- Email: [email protected]
Implications of Early Systemic Therapy on the Incidence of Endogenous Fungal Endophthalmitis
Implications of Early Systemic Therapy on the Incidence of Endogenous Fungal Endophthalmitis Warren ] . Scherer, MD, PhD, Kathy Lee, MD, PhD Objective: In the past, evidence of endogenous fungal endophthalmitis has been used as a guide to initiating potentially toxic antifungal therapy in patients with systemic fungal infections. Recently, however, a trend has developed to provide patients with antifungal therapy at the first evidence of fungal infection. The authors' study evaluates the incidence of endogenous fungal endophthalmitis in this setting. Design: The design is a retrospective review of the medical records of patients examined by the inpatient ophthalmology consultation service to rule out endogenous fungal endophthalmitis between January 1994 and April 1996 at the University of Michigan Hospitals, Ann Arbor, Michigan. Participants: Two hundred fourteen eyes of 107 patients with a diagnosis of systemic fungal infection were studied. Intervention: A review of medical records was performed. Main Outcome Measures: The findings of the ocular examination, the presence of risk factors tor disseminated fungal infection, the type of antifungal therapy, and the source and identity of the isolated fungus were recorded. Results: The majority of patients examined had either fungal growth from blood cultures or evidence of deep tissue fungal infection. All patients in the study were at risk tor fungal disease with each having at least one risk factor tor disseminated fungal infection. Of the patients examined, 93.4% already were receiving systemic antifungal therapy at the time of ophthalmologic consultation. Only 3 (2.8%) of the 107 patients examined had chorioretinal findings consistent with early endogenous fungal endophthalmitis. None had intravitreous involvement, and the ocular findings did not change the course of therapy. Conclusions: Early systemic treatment of deep tissue fungal infection appears to dramatically decrease the incidence of endogenous fungal endophthalmitis. Ophthalmology 1997; 104:1593-1598
Fungi are an increasingly prevalent cause of nosocomial infection. Candida species are important in this regard and currently account for 7% to 15% of all hospitalOriginally received: November 5, 1996. Revision accepted: May 5. 1997. From The W. K. Kellogg Eye Center, University of Michigan, Ann Arbor, Michigan. The authors have no financial support and no proprietary interest in any of the products described in this study. Address correspondence to Warren J. Scherer, MD, PhD, 1436 Bay Harbor Drive, #16-308, Palm Harbor, FL 34685.
acquired infections. 1 Infection either may be blood borne (candidemia) or involve parenchymal tissues (candidiasis). Risk factors for disseminated candidiasis are well established and include broad-spectrum antibiotic use, recent major surgery (particularly of the gastrointestinal tract), a prolonged postoperative course, parenteral hyperalimentation, generalized immunosuppression, indwelling intravenous catheters, and intravenous drug abuse?- 4 Ocular infection with Candida species usually causes a multifocal chorioretinitis. Clinically, the infiltrates are small, creamy-white lesions with overlying vitreous inflammation of varying intensity. With extension into the
1593
Ophthalmology
Volume 104, Number 10, October 1997
vitreous, the lesions may enlarge to form intravitreous fungus balls. The treatment of Candida chorioretinitis is with intravenous amphotericin B alone or in combination with 5-flucytosine or oral fluconazole. Vitreous involvement necessitates the addition of intravitreous amphotericin B to the systemic therapy. Response to treatment typically is prompt with regression of retinal and vitreous involvement within 4 to 7 days. 4•5 Candida endophthalmitis will develop in between 10% and 37% of patients with candidemia if not treated with systemic antifungal therapies. 6 - 8 Furthermore, autopsy series show that ocular fungal involvement correlates highly with concurrent deep tissue infection. 3•9•10 The high incidence of fungal endophthalmitis in patients with candidemia and its correlation with widespread systemic disease led to the recommendation that all patients with candidemia receive a dilated fundoscopic examination at the time of positive blood culture. 5 •9 - 11 When initially proposed, the rationale of the recommendation was two-fold: (1) to prevent the ocular complications of Candida infection by early detection and (2) to use ocular involvement as an indicator of likely widespread invasive fungal disease. Positive findings on ocular examination were used as a guide to initiating potentially toxic systemic antifungal therapy. Recently, a trend has developed to initiate antifungal treatment after a single positive fungal culture to avoid the potentially lethal complications of disseminated fungal infection. Greater physician experience and advances in supportive care have lessened the toxicity risks of systemic antifungal therapy, and it is therefore initiated more readily. It still is customary, however, to perform a fundoscopic examination upon the report of a positive fungal blood culture even though antifungal therapy appropriate to treat disseminated disease already has been initiated. We evaluated the efficacy of screening ophthalmic examinations to rule out endogenous fungal endophthalmitis in patients with known or suspected deep tissue fungal infection. In contrast to previous studies that address this issue, the majority of patients in our study were receiving systemic antifungal therapy at the time of examination. Because early initiation of systemic antifungal therapy rapidly is becoming the standard of care for known or suspected disseminated fungal infection, we thought it important to assess the incidence of endogenous fungal endophthalmitis and the contribution of the ocular examination to the treatment of these patients.
20~-----------------------------------.
male
£}
15
1::
<1>
~ a.
