The frequency of invasive fungal infections has increased over the past 20 years. This is directly related to an expanding immunocompromised population. Although this is primarily due to the acquired immunodeficiency syndrome pandemic, several other factors have contributed to a growing population of susceptible hosts. Such factors include cancer chemotherapy, immunosuppression after organ and bone marrow transplantation, and the treatment of seriously ill patients in intensive care units (ICUs).1
However, conflicting evidence exists regarding risks in the ICU; in general, the incidence of candidemia and other nosocomial fungal infections is greater in high-risk ICUs than in other areas of the hospital.2
Individuals in the ICU include the most sick and severely injured patients in the hospital. These patients often include the youngest and the oldest patients; they are subjected to the most invasive procedures for diagnosis, therapy, monitoring, and life support; they are given more antibiotics; and they often stay in hospital for longer periods. Each of these factors individually is recognized as increasing the risk of infection, especially with opportunistic pathogens—including invasive pulmonary fungal infections (PFI).3
The exact incidence of PFI in the Pediatric Intensive Care Unit (PICU) is unknown, but the attributable prognosis is almost always fatal.4 The reason is greatly related to the difficulty in diagnosing invasive PFI. Autopsy remains the “gold standard,”2 but is not clinically useful. Clinical criteria alone might not be reliable.3,4
Over the years, new diagnostic techniques such as bronchoscopic bronchoalveolar lavage (BAL) and the use of a protected specimen brush have been studied, and these are now widely used in adults.5 The reproducibility of these bronchoscopic methods is reasonable in adults,6 and they are now considered to be the best methods to diagnose PFI in this population. In the PICU, these bronchoscopic methods are not easily applicable, mainly because of the small size of the endotracheal tube and because of complications associated with these procedures.5 Moreover, experience with these methods in the PICU is still limited, and their validity in mechanically ventilated children remains to be determined. They are easy to perform and safe to use. However, they have never been validated by comparison with a gold standard (autopsy), in pediatric patients.6
The primary objective of this study was to evaluate the validity, in critically ill children, of microbiologic and serologic analyses of respiratory samples, collected by bronchoscopic BAL, for the diagnosis of PFI. The second objective was to evaluate the safety and feasibility of this technique in the pediatric population.
This prospective descriptive study was performed over a 26-month period, with patients aged 15 years or younger (average 1 to 15 y, median 4 y, mean 4.7 y), who had been admitted to the PICU of Ain Shams University Hospital (Cairo, Egypt). A sample size of 35 patients was considered appropriate (confidence level of study power=80%—calculated using Epi Info software for windows, version 3.3.2; Atlanta, CDCP; 2005). All critically ill children [assessed and scored by both physiologic stability index (PSI) and pediatric risk of mortality (PRISM) scores] with recognized risk factors for PFI,7 who had been in the PICU for more than 48 hours, were included, if they satisfied any of the following criteria: (1) unexplained persistent fever despite negative bacterial cultures with pulmonary infiltrates; (2) pneumonic patients who were unresponsive to broad-spectrum antibiotics for 3 days (in non-neutropenic patients); and (3) pneumonic patients who were unresponsive to broad-spectrum antibiotics for 2 days (in febrile neutropenic patients). The risk factors for PFI were suspected either by clinical or by laboratory investigations, and they explained the immunocompromised states of the patients, which include (a) neutropenia (neutrophilic count <1500) due to hematologic malignancies or during myeloablative chemotherapy or any other causes; (b) long-term or high-dose corticosteroid therapy (long-term corticosteroid therapy=20 mg/m2/d prednisone for 2 mo and high-dose corticosteroid therapy=60 mg/m2/d prednisone for 2 wk within the last 3 mo); (c) multiorgan failure syndrome; (d) severe malnutrition (marasmus, kwashiorkor, or marasmic-kwashiorkor); and (e) administration of immunosuppressive drugs within the last 6 months, for the treatment either of malignancies or of rheumatologic diseases. Exclusion criteria were as follows: brain death, contraindication to perform a bronchoscopic BAL (severe bronchospasm, pulmonary hypertension, cardiopulmonary instability, and severe hypoxemia), refusal of permission from the attending physician or the parent for participation, loss of specimen, and incomplete BAL procedure. Informed consent was obtained from the parents. This study was approved by the Research Ethics Committee of Ain Shams University Hospital.
