Published ahead of print on September 20, 2007, doi:10.1164/rccm.200704-606OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200704-606OC
Galactomannan in Bronchoalveolar Lavage FluidA Tool for Diagnosing Aspergillosis in Intensive Care Unit Patients1 Medical Intensive Care Unit and Infectious Diseases Unit, Department of General Internal Medicine; 2 Department of Medical Diagnostic Sciences; 3 Department of Hematology, 4 Department of Pharmacy; and 5 Department of Pathology, University Hospital Leuven, Leuven, Belgium Correspondence and requests for reprints should be addressed to Wouter Meersseman, M.D., Medical Intensive Care Unit, University Hospital Leuven, Herestraat 49, B-3000 Leuven, Belgium. E-mail: wouter.meersseman{at}uz.kuleuven.ac.be
Rationale: Invasive aspergillosis (IA) is an important cause of mortality in patients with hematologic malignancies. However, IA appears to be gaining a foothold in the intensive care unit (ICU) in patients without classical risk factors. A recent study described 89 cases of IA in patients in a medical ICU without leukemia or cancer. The diagnosis of IA remains difficult and is often established too late. Galactomannan (GM) is an exo-antigen released from Aspergillus hyphae while they invade host tissue. Objectives: This prospective single-center study was conducted to investigate the role of GM in bronchoalveolar lavage (BAL) fluid as a tool for early diagnosis of IA in the ICU. Methods: All patients with risk factors identified in our earlier study were evaluated. BAL for culture and GM detection, serum GM levels, and computed tomography scan were obtained for all included patients with signs of pneumonia. Patients were classified as having proven, probable, or possible IA. Measurements and Main Results: A total of 110 patients out of 1,109 admissions were eligible. There were 26 proven IA cases. Using a cutoff index of 0.5, the sensitivity and specificity of GM detection in BAL fluid was 88 and 87%, respectively. The sensitivity of serum GM was only 42%. In 11 of 26 proven cases, BAL culture and serum GM remained negative, whereas GM in BAL was positive. Conclusions: IA is common in immunocompromised, critically ill patients. GM detection in BAL fluid seems to be useful in establishing or excluding the diagnosis of IA in the ICU.
Key Words: Aspergillus • bronchoscopy • intensive care unit • galactomannan • immunosuppression
Invasive aspergillosis (IA) has emerged as an important cause of morbidity and mortality in transplant recipients and in patients with hematologic disorders (1–3). Recent data indicate that IA may be an underestimated opportunistic fungal infection in critically ill patients, even in the absence of hematologic malignancy (4–8). Several autopsy studies from intensive care unit (ICU) patients confirm that invasive fungal infections are among the most common missed diagnoses (9, 10). Patients with chronic obstructive pulmonary disease (COPD), patients with cirrhosis, and patients receiving steroids are especially at risk and were not included in the original criteria for definition of proven and probable IA according to the widely accepted European Organization for Research and Treatment of Cancer/Mycoses Study Group guidelines (11, 12). Establishing a diagnosis of IA at an early stage of disease is necessary for successful treatment but challenging in patients with hematologic diseases, and even more difficult to achieve in the poorly defined, broad group of critically ill patients at risk. Definite diagnosis is rarely established before death or before overwhelming fungal proliferation. The utility of radiology (e.g., high-resolution computed tomography [CT] scan with the "halo" sign) is of limited value in mechanically ventilated patients (13). Tissue biopsy as a means of making a definite diagnosis is not without risk in the critically ill patient, and its sensitivity is unknown. Conventional diagnostic tests, such as culture and microcopy of respiratory tract samples, have only a sensitivity and specificity of around 50% (14, 15). Galactomannan (GM) is a polysaccharide fungal cell wall component that is released during tissue invasion by Aspergillus hyphae and that can be detected in body fluids. The presence of β-lactam antibiotics, such as piperacillin–tazobactam may give rise to false-positive results. GM serum levels have been evaluated most extensively in allogeneic bone marrow or stem cell transplant recipients and/or neutropenic patients (16, 17). Data on the performance of GM detection in serum are sparse in nonneutropenic patients but suggest that serum GM is probably not a good marker for IA in these patients (18, 19). Sandwich ELISA methods have also been used to detect GM in bronchoalveolar lavage (BAL) fluid of neutropenic patients, yielding sensitivities between 76 and 100% (20, 21). The use of GM detection in BAL samples may be also useful in the subset of patients without neutropenia, as demonstrated in an experimental model of IA (22). To investigate the role of GM in BAL for the diagnosis of IA in the ICU, we conducted a prospective trial in critically ill patients at risk for IA and compared the diagnostic performances of GM detection in BAL, radiologic signs, culture results, and serum GM detection.