0
.... <1>
10
..Q
E
::J
z
5
0 <1
2-18
19- 35
36-50
5 1-65
>65
Age (Years) Figure 1. Age and gender distribution of the patient population (N = 107).
positive fungal culture result. No patients with recent ocular surgery or trauma were included in the study. Each consultation included a full bedside ophthalmologic examination consisting of visual acuity when possible, external examination, penlight examination of the anterior segment, intraocular pressure, and dilated indirect ophthalmoscopy. Examinations were performed by members of the consultation team on a rotating basis. Negative examinations were not repeated. Examinations with findings consistent with endogenous fungal endophthalmitis were reviewed or repeated or both by retinal specialists. The medical records were reviewed to determine pertinent patient characteristics, including gender and age at time of consultation. We noted whether antifungal therapy had been initiated at the time of consultation and, if so, what agents were used. The sources and identities of fungal isolates cultured on or before the date of consultation were determined as completely as possible in every case. A tally of established positive risk factors for fungal disease was made for each patient. These included recent major surgery, broad-spectrum antibiotic use, the presence of indwelling intravenous catheters, total parenteral nutrition, immunosuppression, diabetes, prematurity or low birth weight, and alcoholism or liver disease. The patient population was analyzed as a whole. The data are expressed as percentages where applicable.
Results Methods The records of all examinations performed by the ophthalmology consultation service at the University of Michigan between January 1994 and April 1996 were reviewed. During this period, a total of 107 consultations were placed to rule out endogenous fungal endophthalmitis in inpatients in the University of Michigan, Mott Children's and Holden Perinatal Hospitals. Consultations typically were requested within 48 hours after notification of a
1594
One hundred seven inpatients with fungal infection were examined by the ophthalmology consult service at the University of Michigan from January 1994 until April 1996. The age and gender distribution of the patient population is presented in Figure 1. Males (63.6%, 681107) outnumbered females (37.4%, 40/107) by almost 2 to 1. The age distribution is bimodal with peaks at young ( < 1 year) and middle ages (36-50 years). The first peak largely represents premature and immunosuppressed infants. The second peak comprises a mixture of diagnoses,
Scherer and Lee · Early Systemic Therapy and Fungal Endophthalmitis Table 1. Treatment of Systemic Fungal Infection
Table 3. Identity of Fungus Isolated
Treatment
n
%
Fungus
Any antifungal Amphoteracin B Fluconazole Flucytosine Multiple antifungals None Unknown
100 79 36 19 35 7 3
93.4 73.8 33.6 17.8 32.7 6.5 2.8
Candida Candida species identified C. albicans C. perapsilosis C. tropicalis C. glabrata C. lusitaniaq
including postoperative patients and those immunosuppressed secondary to cancer or acquired immune deficiency syndrome. The incidence of fungal endophthalmitis in our study population was 2.8% (3/107). These patients in~luded a 34-year-old male with acute myelogenous leukelllla, a 56year-old female with rheumatoid arthritis and pneumonia, and a 65-year-old male who had recent gastrointestinal surgery for a bleeding gastric ulcer. In all three cases, retinal lesions consistent with fungal chorioretinitis were present without vitreous seeding. All three patients already were receiving systemic antifungal therapy adequate to treat disseminated disease, and treatment was not changed as a result of ophthalmologic consultation. Two of the three patients with endogenous fungal endophthalmitis died within days of ophthalmologic evaluation, one from an acute neurologic event and the other from disseminated fungemia. Postmortem examination of the eyes was not performed. The third patient was discharged and lost to follow-up. At the University of Michigan, initial systemic antifungal therapy usually consists of a minimum of amphotericin B (500 mg total dose, intravenous) with or without flucytosine or fluconazole (400 mg, orally, every day) for even a single positive fungal blood culture (personal communication, Department of Infectious Disease, University of Michigan, 1996). In this study, a majority of the patient population (93.4%, 100/107) already was receiving systemic antifungal therapyat the time of oph~h-~ mology consultation (Table 1). Therapy usually was Imtlated 48 to 72 hours before ophthalmic examination. Most patients were treated with intravenous amphotericin B (73.8%, 79/107). Approximately one third received oral
n
%
Antibiotic therapy Recent surgery Indwelling line Immunosuppression Diabetes Prematurity/low birth weight Alcoholism/liver disease
84 49 54 30
78.5 45.4 50.4 27.7 12.0 11.1 3.7
13
12 4
72.9
14.0 0.9 9.2 1.9
0.9
107
Total
fluconazole (33.6%, 36/107). A smaller number of patients (17 .8%, 19/107) were treated with flucytosine, usually in combination with amphotericin B. Many patients (32.7%, 35/107) received multiple antifungal medications. Of the seven patients not receiving systemic antifungal therapies, three had no evidence of fungal infection, and ophthalmologic consultation was requested to rule out fungal infection as a cause of fever of unknown origin. We examined known risk factors for the development of endogenous fungal endophthalmitis in our population. 4 These risk factors for our patient population are reported in Table 2. A history of recent antibiotic therapy was present in the majority of patients (78.5%, 84/107). Approximately half of patients had a recent history of surgery (45.7%, 49/107) or indwelling lines (50.4%, 54/107). Other risk factors evaluated include immunosuppression (28%, 30/107), diabetes (12.1 %, 13/107), prematurity (11.2%, 12/107), and alcoholic liver disease (3.7%, 4/ 107). The distribution of the number of risk factors across the patient population is depicted in Figure 2. All of the 50
c (J")
40
~
30
...... 0 .....
20
E ::J z
10
..0
Risk Factor
78 69 37 14 7 10 1 15 1 10 2 1
Yeast Fungal growth No isolation Other Unknown
0..
Table 2. Risk Factors for the Development of Endogenous Fungal Endophthalmitis
%
n
0
2
3
Number of Risk Factors
4
Figure 2. Histogram of risk factor distribution across the patient population. All patients had at least one risk factor for the development of endogenous fungal endophthalmitis (N = 107).