- BAL: All BALs were carried out by the same operating team, using a Pantex flexible bronchoscope system (PENTAX FB-10X bronchoscope with an external diameter of 3.4 mm, a working channel diameter=1.2 mm, working length=600 mm, and total length=900 mm). The procedure was carried out with patient preparation, sedation, and monitoring, according to the European Respiratory Society guidelines for children.6,8
Prebronchoscopy evaluation of the cardiac condition, blood pressure, and arterial blood gases was done. All subjects studied were fasting for at least 6 hours before bronchoscopy. Conscious patients were referred to the theater to undergo the procedure under general anesthesia using deep Halothane inhalation, monitored by electrocardiogram, pulse oximetry, and noninvasive blood pressure monitor. Unconscious, ventilated patients in the PICU underwent the procedure inside the ICU with meticulous observation of their ventilatory and hemodynamic status under supervision of the operator and another intensivist together with an assistant nurse.6 Flexible bronchoscopy was introduced via endotracheal tube, laryngeal mask, or through face mask using Swivel connector to maintain anesthesia. The tip of the bronchoscope was wedged into the diseased segment seen in chest x-ray, or in the right middle lobe and lingula when pathology is generalized. Aliquot 1 to 3 mL/kg of prewarmed sterile normal saline 0.9% (maximum of 3 aliquots) was used for lavage and rapidly aspirated through the suction port of the bronchoscope.8
- (A) Macroscopic examination: BAL fluid was rapidly transferred to the laboratory and examined for caseous, purulent, or bloody areas, and necrotic material.
- (B) Microscopic examination: It was examined by the use of KOH-Calcofluor White Solution under Immunofluorescent microscope. Fungi stain bright green or blue-white, depending upon the filters used.
- (C) Culture BAL fluid: BAL fluid was cultured on Sabouraud Dextrose Agar as a primary recovery medium and on Brain-Heart Infusion Agar, with and without chloramphenicol, as a selective enriched medium.9
- (D) Fungal identification: Fungal growth was identified by
- Appearance of growth.
- Rate of growth.
- Colony pigmentation.
- Growth on media containing antifungal agents.
- Incubation temperature.
Antifungal sensitivity was determined for Candida growth, only by using Candifast (International Microbio, Stago Group, France).10 It assesses the sensitivity of the cultured fungi to the following antifungal drugs: Amphotericin B, Nystatin, Flucytosine, Boonazole, Ketoconazole, Miconazole, and Fluconazole.
(II) Detection of Fungal Antigens
A BAL sample from every patient was preserved at −80°C for the detection of the Aspergillus galactomannan and Candida mannan antigens by the enzyme-linked immunosorbent assay (ELISA) technique, using Platelia Aspergillus and Platelia Candida; Sanofi Diagnostics Pasteur, Marnes-La Coquette, France). All samples were analyzed by the same microbiologist and the same hematologist, who were blinded as to the origin of the samples.
Descriptive data are expressed as means±SD. Continuous data were compared by the Student paired t test or Wilcoxon signed-rank test. Categorical data were compared by χ2 test analysis. Analysis of variance test was performed for multivariate analysis of risks and for predictors of mortality, and expressed as P value, odds ratio (OR), and 95% confidence interval (CI). Statistical significance was established at P<0.05. The statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS) for Windows (version 11.5; SPSS Inc, Chicago, IL; 2006).
Thirty-five patients were included in the study. Demographic data for the 35 patients are presented in Table 1.