Study Population and Data Collection Between July 2005 and December 2006, all patients older than 18 years admitted to our medical ICU were screened for inclusion in a prospective study. Eligible patients displayed at least one of the following host factors:
In addition, eligible patients could only be enrolled if they had at least two of the three following features:
Three patients who did not having any of the prespecified host factors were also included in the study because of a positive tracheal surveillance culture (performed once weekly) for Aspergillus spp. The following information was stored in a data file: patients' characteristics, including age, sex, medical history, and reason for ICU admission; standard ICU laboratory findings; presence or absence of fever and/or neutropenia (if applicable); and use of immunosuppressive drugs (steroids and others). Disease severity at admission in the ICU was assessed by the Simplified Acute Physiology Score II (SAPS II). The sandwich ELISA assay for GM detection (Platelia Aspergillus; Sanofi Diagnostics Pasteur, Marnes-La-Coquette, France) was performed as per the manufacturer's instructions and modified to a semiautomated protocol (23) (see the online supplement). Fiberoptic bronchoscopy with BAL (2 x 20 ml) was performed upon inclusion and weekly thereafter if feasible. The sampling area was selected based on the infiltrate location on the chest radiograph. The presence of any tracheal or bronchial lesions was recorded by the endoscopist (W.M.). Lavage samples were submitted for direct microscopic examination, GM detection, and bacterial, fungal, and mycobacterial cultures. An optical density (OD) ratio of 0.5 or greater for GM in serum and BAL was considered positive. Serum sampling for GM detection was done twice weekly. Clinicians remained blinded throughout the study from the BAL results for GM, but serum indices of 0.5 or more were reported. Antifungal treatment was started at the discretion of the attending physician and was not protocol defined. Positive GM levels in patients taking piperacillin–tazobactam or amoxicillin–clavulanate were analyzed separately. A pulmonary CT scan was done (if feasible) between Day 1 and 3 after inclusion or, in case of clinical deterioration, 1 week after enrollment. Autopsies were pursued in all fatal cases unless there was an explicit refusal of the family. The study was approved by the ethics committee and written, informed consent was obtained from the patient or next of kin.
Case Definitions of IA
Statistical Analysis
Patient Characteristics From July 2005 to December 2006, 1,109 patients were admitted to our ICU. All 110 patients (10%) fulfilling the inclusion criteria were enrolled. The characteristics of the overall study group are shown in Table 1. Thirty-six (33%) patients suffered from a hematologic malignancy, whereas 74 (67%) had other immunocompromising factors. Only 24 patients (22%) were neutropenic. Mean SAPS II score was 53 (±15). The ICU mortality rate of 66% was higher than the predicted risk of death based on the SAPS II score (52%). After post mortem examination in 69 (95%) of 73 fatalities, study patients were classified as proven IA (n = 26, 24%), probable IA (n = 8, 7%), possible IA (n = 27, 24%), proven non-Aspergillus mold infection (n = 3, 3%), colonization (n = 3, 3%), and no IA (n = 43, 39%). Only 4 of the 26 proven cases were diagnosed as proven IA before death (Figure 1).
Test Characteristics BAL GM levels. On average, bronchoscopy with BAL was performed 6 days after admission to the ICU. A total of 156 procedures were performed. All 26 patients with proven IA had at least one BAL GM index  0.5 (range, 0.6–7.9). In 23 patients, the first BAL sample yielded a positive value, while in the remaining 3 patients, only the second BAL GM index was 0.5 (Table 2). The number of BAL GM-positive patients in the truly negative group (n = 46) was 6. As depicted in Figure 2A, the levels of GM in BAL were significantly higher in the proven group (median value, 4.1) than in the truly negative group (median value, 0.1; P < 0.0005).
Serum GM levels. A total of 397 serum samples were tested for GM (mean, 3.6 samples per patient). A total of 125 samples were analyzed from 26 patients with proven IA; 44% of these serially analyzed samples tested positive. The median value of GM in serum in the proven cases was 0.3 (Figure 2B). Fourteen patients did not have a serum index 0.5 at the time of BAL positivity.
Culture results and direct microscopic examination of BAL fluid.
Diagnostic value of GM assay in BAL fluid.
False positivity of GM in serum and/or BAL. Overall, six patients had false-positive tests in the truly negative group (13%). Four of these six patients were treated with antifungals before death. GM values were higher in BAL than in serum. When an index cutoff of 0.5 was applied, 4 of 33 patients treated with piperacillin–tazobactam had a positive serum and BAL GM result. The false-positive values in BAL ranged from 1.3 to 5.8, and from 0.7 to 2.8 in serum. Results remained positive until 4 days after the antibiotic had been stopped. No patients were treated with amoxicillin–clavulanate.
Comparison of proven cases with and without neutropenia (n = 26). Thoracic CT scan performed in 15 proven cases displayed no halo or air-crescent sign, not even in neutropenic patients. Most patients had nonspecific radiologic findings (nodular lesions, necrotizing cavities) (see the online supplement for illustrations). All deceased patients in the proven group underwent autopsy (n = 24). The histopathologic pattern of IA in steroid-treated patients and patients with cirrhosis differed from neutropenic patients. In the latter group, inflammation in the lung was less pronounced and fungal burden was higher than in patients without neutropenia.
