommon complication of disseminated aspergillosis is endogenous endophthalmitis, and even more uncommon is isolated endogenous endophthalmitis without other organ involvement. We report 3 cases of endogenous Aspergillus endophthalmitis and present a detailed review of the literature to illustrate the difficulties of diagnosis and treatment associated with this entity. Because endophthalmitis caused by Aspergillus tends to present abruptly with nonspecific symptoms and progresses quickly to irreversible retinal damage, rapid recognition is important. Among the 86 total cases reviewed, certain diagnostic methods and treatment modalities emerged as being more useful than others. Early diagnosis and appropriate treatment may increase the likelihood of a favorable outcome. Methods We searched the English-language literature for cases of Aspergillus endogenous endophthalmitis, through an electronic search (MEDLINE, National Library of Medicine, Bethesda, MD) using key words “Aspergillus,” “endogenous,” and “endophthalmitis” and cross-referencing articles to include older articles, going back to 1949. Cases were included if Aspergillus was confirmed to be present within the eye by growth of the organism from samples of intraocular fluid or by identification of fungal elements typical of Aspergillus species on histopathologic examination of methenamine silver-stained ocular tissue. Cases of exogenous posttraumatic or postoperative endophthalmitis and those with contiguous spread from sinuses were excluded. All cases were assessed for demographic information, risk factors, diagnostic methods, treatment strategies, and outcomes. Data were analyzed using the Student t test for continuous variables and the Fisher exact test for dichotomous variables. Case Reports Case 1 A 56-year-old man presented with a 4-day history of left eye pain, swelling, and decreased visual acuity. These symptoms occurred 15 days after admission to hospital for cough, fever, and dyspnea for which he required endotracheal intubation and mechanical ventilation. He was treated for pneumonia with broad-spectrum antibiotics and received high doses of corticosteroids for management of chronic obstructive pulmonary disease. Past medical history included end-stage renal disease requiring chronic hemodialysis following failure of a renal transplant, chronic obstructive pulmonary disease, hypertension, coronary artery disease, and congestive heart failure. At the time of admission for the ocular symptoms, the patient’s medications included prednisone 15 mg daily and amoxicillin/clavulanic acid 875 mg twice daily. His temperature was 97.1 °F (36.1 °C), respiratory rate 16/min, blood pressure 176/86 mmHg. The left eye had extensive scleral injection, subconjunctival edema inferiorly, and a 1-mm hypopyon in the anterior chamber (Figure 1). Visual acuity was 20/200. Extraocular motions and pupillary reflexes were intact. Funduscopic examination revealed a cloudy vitreous; the retina could not be visualized. The right eye was normal. Chest examination revealed decreased breath sounds; no heart murmur was heard. There were no peripheral stigmata of infective endocarditis, and no neurologic deficits were noted. Laboratory evaluation revealed a white blood cell count of 17,500/L with 88% segmented neutrophils. A right upper lobe infiltrate was noted on chest radiograph.Fig. 1: Case 1, left eye. Subconjunctival edema and scleral injection obscure the 1-mm hypopyon.A vitreous aspirate was obtained; intravitreal vancomycin and ceftazidime were given; and ciprofloxacin, atropine, and prednisolone were administered topically to the left eye. Over the next 2 days, the patient’s cough worsened, he became short of breath and developed hemoptysis, and the pulmonary infiltrate worsened. Computed tomography (CT) of the chest showed bilateral airspace disease without nodules or cavitation. Repeated cultures from blood were negative. Broad-spectrum antibiotics were given for nosocomial pneumonia. Transthoracic echocardiography revealed no evidence of valvular vegetations. All vitreous cultures remained sterile. Cultures taken from the right upper lobe at bronchoscopy yielded yeast, Aspergillus fumigatus, and oral flora; his physicians felt that the Aspergillus was a contaminant. The patient was discharged with topical ciprofloxacin and prednisolone ophthalmic solution for presumed bacterial endophthalmitis and oral levofloxacin for presumed bacterial pneumonia. Oral corticosteroids were tapered over the following 2 weeks. He continued to have dense vitritis, visual loss, and pain. Three weeks later, he underwent pars plana vitrectomy for persistent vitritis. Retinal vascular occlusions and necrotic retinal tissue were found. Eight weeks