Significant severity of illness was present in these patients, who suffered from various underlying disease processes (critical illness scored with PRISM and PSI). The primary diagnoses were mainly cancer complications (n=15), severe gastroenteritis with shock (n=10), and severe malnutrition (n=3). Eight patients (23%) had a multiple organ-dysfunction syndrome on the day of entry in the study. Twenty-seven deaths occurred during the study and were unrelated to the BAL.
Imaging was limited to the chest x-ray, which was done inside the PICU: the most frequent radiographic findings were patchy consolidations (n=25, 71%), followed by multiple nodular lesions (n=8, 23%); 2 patients (6%) had cavitary lesions (bronchiectasis). Four patients had pleural effusions. A total of 29 patients (82.9%) had bilateral lesions and 6 (17.1%) had unilateral lesions. For patients with unilateral involvement, the upper lobes (right upper lobe; n=3, left upper lobe; n=2) were the most common lobes of distribution, followed by the lower lobes (right lower lobe; n=1).
Computerized tomography could not be carried out, except in the case of 2 patients, as the critical conditions of the other patients did not permit their transfer to the Radiology department. It showed patches of consolidation and bronchiectatic changes in both the patients.
Thirty-five samples of BAL were analyzed. All specimens were considered satisfactory (<1% squamous epithelial cells). The amount of fluid injected was 25±10.6 mL, the amount of aspirated fluid was 6.0±2 mL, and the ratio of fluid aspirated to fluid injected was appropriate (31±11%). Bronchoscopic visualization of the airways manifested different degrees of inflamed mucosa and contained different morphologies of the pathologic secretions: these findings directed the BAL samples.
The diagnosis of PFI is only definite when lung tissue histopathology shows the fungus,1,11 with or without a positive culture from the same site (but this was not done in this study).
The diagnosis of PFI is probable when the clinical picture is consistent with the diagnosis, and when one or more of the following are positive for fungi: BAL fluid culture, a cytologic examination of the BAL fluid showing the characteristic fungal morphology, or positive fungal antigen in the serum or in the BAL fluid.1 In our study population, only 27/35 (77%) children were diagnosed as having probable IPA and were assigned to group 1. These patients were further subdivided into 2 subgroups: highly probable subgroup, consisting of 14 patients out of 27 (52%), characterized by having both positive fungal antigen and positive fungal culture. The less probable subgroup consisted of 13 patients out of 27 (48%), and was characterized by having single positive test for PFI, either antigen or culture.
Possible PFI was considered when the clinical picture was compatible, but without any positive laboratory mycologic evidence.1 In this study, 8/35 (23%) children had possible PFI and were assigned to group 2.
All the microscopic assessments of the samples were negative for fungi. Only BAL samples showed positive fungal cultures (15/35) (43%), whereas all blood samples were negative for fungal culture in the studied children. Isolated Candidal growth was found in 34% of the studied population (n=12); growth of both Candida and Aspergillus was found in 6% of the studied population (n=2); isolated growth of Aspergillus was seen in 3% of the studied population (n=1); whereas, no growth was found in 57% of the studied population (n=20). Six of the Candidal isolates (43%) were of the Candida albicans type and 8 (57%) were of the Candida non-albicans type. However, 2 patients (66.7%) showed growth of Aspergillus fumigatus and 1 patient (33.3%) showed Aspergillus niger.
The isolated Candida growth from cultured BAL fluid was further tested for antifungal sensitivity. The tested antifungal drugs are fluconazole, ketoconazole, miconazole, echonazole, flucytosine, nystatin, and amphotericin.
Isolates of Candida albicans spp. (n=6) were all sensitive to amphotericin B. Three were sensitive to fluconazole, ketoconazole, and miconazole; 1 was sensitive to each of the drugs, echonazole, flucytosine, and nystatin. Although isolates of Candida non-albicans (n=8) were all sensitive to amphotericin B, 6 were sensitive to nystatin, 6 were sensitive to flucytosine, 4 were sensitive to ketoconazole, and 1 was sensitive to each of fluconazole and for miconazole.