To the best of our knowledge, this is the first study that analyzes the diagnostic contribution of GM detection in BAL in ICU. Our findings support the utility of determining GM levels in BAL fluid of critically ill patients at risk for IA. The sensitivity of GM detection was 88% in proven cases when calculated on the first BAL and applying a cutoff value of 0.5. The contribution of serum GM to the diagnosis of IA in our ICU population was much lower (sensitivity, 42%). The specificity of GM detection in serum and BAL was high even in patients treated with piperacillin–tazobactam (96 and 87%, respectively; cutoff, 0.5 ng/ml). Overall, the performance of fungal culture and/or direct examination on BAL was only moderate for the diagnosis of IA (sensitivity, 58% for the proven cases). CT features, such as the halo sign, proved to be of no value in the ICU, even in the neutropenic patients. The histopathologic pattern of IA in autopsy specimens confirmed earlier reports that steroid patients had more inflammation and less fungal burden in the lungs than neutropenic patients (24). Recent studies indicate that IA may be considered an emerging problem in critically ill patients (4–8). Patients receiving steroids and patients with COPD or liver cirrhosis are especially at risk. The diagnosis of IA remains difficult. Clinical signs are frequently lacking in mildly immunosuppressed patients. The main problem with the fungal culture as a diagnostic tool is its limited performance and the delay in diagnosis of 48 to 72 hours. Early diagnosis is of great importance because early start of antifungal treatment improves survival (25). The use of antigen markers have been studied in an effort to improve rapidity and performance of the diagnostic procedure. GM is a polysaccharide cell wall component that is released by Aspergillus during growth. Several large, prospective studies in neutropenic patients reported sensitivities above 90% for GM detection by ELISA in serum (17). However, most ICU patients who are at risk for IA are not neutropenic (78% of the patients in the current study). In addition, it has been hypothesized that neutrophils are capable of clearing GM from the blood by its mannose-binding receptors (17). This might explain the lower sensitivity of serum GM levels (50%) for the diagnosis of IA in our previous study (6). Therefore, we decided to focus on the lungs, reflecting the primary site of entry of Aspergillus conidia. Other studies examining GM values in BAL fluid have mainly been done in liver and lung transplant recipients and showed rather poor sensitivities (30%) (18, 19). A possible explanation is the less severe disease status of the patients included in these studies and the lack of histologic data in most patients. In patients with hematologic malignancies, better sensitivities have been reported (20, 21). The excellent sensitivity of GM detection in the BAL fluid might be counterbalanced by a loss of specificity due to a greater occurrence of false-positive results. Although several sources of false positivity are frequently encountered in the ICU (piperacillin–tazobactam or amoxicillin–clavulanate, plasma–lyte infusions) (17), the specificity was still above 85%, both for serum and BAL. CT scan is an important tool for diagnosis of IA in neutropenic patients (13). Typical lesions, such as the halo sign, were not seen in our study. This observation was confirmed by another study, reflecting the fact that the halo sign is only seen in early IA (4). Moreover, many confounding factors (e.g., pleural effusions and ventilator-associated pneumonia) are present in ICU patients and obscure the presence of characteristic signs. One of the strengths of our study is the high autopsy rate of 95%. This enabled us to discriminate between invasive disease and colonization. As such, we could demonstrate that in 9 of 26 patients with proven IA who were screened before death, antifungal therapy might have been given if physicians had been aware of the GM value in the BAL fluid. We do realize that the study design contains limitations. First, the optimal timing for sampling the patient at risk is prone to subjective interpretation and is much more difficult to determine in the broad group of critically ill patients than in previous studies confined to patients with hematologic diseases. Obvious entry criteria, such as 10 days of neutropenia or presence of graft-versus-host disease, used in previous studies cannot be implemented in the broad group of patients admitted to the ICU. Second, because our referral unit cares for patients with severe medical disease only, it is difficult to extrapolate our results to other ICUs. Finally, we cannot provide data on the influence of antifungal therapy on the performance of GM detection in BAL fluid, because the patients did not receive antifungals at the time of the first bronchoscopy. In summary, IA appears to be an important problem in patients admitted to a medical ICU. The use of GM in BAL fluid as a means of establishing early diagnosis of IA in critically ill patients at risk is very promising. The validity of the data needs to be confirmed in other ICUs. If so, critical care physicians do have a helpful instrument to decide in which circumstances antifungal therapy should be initiated early.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200704-606OC on September 20, 2007 Conflict of Interest Statement: W.M. has participated as a speaker for Pfizer and Merck, Sharp and Dohme (MSD). K.L. has participated as a speaker in a scientific meeting organized and financed by Pfizer. J.M. has been a consultant and received lecture fees from AstraZeneca, Schering Plough, MSD, BioRad Laboratories, Pfizer, Johnson and Johnson, and Amgen. A.W. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. G.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. S.V. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. I.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. E.V. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. E.V.W. has participated as a speaker and consultant for Pfizer and MSD. Received in original form April 20, 2007; accepted in final form September 14, 2007
Related articles in AJRCCM:
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