1595
Ophthalmology
Volume 104, Number 10, October 1997
Table 4. Source of Fungal Infection Site
n
%
Blood Catheter tip Urine Surgical wound or drain Diagnostic test (CT or ultrasound) Cerebrospinal fluid Other
66
61.7
10 13 6
9.3 12.1
~~
Multiple sites
5.6 5.6
6 2 14
1.9 13.1
18
16.8
6
5.6
patients (100%, 107/107) had at least 1 risk factor. No patient was found to have more than four risk factors. The average number of risk factors per patient was 2.2 ::±:: 0.9 standard deviation. Of the three patients with chorioretinal lesions, all were receiving intravenous antibiotics. Additionally, the 34-year-old male was immunocompromised secondary to acute myelogenous leukemia, the 56-year-old female was receiving oral steroids for treatment of rheumatoid arthritis, and the 65-year-old male was status postgastrointestinal surgery, had a J-tube in place, and had a history of diabetes. The fungus isolates for the patient population are indicated in Table 3. The most common fungus isolated was Candida species (72.9%, 78/107). Specific species of Candida were reported in 69 patients. Of this group, the most common species was C. albicans (53.6%, 37/69), followed by C. parapsilosis (20.2%, 14/69), C. tropicalis (10.1 %, 7/69), C. glabrata (14.5%, 10/69), and C. lusitaniq (1.5%, 11107). The second most commonly reported entity was "yeast" (14%, 15/107). In a smaller subset of patients (9.3%, 10/107), fungus was not isolated; however, fungal infection was either suspected secondary to suspicious lesions on computed tomography or ultrasound scan or strongly was considered under the differential diagnosis of fever of unknown origin. One case each of Aspergillus pneumonia and of Histoplasma capsulatum fungemia also were encountered. Neither patient had ocular findings. All three patients with endogenous fungal endophthalmitis had blood cultures positive for C. albicans. No additional tissue sites were found to be positive. The source of fungal growth also was examined (Table 4). The majority of patients had culture positive fungemia (61.7%, 66/107). The next most common site of infection was urine (12.1 %, 13/107). The cultured tips of indwelling catheters such as total parenteral nutrition or Broviac lines (9.3%, 10/107) also were a significant source of fungal cultures. Nine of the 13 patients with positive urine cultures also had other positive tissue fungal cultures, mainly blood. Additional sites included surgical wounds (5.6%, 6/107), suspicious soft tissue lesions on abdominal computed tomographic scan (5.6%, 6/107), and cerebrospinal fluid (1.9%, 2/107). Sites of infection listed as "other" include sputum, pleural fluid, heart valves, and pericardia! fluid. A proportion of patients (16.8%, 18/107) had fungal growth from more than 1 site.
1596
We reviewed the visual acuity of our patient population (Table 5). In almost half of the patients (47.6%, 102/214 eyes) evaluated, visual acuity was not able to be assessed reliably secondary to obtunded mental status or poor fixation. Of the patients whose vision could be measured, approximately one third (36.9%, 79/214 eyes) were 20/ 20 to 20/40, whereas a smaller number were 20/50 to 20/ 200 (6.5%, 14/214 eyes), count fingers (2.3%, 5/214 eyes), or were able to fix and follow (6.5%, 14/214 eyes). Of the three patients with fungal chorioretinitis, visual acuity of two patients could not be measured and the other had acuity of 20/20 in both eyes. Visual acuity is not a helpful prognostic indicator of the presence of ocular fungal infection.
Discussion Endogenous fungal endophthalmitis is a relatively common sequela of nosocomial systemic fungal infection. In previous studies, the incidence of endogenous fungal endophthalmitis in patients with candidemia or candidiasis or both has ranged from approximately 10% to 40%. 5 - 9 We report a much lower incidence (2.8%) in our study population. This low incidence of endogenous fungal endophthalmitis may be explained by the recent change in philosophy regarding the treatment of systemic fungal infections. In the past, physicians were reluctant to treat disseminated fungal infections because of the hematologic and nephrotoxic side effects of the antifungal agent amphotericin B 12 and because some patients were reported to recover spontaneously. 2 ' 13 Ophthalmologic consultation was obtained to confirm the presence of endogenous fungal endophthalmitis as an indicator of disseminated fungal infection and as justification for initiating systemic antifungal therapy. Recently, however, physicians have become less reluctant to start antifungal therapy. Currently, at the University of Michigan, almost all patients with fungal infections promptly receive a systemic antifungal agent at a dose effective against ocular involvement. 4 In the current study, approximately 90% of patients were being treated with a systemic antifungal medication at the time of ophthalmologic consultation. Most patients were receiving amphotericin B; others were prescribed oral fluconazole, which also is considered to represent adequate treatment. 14
Table 5. Visual Acuity of Patients with Systemic Fungal Infection No. of Eyes
Vision
79
20/20-20/40 20/50-20/200
14
5
CF F/F Unable to assess CF
= counting fingers;
14 102 F/F
= fix
and follow.
