- Email: [email protected]
Pulmonary Fungal Infection
Pulmonary Fungal Infection* Emphasis on Microbiological Spectra, Patient Outcome, and Prognostic Factors Kuan-Yu Chen, MD; Shiann-Chin Ko, MD; Po-Ren Hsueh, MD; Kwen-Tay Luh, MD, FCCP; and Pan-Chyr Yang, MD, PhD, FCCP
Study objectives: To investigate the microbiological spectra, patient outcome, and prognostic factors of pulmonary fungal infection. Design: The medical and microbiological records of patients with pulmonary fungal infection were retrospectively analyzed. Setting: A university-affiliated tertiary medical center. Patients and methods: From January 1988 to December 1997, all cases of pulmonary fungal infection were reviewed. The criteria for inclusion were obvious lung lesion shown on chest radiographs and one of the following: (1) the presence of fungi in or isolation of fungi from the biopsy specimen of open thoracotomy, thoracoscopy, transbronchial lung biopsy, or ultrasoundguided percutaneous needle aspiration/biopsy; or (2) isolation of fungi from pleural effusion or blood, with no evidence of extrapulmonary infection. Results: A total of 140 patients were included. Ninety-four cases of pulmonary fungal infection (67%) were community acquired. The most frequently encountered fungi were Aspergillus species (57%), followed by Cryptococcus species (21%) and Candida species (14%). There were 72 patients with acute invasive fungal infection, with a mortality rate of 67%. Multivariate logistic regression analysis showed that nosocomial infection (p ⴝ 0.014) and respiratory failure (p ⴝ 0.001) were significantly and independently associated with death of acute invasive fungal infection. Conclusions: Pulmonary fungal infection of community-acquired origins is becoming a serious problem. It should be taken into consideration for differential diagnosis of community-acquired pneumonia. Furthermore, acute invasive fungal infection is associated with a much higher mortality rate for patients with nosocomial infection or complicating respiratory failure. Early diagnosis with prompt antifungal therapy, or even with surgical intervention, might be warranted to save patients’ lives. (CHEST 2001; 120:177–184) Key words: community-acquired; fungal pneumonia; nosocomial pneumonia Abbreviation: RR ⫽ relative risk
infection has emerged as a worldwide F ungal health-care problem in the last decade. The 1– 4
increasing incidence of fungal infection has also become a serious problem in Taiwan. From 1980 to 1994, a 27-fold increase in bloodstream infections due to Candida species was observed at National Taiwan University Hospital.5 However, the prevalence and prognosis of pulmonary fungal *From the Departments of Internal Medicine (Drs. Chen, Ko, Hsueh, and Yang) and Laboratory Medicine (Dr. Luh), National Taiwan University Hospital, Taipei, Taiwan. Manusript received August 1, 2000; revision accepted February 14, 2001. Correspondence to: Pan-Chyr Yang, MD, PhD, FCCP, Department of Internal Medicine, National Taiwan University Hospital, No. 7 Chung-Shan South Rd, Taipei 100, Taiwan; e-mail: [email protected]
infection has been difficult to evaluate since diagnoses were seldom confirmed. Fungi isolated from sputum may represent either pathogens or saprophytes. A previous report demonstrated that Candida colonization could be found in respiratory samples obtained by BAL, endotracheal aspirate, or protected specimen brushing in critically ill patients.6 Therefore, identification of fungi in biopsy specimens obtained by invasive diagnostic procedures such as bronchoscopy, thoracoscopy, or thoracotomy was usually necessary to confirm the diagnosis. In this study, we reviewed the clinical features of patients with pulmonary fungal infection using strict criteria to analyze the microbiological spectra, patient outcome, and prognostic factors of this disease. CHEST / 120 / 1 / JULY, 2001
177
Materials and Methods The medical records of consecutive patients with diagnoses of pulmonary fungal infection treated at National Taiwan University Hospital during the period of January 1988 through December 1997 were reviewed. The inclusion of a patient required that the following criteria were met: obvious lung lesion shown on chest radiography and at least one of the following: (1) identification of fungi in biopsy specimens obtained from open thoracotomy, thoracoscopy, transbronchial lung biopsy, or ultrasound-guided percutaneous needle biopsy; (2) isolation of fungi species from specimens of open thoracotomy biopsy, thoracoscopic biopsy, transbronchial lung biopsy, or ultrasound-guided percutaneous needle aspiration/biopsy; (3) isolation of fungal species from pleural effusion or blood, with no evidence of extrapulmonary infection. Pulmonary cryptococcal infection was also defined as cryptococci identified by microscopic examination of percutaneous needle aspiration fluid or positive cryptococcal antigen (an antigen titer equal to or ⬎ 1:8) in percutaneous needle aspiration fluid,7 with resolution of pulmonary lung lesion after antifungal therapy. Acute invasive fungal infection was defined as patients with rapid progression of disease clinically or radiographically, and at least one of the followings: (1) fungal invasion of lung parenchyma noted by pathologic examination or (2) fungal isolation from sterile site such as blood or pleural effusion. Pulmonary fungal infection developing after 48 h of hospital admission was regarded as nosocomial; otherwise, the fungal infection was considered community acquired. Previous use of antibiotics was defined as receiving antibiotic therapy before hospital admission or for ⬎ 48 h during hospital admission before a clinical diagnosis of fungal pneumonia was made. Antecedent chemotherapy was defined as receiving anticancer chemotherapy ⬍ 1 month before clinical diagnosis of fungal pneumonia. Long-term steroid use was defined as receiving steroid therapy for ⬎ 1 month before clinical diagnosis of fungal pneumonia. Fever was defined as body temperature ⬎ 37.5°C. Neutropenia was defined as absolute neutrophil count ⬍ 1,000/L in peripheral blood. Complicating fungemia was defined as the same fungal species were identified both in the lung and blood. Respiratory failure was defined as Pao2 ⬍ 60 mm Hg and/or Paco2 ⱖ 50 mm Hg while breathing room air. For isolation of fungi, specimens were inoculated onto Sabouraud dextrose agar plates (BBL Microbiology Systems; Cockeysville, MD). Identification of molds was based on gross colony morphologies and microscopic pictures. Cornmeal agar (BBL Microbiology Systems) slide cultures were used to identify molds. Yeasts were identified to species level by standard method, and their identity was confirmed by the API ID 32C (bioMerieux; Marcy-I’Etoile, France).8 The following data were collected for each patient: age and gender; predisposing factors, including underlying diseases and associated medical conditions; antimicrobial agents administered; clinical symptoms; peripheral WBC and differential cell counts; findings in chest radiographs; strains of pathogens isolated or found in biopsy specimens; regimens and durations of antifungal therapy; invasive or surgical procedures; durations of hospitalization; and patient outcome. Differences in survival among subgroups of variables were analyzed by 2 test, or Fisher’s exact test when necessary. A multivariate logistic regression model was applied (SPSS 8.0 for Windows; SPSS; Chicago, IL) with only variables that were significantly associated with survival in the univariate analysis. A p value of ⱕ 0.05 was considered significant. 178
Results Clinical Characteristics From January 1988 to December 1997, medical and microbiological records of 187 patients with pulmonary fungal infection were reviewed. Of these, 140 patients fulfilled the inclusion criteria by various diagnostic methods (Table 1). There were 74 patients included by pathologic examinations of biopsy specimens. Fifty-eight patients had fungi isolated from biopsy specimens, aspirated fluids, pleural effusions, or blood. Three patients had cryptococci confirmed by microscopic examinations of aspirated fluids, and five patients had positive cryptococcal antigen detected in the aspirated fluids. Of the 140 cases included in the analysis, there were 5 cases in 1988, 8 in 1989, and 6 in 1990. There were 10, 11, and 8 cases annually in 1991, 1992, and 1993, respectively. However, there were ⬎ 15 cases annually from 1994 through 1997 (16 in 1994, 26 in 1995, 20 in 1996, and 30 in 1997), with increases in both nosocomial and community-acquired infections (Fig 1). The demographic features of the 140 patients are summarized in Table 2, including 72 patients with acute invasive fungal infection and 68 with mycetoma. The majority of patients were male in both patient groups. No significant difference in age and gender ratio was demonstrated between both groups. There were more nosocomial infections in patients with acute invasive fungal infection than in those with mycetoma (54% vs 10%; p ⬍ 0.001). One hundred nineteen patients (85%) had underlying diseases or associated medical conditions. Eighty-
Table 1—Diagnostic Methods for the 140 Patients With Pulmonary Fungal Infection Methods Pathologic Thoracotomy or thoracoscopy Percutaneous ultrasound-guided Transbronchial lung biopsy Necropsy† Microbiological Thoracotomy or thoracoscopy Percutaneous ultrasound-guided Transbronchial lung biopsy Pleural effusion Blood Microscopic‡ Percutaneous ultrasound-guided Serologic‡ Percutaneous ultrasound-guided
Data
biopsy
biopsy
aspiration aspiration
74 (53) 50 (36) 9 (6) 10 (7) 5 (4) 58 (41) 8 (6) 26 (19) 3 (2) 16 (11) 5 (4) 3 (2) 3 (2) 5 (4) 5 (4)
*Data are presented as No. (%). †Including one patient who coughed up a piece of necrotic lung tissue that proved to be aspergillosis by pathologic examination. ‡Diseases diagnosed by microscopy or serology were all cryptococcosis. Clinical Investigations
Figure 1. Annual incidence of pulmonary fungal infection during the 10-year period in National Taiwan University Hospital.