after the onset of symptoms he underwent enucleation of the left eye for refractory pain. Pathologic examination of the removed globe revealed acute suppurative and chronic granulomatous endophthalmitis with extensive necrosis of the retina, reactive retinal gliosis, severe panuveitis, intraocular hemorrhage, and episcleral inflammation and scarring. The methenamine silver stain revealed branching septate hyphae consistent with Aspergillus species (Figure 2), which prompted the initiation of a 6-week course of treatment with liposomal amphotericin B, 5 mg/kg per day, followed by oral itraconazole solution 200 mg twice daily. During treatment, the patient developed low back pain. Magnetic resonance imaging (MRI) showed inflammation of the L1-L2 disc space with increased signal on T2-weighted images. A CT-guided biopsy of the intervertebral disc showed only chronic inflammation; the material yielded no growth on culture. Two months later he presented with respiratory distress and died shortly thereafter. No autopsy was performed.Fig. 2: Case 1, enucleation specimen of the left eye. Fragments of branching septate hyphae are seen within the vitreous. (Gomori methenamine silver stain; original magnification × 630.)Case 2 A 57-year-old woman presented with a 7-day history of right eye pain and redness. She had a 10-year history of bronchiolitis obliterans organizing pneumonia (BOOP) which had required intermittent prolonged courses of oral corticosteroids. Three months before admission she was treated with ceftazidime and tobramycin and then ofloxacin and trimethoprim/sulfamethoxazole for cough and shortness of breath. When her symptoms persisted after 6 weeks, a course of prednisone was initiated. Ten days later, she developed right eye pain with movement, redness, photophobia, and decreased visual acuity. An aspiration of the aqueous humor and vitreous body yielded rare Staphylococcus epidermidis after 48 hours. With the development of an anterior chamber hypopyon and increasing vitritis, an intravitreal injection of vancomycin and ceftazidime was performed. Visual acuity was 20/300 in the right eye. Because of persistent symptoms, a second intravitreal injection of vancomycin was given 6 days later. On admission, she had a temperature of 99.0 °F (37.2 °C) and a heart rate of 110 beats/min. There was severe right conjunctival injection, and a hypopyon was present in the anterior chamber (Figure 3). Bilateral rhonchi were present, and a previously described II/VI systolic murmur at the apex was heard. On the right lateral thigh there were a few firm, nontender erythematous nodules. Laboratory evaluation revealed a white blood cell count of 25,000/L with 87% segmented neutrophils. The chest radiograph showed changes consistent with BOOP as well as a right upper lobe nodular opacity that was new compared with a radiograph performed the preceding month.Fig. 3: Case 2, right eye. Photograph taken under slit lamp examination demonstrates a small hypopyon and conjunctival injection with a lateral subconjunctival hemorrhage.Initial management included an increase in prednisone dosage to 60 mg daily and treatment with intravenous vancomycin and levofloxacin. A CT of the sinuses and orbits disclosed no source of infection contiguous with the eye. Pars plana vitrectomy with vitreous biopsy was performed, and vancomycin, gentamicin, and amphotericin B were injected into the eye. A. fumigatus was obtained on culture of the vitreous sample. The patient subsequently developed pain in the other eye, and further evaluation revealed changes consistent with endophthalmitis. Transthoracic echocardiography revealed an 8 × 13 mm mobile vegetation on the mitral valve and severe mitral regurgitation. Blood cultures yielded no organisms. Therapy with amphotericin B was initiated. A. fumigatus was isolated from sputum, urine, and a skin biopsy specimen. She developed progressive visual loss, expressive aphasia, and right-sided hemiplegia, and died on the 12th hospital day. Case 3 A 73-year-old man was admitted with loss of vision in the left eye. Studies performed during a 34-day hospital admission in the preceding 2 months had suggested BOOP, for which he was treated with corticosteroids. On admission, a chest radiograph showed new changes consistent with a fungus ball and airspace disease in the right lower and left upper lobes. Pars plana vitrectomy was performed because of concern for possible fungal endophthalmitis. At the time of the procedure, 10 μg of intravitreous amphotericin B and 400 μg of dexamethasone were injected into the vitreous. The vitrectomy specimen grew A. fumigatus. Sputum samples also yielded Aspergillus. The prednisone was tapered, intravenous amphotericin B was begun, and