% 36.9
6.5 2.3 6.5 47.6
Scherer and Lee · Early Systemic Therapy and Fungal Endophthalmitis The risk factors that predispose inpatients with systemic fungal infection to endogenous fungal endophthalmitis are well known. These include recent history of antibiotic therapy, surgery, indwelling catheters, immunosuppression, diabetes, prematurity, alcoholism, and liver disease. 3 To determine whether the low incidence of endogenous fungal endophthalmitis in the current study was related to our patient population somehow being at less risk than those patients in previous studies, we examined known risk factors for the development of endogenous fungal endophthalmitis. We found that all of our patients had at least one or more of known risk factors for ocular fungal infection. The average number of risk factors per patient was approximately two (range, one to four). Our conclusion is that the patients in our study were at significant risk for the development of endogenous fungal endophthalmitis and did not differ in this regard from prior study populations. Although other fungal species can infect the eye, 15 •16 Candida species are the most commonly reported cause of endogenous fungal endophthalmitis. 4 Of the various species of Candida, C. albicans is thought to be more likely to infect the eye. 9 The mechanism underlying this predilection of C. albicans to infect the eye is not understood completely but may involve the ability of this species to form germ tubes in serum that could embolize and lodge in the choriocapillaris in vivo. 17 We also found Candida species to represent the greatest percentage of fungal isolates among our patient population. The majority of the identified species were C. albicans, although a mixture of other Candida species also were isolated. Nonalbicans species of Candida have been reported and may be increasing in prevalence. 18 All three patients in our study with fungal chorioretinitis had C. albicans fungemia. We examined visual acuity to determine whether acuity would be a useful indicator of ocular fungal involvement in our patient population. In accord with previous studies, we found that visual acuity was not a useful prognostic factor in assessing the presence of endogenous fungal endophthalmitis. There are limitations of the current study. The first is the retrospective nature of the study. A prospective study in which treatment would be withheld from a random control group of patients would not be considered ethical, given the life-threatening nature of systemic fungemia. The second is that most of the consultations consisted of a single fundus examination, and we cannot be sure that fungal endophthalmitis did not develop in some patients subsequently. However, in a recent study examining the development of fungal endophthalmitis in nontreated, hospitalized patients with candidemia, ocular lesions developed within 72 hours of the suspected onset of fungemia in approximately 90% of patients. 6 Consultations typically were completed within this timeframe in our study population. If this assumption is applied to the current study, the additional cases missed because of a single evaluation would increase the incidence to approximately 3.1%. This number still is much lower than reported previously. Finally, because the consultation ser-
vice at our institution is staffed on a rotating basis, examinations were performed by multiple examiners during the period studied. In conclusion, we report the incidence of endogenous fungal endophthalmitis in a systemically infected inpatient population already receiving adequate antifungal therapy to be 2.8% (3 of 107). This figure is substantially lower than that reported in all previous studies. We thoroughly examined factors that could be responsible for this low incidence of ocular fungal infection. Our patients were found to be at equal risk for the development of endogenous fungal endophthalmitis when compared to those of previous studies. As in previous studies, fungal involvement was either known or believed to be systemic in all patients and was secondary to Candida species in most patients. The main discernible difference between our study and past investigations was the prompt initiation of early systemic antifungal therapy. We therefore conclude that early recognition and adequate treatment of systemic fungal infection substantially decrease the incidence of endogenous fungal endophthalmitis. However, given the lack of prospective data and the potentially lethal nature of disseminated candidiasis, we continue to recommend ophthalmologic examination of any patient with evidence of systemic fungal infection.
References 1. Wade RP. Epidemiology of Candida infections In: Bodey GP, ed. Candidiasis: Pathogenesis, Diagnosis, and Treatment, 2nd ed. New York: Raven Press, 1993;85-107. 2. Klein JJ, Watanakunakorn CC. Hospital-acquired fungemia. Its natural course and clinical significance. Am J Med 1979;67:51-8. 3. Edwards JE Jr, Foos RY, Montgomerie JZ, Guze LB. Ocular manifestations of Candida septicemia: review of seventy-six cases of hematogenous Candida endophthalmitis. Medicine (Baltimore) 1974;53:47 -75. 4. Pettit TH, Edwards JE Jr, Purdy EP, Bullock JD. Endogenous fungal endophthalmitis. In: Pepose JS, Holland GN, Wilhelmus KR, eds. Ocular Infection and Immunity. St. Louis: Mosby, 1996; 1262-85. 5. Fishman LS, Griffin JR, Sapico FL, Hecht R. Hematogenous Candida endophthalmitis-a complication of candidemia. N Eng! J Med 1972;286:675-81. 6. Brooks RG. Prospective study of Candida endophthalmitis in hospitalized patients with candidemia. Arch Intern Med 1989; 149:2226-8. 7. Henderson DK, Edwards JE Jr, Montgomerie JZ. Hematogenous Candida endophthalmitis in patients receiving parenteral hyperalimentation fluids. J Infect Dis 1981; 143: 655-61. 8. Parke DW II, Jones DB, Gentry LO. Endogenous endophthalmitis among patients with candidemia. Ophthalmology 1982; 89:789-96. 9. Griffin JR, Petit TH, Fishman LS, et al. Blood-borne Candida endophthalmitis: a clinical and pathologic study of 21 cases. Arch Ophthalmol 1973;89:450-6. 10. Griffin JR, Foos RY, Petit TH. Relationship between candida! endophtha1mitis, candidemia, and disseminated candidiasis. In: Concilium Ophthalmologicum, 22nd Annual Meeting, Paris: Masson, 1974;2:661-4.
1597
Ophthalmology
Volume 104, Number 10, October 1997
11. Gold JWM. Infections due to fungi, Actinomyces and Nocardia. In: Reece GE, Betts RG, eds. A Practical Approach to Infectious Disease. Boston: Little, Brown, 1991; 512-27. 12. Burgess JL, Birchall R. Nephrotoxicity of amphotericin B, with emphasis on changes in tubular function. Am J Med 1972;53:77-84. 13. Ellis CA, Spivack ML. The significance of candidemia. Ann Intern Med 1967;67:511-22. 14. Friedberg MA, Rapuano CJ, eds. Wills Eye Hospital office and emergency room diagnosis and treatment of eye dis. ease. Philadelphia: Lippincott, 1990;383-5.