five patients had compromised immunity due to (in order of decreasing frequency) malignancy, diabetes mellitus, long-term steroid use, cirrhosis of the liver, organ transplantation, uremia, alcoholism, and AIDS. The rate of immunocompromised status was higher in patients with acute invasive fungal infection (89% vs 31%; p ⬍ 0.001). However, previous tuberculous infection was more frequent in patients
with mycetoma, mostly aspergilloma (53% vs 11%; p ⬍ 0.001). The most common presentations among the 140 patients were cough (63%), fever (54%), and dyspnea (47%). Peripheral WBC counts were ⬎ 20,000/L in 17 patients (12%). Pulmonary fungal infection developed in 28 patients (20%) during granulocytopenia after chemotherapy. The most frequent radiographic manifestations were single nodular lesions (n ⫽ 74; 53%), followed by consolidations (n ⫽ 45; 32%), multiple nodular lesions (n ⫽ 14; 10%), and interstitial pattern (n ⫽ 7; 5%). Seventy-two patients (51%) had cavitary lesions, including 6 patients with abscess formation confirmed by percutaneous aspiration. Thirty-four patients (24%) had pleural effusions. There were 99 patients with unilateral lesions and 41 with bilateral lesions. For patients with unilateral involvement of pulmonary fungal infection, upper lobes (right upper lobe, 21%; left upper lobe, 12%) were the most common lobes of distribution, followed by lower lobes (right lower lobe, 15%; left lower lobe, 9%). The most common radiologic manifestation in pa-
Table 2—Demographic Data, Clinical Courses, Treatments, and Outcomes of Different Patient Groups With Pulmonary Fungal Infection*
Variables Mean age, yr Male/female gender, No. Male patients, % Hospital acquired ICU acquired Underlying conditions Immunocompromised Malignancy Hematologic (leukemia and lymphoma) Lung cancer Others† Diabetes mellitus Chemotherapy Long-term steroid use Organ transplantation AIDS Uremia Cirrhosis of the liver COPD Previous TB infection None Mean (⫾ SD) hospital stay, d Antifungal therapy Amphotericin B Fluconazole Itraconazole Operation ICU admission Death
Total (n ⫽ 140)
Acute Invasive Fungal Infection (n ⫽ 72)
Mycetoma (n ⫽ 68)
45 82/58 59 46 (33) 38 (27)
41 44/28 60 39 (54) 33 (46)
49 38/30 56 7 (10) 5 (7)
85 (61) 38 (27) 29 (21) 3 (2) 6 (4) 27 (19) 23 (16) 22 (15) 8 (6) 7 (5) 5 (3) 1 (1) 7 (5) 44 (31) 21 (15) 47 ⫾ 57 98 (70) 65 (46) 61 (44) 10 (7) 63 (45) 38 (27) 51 (36)
64 (89) 33 (46) 27 (38) 1 (1) 4 (6) 11 (15) 21 (29) 19 (26) 8 (11) 7 (10) 5 (7) 1 (1) 5 (7) 8 (11) 2 (3) 69 ⫾ 67 68 (94) 57 (79) 37 (51) 8 (11) 12 (17) 33 (46) 48 (67)
21 (31) 5 (7) 2 (3) 2 (3) 2 (3) 16 (24) 2 (3) 3 (4) 0 (0) 0 (0) 0 (0) 0 (0) 2 (3) 36 (53) 19 (28) 24 ⫾ 29 30 (44) 8 (12) 24 (35) 2 (3) 51 (75) 5 (7) 3 (4)
*Data are presented as No. (%) unless otherwise indicated. TB ⫽ tuberculosis. †Other malignancies included breast (n ⫽ 2), urinary bladder (n ⫽ 1), nasopharyngeal (n ⫽ 1), hepatocellular (n ⫽ 1), and esophageal (n ⫽ 1). CHEST / 120 / 1 / JULY, 2001
179
tients with Aspergillus infection was nodular type (n ⫽ 60; 74.1%; Fig 2, top, left), followed by consolidation (n ⫽ 21; 25.9%; Fig 2, top, right). One patient with invasive aspergillosis had concurrent
interstitial pattern of chest radiographs. Nodular pattern was most commonly found in chest radiography of patients with cryptococcal infection (n ⫽ 23; 76.7%; Fig 2, middle, left), followed by
Figure 2. Top, left: Aspergilloma. A 22-year-old man with diabetes mellitus presented with hemoptysis. Chest radiography shows a cavitary nodular lesion with air-crescent in the right lower lung. He underwent wedge resection and Aspergillus species was found. Top, right: Invasive aspergillosis. A 26-year-old woman with acute lymphoblastic leukemia status post bone marrow transplantation presented with fever. Chest radiography shows a consolidation in the left upper lung field. Aspergillus niger was isolated from specimen by percutaneous fine-needle aspiration. Middle, left: Cryptococcoma. A 53-year-old woman without any underlying diseases presented with cough. A posteroanterior chest radiograph shows a well-defined nodule in lateral aspect of the right middle lung area. She underwent thoracoscopic resection, and pathology findings revealed a cryptococcoma. Middle, right: Cryptococcal pneumonia. A 34-year-old man without any underlying diseases presented with dry cough, fever, and dyspnea. Chest radiography shows consolidations in the bilateral lower lung fields. The cryptococcal antigen titer of the aspiration specimen was 1:2,560. Bottom, center: Candida pneumonia. A 45-year-old woman with acute myeloblastic leukemia presented with fever. Chest radiography shows a lobar consolidation in the right upper lung field. C albicans was isolated from specimen of percutaneous fine-needle biopsy. 180
Clinical Investigations
consolidation (n ⫽ 5; 16.7%; Fig 2, middle, right) and interstitial pattern (n ⫽ 2; 6.7%). As for patients with Candida infection, consolidation was the most frequent pattern found (n ⫽ 17; 87%; Fig 2, bottom, center), followed by nodular pattern (n ⫽ 1; 5%) and interstitial pattern (n ⫽ 2; 10%). Microbiology The fungi identified from the 140 patients are shown in Table 3. The major fungal species encountered were Aspergillus (n ⫽ 80, 57%), Cryptococcus (n ⫽ 30, 20%), and Candida species (n ⫽ 20; 14%). Various types of pulmonary infection caused by different fungal species are listed in Table 4. For patients with nosocomial pulmonary infections, the most common pathogen isolated was Aspergillus species (n ⫽ 27), followed by Candida species (n ⫽ 16) and Cryptococcus species (n ⫽ 1). Eighteen patients had complicating fungemia. The major fungi isolated from blood were Candida albicans, Aspergillus fumigatus, and Aspergillus flavus (respectively, n ⫽ 3; 16%). Aspergillus species were the leading pathogen in both patient groups with acute invasive infection and mycetoma. Among patients with Aspergillus infection, there was no difference in the incidences of nosocomial and community-acquired infections (p ⫽ 0.722). Aspergilloma (n ⫽ 41; 29%) was commonly encountered. For these patients, most (n ⫽ 34) had previous history of pulmonary tuberculosis. Five patients had aspergilloma complicated by active tuberculosis, verified by microscopic examina-
Table 3—Fungal Species Identified From 140 Patients With Pulmonary Fungal Infections Fungal species
No.