a second intravitreal injection of amphotericin B, 10 μg, and dexamethasone, 400 μg, was performed. A transesophageal echocardiogram demonstrated no evidence of endocarditis. He was prescribed itraconazole 200 mg twice daily on discharge after having received 1 g of amphotericin B. The following month he developed low back pain. An MRI demonstrated post-Gadolinium enhancing thoracic spine lesions with increased signal on T2-weighted images at T7, T9, and T10. Discitis was also seen at the L2-L3 disc space. A CT-guided biopsy of the L2 intervertebral lesion was performed, but cultures yielded no growth. Four months later, visual acuity in the right eye remained only finger counting at 1–6 feet. There were posterior synechiae for 360° extending to the lens with a fibrovascular retrolenticular membrane. On ocular ultrasound there was no retinal or choroid detachment, and only mild diffuse mobile vitreous opacities were found. Itraconazole therapy was stopped after a total of 3 years of treatment. Results Demographics Case reports of 86 patients (109 eyes) with endogenous Aspergillus endophthalmitis, including the 3 patients in the present report, were reviewed (1,4,6,7,9,11–16,20,21,26,29,30,33,35–37,39–41,43–46,48,50,51,53–55,57–60,63,65,68,75,78,81–87,89). Cases were reported from 1949 to 2001. The mean age (± SD) was 42.3 ± 19.8 years (range, 10 d to 91 yr); 46 (55%) patients were male and 37 (45%) were female (no sex listed for 3 patients). Underlying diseases and risk factors The majority of the 84 patients for whom data were available had serious underlying illnesses, and many were immunosuppressed (Table 1). The most common underlying conditions were intravenous drug use, which was present in 23 (27%) patients (4,16,20,21,33,35,39,40,48,57,68,78,85,87); receipt of a solid organ transplant (23%) (1,6,11,30,37,45,59,65,86); and lung disease, including chronic obstructive pulmonary disease, asthma, and BOOP (17%) (29,44,50,51,65,85). Exposure to corticosteroids proved to be a strong risk factor. A total of 36 (43%) patients had received corticosteroid therapy; 36% of these had received corticosteroids for only a brief time (<3 mo). The reporting authors could identify no risk factor for development of Aspergillus endophthalmitis in 5 patients (12,54,58,75,81).TABLE 1: Underlying conditions in 84 patients with endogenous Aspergillus endophthalmitis*Clinical characteristics Of the 86 patients, 63 (73%) had unilateral involvement. Of those, the left eye was involved in 30 (48%), the right in 25 (40%), and in 8 (12%) only “unilateral involvement” was reported. In 23 (27%) of cases, both eyes eventually became infected. Information regarding the onset of symptoms and physical findings were available for 13 of these 23 cases with bilateral involvement. In 6 there was bilateral involvement at the onset of disease, and in 7 the fellow eye manifested involvement from 2 to 70 days later. The mean time that elapsed before seeking medical advice was 7.4 ± 19 days after the onset of symptoms for 53 patients for whom data were available. Of these 53, 46 (87%) presented within 7 days of the onset of symptoms, and 7 (13%) presented after 10 or more days (range, 10–105 d). For the 55 patients for whom presenting symptoms were reported, 46 (84%) presented with complaints of blurred vision or were found to have diminished visual acuity. Eye pain (44%) and redness (24%) also were common presenting symptoms. Other symptoms, such as eye swelling, photophobia, floaters, and headache, were reported less frequently (Table 2).TABLE 2: Ocular symptoms in 55 patients with endogenous Aspergillus endophthalmitis*Ophthalmologic examination showed vitritis in 45 of 73 cases (62%), retinal exudative lesions or chorioretinitis in 28 (38%), and an exudative mass in the vitreous in 14 (19%). Of the 14 patients with a vitreous exudative mass, 8 were described as having a pseudohypopyon or layering preretinal exudate. Other physical findings included conjunctival injection and chemosis, as well as anterior chamber cells and hypopyon (Table 3). In 17 (23%) cases, retinal details could not be visualized because of dense vitritis. Only 9 patients presented with visual acuity of 20/70 or better. Two patients had only anterior chamber involvement (20,44).TABLE 3: Ocular findings in 73 patients with endogenous Aspergillus endophthalmitis*Thirty (35%) patients had isolated endophthalmitis, and 56 (65%) had involvement of organs other than the eye (Table 4). Of the 30 patients with isolated eye involvement, 22 (73%) admitted to intravenous drug use. Only 1 of the 23 patients with intravenous drug use as a risk factor had Aspergillus infection elsewhere (right anterior seventh rib osteomyelitis) (57). The