1598
15. Gross JG. Endogenous Aspergillus-induced endophthalmitis: successful treatment without systemic antifungal medication. Retina 1992;12:341-5. 16. Hunt KE, Glasgow BJ. Aspergillus endophthalmitis. An unrecognized endemic disease in orthotopic liver transplantation. Ophthalmology 1996; 103:757-67. 17. Moyer DY, Edwards JE. Candida! endophthalmitis and central nervous system infection. In: Bodey GP, ed. Candidiasis: Pathogenesis, Diagnosis, and Treatment, 2nd ed. New York: Raven Press, 1993;331-55. 18. Joshi N, Hamory BH. Endophthalmitis caused by non-albicans species of Candida. Rev Infect Dis 1991; 13:281-7 .
Fungi are an increasingly prevalent cause of nosocomial infection. Candida species are important in this regard and currently account for 7% to 15% of all hospitalOriginally received: November 5, 1996. Revision accepted: May 5. 1997. From The W. K. Kellogg Eye Center, University of Michigan, Ann Arbor, Michigan. The authors have no financial support and no proprietary interest in any of the products described in this study. Address correspondence to Warren J. Scherer, MD, PhD, 1436 Bay Harbor Drive, #16-308, Palm Harbor, FL 34685.
acquired infections. 1 Infection either may be blood borne (candidemia) or involve parenchymal tissues (candidiasis). Risk factors for disseminated candidiasis are well established and include broad-spectrum antibiotic use, recent major surgery (particularly of the gastrointestinal tract), a prolonged postoperative course, parenteral hyperalimentation, generalized immunosuppression, indwelling intravenous catheters, and intravenous drug abuse?- 4 Ocular infection with Candida species usually causes a multifocal chorioretinitis. Clinically, the infiltrates are small, creamy-white lesions with overlying vitreous inflammation of varying intensity. With extension into the
1593
Ophthalmology
Volume 104, Number 10, October 1997
vitreous, the lesions may enlarge to form intravitreous fungus balls. The treatment of Candida chorioretinitis is with intravenous amphotericin B alone or in combination with 5-flucytosine or oral fluconazole. Vitreous involvement necessitates the addition of intravitreous amphotericin B to the systemic therapy. Response to treatment typically is prompt with regression of retinal and vitreous involvement within 4 to 7 days. 4•5 Candida endophthalmitis will develop in between 10% and 37% of patients with candidemia if not treated with systemic antifungal therapies. 6 - 8 Furthermore, autopsy series show that ocular fungal involvement correlates highly with concurrent deep tissue infection. 3•9•10 The high incidence of fungal endophthalmitis in patients with candidemia and its correlation with widespread systemic disease led to the recommendation that all patients with candidemia receive a dilated fundoscopic examination at the time of positive blood culture. 5 •9 - 11 When initially proposed, the rationale of the recommendation was two-fold: (1) to prevent the ocular complications of Candida infection by early detection and (2) to use ocular involvement as an indicator of likely widespread invasive fungal disease. Positive findings on ocular examination were used as a guide to initiating potentially toxic systemic antifungal therapy. Recently, a trend has developed to initiate antifungal treatment after a single positive fungal culture to avoid the potentially lethal complications of disseminated fungal infection. Greater physician experience and advances in supportive care have lessened the toxicity risks of systemic antifungal therapy, and it is therefore initiated more readily. It still is customary, however, to perform a fundoscopic examination upon the report of a positive fungal blood culture even though antifungal therapy appropriate to treat disseminated disease already has been initiated. We evaluated the efficacy of screening ophthalmic examinations to rule out endogenous fungal endophthalmitis in patients with known or suspected deep tissue fungal infection. In contrast to previous studies that address this issue, the majority of patients in our study were receiving systemic antifungal therapy at the time of examination. Because early initiation of systemic antifungal therapy rapidly is becoming the standard of care for known or suspected disseminated fungal infection, we thought it important to assess the incidence of endogenous fungal endophthalmitis and the contribution of the ocular examination to the treatment of these patients.
20~-----------------------------------.
male
£}
15
1::
<1>
~ a.
0
.... <1>
10
..Q
E
::J
z
5
0 <1
2-18
19- 35
36-50
5 1-65
>65
Age (Years) Figure 1. Age and gender distribution of the patient population (N = 107).
positive fungal culture result. No patients with recent ocular surgery or trauma were included in the study. Each consultation included a full bedside ophthalmologic examination consisting of visual acuity when possible, external examination, penlight examination of the anterior segment, intraocular pressure, and dilated indirect ophthalmoscopy. Examinations were performed by members of the consultation team on a rotating basis. Negative examinations were not repeated. Examinations with findings consistent with endogenous fungal endophthalmitis were reviewed or repeated or both by retinal specialists. The medical records were reviewed to determine pertinent patient characteristics, including gender and age at time of consultation. We noted whether antifungal therapy had been initiated at the time of consultation and, if so, what agents were used. The sources and identities of fungal isolates cultured on or before the date of consultation were determined as completely as possible in every case. A tally of established positive risk factors for fungal disease was made for each patient. These included recent major surgery, broad-spectrum antibiotic use, the presence of indwelling intravenous catheters, total parenteral nutrition, immunosuppression, diabetes, prematurity or low birth weight, and alcoholism or liver disease. The patient population was analyzed as a whole. The data are expressed as percentages where applicable.