Aspergillus spp Aspergillus fumigatus Aspergillus flavus Aspergillus niger Aspergillus terreus Cryptococcus spp Cryptococcus neoformans Candida spp Candida albicans Candida tropicalis Candida glabrata Candida parapsilosis Others Mycelium sterila Penicillium marneffei Coccidioides immitis Mucor spp Rhizopus spp Histoplasma capsulatum Rhodotorula glutinis
80 12 9 2 1 30 30 20 9 4 4 2 3 2 1 1 1 1 1
Table 4 —Patients and Mortality Rate With Different Patterns of Pulmonary Fungal Infections*
Total Aspergillus spp Aspergilloma Invasive aspergillosis† Systemic Cryptococcus spp Cryptococcoma Pneumonia Systemic Candida spp Pneumonia‡ Systemic Others Mycetoma Pneumonia Systemic
Patients, No.
Mortality, No. (%)
140 80 41 33 6 30 23 5 2 20 12 8 10 4 4 2
51 (36.4) 32 (40.0) 3 (7.3) 24 (72.7) 5 (83.3) 2 (6.7) 0 (0) 0 (0) 2 (100) 14 (70.0) 9 (75.0) 5 (62.5) 3 (30.0) 0 (0) 2 (50.0) 1 (50.0)
*Systemic ⫽ pneumonia plus concurrent fungemia. †Including two patients with lung abscess and two patients with endobronchial lesion. ‡Candida pneumonia, including three patients with lung abscess.
tion of biopsy specimens. Hemoptysis (n ⫽ 32) was the main symptom of patients with aspergilloma, with nine patients having life-threatening episodes. Five patients underwent transarterial embolization. Even after surgical intervention, four patients still had episodes of hemoptysis. Most patients (n ⫽ 29; 97%) had cryptococcal infection that was considered community acquired. Only one patient acquired cryptococcal infection in the hospital. Cryptococcoma (n ⫽ 23; 76%) was the most common presentation of cryptococcal infection. No patients with cryptococcoma died. For Candida infection, nosocomial infection was more prevalent than community-acquired infection (n ⫽ 16 vs n ⫽ 4; p ⬍ 0.001). Primary Candida pneumonia was the most frequent manifestation (pulmonary consolidation as the initial radiographic presentation), followed by disseminated infection (pneumonia plus candidemia). The mortality rate (70%) of patients with pulmonary Candida infection was higher than that of any other group of patients (Table 4). During hospital admission, 39 bacterial isolates were concomitantly found in the blood of 24 patients. The majority of bacterial isolates were Gramnegative bacilli (24 isolates; 61%), of which the most common isolates were Pseudomonas aeruginosa (5 isolates; 15%), Klebsiella pneumoniae (5 isolates; 15%), and Enterobacter cloacae (5 isolates; 15%). The most common Gram-positive isolates were methicillin-resistant Staphylococcus aureus (4 isolates; 10%), followed by methicillin-resistant Staphylococcus epidermidis (3 isolates; 8%), and Enterococcus faecalis (3 isolates; 8%). CHEST / 120 / 1 / JULY, 2001
181
Treatment and Outcome The clinical courses, treatments and outcomes of different patient groups with pulmonary fungal infections were also demonstrated in Table 2. The mean hospital stay is longer in patients with acute invasive fungal infection than those with mycetoma (69 ⫾ 67 days vs 24 ⫾ 29 days; p ⬍ 0.001). A total of 98 patients received systemic antifungal therapy (acute invasive fungal infection vs mycetoma, 94% vs 44%, respectively; p ⬍ 0.001), with amphotericin B (79% vs 12%; mean duration, 33 days vs 27 days), fluconazole (51% vs 35%; mean duration, 47 days vs 55 days), or itraconazole (11% vs 3%; mean duration, 74 days vs 93 days). The use of antifungal agents did not influence the outcome (mortality rates, with vs without antifungal agents, 66% vs 75%) of patients with acute invasive fungal infection. Sixty-three patients underwent surgical resection of pulmonary lesions (acute invasive fungal infection vs mycetoma, 19% vs 81%). Thirty-eight patients were admitted to the ICU (acute invasive fungal infection vs mycetoma, 45.8% vs 7.4%). Fifty-one patients died, and 47 of these had compromised immunity. The mortality rate is much higher in patients with acute invasive fungal infection than those with mycetoma (67% vs 4%, respectively; p ⬍ 0.001). Because of the relatively higher mortality rate (67%), we studied the clinical characteristics and outcomes of patients with acute invasive fungal infection. Table 5 summarizes clinical, epidemiologic, microbiological, and laboratory variables of patients with acute invasive fungal infection studied,
as well as their relative influence on mortality (univariate analysis). Patients with compromised immunity (relative risk [RR], 7.67), absolute neutrophil count ⬍ 1,000/L (RR, 3.50), previous antibiotic use (RR, 9.23), and nosocomial fungal infection (RR, 9.23) had a higher risk of mortality. Patients who had chest radiographic manifestations of bilateral involvement (RR, 3.80) also had a higher risk of death. Patients who had complications of respiratory failure (RR, 20.9) were more likely to die than those who did not have these complications. Multivariate logistic regression analysis showed that hospital-acquired infection (p ⫽ 0.014) and respiratory failure (p ⫽ 0.001) were significantly and independently associated with death. Discussion There has been an increasing incidence of pulmonary fungal infection at our hospital in the past decade. A large portion was acquired from sources other than the hospital. The most frequently encountered pathogens were Aspergillus species, followed by Cryptococcus species and Candida species. The major prognostic factors of acute invasive fungal infection were the source of fungal infection and the occurrence of respiratory failure. Previous literature was concerned mostly with invasive and systemic fungal infections in patients with cancer,2 organ transplantation,9 –11 and AIDS.12 To our knowledge, there has been no large-scale study on the incidence, microbiological spectra, pa-
Table 5—Prognostic Factors and RRs of Death in the 72 Patients With Active Invasive Fungal Infection* Variables Underlying conditions Immunocompetent Immunocompromised ANC ⬎ 1,000/L ANC ⬍ 1,000/L Previous antibiotic use No Yes Clinical features Nosocomial No Yes Chest radiography Unilateral Bilateral Complication Respiratory failure No Yes
Patients in Each Category, No.