respiratory tract was the presumed portal of entry for patients who did not use intravenous drugs. Thirty-four (40%) patients had clinical evidence of pulmonary involvement and 33 (38%) had disseminated aspergillosis (≥2 noncontiguous organs infected). Aspergillus endocarditis was present in 14 (16%) patients; 8 had native valve endocarditis, and 6 had prosthetic valve endocarditis (7,14,29,45,65,83,84,86).TABLE 4: Organ involvement in 86 patients with endogenous Aspergillus endophthalmitisDiagnosis The diagnosis of Aspergillus endophthalmitis proved to be difficult. The diagnosis was made only after enucleation or at autopsy in 44 (51%) patients. For those patients in whom the diagnosis was made antemortem, the highest yield was obtained by culture of vitrectomy specimens. The diagnosis was made by culture or fungal stain in 28 of 31 (90%) vitrectomy specimens, compared with 6 of 12 (50%) vitreous aspirate specimens (p = 0.013). Only 1 of 12 anterior chamber aspirates yielded Aspergillus species (20). Only 1 patient had Aspergillus isolated from a culture taken from blood (6). Microbiology A. fumigatus was isolated in 36 (51%), and A. flavus in 18 (26%) of the 70 patients for whom culture data were available. Of the 18 patients in whom A. flavus was isolated, 14 (78%) were intravenous drug users with no other identifiable risk factors. The association of the intravenous drug injection and A. flavus was statistically significant (p < 0.0001). The remaining species were A. terreus (3), A. niger (3), A. glaucus (1), A. nidulans (1), and A. candidus (1). In 7 cases, Aspergillus was cultured but not speciated. In 16 cases, the diagnosis was based on histologic features in ocular tissue (15,21,43,48,53,59,74,86,89). Treatment Seventeen patients, all of whom had the diagnosis made postenucleation or postmortem, received no antifungal therapy. For another 15, no data on therapy were reported. Of the 54 patients who were noted to have received antifungal treatment, 41 (76%) were treated with intravenous amphotericin B for varying durations of time. Twenty-two of these 41 (54%) received intravitreal amphotericin B, with or without vitrectomy, in addition to intravenous amphotericin B. Of the 19 patients who received only intravenous amphotericin B, 9 were diagnosed before autopsy or enucleation; 5 of these patients had undergone vitrectomy, and another had resection of an anterior chamber mass. Thirty-one (57%) patients received intravitreal injections of amphotericin B; single doses ranged from 5 μg to 100 μg (Table 5). Repeated injections were typically administered. The 17 patients who received the 5 μg dose on average were administered 2 ± 1.2 injections (range, 1–6). Four patients received intravitreal therapy with amphotericin B without the systemic administration of an active agent against Aspergillus. Oral flucytosine was given with amphotericin B in 14 patients, and 4 patients received itraconazole following treatment with amphotericin B. Intravitreal injection of dexamethasone (400–800 μg) was administered along with amphotericin B to 4 patients. Subconjunctival administration of amphotericin B (200–2,000 μg) and corticosteroids occurred in 5 and 3 patients, respectively; and topical antifungal agents and corticosteroids were also used in others.TABLE 5: Antifungal treatment for endogenous Aspergillus endophthalmitis in 54 patients*Outcomes Only 9 of 86 (10%) patients regained useful vision. Two of those 9 patients had only anterior chamber involvement (20,44). Thus, only 7 (8%) patients with posterior chamber involvement regained meaningful vision in the infected eye. All of these 7 patients had received intravenous amphotericin B, and 6 of the 7 also had received intravitreal amphotericin B and had undergone vitrectomy. Forty (47%) patients died, 19 (22%) required enucleation, and 18 (21%) patients were left with a visual acuity of 20/200 or worse in the affected eye. Patients who had good outcomes presented earlier after the onset of symptoms (4.6 ± 4.8 d) than patients who had poor outcomes (7.8 ± 20 d); however, this difference failed to reach statistical significance. Twelve patients were reported to have had relapse or progression despite therapy. Four relapsed in the same eye, 4 in the contralateral eye, 2 developed vertebral osteomyelitis, and 1 patient each had progressive lesions involving the skin and a prosthetic heart valve, respectively. The average number of days to relapse or diagnosis of metastatic infection was 64.2 ± 55.6 days (range, 15–180 d). Discussion Endophthalmitis is a rare manifestation of aspergillosis that has been reported infrequently in those patients traditionally thought to be at risk for invasive aspergillosis. In 1 series from the