Results Methods The records of all examinations performed by the ophthalmology consultation service at the University of Michigan between January 1994 and April 1996 were reviewed. During this period, a total of 107 consultations were placed to rule out endogenous fungal endophthalmitis in inpatients in the University of Michigan, Mott Children's and Holden Perinatal Hospitals. Consultations typically were requested within 48 hours after notification of a
1594
One hundred seven inpatients with fungal infection were examined by the ophthalmology consult service at the University of Michigan from January 1994 until April 1996. The age and gender distribution of the patient population is presented in Figure 1. Males (63.6%, 681107) outnumbered females (37.4%, 40/107) by almost 2 to 1. The age distribution is bimodal with peaks at young ( < 1 year) and middle ages (36-50 years). The first peak largely represents premature and immunosuppressed infants. The second peak comprises a mixture of diagnoses,
Scherer and Lee · Early Systemic Therapy and Fungal Endophthalmitis Table 1. Treatment of Systemic Fungal Infection
Table 3. Identity of Fungus Isolated
Treatment
n
%
Fungus
Any antifungal Amphoteracin B Fluconazole Flucytosine Multiple antifungals None Unknown
100 79 36 19 35 7 3
93.4 73.8 33.6 17.8 32.7 6.5 2.8
Candida Candida species identified C. albicans C. perapsilosis C. tropicalis C. glabrata C. lusitaniaq
including postoperative patients and those immunosuppressed secondary to cancer or acquired immune deficiency syndrome. The incidence of fungal endophthalmitis in our study population was 2.8% (3/107). These patients in~luded a 34-year-old male with acute myelogenous leukelllla, a 56year-old female with rheumatoid arthritis and pneumonia, and a 65-year-old male who had recent gastrointestinal surgery for a bleeding gastric ulcer. In all three cases, retinal lesions consistent with fungal chorioretinitis were present without vitreous seeding. All three patients already were receiving systemic antifungal therapy adequate to treat disseminated disease, and treatment was not changed as a result of ophthalmologic consultation. Two of the three patients with endogenous fungal endophthalmitis died within days of ophthalmologic evaluation, one from an acute neurologic event and the other from disseminated fungemia. Postmortem examination of the eyes was not performed. The third patient was discharged and lost to follow-up. At the University of Michigan, initial systemic antifungal therapy usually consists of a minimum of amphotericin B (500 mg total dose, intravenous) with or without flucytosine or fluconazole (400 mg, orally, every day) for even a single positive fungal blood culture (personal communication, Department of Infectious Disease, University of Michigan, 1996). In this study, a majority of the patient population (93.4%, 100/107) already was receiving systemic antifungal therapyat the time of oph~h-~ mology consultation (Table 1). Therapy usually was Imtlated 48 to 72 hours before ophthalmic examination. Most patients were treated with intravenous amphotericin B (73.8%, 79/107). Approximately one third received oral
n
%
Antibiotic therapy Recent surgery Indwelling line Immunosuppression Diabetes Prematurity/low birth weight Alcoholism/liver disease
84 49 54 30
78.5 45.4 50.4 27.7 12.0 11.1 3.7
13
12 4
72.9
14.0 0.9 9.2 1.9
0.9
107
Total
fluconazole (33.6%, 36/107). A smaller number of patients (17 .8%, 19/107) were treated with flucytosine, usually in combination with amphotericin B. Many patients (32.7%, 35/107) received multiple antifungal medications. Of the seven patients not receiving systemic antifungal therapies, three had no evidence of fungal infection, and ophthalmologic consultation was requested to rule out fungal infection as a cause of fever of unknown origin. We examined known risk factors for the development of endogenous fungal endophthalmitis in our population. 4 These risk factors for our patient population are reported in Table 2. A history of recent antibiotic therapy was present in the majority of patients (78.5%, 84/107). Approximately half of patients had a recent history of surgery (45.7%, 49/107) or indwelling lines (50.4%, 54/107). Other risk factors evaluated include immunosuppression (28%, 30/107), diabetes (12.1 %, 13/107), prematurity (11.2%, 12/107), and alcoholic liver disease (3.7%, 4/ 107). The distribution of the number of risk factors across the patient population is depicted in Figure 2. All of the 50
c (J")
40
~
30
...... 0 .....
20
E ::J z
10
..0
Risk Factor
78 69 37 14 7 10 1 15 1 10 2 1
Yeast Fungal growth No isolation Other Unknown
0..
Table 2. Risk Factors for the Development of Endogenous Fungal Endophthalmitis
%
n
0
2
3
Number of Risk Factors
4
Figure 2. Histogram of risk factor distribution across the patient population. All patients had at least one risk factor for the development of endogenous fungal endophthalmitis (N = 107).