Patients Who Died, No. (%)
Univariate analysis: RR of Death Within Each Category (95% CI)
Multivariate Logistic Regression p Value
8 64 44 28
2 (25) 46 (72) 25 (57) 23 (82)
1.0 7.67 (1.41, 41.57) 1.0 3.50 (1.12, 10.89)
27 45
10 (37) 38 (84)
1.0 9.23 (3.00, 28.35)
0.343
33 39
14 (42) 34 (87)
1.0 9.23 (2.88, 29.59)
0.014
43 29
24 (56) 24 (83)
1.0 3.80 (1.22, 11.84)
0.232
22 50
5 (23) 43 (86)
1.0 20.89 (5.82, 74.94)
0.001
0.053 0.369
*ANC ⫽ absolute neutrophil count; CI ⫽ confidence interval. 182
Clinical Investigations
tient outcome, and prognosis of pulmonary fungal infection. Thus, we reviewed the patients with pulmonary fungal infection during a 10-year period and found the incidence of the disease rising in the recent years. A number of factors may be associated with an increasing incidence of pulmonary fungal infection. Impaired T-lymphocyte function due to high-dose steroid therapy, chemotherapy, or AIDS, as well as depressed neutrophil count or function due to hematologic malignancies, or anticancer chemotherapy, may increase the risk of fungal infection.13 The use of broad-spectrum antibiotics in critically ill patients may change normal flora, especially overgrowth of Candida species in the GI tract, which may translocate into the bloodstream.14 Hyperalimentation and invasive devices, including central vascular catheter, urinary catheter, and chest tube, may also contribute to the development of fungal infection.15 The main problem in dealing with pulmonary fungal infection is in distinguishing simple colonization from invasive or disseminated infection. A diagnosis of invasive disease requires the presence of the fungus in normally sterile tissues, while dissemination is defined as invasion of noncontiguous organs secondary to hematogenous spread.16 In our study, we used strict criteria to define pulmonary fungal infection and excluded patients who had fungi isolated from airway secretion. Therefore, the prevalence of fungal pneumonia may be underestimated. A large portion (68%) of pulmonary fungal infection was community acquired in our study. This suggests that, in addition to nosocomial infection,17 community-acquired fungal infections may also cause significant morbidity and mortality. We demonstrated that Aspergillus, Cryptococcus, and Candida species, often responsible for opportunistic infection in an immune-compromised host,18,19 were the most common pathogens. However, among the patients included, there were 57 apparently immune-competent patients with pulmonary fungal infections, including 23 patients with no underlying disease. This implies that these fungi may be also an important cause of pulmonary infection in relatively healthy patients. Aspergillus species are found worldwide in the environment and are acquired primarily through the respiratory tract. Though known as a soil fungus, Asperillus spores have often been found in hospital air, including inside the hospital operating rooms.20 Thus, exposure to Aspergillus is almost universal. The National Nosocomial Infections Surveillance study of the Centers for Disease Control and Prevention reveals that Candida species accounted for 19,621 infections (72%) and represented the sixth most common nosocomial pathogens in the United
States. Aspergillus species, however, accounted for only 1.1% of fungal infections.21 A population-based active laboratory surveillance for invasive fungal infection22 also reveals that Candida species (72.8 per million per year) was the most common pathogen, followed by Cryptococcus species (65.5), Coccidioides species (15.3), Aspergillus species (12.4), and Histoplasma species (7.1). For pulmonary fungal infection, Aspergillus species was considered as a leading cause of nosocomial fungal pneumonia.23 Our study demonstrated that Aspergillus species were the most common pathogen in patient groups with acute invasive infections. Aspergillus species were also most frequently encountered in either nosocomial or community-acquired patient groups, which indicates that Aspergillus species could be important pathogens of severe pulmonary fungal infection acquired from sources other than hospitals. Cryptococcus species was the second most common pathogen of pulmonary fungal infection in our study. In the majority of patients, cryptococcal infection was community acquired (97%). Only two patients died of pulmonary cryptococcal infection. In previous literature, pulmonary cryptococcosis may occur mostly in immune-compromised hosts.24 –26 The outcome of patients with cryptococcal infection in our study is relatively good. This may be due to the low frequency of immune-compromised status (38%). Candida infection accounted for the third most common cause of fungal pneumonia. Similar to a previous report,27 the mortality rate of Candida pneumonia (70%) is higher than that of the patients with pulmonary infection caused by other fungal species. Multivariate logistic regression analysis showed that nosocomial fungal infection and respiratory failure were significantly and independently associated with death of acute invasive pulmonary fungal infection. Patients with nosocomial fungal infection had a higher frequency of immune-compromised status (83%), which might be a contributory factor of mortality. They had a higher proportion of Candida pneumonia than the community-acquired patient group (26.1% vs 8.2%). The high mortality rate of Candida pneumonia (70%) might have also contributed to the higher mortality risk of patients with nosocomial fungal infection. The complication of respiratory failure might reflect the severity of pulmonary fungal infection or deterioration of underlying pulmonary function, which might be associated with mortality. In conclusion, pulmonary fungal infection, particularly community-acquired fungal infection, has become an emerging problem that deserves more clinical attention. Aspergillus species was a major pathogen of pulmonary fungal infection in Taiwan, in CHEST / 120 / 1 / JULY, 2001
183
either nosocomial or community-acquired patient groups. Patients who had nosocomial fungal infection or complicating respiratory failure had a high risk of death. Acute invasive fungal infection in almost all patients was treated as bacterial pneumonia initially, with poor outcome even with later antifungal therapy. For these patients, early diagnosis and more aggressive treatment for infection control, such as resection of the pulmonary lesion, might be necessary to improve their outcome. References 1 Harvey RL, Myers JP. Nosocomial fungemia in a large community teaching hospital. Arch Intern Med 1987; 147: 2117–2120 2 Horn R, Wong B, Kiehn TE, et al. Fungemia in a cancer hospital: changing frequency, earlier onset, and results of therapy. Rev Infect Dis 1985; 7:646 – 655 3 Fraser V J, Jones M, Dunkel J, et al. Candidemia in a tertiary care hospital: epidemiology, risk factors, and predictors of mortality. Clin Infect Dis 1992; 15:414 – 421 4 Henderson VJ, Hirvela ER. Emerging and reemerging microbial threats: nosocomial fungal infections. Arch Surg 1996; 131:330 –337 5 Hung CC, Chen YC, Chang SC, et al. Nosocomial candidemia in a university hospital in Taiwan. J Formos Med Assoc 1996; 95:19 –28 6 El-Ebiary M, Torres A, Fabregas NJ, et al. Significance of the isolation of Candida species from respiratory samples in critical ill, non-neutropenic patients. Am J Respir Crit Care Med 1997; 156:583–590 7 Liaw YS, Yang PC, Yu CJ, et al. Direct determination of cryptococcal antigen in transthoracic needle aspirate for diagnosis of pulmonary cryptococcosis. J Clin Microbiol 1995; 33:1588 –1591 8 Merz WG, Roberts GD. Detection and recovery of fungi from clinical specimens. In: Murray PR, Baron EJ, Pfaller MA, et al, eds. Manual of clinical microbiology. 6th ed. Washington, DC: American Society for Microbiology, 1995; 709 –722 9 Crawford SW. Bone-marrow transplantation and related infections. Semin Respir Infect 1993; 8:183–190 10 Collins LA, Samore MH, Roberts MS, et al. Risk factor for invasive fungal infections complicating orthotopic liver transplantation. J Infect Dis 1994; 170:644 – 652
184
11 Chugh KS, Sakhuja V, Jain S, et al. High mortality in systemic fungal infections following renal transplantation in thirdworld countries. Nephrol Dial Transplant 1993; 8:168 –172 12 Minamoto GY, Rosenberg AS. Fungal infection in patients with acquired immunodeficiency syndrome. Med Clin North Am 1997; 81:381– 409 13 Davies SF. Fungal pneumonia. Med Clin North Am 1994; 78:1049 –1065 14 Samonis G, Gikas A, Toloudis P, et al. Prospective study of the impact of broad-spectrum antibiotics on yeast flora of the human gut. Eur J Clin Microbiol 1994; 13:665– 667 15 Anaissie E, Solomkin JS. Fungal infection. In: Meakins JL, ed. Surgical infections: diagnosis and treatment (vol 1) New York, NY: Scientific American, 1994; 411– 425 16 Dean DA, Burchard KW. Fungal infection in surgical patients. Am J Surg 1996; 171:374 –382 17 Wenzel RP. Nosocomial candidemia: risk factors and attributable mortality. Clin Infect Dis 1995; 20:1531–1534 18 Guiot HF, Fibbe WE, van’t Wout JW. Risk factor for fungal infection in patients with malignant hematologic disorders: implications for empirical therapy and prophylaxis. Clin Infect Dis 1994; 18:525–532 19 Denning DW, Stevens DA. Antifungal and surgical treatment of invasive aspergillosis: review of 2121 published cases. Rev Infect Dis 1990; 12:1147–1201 20 Andriole VT. Infection with Aspergillus species. Clin Infect Dis 1993; 17(suppl):481– 486 21 Banerjee SN, Emori TG, Culver DH, et al. Secular trends in nosocomial primary bloodstream infections in the United States: the National Nosocomial Infections Surveillance System 1980 –1990. Am J Med 1991; 92(suppl 3B):86 – 89 22 Rees JR, Pinner RW, Hajjeh RA, et al. The epidemiologic features of invasive mycotic infections in the San Francisco Bay Area, 1992–1993: results of population-based laboratory active surveillance. Clin Infect Dis 1998; 27:1138 –1147 23 Saubolle MA. Mycology and the clinical laboratory in the diagnosis of respiratory mycoses. In: Sarosi GA, Davies SF, eds. Fungal diseases of the lung. 3rd ed. Philadelphia, PA: Lippincott William and Wilkins, 2000; 1–16 24 Haque AK. Pathology of common pulmonary fungal infections. J Thorac Imaging 1992; 7:1–11 25 Boyars MC, Zwischenberger JB, Cox CS Jr. Clinical manifestations of pulmonary fungal infections. J Thorac Imaging 1992; 7:12–22 26 Woodring JH, Ciporkin G, Lee C, et al. Pulmonary cryptococcosis. Semin Roentgenol 1996; 31:67–75 27 Haron E, Vartivarian S, Anaissie E, et al. Primary candida pneumonia: experience at a large cancer center and review of the literature. Medicine 1993; 72:137–142