United Kingdom, only 1 patient among 87 hematology patients with invasive pulmonary aspergillosis developed endophthalmitis (91); in several other series among patients with hematologic malignancies and transplants, this complication was not reported (8,34,42). Corticosteroid use emerged as a major risk factor for the development of Aspergillus endophthalmitis. We found that 43% of patients had been treated with corticosteroids before the development of Aspergillus endophthalmitis. It is noteworthy that corticosteroid use often was not prolonged; over one-third of those treated with corticosteroids received therapy for less than 3 months’ time. Corticosteroids, a well-known risk factor for invasive aspergillosis, impair macrophage nonoxidative killing mechanisms and neutrophil recruitment, adherence, and oxidative and nonoxidative killing mechanisms (49). The large proportion of cases (27%) that occurred among intravenous drug users is intriguing. A variety of mechanisms have been proposed to explain the route of entry of Aspergillus in drug users (4,20,57,73,80). Aspergillus species do not commonly colonize the skin. Although contamination of the injection paraphernalia has been described as the source of an outbreak of Candida albicans endophthalmitis in heroin users (73), it is more likely that the drug itself or the material used to filter the drug is contaminated with Aspergillus. In 1 study (80) from the District of Columbia, Aspergillus species were isolated from 26% of 100 heroin samples and only 3% of 100 sets of injection paraphernalia. Isolation of Aspergillus species was second only to that of Bacillus species (32%). Unfortunately, the Aspergillus isolates were not identified to species level. A. flavus was isolated in 64% of the cases in which intravenous drug use was the risk factor for acquisition of disease, and 78% of all A. flavus isolates were from intravenous drug users. In this review, 26% of all isolates causing endophthalmitis were A. flavus, compared with 13% of isolates identified as A. flavus in a large multicenter hospital survey that reported on all forms of aspergillosis (64). Different Aspergillus species have been shown to occupy different niches in the environment. A. flavus is frequently identified as the cause of fluorescent yellow spots on raw cotton and has been found in 9%–54% of raw cotton samples (66). Other species of Aspergillus also have been found to contaminate cotton. A. terreus has been isolated from cotton exposed to moisture, A. niger causes cotton boll rot, and A. glaucus and A. candidus have been reported to contaminate stored cotton seed (66). Given that at least 4 of the patients in this study were reported to have filtered their drug through cotton balls and that 3 others injected cocaine, which also is sometimes filtered through cotton balls, contaminated cotton could be a potential source of A. flavus in intravenous drug users. The diagnosis of Aspergillus endophthalmitis is frequently made only at autopsy, and progression to severe visual impairment is the usual outcome in the majority of cases. Most of the diagnostic difficulties occur because there are so few features that distinguish Aspergillus endophthalmitis from other causes of endophthalmitis. Decreased visual acuity, pain, and redness are common to endophthalmitis caused by a variety of organisms. There are a few clues that should direct clinicians to consider Aspergillus as a possible cause of endogenous endophthalmitis. Intravenous drug use appears to be an important risk factor. This is also a risk factor for Candida endophthalmitis (18,19,52,71,73), but is noted infrequently as a risk factor for bacterial endophthalmitis (23,27,62,90). Corticosteroid use for any duration also should raise the index of suspicion that Aspergillus species are the cause of endophthalmitis (34). Endogenous endophthalmitis that occurs in an immunosuppressed patient who has persistently negative blood cultures is more likely due to Aspergillus than Candida. The development of characteristic findings of invasive pulmonary aspergillosis, such as a halo sign or cavitation on a high-resolution CT scan, should also point toward the possibility of Aspergillus species as the cause of endogenous endophthalmitis. Although diabetes has been shown to be a strong risk factor for bacterial endophthalmitis (62,90), it is not commonly noted as a risk factor for fungal endophthalmitis (17,25,32,47). Certain ocular findings may help distinguish between Candida and Aspergillus endophthalmitis. A “pseudohypopyon” or preretinal layering exudate has been said to be pathognomonic for Aspergillus endophthalmitis. This finding was observed in only 8 of 73 (11%) reported cases; thus, this sign is helpful