1595
Ophthalmology
Volume 104, Number 10, October 1997
Table 4. Source of Fungal Infection Site
n
%
Blood Catheter tip Urine Surgical wound or drain Diagnostic test (CT or ultrasound) Cerebrospinal fluid Other
66
61.7
10 13 6
9.3 12.1
~~
Multiple sites
5.6 5.6
6 2 14
1.9 13.1
18
16.8
6
5.6
patients (100%, 107/107) had at least 1 risk factor. No patient was found to have more than four risk factors. The average number of risk factors per patient was 2.2 ::±:: 0.9 standard deviation. Of the three patients with chorioretinal lesions, all were receiving intravenous antibiotics. Additionally, the 34-year-old male was immunocompromised secondary to acute myelogenous leukemia, the 56-year-old female was receiving oral steroids for treatment of rheumatoid arthritis, and the 65-year-old male was status postgastrointestinal surgery, had a J-tube in place, and had a history of diabetes. The fungus isolates for the patient population are indicated in Table 3. The most common fungus isolated was Candida species (72.9%, 78/107). Specific species of Candida were reported in 69 patients. Of this group, the most common species was C. albicans (53.6%, 37/69), followed by C. parapsilosis (20.2%, 14/69), C. tropicalis (10.1 %, 7/69), C. glabrata (14.5%, 10/69), and C. lusitaniq (1.5%, 11107). The second most commonly reported entity was "yeast" (14%, 15/107). In a smaller subset of patients (9.3%, 10/107), fungus was not isolated; however, fungal infection was either suspected secondary to suspicious lesions on computed tomography or ultrasound scan or strongly was considered under the differential diagnosis of fever of unknown origin. One case each of Aspergillus pneumonia and of Histoplasma capsulatum fungemia also were encountered. Neither patient had ocular findings. All three patients with endogenous fungal endophthalmitis had blood cultures positive for C. albicans. No additional tissue sites were found to be positive. The source of fungal growth also was examined (Table 4). The majority of patients had culture positive fungemia (61.7%, 66/107). The next most common site of infection was urine (12.1 %, 13/107). The cultured tips of indwelling catheters such as total parenteral nutrition or Broviac lines (9.3%, 10/107) also were a significant source of fungal cultures. Nine of the 13 patients with positive urine cultures also had other positive tissue fungal cultures, mainly blood. Additional sites included surgical wounds (5.6%, 6/107), suspicious soft tissue lesions on abdominal computed tomographic scan (5.6%, 6/107), and cerebrospinal fluid (1.9%, 2/107). Sites of infection listed as "other" include sputum, pleural fluid, heart valves, and pericardia! fluid. A proportion of patients (16.8%, 18/107) had fungal growth from more than 1 site.
1596
We reviewed the visual acuity of our patient population (Table 5). In almost half of the patients (47.6%, 102/214 eyes) evaluated, visual acuity was not able to be assessed reliably secondary to obtunded mental status or poor fixation. Of the patients whose vision could be measured, approximately one third (36.9%, 79/214 eyes) were 20/ 20 to 20/40, whereas a smaller number were 20/50 to 20/ 200 (6.5%, 14/214 eyes), count fingers (2.3%, 5/214 eyes), or were able to fix and follow (6.5%, 14/214 eyes). Of the three patients with fungal chorioretinitis, visual acuity of two patients could not be measured and the other had acuity of 20/20 in both eyes. Visual acuity is not a helpful prognostic indicator of the presence of ocular fungal infection.
Discussion Endogenous fungal endophthalmitis is a relatively common sequela of nosocomial systemic fungal infection. In previous studies, the incidence of endogenous fungal endophthalmitis in patients with candidemia or candidiasis or both has ranged from approximately 10% to 40%. 5 - 9 We report a much lower incidence (2.8%) in our study population. This low incidence of endogenous fungal endophthalmitis may be explained by the recent change in philosophy regarding the treatment of systemic fungal infections. In the past, physicians were reluctant to treat disseminated fungal infections because of the hematologic and nephrotoxic side effects of the antifungal agent amphotericin B 12 and because some patients were reported to recover spontaneously. 2 ' 13 Ophthalmologic consultation was obtained to confirm the presence of endogenous fungal endophthalmitis as an indicator of disseminated fungal infection and as justification for initiating systemic antifungal therapy. Recently, however, physicians have become less reluctant to start antifungal therapy. Currently, at the University of Michigan, almost all patients with fungal infections promptly receive a systemic antifungal agent at a dose effective against ocular involvement. 4 In the current study, approximately 90% of patients were being treated with a systemic antifungal medication at the time of ophthalmologic consultation. Most patients were receiving amphotericin B; others were prescribed oral fluconazole, which also is considered to represent adequate treatment. 14
Table 5. Visual Acuity of Patients with Systemic Fungal Infection No. of Eyes
Vision
79
20/20-20/40 20/50-20/200
14
5
CF F/F Unable to assess CF
= counting fingers;
14 102 F/F
= fix
and follow.