Clinical Investigations
Study objectives: To investigate the microbiological spectra, patient outcome, and prognostic factors of pulmonary fungal infection. Design: The medical and microbiological records of patients with pulmonary fungal infection were retrospectively analyzed. Setting: A university-affiliated tertiary medical center. Patients and methods: From January 1988 to December 1997, all cases of pulmonary fungal infection were reviewed. The criteria for inclusion were obvious lung lesion shown on chest radiographs and one of the following: (1) the presence of fungi in or isolation of fungi from the biopsy specimen of open thoracotomy, thoracoscopy, transbronchial lung biopsy, or ultrasoundguided percutaneous needle aspiration/biopsy; or (2) isolation of fungi from pleural effusion or blood, with no evidence of extrapulmonary infection. Results: A total of 140 patients were included. Ninety-four cases of pulmonary fungal infection (67%) were community acquired. The most frequently encountered fungi were Aspergillus species (57%), followed by Cryptococcus species (21%) and Candida species (14%). There were 72 patients with acute invasive fungal infection, with a mortality rate of 67%. Multivariate logistic regression analysis showed that nosocomial infection (p ⴝ 0.014) and respiratory failure (p ⴝ 0.001) were significantly and independently associated with death of acute invasive fungal infection. Conclusions: Pulmonary fungal infection of community-acquired origins is becoming a serious problem. It should be taken into consideration for differential diagnosis of community-acquired pneumonia. Furthermore, acute invasive fungal infection is associated with a much higher mortality rate for patients with nosocomial infection or complicating respiratory failure. Early diagnosis with prompt antifungal therapy, or even with surgical intervention, might be warranted to save patients’ lives. (CHEST 2001; 120:177–184) Key words: community-acquired; fungal pneumonia; nosocomial pneumonia Abbreviation: RR ⫽ relative risk
infection has emerged as a worldwide F ungal health-care problem in the last decade. The 1– 4
increasing incidence of fungal infection has also become a serious problem in Taiwan. From 1980 to 1994, a 27-fold increase in bloodstream infections due to Candida species was observed at National Taiwan University Hospital.5 However, the prevalence and prognosis of pulmonary fungal *From the Departments of Internal Medicine (Drs. Chen, Ko, Hsueh, and Yang) and Laboratory Medicine (Dr. Luh), National Taiwan University Hospital, Taipei, Taiwan. Manusript received August 1, 2000; revision accepted February 14, 2001. Correspondence to: Pan-Chyr Yang, MD, PhD, FCCP, Department of Internal Medicine, National Taiwan University Hospital, No. 7 Chung-Shan South Rd, Taipei 100, Taiwan; e-mail: [email protected]
infection has been difficult to evaluate since diagnoses were seldom confirmed. Fungi isolated from sputum may represent either pathogens or saprophytes. A previous report demonstrated that Candida colonization could be found in respiratory samples obtained by BAL, endotracheal aspirate, or protected specimen brushing in critically ill patients.6 Therefore, identification of fungi in biopsy specimens obtained by invasive diagnostic procedures such as bronchoscopy, thoracoscopy, or thoracotomy was usually necessary to confirm the diagnosis. In this study, we reviewed the clinical features of patients with pulmonary fungal infection using strict criteria to analyze the microbiological spectra, patient outcome, and prognostic factors of this disease. CHEST / 120 / 1 / JULY, 2001
177
Materials and Methods The medical records of consecutive patients with diagnoses of pulmonary fungal infection treated at National Taiwan University Hospital during the period of January 1988 through December 1997 were reviewed. The inclusion of a patient required that the following criteria were met: obvious lung lesion shown on chest radiography and at least one of the following: (1) identification of fungi in biopsy specimens obtained from open thoracotomy, thoracoscopy, transbronchial lung biopsy, or ultrasound-guided percutaneous needle biopsy; (2) isolation of fungi species from specimens of open thoracotomy biopsy, thoracoscopic biopsy, transbronchial lung biopsy, or ultrasound-guided percutaneous needle aspiration/biopsy; (3) isolation of fungal species from pleural effusion or blood, with no evidence of extrapulmonary infection. Pulmonary cryptococcal infection was also defined as cryptococci identified by microscopic examination of percutaneous needle aspiration fluid or positive cryptococcal antigen (an antigen titer equal to or ⬎ 1:8) in percutaneous needle aspiration fluid,7 with resolution of pulmonary lung lesion after antifungal therapy. Acute invasive fungal infection was defined as patients with rapid progression of disease clinically or radiographically, and at least one of the followings: (1) fungal invasion of lung parenchyma noted by pathologic examination or (2) fungal isolation from sterile site such as blood or pleural effusion. Pulmonary fungal infection developing after 48 h of hospital admission was regarded as nosocomial; otherwise, the fungal infection was considered community acquired. Previous use of antibiotics was defined as receiving antibiotic therapy before hospital admission or for ⬎ 48 h during hospital admission before a clinical diagnosis of fungal pneumonia was made. Antecedent chemotherapy was defined as receiving anticancer chemotherapy ⬍ 1 month before clinical diagnosis of fungal pneumonia. Long-term steroid use was defined as receiving steroid therapy for ⬎ 1 month before clinical diagnosis of fungal pneumonia. Fever was defined as body temperature ⬎ 37.5°C. Neutropenia was defined as absolute neutrophil count ⬍ 1,000/L in peripheral blood. Complicating fungemia was defined as the same fungal species were identified both in the lung and blood. Respiratory failure was defined as Pao2 ⬍ 60 mm Hg and/or Paco2 ⱖ 50 mm Hg while breathing room air. For isolation of fungi, specimens were inoculated onto Sabouraud dextrose agar plates (BBL Microbiology Systems; Cockeysville, MD). Identification of molds was based on gross colony morphologies and microscopic pictures. Cornmeal agar (BBL Microbiology Systems) slide cultures were used to identify molds. Yeasts were identified to species level by standard method, and their identity was confirmed by the API ID 32C (bioMerieux; Marcy-I’Etoile, France).8 The following data were collected for each patient: age and gender; predisposing factors, including underlying diseases and associated medical conditions; antimicrobial agents administered; clinical symptoms; peripheral WBC and differential cell counts; findings in chest radiographs; strains of pathogens isolated or found in biopsy specimens; regimens and durations of antifungal therapy; invasive or surgical procedures; durations of hospitalization; and patient outcome. Differences in survival among subgroups of variables were analyzed by 2 test, or Fisher’s exact test when necessary. A multivariate logistic regression model was applied (SPSS 8.0 for Windows; SPSS; Chicago, IL) with only variables that were significantly associated with survival in the univariate analysis. A p value of ⱕ 0.05 was considered significant. 178
Results Clinical Characteristics From January 1988 to December 1997, medical and microbiological records of 187 patients with pulmonary fungal infection were reviewed. Of these, 140 patients fulfilled the inclusion criteria by various diagnostic methods (Table 1). There were 74 patients included by pathologic examinations of biopsy specimens. Fifty-eight patients had fungi isolated from biopsy specimens, aspirated fluids, pleural effusions, or blood. Three patients had cryptococci confirmed by microscopic examinations of aspirated fluids, and five patients had positive cryptococcal antigen detected in the aspirated fluids. Of the 140 cases included in the analysis, there were 5 cases in 1988, 8 in 1989, and 6 in 1990. There were 10, 11, and 8 cases annually in 1991, 1992, and 1993, respectively. However, there were ⬎ 15 cases annually from 1994 through 1997 (16 in 1994, 26 in 1995, 20 in 1996, and 30 in 1997), with increases in both nosocomial and community-acquired infections (Fig 1). The demographic features of the 140 patients are summarized in Table 2, including 72 patients with acute invasive fungal infection and 68 with mycetoma. The majority of patients were male in both patient groups. No significant difference in age and gender ratio was demonstrated between both groups. There were more nosocomial infections in patients with acute invasive fungal infection than in those with mycetoma (54% vs 10%; p ⬍ 0.001). One hundred nineteen patients (85%) had underlying diseases or associated medical conditions. Eighty-
Table 1—Diagnostic Methods for the 140 Patients With Pulmonary Fungal Infection Methods Pathologic Thoracotomy or thoracoscopy Percutaneous ultrasound-guided Transbronchial lung biopsy Necropsy† Microbiological Thoracotomy or thoracoscopy Percutaneous ultrasound-guided Transbronchial lung biopsy Pleural effusion Blood Microscopic‡ Percutaneous ultrasound-guided Serologic‡ Percutaneous ultrasound-guided
Data
biopsy
biopsy
aspiration aspiration
74 (53) 50 (36) 9 (6) 10 (7) 5 (4) 58 (41) 8 (6) 26 (19) 3 (2) 16 (11) 5 (4) 3 (2) 3 (2) 5 (4) 5 (4)
*Data are presented as No. (%). †Including one patient who coughed up a piece of necrotic lung tissue that proved to be aspergillosis by pathologic examination. ‡Diseases diagnosed by microscopy or serology were all cryptococcosis. Clinical Investigations
Figure 1. Annual incidence of pulmonary fungal infection during the 10-year period in National Taiwan University Hospital.