From the fungal antigens in BAL fluid, Candida mannan antigen was detected in 22 patients (62.3%), Aspergillus galactomannan antigen in 4 patients (11%), and no antigen was detected in 9 patients (26.7%). In contrast, fungal antigens in serum were positive only for the Candida mannan antigen in 19 patients (54%), for the Aspergillus galactomannan antigen in 3 patients (9%), and no antigen was detected in 13 patients (37%). This shows that BAL fluid has a significantly higher diagnostic value and sensitivity than serum for the serologic detection of PFI (χ2=20.503, P<0.001).
Moreover, fungal antigens in BAL fluid showed a highly significant increase in the ability to detect fungal infection, compared with fungal culture, among the clinically suspected patients (Table 2).
Risk Factors for Fungal Pulmonary Infection
Multivariate analysis for risk factors for fungal pneumonia showed that neutropenia (95%CI: 0.218-0.565; P<0.001), degree of neutropenia (95%CI: 0.001-0.119; P=0.046), and thrombocytopenia (95%CI: 0.017-0.258; P=0.027) were independent risk factors for PFI. Prolonged PICU stay (≥1 wk) (OR=24; 95%CI: 2.14-650, P=0.002), invasive procedures (P=0.009), MODS (P<0.001), high PSI (cutoff value=67; P<0.001), and high PRISM (cutoff value=35; P<0.001) were significant PICU risk factors for probable fungal infection. Intake of corticosteroids (OR=1.57; 95%CI: 1.12-2.84, P=0.046) and cytotoxic drugs (P=0.031) have been proved to constitute significant drug-related risk factors for infection.
Observed mortality was 100% (27 patient) among the group 1 patients, despite their receiving antifungal treatment in the form of fluconazole. Multivariate analyses for the predictors of mortality within the studied population showed that neutropenia (95%CI: −0.711 to 0.194; P<0.001) and multiorgan failure (95%CI: −0.243 to 0.001; P=0.048) were independent predictors of mortality.
Adverse events during the procedure are detailed in Table 3. Most adverse events were benign, transient, and did not require any treatment: the lowest oxygen saturation noted was 80% and lasted for <2 minutes; bradycardia and a drop in blood pressure occurred in 2 patients, but resolved spontaneously within seconds.
BAL Diagnostic Specimen
In adults, bronchoscopic BAL is considered to be one of the most reliable diagnostic tests for respiratory infections, and it is used to obtain quantitative cultures and cytologic results.5,6 Nonbronchoscopic BAL has also been used in the past 10 years with good results. The advantages of doing a blind (nonbronchoscopic) BAL include ease of performance at the bedside, feasibility through small endotracheal tubes (<4.0 mm), less discomfort for the patient, and low cost. The disadvantage is that, without bronchoscopy, it is difficult to predict from which part of the lung the sample will be taken.11
Some authors consider that a bronchial aspirate is reliable only if directed toward the affected lung region and that this is only feasible with visualization by a bronchoscope.8,9 In this study, bronchoscopic BAL showed a significantly higher diagnostic yield than blood samples.
Therefore, bronchoscopic BAL specimens seem ideal for diagnoses in the pediatric population, regarding reproducibility, safety, and costs; however, its validity as a diagnostic marker of PFI in critically ill children still remains to be estimated, with autopsy as the reference standard.12
Even though bronchoscopic BAL is considered fairly safe, complications of BAL were frequent (in 54% of children) in this study; however, most were transitory and benign. Meticulous monitoring of the patients' hemodynamics while performing the bronchoscopic lavage, shortening the length of the procedure, and using appropriately sized scopes are important precautions for performing a safe BAL for the critically ill patients. Complications could probably be totally avoided if the procedure is done by a qualified operator, and if the patient is monitored carefully during and after the procedure.6,8
Diagnosis of PFIs
Infectious fungal respiratory diseases can be divided into those that occur in generally healthy individuals and those that occur opportunistically in immunosuppressed patients. In the first group, organisms such as Histoplasma capsulatum, Coccidioides immitis, and Blastomyces dermatitidis are frequent pathogens in certain endemic regions. The fungi that affect immunosuppressed individuals are frequently species of the Aspergillus spp. and Candida spp., as well as of the Cryptococcus neoformans.4,13 In our study, the predominant isolated species were Candida (evident by one or more positive test in 66% of the children), followed by Aspergillus (evident by one or more positive test in 11% of the children).