when present, but appears to be an uncommon finding. Retinal exudative lesions are seen in both Candida and Aspergillus endophthalmitis. However, the lesions associated with Candida infection tend to be more discrete. This may be because endophthalmitis is routinely sought when blood cultures yield Candida; therefore, early infection limited to chorioretinal disease is now frequently noted before extension to the vitreous has developed (17,70). In the case of aspergillosis, vitreous involvement is almost always present at the time of the ophthalmologic examination. A recent study compared histopathologic changes associated with Candida and Aspergillus endogenous endophthalmitis from autopsy or enucleation specimens (65). Of the 13 cases of Aspergillus endophthalmitis, all demonstrated involvement of the vitreous, and 8 were found to have involvement of multiple posterior compartment structures. Septate hyphae were seen predominantly in the subretinal space and prominently invaded through the walls of blood vessels leading to necrosis. Candida involvement was more limited with 7 of 12 cases demonstrating abscesses in the vitreous with less damage to retinal structures. A definitive diagnosis of Aspergillus endophthalmitis requires growth of the organism. The sensitivity of culturing Aspergillus from aspirates of the aqueous and vitreous was low. The highest yield was found with vitrectomy specimens. Blood cultures are almost always negative; these yielded Aspergillus in only 1 case (6). Because endophthalmitis due to Aspergillus is often a sequel to disseminated infection, biopsy of skin, lung, or other organs may yield a diagnosis by histopathology or culture methods. Histopathologic findings alone may be difficult to interpret given the similarity in appearance of Aspergillus species to those of other fungi, such as Pseudallescheria boydii (56). Newer diagnostic methods that use polymerase chain reaction technology on ocular samples or blood may prove helpful in the future (10,38). Many different strategies have been used for the treatment of endogenous Aspergillus endophthalmitis. Intravenous amphotericin B, the standard of care for most patients with aspergillosis (77), has poor intraocular penetration (24,31,61). Therefore, other methods of delivery, such as topical application, subconjunctival injection, and intravitreal injection, have been used. The topical application of amphotericin B is irritating and leads to an abrupt and vigorous inflammatory reaction in the conjunctiva (28). Subconjunctival administration does not produce significant drug levels within the vitreous (31). Injection of amphotericin B directly into the vitreous, however, has been shown to be safe and efficacious (2,3,52). The optimal dose of intravitreal amphotericin B in humans has not yet been determined. Based on experimental animal models, 5–10 μg appears to be safe and to maintain adequate intraocular concentrations (2,88). The 4 patients who had higher doses of intravitreal amphotericin B administered (20– 100 μg) did not have adverse events associated with the injection (59,68,85,87); however, doses of ≥25 μg have led to toxic effects including localized retinal necrosis and detachment in rabbits (2). Other earlier studies demonstrated toxicity with doses as low as 1–2 μg when the drug was delivered in proximity to retinal tissue (76); however, this direct toxicity can be avoided by injection into the center of the vitreous body. In an effort to minimize toxicity, experiments with liposomal preparations of amphotericin B have shown promising results in rabbits and rhesus monkeys (5,79). The half-life and distribution of intravitreal amphotericin B varies depending on whether vitrectomy has been performed. Without vitrectomy the half-life is 7–14 days and there appears to be accumulation of the drug in the sclera, choroid, and retina (88). In eyes that have undergone vitrectomy, the half-life is shortened to 2 days without evidence of accumulation in ocular tissues (88). Fluconazole and itraconazole are not useful in the treatment of Aspergillus endophthalmitis. Fluconazole is not active against Aspergillus species, and itraconazole penetrates poorly into the eye (69). The 4 patients who were treated with itraconazole in this series received the drug for treatment of other foci of aspergillosis after initial therapy with systemic amphotericin B. Other new antifungal agents may prove useful for treating Aspergillus endophthalmitis. Voriconazole, a new triazole, is fungicidal for Aspergillus and attains measurable concentrations in the vitreous (personal communication, Christopher Hitchcock). At this time, however, there is no reported clinical experience using voriconazole for treating