five patients had compromised immunity due to (in order of decreasing frequency) malignancy, diabetes mellitus, long-term steroid use, cirrhosis of the liver, organ transplantation, uremia, alcoholism, and AIDS. The rate of immunocompromised status was higher in patients with acute invasive fungal infection (89% vs 31%; p ⬍ 0.001). However, previous tuberculous infection was more frequent in patients
with mycetoma, mostly aspergilloma (53% vs 11%; p ⬍ 0.001). The most common presentations among the 140 patients were cough (63%), fever (54%), and dyspnea (47%). Peripheral WBC counts were ⬎ 20,000/L in 17 patients (12%). Pulmonary fungal infection developed in 28 patients (20%) during granulocytopenia after chemotherapy. The most frequent radiographic manifestations were single nodular lesions (n ⫽ 74; 53%), followed by consolidations (n ⫽ 45; 32%), multiple nodular lesions (n ⫽ 14; 10%), and interstitial pattern (n ⫽ 7; 5%). Seventy-two patients (51%) had cavitary lesions, including 6 patients with abscess formation confirmed by percutaneous aspiration. Thirty-four patients (24%) had pleural effusions. There were 99 patients with unilateral lesions and 41 with bilateral lesions. For patients with unilateral involvement of pulmonary fungal infection, upper lobes (right upper lobe, 21%; left upper lobe, 12%) were the most common lobes of distribution, followed by lower lobes (right lower lobe, 15%; left lower lobe, 9%). The most common radiologic manifestation in pa-
Table 2—Demographic Data, Clinical Courses, Treatments, and Outcomes of Different Patient Groups With Pulmonary Fungal Infection*
Variables Mean age, yr Male/female gender, No. Male patients, % Hospital acquired ICU acquired Underlying conditions Immunocompromised Malignancy Hematologic (leukemia and lymphoma) Lung cancer Others† Diabetes mellitus Chemotherapy Long-term steroid use Organ transplantation AIDS Uremia Cirrhosis of the liver COPD Previous TB infection None Mean (⫾ SD) hospital stay, d Antifungal therapy Amphotericin B Fluconazole Itraconazole Operation ICU admission Death
Total (n ⫽ 140)
Acute Invasive Fungal Infection (n ⫽ 72)
Mycetoma (n ⫽ 68)
45 82/58 59 46 (33) 38 (27)
41 44/28 60 39 (54) 33 (46)
49 38/30 56 7 (10) 5 (7)
85 (61) 38 (27) 29 (21) 3 (2) 6 (4) 27 (19) 23 (16) 22 (15) 8 (6) 7 (5) 5 (3) 1 (1) 7 (5) 44 (31) 21 (15) 47 ⫾ 57 98 (70) 65 (46) 61 (44) 10 (7) 63 (45) 38 (27) 51 (36)
64 (89) 33 (46) 27 (38) 1 (1) 4 (6) 11 (15) 21 (29) 19 (26) 8 (11) 7 (10) 5 (7) 1 (1) 5 (7) 8 (11) 2 (3) 69 ⫾ 67 68 (94) 57 (79) 37 (51) 8 (11) 12 (17) 33 (46) 48 (67)
21 (31) 5 (7) 2 (3) 2 (3) 2 (3) 16 (24) 2 (3) 3 (4) 0 (0) 0 (0) 0 (0) 0 (0) 2 (3) 36 (53) 19 (28) 24 ⫾ 29 30 (44) 8 (12) 24 (35) 2 (3) 51 (75) 5 (7) 3 (4)
*Data are presented as No. (%) unless otherwise indicated. TB ⫽ tuberculosis. †Other malignancies included breast (n ⫽ 2), urinary bladder (n ⫽ 1), nasopharyngeal (n ⫽ 1), hepatocellular (n ⫽ 1), and esophageal (n ⫽ 1). CHEST / 120 / 1 / JULY, 2001
179
tients with Aspergillus infection was nodular type (n ⫽ 60; 74.1%; Fig 2, top, left), followed by consolidation (n ⫽ 21; 25.9%; Fig 2, top, right). One patient with invasive aspergillosis had concurrent
interstitial pattern of chest radiographs. Nodular pattern was most commonly found in chest radiography of patients with cryptococcal infection (n ⫽ 23; 76.7%; Fig 2, middle, left), followed by
Figure 2. Top, left: Aspergilloma. A 22-year-old man with diabetes mellitus presented with hemoptysis. Chest radiography shows a cavitary nodular lesion with air-crescent in the right lower lung. He underwent wedge resection and Aspergillus species was found. Top, right: Invasive aspergillosis. A 26-year-old woman with acute lymphoblastic leukemia status post bone marrow transplantation presented with fever. Chest radiography shows a consolidation in the left upper lung field. Aspergillus niger was isolated from specimen by percutaneous fine-needle aspiration. Middle, left: Cryptococcoma. A 53-year-old woman without any underlying diseases presented with cough. A posteroanterior chest radiograph shows a well-defined nodule in lateral aspect of the right middle lung area. She underwent thoracoscopic resection, and pathology findings revealed a cryptococcoma. Middle, right: Cryptococcal pneumonia. A 34-year-old man without any underlying diseases presented with dry cough, fever, and dyspnea. Chest radiography shows consolidations in the bilateral lower lung fields. The cryptococcal antigen titer of the aspiration specimen was 1:2,560. Bottom, center: Candida pneumonia. A 45-year-old woman with acute myeloblastic leukemia presented with fever. Chest radiography shows a lobar consolidation in the right upper lung field. C albicans was isolated from specimen of percutaneous fine-needle biopsy. 180
Clinical Investigations
consolidation (n ⫽ 5; 16.7%; Fig 2, middle, right) and interstitial pattern (n ⫽ 2; 6.7%). As for patients with Candida infection, consolidation was the most frequent pattern found (n ⫽ 17; 87%; Fig 2, bottom, center), followed by nodular pattern (n ⫽ 1; 5%) and interstitial pattern (n ⫽ 2; 10%). Microbiology The fungi identified from the 140 patients are shown in Table 3. The major fungal species encountered were Aspergillus (n ⫽ 80, 57%), Cryptococcus (n ⫽ 30, 20%), and Candida species (n ⫽ 20; 14%). Various types of pulmonary infection caused by different fungal species are listed in Table 4. For patients with nosocomial pulmonary infections, the most common pathogen isolated was Aspergillus species (n ⫽ 27), followed by Candida species (n ⫽ 16) and Cryptococcus species (n ⫽ 1). Eighteen patients had complicating fungemia. The major fungi isolated from blood were Candida albicans, Aspergillus fumigatus, and Aspergillus flavus (respectively, n ⫽ 3; 16%). Aspergillus species were the leading pathogen in both patient groups with acute invasive infection and mycetoma. Among patients with Aspergillus infection, there was no difference in the incidences of nosocomial and community-acquired infections (p ⫽ 0.722). Aspergilloma (n ⫽ 41; 29%) was commonly encountered. For these patients, most (n ⫽ 34) had previous history of pulmonary tuberculosis. Five patients had aspergilloma complicated by active tuberculosis, verified by microscopic examina-
Table 3—Fungal Species Identified From 140 Patients With Pulmonary Fungal Infections Fungal species
No.