The fungal agents responsible for fungal respiratory diseases are various, but infections due to the Aspergillus spp. and other non-Aspergillus molds, in immunocompromised patients, are associated with high mortality. The appropriate use of the diagnostic tools and therapeutic regimens available at present might help speed up the diagnosis and improve treatment. Of course, there is a long way to go before we can reach early diagnosis and optimal treatment.11
The clinical presentations of fungal respiratory infections are nonspecific, and they often overlap with those of other infectious and noninfectious processes. The specific diagnosis is often long delayed.14
Radiographic features are part of the evidence contributing to the diagnosis of fungal respiratory infections. Chest x-rays of the majority of respiratory fungal infections mimic other lung diseases and have limited value in predicting the causative organism. These signs include interstitial infiltrates, bronchopneumonia, consolidation, segmental pneumonia, multiple nodules, masses, cavitary lesions, and pleural effusion. Cavitary lesions might be observed on the chest radiograph.13–15 However, computed tomography scanning can reveal crescentic cavitation much earlier than a chest x-ray can. Only a few signs from chest x-rays are highly suggestive in predicting special fungal infections; radiographic signs of a rounded density within a cavity, partially surrounded by a radiolucent halo, are characteristic of an asperilloma.3
Nonetheless, the radiologic manifestations in this study population with probable infection were mostly bilateral lesions in 24 patients (89%) and unilateral lesions in 3 patients (11%). Twenty-five patients had patchy consolidation; 1 patient had a multinodular pattern of infiltration, whereas, the last one had bronchiectasis.
Accordingly, diagnoses of deep-tissue mycotic infections such as invasive candidiasis or aspergillosis present significant challenges. The manifestations of infection can vary, depending on the site of the infection and the host's immune status. Moreover, clinical signs and symptoms are often nonspecific, and are difficult to distinguish from those seen with bacterial infections.16 Commonly, a fever that persists or develops while the patient is receiving a broad-spectrum antibiotic is the first indication of fungal infection. Collection of clinical specimens, to isolate and identify fungal pathogens, might be difficult, especially when the focus of the infection is unknown. Nevertheless, histologic isolation remains the gold standard for deep fungal infections including pulmonary lesions.3,11
Diagnosis is further complicated by the fact that there are few signs and symptoms for systemic fungal infections, and immunosuppression masks the typical clinical markers of systemic infection. Often sputum and blood cultures are negative, and invasive diagnostic procedures are associated with greater risk in patients who are hypoxic and/or thrombocytopenic.17
When the clinical, radiologic, and microbiologic data do not help to differentiate between infection and colonization, flexible bronchoscopy with BAL and transbronchial biopsies,18 or computed tomography-guided percutaneous transthoracic lung biopsies15 are undoubtedly the most reliable methods, for establishing a certain diagnosis and for defining the species of fungi responsible for infection in suspected respiratory fungal infections. These methods are not problematic to perform in immunocompetent individuals. However, they have limited use in patients with poor general health and with a tendency to bleed, such as hematologic patients and recipients of bone marrow transplantation.4
Promising Diagnostic Tool
The old standing concept of the importance of fungal cultures for diagnosis of fungal pneumonia is debatable nowadays. The evolving methods of the diagnosis of invasive fungal infections, using the simple, sensitive, and rapid ELISA technique for selected fungal antigens, showed a new hope for early diagnosis; hence, early treatment of fungal pneumonia is possible, especially among critically ill patients.18
In a study by Jones et al,19 the Aspergillus mitochondrial gene polymerase chain reaction (PCR)-ELISA was 100% sensitive and 100% specific for aspergillosis in neutropenic patients; all 12 neutropenic patients with definite or probable invasive aspergillosis had PCR-positive BAL fluids. PCR and ELISA have been used to detect serum galactomannan antigen in patients with invasive aspergillosis.19 Bretagne et al20 found 19 of the 20 PCR-positive serum samples to be positive for galactomannan antigen. The results of these studies showed these methods to be highly predictive of infection, yielding an earlier diagnosis by noninvasive means.20
However, in the literature, some data showed that the administration of antifungal treatment before BAL can modulate the results of fungal culture or antigen detection.17 In our study, the fact that all the patients were not treated with antifungal treatment before BAL did not seem to influence the results.