endophthalmitis. Caspofungin, an echinocandin antifungal agent, also appears to be fungicidal for Aspergillus, but it is unclear whether effective intraocular concentrations are attained. The outcome of endogenous Aspergillus endophthalmitis is poor. Of the 84 patients with retinal disease, only 7 (8%) regained meaningful visual acuity (48,57,68,75,81,85). This contrasts with endogenous Candida endophthalmitis, for which recent studies show recovery of vision in 76%–100% of patients (17,22,52). All 7 patients with Aspergillus endophthalmitis who responded favorably received intravenous amphotericin B, and 6 also received intravitreal amphotericin B and had a vitrectomy performed. In addition to its diagnostic benefit, vitrectomy may also be of therapeutic benefit as it removes infective fungal elements. This procedure has been shown to improve outcomes in heroin users who had vitritis due to endogenous Candida endophthalmitis (52). Six of the 7 patients who regained meaningful vision presented with poor visual acuity (20/400 or worse) and had macular retinal lesions. Patients with good outcomes in general presented earlier than those with poor outcomes, demonstrating the importance of early recognition and treatment. The overall mortality of 44% seen in this group of patients with endogenous Aspergillus endophthalmitis is less than the 60%–90% overall crude mortality reported for invasive aspergillosis (49,64). The lower mortality observed in this group is likely related to the high proportion of intravenous drug users who had no underlying immunosuppression and who had infection limited to the eye. A rational approach to the diagnosis and treatment of endogenous Aspergillus endophthalmitis might be as follows: 1) Patients who present with symptoms suggestive of endogenous endophthalmitis with a history of intravenous drug use, recent corticosteroid use, or immunosuppression with negative blood cultures should be considered at risk. 2) Physical examination findings of posterior compartment disease with pseudohypopyon with or without dense vitritis suggest aspergillosis. 3) Given the above risk factors and findings, prompt pars plana vitrectomy for both therapeutic drainage and diagnostic culture purposes is imperative. The empiric administration of 5–10 μg of amphotericin B via intravitreal injection at the time of vitrectomy is indicated to provide the earliest possible therapy while awaiting results of cultures and special stains for fungal organisms as well as other pathogens. 4) A careful evaluation for other sites of infection should be carried out, tissue biopsies taken to help establish a diagnosis of aspergillosis, and intravenous amphotericin B therapy initiated if there is any possibility of invasive aspergillosis. 5) If an Aspergillus species is confirmed as the pathogen, repeat intravitreal administration of amphotericin B is indicated, given the short half-life of amphotericin B in the posterior compartment after vitrectomy. Repeat therapeutic vitrectomy, with or without repeated intravitreal amphotericin B, may also be needed depending on the subsequent observed clarity of the posterior chamber. The total dose of intravenous amphotericin B will be determined not only by the ocular findings, but also by the presence of extraocular aspergillosis. Repeated examinations by an ophthalmologist are crucial, particularly early in the course of treatment, to detect relapse in the involved eye or the development of disease in the fellow eye. Summary We reviewed 86 cases of endogenous Aspergillus endophthalmitis reported since 1949. Predisposing medical conditions were present in 94%. Intravenous drug use was noted in 27%, solid organ transplantation in 23%, underlying lung disease in 17%, and 43% had received corticosteroids. The predominant organism causing endophthalmitis was A. fumigatus (51% of cultures). However, 26% of patients had A. flavus isolated. This species was strongly linked to the use of intravenous drugs. Of the 18 patients with endophthalmitis due to A. flavus, 14 (78%) were intravenous drug users. The diagnosis of endogenous Aspergillus endophthalmitis is difficult and frequently made only at autopsy. Pars plana vitrectomy appears to be the highest-yield procedure for establishing the diagnosis; 90% of vitrectomy specimens yielded Aspergillus compared with 50% of vitreous aspirates. Outcomes were generally poor; only 7 of the 84 patients (8%) with posterior chamber involvement regained useful vision. Treatment with intravenous amphotericin B combined with intravitreal injection of amphotericin B and vitrectomy appears to be the most efficacious therapy. Six of the 7 patients reported to have regained useful vision had received such treatment.