Aspergillus spp Aspergillus fumigatus Aspergillus flavus Aspergillus niger Aspergillus terreus Cryptococcus spp Cryptococcus neoformans Candida spp Candida albicans Candida tropicalis Candida glabrata Candida parapsilosis Others Mycelium sterila Penicillium marneffei Coccidioides immitis Mucor spp Rhizopus spp Histoplasma capsulatum Rhodotorula glutinis
80 12 9 2 1 30 30 20 9 4 4 2 3 2 1 1 1 1 1
Table 4 —Patients and Mortality Rate With Different Patterns of Pulmonary Fungal Infections*
Total Aspergillus spp Aspergilloma Invasive aspergillosis† Systemic Cryptococcus spp Cryptococcoma Pneumonia Systemic Candida spp Pneumonia‡ Systemic Others Mycetoma Pneumonia Systemic
Patients, No.
Mortality, No. (%)
140 80 41 33 6 30 23 5 2 20 12 8 10 4 4 2
51 (36.4) 32 (40.0) 3 (7.3) 24 (72.7) 5 (83.3) 2 (6.7) 0 (0) 0 (0) 2 (100) 14 (70.0) 9 (75.0) 5 (62.5) 3 (30.0) 0 (0) 2 (50.0) 1 (50.0)
*Systemic ⫽ pneumonia plus concurrent fungemia. †Including two patients with lung abscess and two patients with endobronchial lesion. ‡Candida pneumonia, including three patients with lung abscess.
tion of biopsy specimens. Hemoptysis (n ⫽ 32) was the main symptom of patients with aspergilloma, with nine patients having life-threatening episodes. Five patients underwent transarterial embolization. Even after surgical intervention, four patients still had episodes of hemoptysis. Most patients (n ⫽ 29; 97%) had cryptococcal infection that was considered community acquired. Only one patient acquired cryptococcal infection in the hospital. Cryptococcoma (n ⫽ 23; 76%) was the most common presentation of cryptococcal infection. No patients with cryptococcoma died. For Candida infection, nosocomial infection was more prevalent than community-acquired infection (n ⫽ 16 vs n ⫽ 4; p ⬍ 0.001). Primary Candida pneumonia was the most frequent manifestation (pulmonary consolidation as the initial radiographic presentation), followed by disseminated infection (pneumonia plus candidemia). The mortality rate (70%) of patients with pulmonary Candida infection was higher than that of any other group of patients (Table 4). During hospital admission, 39 bacterial isolates were concomitantly found in the blood of 24 patients. The majority of bacterial isolates were Gramnegative bacilli (24 isolates; 61%), of which the most common isolates were Pseudomonas aeruginosa (5 isolates; 15%), Klebsiella pneumoniae (5 isolates; 15%), and Enterobacter cloacae (5 isolates; 15%). The most common Gram-positive isolates were methicillin-resistant Staphylococcus aureus (4 isolates; 10%), followed by methicillin-resistant Staphylococcus epidermidis (3 isolates; 8%), and Enterococcus faecalis (3 isolates; 8%). CHEST / 120 / 1 / JULY, 2001
181
Treatment and Outcome The clinical courses, treatments and outcomes of different patient groups with pulmonary fungal infections were also demonstrated in Table 2. The mean hospital stay is longer in patients with acute invasive fungal infection than those with mycetoma (69 ⫾ 67 days vs 24 ⫾ 29 days; p ⬍ 0.001). A total of 98 patients received systemic antifungal therapy (acute invasive fungal infection vs mycetoma, 94% vs 44%, respectively; p ⬍ 0.001), with amphotericin B (79% vs 12%; mean duration, 33 days vs 27 days), fluconazole (51% vs 35%; mean duration, 47 days vs 55 days), or itraconazole (11% vs 3%; mean duration, 74 days vs 93 days). The use of antifungal agents did not influence the outcome (mortality rates, with vs without antifungal agents, 66% vs 75%) of patients with acute invasive fungal infection. Sixty-three patients underwent surgical resection of pulmonary lesions (acute invasive fungal infection vs mycetoma, 19% vs 81%). Thirty-eight patients were admitted to the ICU (acute invasive fungal infection vs mycetoma, 45.8% vs 7.4%). Fifty-one patients died, and 47 of these had compromised immunity. The mortality rate is much higher in patients with acute invasive fungal infection than those with mycetoma (67% vs 4%, respectively; p ⬍ 0.001). Because of the relatively higher mortality rate (67%), we studied the clinical characteristics and outcomes of patients with acute invasive fungal infection. Table 5 summarizes clinical, epidemiologic, microbiological, and laboratory variables of patients with acute invasive fungal infection studied,
as well as their relative influence on mortality (univariate analysis). Patients with compromised immunity (relative risk [RR], 7.67), absolute neutrophil count ⬍ 1,000/L (RR, 3.50), previous antibiotic use (RR, 9.23), and nosocomial fungal infection (RR, 9.23) had a higher risk of mortality. Patients who had chest radiographic manifestations of bilateral involvement (RR, 3.80) also had a higher risk of death. Patients who had complications of respiratory failure (RR, 20.9) were more likely to die than those who did not have these complications. Multivariate logistic regression analysis showed that hospital-acquired infection (p ⫽ 0.014) and respiratory failure (p ⫽ 0.001) were significantly and independently associated with death. Discussion There has been an increasing incidence of pulmonary fungal infection at our hospital in the past decade. A large portion was acquired from sources other than the hospital. The most frequently encountered pathogens were Aspergillus species, followed by Cryptococcus species and Candida species. The major prognostic factors of acute invasive fungal infection were the source of fungal infection and the occurrence of respiratory failure. Previous literature was concerned mostly with invasive and systemic fungal infections in patients with cancer,2 organ transplantation,9 –11 and AIDS.12 To our knowledge, there has been no large-scale study on the incidence, microbiological spectra, pa-
Table 5—Prognostic Factors and RRs of Death in the 72 Patients With Active Invasive Fungal Infection* Variables Underlying conditions Immunocompetent Immunocompromised ANC ⬎ 1,000/L ANC ⬍ 1,000/L Previous antibiotic use No Yes Clinical features Nosocomial No Yes Chest radiography Unilateral Bilateral Complication Respiratory failure No Yes
Patients in Each Category, No.