Fungal Pulmonary Infection Management
With systemic mycoses, a definitive diagnosis might not be made until late in the course of the infection or at autopsy. Because a delay in treatment can be fatal, antifungal therapy is commonly initiated empirically. The decision to treat is based on clinical suspicion, which takes into account the patient's clinical status, the failure of antibiotic therapy, and the presence of known predisposing factors for fungal infection.2,3
For possible fungal infections (without laboratory evidence), empiric therapy is recommended. For patients with proven fungal infections or clinically documented fungal infections, targeted therapy is warranted.17
Although the diagnosis of aspergillosis can be difficult, its early detection and treatment are important factors in improving the clinical outcomes and survival rates of patients with invasive aspergillosis. The mortality rate for untreated invasive aspergillosis is nearly 100%. Ideally, the diagnosis should be based on biopsy, which would histologically prove the presence of Aspergillus hyphae. However, many patients are not stable enough to undergo such an invasive procedure.16
However, because like humans, fungi are eukaryotes, it has been difficult to find suitable specific targets on fungi for antifungal drug action, which do not harm humans.1
Despite advances in antifungal treatment options, the best use and timing of these agents are unclear. A significant challenge in the treatment of fungal infections is that there are limited data available regarding the efficacy of antifungal medications. The biggest challenges in the treatment of fungal infections are early and accurate diagnosis and the selection and timing of appropriate therapy.1,3,13
Limitations and Strengths of the Study
Our study shows the feasibility of using bronchoscopic BAL samples to assess for PFI in critically ill infections. This technique, even if good, is not perfect, as the gold standard remains a lung biopsy or an autopsy isolation of fungi.
However, the BAL laboratory results for the probable PFI group were positive, and all the children died in that group (27 patients). The reason for this fatal outcome can be related either to the delayed timing of BAL specimens, or to the delayed decision for appropriate antifungal management or prophylaxis.
Even though this study has a number of strengths, by assessing all consecutive patients prospectively, and by using strict inclusion and exclusion criteria, the possibility of selection bias is low for all the included children, as evident.
This study spotlights the feasibility of the BAL specimen technique. Although it is more invasive than blood testing, it carries the highest diagnostic yield for PFI among the critically ill children and is, nevertheless, safe.
All BAL samples were taken by the same operator team to avoid interobserver variability. All analyses were carried out by the same technicians, and all the results were evaluated by the same experts (1 hematologist and 1 microbiologist), to exclude the possibility of variability due to the technique or interpretation of the results.
Lower respiratory tract secretions, collected by bronchoscopic BAL for the diagnosis of PFI, are applicable in critically ill children. This procedure is easily accomplished at the bedside, yields adequate samples of bronchoalveolar secretions, and is usually safe. Invasive fungal infections of the lung remain important causes of death in immunocompromised critically ill patients. The development of new early diagnostic tools and of well-designed multicenter evaluations of diagnostic methods and therapeutic regimens are important future goals. However, before the implementation of routine antifungal prophylaxis in the ICU, data are required to define the population that would receive the most benefit from prophylaxis and to predict the long-term effects on fungal epidemiology.
The authors thank Ain Shams University PICU staff for their help and support, and the participant children's guardians for their cooperation and understanding.