% 36.9
6.5 2.3 6.5 47.6
Scherer and Lee · Early Systemic Therapy and Fungal Endophthalmitis The risk factors that predispose inpatients with systemic fungal infection to endogenous fungal endophthalmitis are well known. These include recent history of antibiotic therapy, surgery, indwelling catheters, immunosuppression, diabetes, prematurity, alcoholism, and liver disease. 3 To determine whether the low incidence of endogenous fungal endophthalmitis in the current study was related to our patient population somehow being at less risk than those patients in previous studies, we examined known risk factors for the development of endogenous fungal endophthalmitis. We found that all of our patients had at least one or more of known risk factors for ocular fungal infection. The average number of risk factors per patient was approximately two (range, one to four). Our conclusion is that the patients in our study were at significant risk for the development of endogenous fungal endophthalmitis and did not differ in this regard from prior study populations. Although other fungal species can infect the eye, 15 •16 Candida species are the most commonly reported cause of endogenous fungal endophthalmitis. 4 Of the various species of Candida, C. albicans is thought to be more likely to infect the eye. 9 The mechanism underlying this predilection of C. albicans to infect the eye is not understood completely but may involve the ability of this species to form germ tubes in serum that could embolize and lodge in the choriocapillaris in vivo. 17 We also found Candida species to represent the greatest percentage of fungal isolates among our patient population. The majority of the identified species were C. albicans, although a mixture of other Candida species also were isolated. Nonalbicans species of Candida have been reported and may be increasing in prevalence. 18 All three patients in our study with fungal chorioretinitis had C. albicans fungemia. We examined visual acuity to determine whether acuity would be a useful indicator of ocular fungal involvement in our patient population. In accord with previous studies, we found that visual acuity was not a useful prognostic factor in assessing the presence of endogenous fungal endophthalmitis. There are limitations of the current study. The first is the retrospective nature of the study. A prospective study in which treatment would be withheld from a random control group of patients would not be considered ethical, given the life-threatening nature of systemic fungemia. The second is that most of the consultations consisted of a single fundus examination, and we cannot be sure that fungal endophthalmitis did not develop in some patients subsequently. However, in a recent study examining the development of fungal endophthalmitis in nontreated, hospitalized patients with candidemia, ocular lesions developed within 72 hours of the suspected onset of fungemia in approximately 90% of patients. 6 Consultations typically were completed within this timeframe in our study population. If this assumption is applied to the current study, the additional cases missed because of a single evaluation would increase the incidence to approximately 3.1%. This number still is much lower than reported previously. Finally, because the consultation ser-
vice at our institution is staffed on a rotating basis, examinations were performed by multiple examiners during the period studied. In conclusion, we report the incidence of endogenous fungal endophthalmitis in a systemically infected inpatient population already receiving adequate antifungal therapy to be 2.8% (3 of 107). This figure is substantially lower than that reported in all previous studies. We thoroughly examined factors that could be responsible for this low incidence of ocular fungal infection. Our patients were found to be at equal risk for the development of endogenous fungal endophthalmitis when compared to those of previous studies. As in previous studies, fungal involvement was either known or believed to be systemic in all patients and was secondary to Candida species in most patients. The main discernible difference between our study and past investigations was the prompt initiation of early systemic antifungal therapy. We therefore conclude that early recognition and adequate treatment of systemic fungal infection substantially decrease the incidence of endogenous fungal endophthalmitis. However, given the lack of prospective data and the potentially lethal nature of disseminated candidiasis, we continue to recommend ophthalmologic examination of any patient with evidence of systemic fungal infection.
References 1. Wade RP. Epidemiology of Candida infections In: Bodey GP, ed. Candidiasis: Pathogenesis, Diagnosis, and Treatment, 2nd ed. New York: Raven Press, 1993;85-107. 2. Klein JJ, Watanakunakorn CC. Hospital-acquired fungemia. Its natural course and clinical significance. Am J Med 1979;67:51-8. 3. Edwards JE Jr, Foos RY, Montgomerie JZ, Guze LB. Ocular manifestations of Candida septicemia: review of seventy-six cases of hematogenous Candida endophthalmitis. Medicine (Baltimore) 1974;53:47 -75. 4. Pettit TH, Edwards JE Jr, Purdy EP, Bullock JD. Endogenous fungal endophthalmitis. In: Pepose JS, Holland GN, Wilhelmus KR, eds. Ocular Infection and Immunity. St. Louis: Mosby, 1996; 1262-85. 5. Fishman LS, Griffin JR, Sapico FL, Hecht R. Hematogenous Candida endophthalmitis-a complication of candidemia. N Eng! J Med 1972;286:675-81. 6. Brooks RG. Prospective study of Candida endophthalmitis in hospitalized patients with candidemia. Arch Intern Med 1989; 149:2226-8. 7. Henderson DK, Edwards JE Jr, Montgomerie JZ. Hematogenous Candida endophthalmitis in patients receiving parenteral hyperalimentation fluids. J Infect Dis 1981; 143: 655-61. 8. Parke DW II, Jones DB, Gentry LO. Endogenous endophthalmitis among patients with candidemia. Ophthalmology 1982; 89:789-96. 9. Griffin JR, Petit TH, Fishman LS, et al. Blood-borne Candida endophthalmitis: a clinical and pathologic study of 21 cases. Arch Ophthalmol 1973;89:450-6. 10. Griffin JR, Foos RY, Petit TH. Relationship between candida! endophtha1mitis, candidemia, and disseminated candidiasis. In: Concilium Ophthalmologicum, 22nd Annual Meeting, Paris: Masson, 1974;2:661-4.
1597
Ophthalmology
Volume 104, Number 10, October 1997
11. Gold JWM. Infections due to fungi, Actinomyces and Nocardia. In: Reece GE, Betts RG, eds. A Practical Approach to Infectious Disease. Boston: Little, Brown, 1991; 512-27. 12. Burgess JL, Birchall R. Nephrotoxicity of amphotericin B, with emphasis on changes in tubular function. Am J Med 1972;53:77-84. 13. Ellis CA, Spivack ML. The significance of candidemia. Ann Intern Med 1967;67:511-22. 14. Friedberg MA, Rapuano CJ, eds. Wills Eye Hospital office and emergency room diagnosis and treatment of eye dis. ease. Philadelphia: Lippincott, 1990;383-5.
1598
15. Gross JG. Endogenous Aspergillus-induced endophthalmitis: successful treatment without systemic antifungal medication. Retina 1992;12:341-5. 16. Hunt KE, Glasgow BJ. Aspergillus endophthalmitis. An unrecognized endemic disease in orthotopic liver transplantation. Ophthalmology 1996; 103:757-67. 17. Moyer DY, Edwards JE. Candida! endophthalmitis and central nervous system infection. In: Bodey GP, ed. Candidiasis: Pathogenesis, Diagnosis, and Treatment, 2nd ed. New York: Raven Press, 1993;331-55. 18. Joshi N, Hamory BH. Endophthalmitis caused by non-albicans species of Candida. Rev Infect Dis 1991; 13:281-7 .