Patients Who Died, No. (%)
Univariate analysis: RR of Death Within Each Category (95% CI)
Multivariate Logistic Regression p Value
8 64 44 28
2 (25) 46 (72) 25 (57) 23 (82)
1.0 7.67 (1.41, 41.57) 1.0 3.50 (1.12, 10.89)
27 45
10 (37) 38 (84)
1.0 9.23 (3.00, 28.35)
0.343
33 39
14 (42) 34 (87)
1.0 9.23 (2.88, 29.59)
0.014
43 29
24 (56) 24 (83)
1.0 3.80 (1.22, 11.84)
0.232
22 50
5 (23) 43 (86)
1.0 20.89 (5.82, 74.94)
0.001
0.053 0.369
*ANC ⫽ absolute neutrophil count; CI ⫽ confidence interval. 182
Clinical Investigations
tient outcome, and prognosis of pulmonary fungal infection. Thus, we reviewed the patients with pulmonary fungal infection during a 10-year period and found the incidence of the disease rising in the recent years. A number of factors may be associated with an increasing incidence of pulmonary fungal infection. Impaired T-lymphocyte function due to high-dose steroid therapy, chemotherapy, or AIDS, as well as depressed neutrophil count or function due to hematologic malignancies, or anticancer chemotherapy, may increase the risk of fungal infection.13 The use of broad-spectrum antibiotics in critically ill patients may change normal flora, especially overgrowth of Candida species in the GI tract, which may translocate into the bloodstream.14 Hyperalimentation and invasive devices, including central vascular catheter, urinary catheter, and chest tube, may also contribute to the development of fungal infection.15 The main problem in dealing with pulmonary fungal infection is in distinguishing simple colonization from invasive or disseminated infection. A diagnosis of invasive disease requires the presence of the fungus in normally sterile tissues, while dissemination is defined as invasion of noncontiguous organs secondary to hematogenous spread.16 In our study, we used strict criteria to define pulmonary fungal infection and excluded patients who had fungi isolated from airway secretion. Therefore, the prevalence of fungal pneumonia may be underestimated. A large portion (68%) of pulmonary fungal infection was community acquired in our study. This suggests that, in addition to nosocomial infection,17 community-acquired fungal infections may also cause significant morbidity and mortality. We demonstrated that Aspergillus, Cryptococcus, and Candida species, often responsible for opportunistic infection in an immune-compromised host,18,19 were the most common pathogens. However, among the patients included, there were 57 apparently immune-competent patients with pulmonary fungal infections, including 23 patients with no underlying disease. This implies that these fungi may be also an important cause of pulmonary infection in relatively healthy patients. Aspergillus species are found worldwide in the environment and are acquired primarily through the respiratory tract. Though known as a soil fungus, Asperillus spores have often been found in hospital air, including inside the hospital operating rooms.20 Thus, exposure to Aspergillus is almost universal. The National Nosocomial Infections Surveillance study of the Centers for Disease Control and Prevention reveals that Candida species accounted for 19,621 infections (72%) and represented the sixth most common nosocomial pathogens in the United
States. Aspergillus species, however, accounted for only 1.1% of fungal infections.21 A population-based active laboratory surveillance for invasive fungal infection22 also reveals that Candida species (72.8 per million per year) was the most common pathogen, followed by Cryptococcus species (65.5), Coccidioides species (15.3), Aspergillus species (12.4), and Histoplasma species (7.1). For pulmonary fungal infection, Aspergillus species was considered as a leading cause of nosocomial fungal pneumonia.23 Our study demonstrated that Aspergillus species were the most common pathogen in patient groups with acute invasive infections. Aspergillus species were also most frequently encountered in either nosocomial or community-acquired patient groups, which indicates that Aspergillus species could be important pathogens of severe pulmonary fungal infection acquired from sources other than hospitals. Cryptococcus species was the second most common pathogen of pulmonary fungal infection in our study. In the majority of patients, cryptococcal infection was community acquired (97%). Only two patients died of pulmonary cryptococcal infection. In previous literature, pulmonary cryptococcosis may occur mostly in immune-compromised hosts.24 –26 The outcome of patients with cryptococcal infection in our study is relatively good. This may be due to the low frequency of immune-compromised status (38%). Candida infection accounted for the third most common cause of fungal pneumonia. Similar to a previous report,27 the mortality rate of Candida pneumonia (70%) is higher than that of the patients with pulmonary infection caused by other fungal species. Multivariate logistic regression analysis showed that nosocomial fungal infection and respiratory failure were significantly and independently associated with death of acute invasive pulmonary fungal infection. Patients with nosocomial fungal infection had a higher frequency of immune-compromised status (83%), which might be a contributory factor of mortality. They had a higher proportion of Candida pneumonia than the community-acquired patient group (26.1% vs 8.2%). The high mortality rate of Candida pneumonia (70%) might have also contributed to the higher mortality risk of patients with nosocomial fungal infection. The complication of respiratory failure might reflect the severity of pulmonary fungal infection or deterioration of underlying pulmonary function, which might be associated with mortality. In conclusion, pulmonary fungal infection, particularly community-acquired fungal infection, has become an emerging problem that deserves more clinical attention. Aspergillus species was a major pathogen of pulmonary fungal infection in Taiwan, in CHEST / 120 / 1 / JULY, 2001
183
either nosocomial or community-acquired patient groups. Patients who had nosocomial fungal infection or complicating respiratory failure had a high risk of death. Acute invasive fungal infection in almost all patients was treated as bacterial pneumonia initially, with poor outcome even with later antifungal therapy. For these patients, early diagnosis and more aggressive treatment for infection control, such as resection of the pulmonary lesion, might be necessary to improve their outcome. References 1 Harvey RL, Myers JP. Nosocomial fungemia in a large community teaching hospital. Arch Intern Med 1987; 147: 2117–2120 2 Horn R, Wong B, Kiehn TE, et al. Fungemia in a cancer hospital: changing frequency, earlier onset, and results of therapy. Rev Infect Dis 1985; 7:646 – 655 3 Fraser V J, Jones M, Dunkel J, et al. Candidemia in a tertiary care hospital: epidemiology, risk factors, and predictors of mortality. Clin Infect Dis 1992; 15:414 – 421 4 Henderson VJ, Hirvela ER. Emerging and reemerging microbial threats: nosocomial fungal infections. Arch Surg 1996; 131:330 –337 5 Hung CC, Chen YC, Chang SC, et al. Nosocomial candidemia in a university hospital in Taiwan. J Formos Med Assoc 1996; 95:19 –28 6 El-Ebiary M, Torres A, Fabregas NJ, et al. Significance of the isolation of Candida species from respiratory samples in critical ill, non-neutropenic patients. Am J Respir Crit Care Med 1997; 156:583–590 7 Liaw YS, Yang PC, Yu CJ, et al. Direct determination of cryptococcal antigen in transthoracic needle aspirate for diagnosis of pulmonary cryptococcosis. J Clin Microbiol 1995; 33:1588 –1591 8 Merz WG, Roberts GD. Detection and recovery of fungi from clinical specimens. In: Murray PR, Baron EJ, Pfaller MA, et al, eds. Manual of clinical microbiology. 6th ed. Washington, DC: American Society for Microbiology, 1995; 709 –722 9 Crawford SW. Bone-marrow transplantation and related infections. Semin Respir Infect 1993; 8:183–190 10 Collins LA, Samore MH, Roberts MS, et al. Risk factor for invasive fungal infections complicating orthotopic liver transplantation. J Infect Dis 1994; 170:644 – 652
184
11 Chugh KS, Sakhuja V, Jain S, et al. High mortality in systemic fungal infections following renal transplantation in thirdworld countries. Nephrol Dial Transplant 1993; 8:168 –172 12 Minamoto GY, Rosenberg AS. Fungal infection in patients with acquired immunodeficiency syndrome. Med Clin North Am 1997; 81:381– 409 13 Davies SF. Fungal pneumonia. Med Clin North Am 1994; 78:1049 –1065 14 Samonis G, Gikas A, Toloudis P, et al. Prospective study of the impact of broad-spectrum antibiotics on yeast flora of the human gut. Eur J Clin Microbiol 1994; 13:665– 667 15 Anaissie E, Solomkin JS. Fungal infection. In: Meakins JL, ed. Surgical infections: diagnosis and treatment (vol 1) New York, NY: Scientific American, 1994; 411– 425 16 Dean DA, Burchard KW. Fungal infection in surgical patients. Am J Surg 1996; 171:374 –382 17 Wenzel RP. Nosocomial candidemia: risk factors and attributable mortality. Clin Infect Dis 1995; 20:1531–1534 18 Guiot HF, Fibbe WE, van’t Wout JW. Risk factor for fungal infection in patients with malignant hematologic disorders: implications for empirical therapy and prophylaxis. Clin Infect Dis 1994; 18:525–532 19 Denning DW, Stevens DA. Antifungal and surgical treatment of invasive aspergillosis: review of 2121 published cases. Rev Infect Dis 1990; 12:1147–1201 20 Andriole VT. Infection with Aspergillus species. Clin Infect Dis 1993; 17(suppl):481– 486 21 Banerjee SN, Emori TG, Culver DH, et al. Secular trends in nosocomial primary bloodstream infections in the United States: the National Nosocomial Infections Surveillance System 1980 –1990. Am J Med 1991; 92(suppl 3B):86 – 89 22 Rees JR, Pinner RW, Hajjeh RA, et al. The epidemiologic features of invasive mycotic infections in the San Francisco Bay Area, 1992–1993: results of population-based laboratory active surveillance. Clin Infect Dis 1998; 27:1138 –1147 23 Saubolle MA. Mycology and the clinical laboratory in the diagnosis of respiratory mycoses. In: Sarosi GA, Davies SF, eds. Fungal diseases of the lung. 3rd ed. Philadelphia, PA: Lippincott William and Wilkins, 2000; 1–16 24 Haque AK. Pathology of common pulmonary fungal infections. J Thorac Imaging 1992; 7:1–11 25 Boyars MC, Zwischenberger JB, Cox CS Jr. Clinical manifestations of pulmonary fungal infections. J Thorac Imaging 1992; 7:12–22 26 Woodring JH, Ciporkin G, Lee C, et al. Pulmonary cryptococcosis. Semin Roentgenol 1996; 31:67–75 27 Haron E, Vartivarian S, Anaissie E, et al. Primary candida pneumonia: experience at a large cancer center and review of the literature. Medicine 1993; 72:137–142
Clinical Investigations