- Email: [email protected]
Bronchoalveolar Lavage in the Diagnosis of Pulmonary Infiltrates in Patients with Acute Leukemia
Bronchoalveolar Lavage in the Diagnosis of Pulmonary Infiltrates in Patients with Acute Leukemia* Hiroshi Saito, M.D.;t Elias] Anaissie, M.D.;§ Rodolfo C. Morice, M.D.;+ Roupen Dekmezian, M.D.;+ and Gerald e Bodey, M.D., F.C.C.~t 'The utility of bronchoalveolar lavage (BAL) in determining the causative agent of pulmonary inflltrates in patients with acute leukemia is not known. We retrospectively evaluated the diagnostic yield of HAL in 22 adults with acute leukemia and compared the results with those at autopsy performed within three weeks of BAL. All patients had neutropenia and thrombocytopenia at the time of BAL, were receiving broad-spectrum antibacterial agents, and 15 were also receiving amphotericin B before BAL. The median interval between the detection of pulmonary infiltrates and HAL was seven days (range, 0 to 23 days); the median interval between HAL and autopsy was nine days (range, 1 to 20 days), The diagnostic yield of HAL was 15 percent (3 of 20
speci6c diseases); all three were Candida pneumonia. The sensitivity of HAL was 75 percent and its speci6city 100 percent, for Candida pneumonia. HAL did not result in a speci6c diagnosis for the 17 remaining diseases, nine of which were Aspergillus pneumonia. In seven patients in whom autopsy was performed within 72 hours of BAL, lavage results correlated with those of autopsy in only one who had Candida pneumonia. All HAL cultures were falsely positive, except in four cases of Candida pneumonia. The therapeutic regimen was not m0di6ed according to the HAL results in any of the 22 patients. There were no major complications associated with the procedure. (Cheat 1988; 94:74&-49)
The diagnosis and management of pulmonary infiltrates in patients with acute leukemia is a challenge to clinicians caring for these patients. Pulmonary infiltrates in this population can have a variety of infectious and noninfectious causes and are often fatal. l -4 Many studies have reported on the utility of various diagnostic procedures in this clinical setting, including open lung biops~5-7 bronchoscopic exami-
serious complications.P'" As a result, BAL has been advocated as a safe and expeditious way to obtain a diagnosis in immunocompromised patients with pulmonary infiltrates. However, with the exception of two studies on patients undergoing bone marrow transplantation,20·21 the reported series included heterogeneous groups of patients with different underlying diseases and immunosuppressive drug regimens, 17-19.22 and a final diagnostic confirmation by histopathology was not available for most patients. Histopathologic examination obtained by open lung biopsy has been misleading in a high proportion of patients with acute leukemia." As a result, autopsy examination remains the most definitive end point for assessing the value of any diagnostic procedure, and even then a significant number will not have a specific diagnosis. The utility of BAL for the diagnosis of pulmonary infiltrates in a homogeneous group of patients with acute leukemia has not been determined. We attempt to do so, reporting our experience with BAL in 22 consecutive adults with acute leukemia and pulmonary infiltrates, all of whom had histopathologic confirmation at autopsy
For editorial comment see page 676 nation with bronchial brushing, washing, and transbronchial biopsy 8-11 and percutaneous needle aspiration or biopsy 12.13 These procedures, however, have been associated with a low diagnostic yield or a number of complications. They also cannot be routinely performed in patients with acute leukemia, because many have coagulation abnormalities or compromised cardiopulmonary systems. 14-16 Recently bronchoalveolar lavage (BAL) has been evaluated for the diagnosis of pulmonary infiltrates in immunocompromised patients, 17-24 where its diagnostic yield in immunocompromised patients without acquired immunodeficiency syndrome (AIDS) has been reported to range from 50 to 93 percent, without ·From the Department of Medical Specialties, and Department of Pathology The University of Texas System Cancer Center, M. D. Anderson Hospital and Tumor Institute, Houston. tSection of Infectious Diseases. § Department of Pathology *Section of Cardiopulmonary Medicine, Department of Pathology Manuscript received February 15, revision accepted March 24 Reprint requests: Dr. Anai8.rie, lnfectiow Diseae&, Box 47, 1515
Holcombe Blvd, Houston 77030
MATERIALS AND METHODS
reviewed the medical records of all adults with acute leukemia cared for at the University of Texas M. D. Anderson Hospital and Tumor Institute between October 1984 and January 1987 and examined by BAL for pulmonary infiltrates. Patients were included only if they met the following criteria: (1) a diagnosis of acute leukemia or chronic myelogenous leukemia (CML) in the terminal phase being treated with intensive chemotherapy; (2) a neutrophil ~
CHEST I 94 I 4 I OClOBER. 1988
745
count of less than I,OOO/fJ.I at the time of BAL; (3) an autopsy examination performed within three weeks of BAL; and (4)a similar location for the disease found at autopsy the pulmonary infiltrates shown on chest x-ray film, and the BAL sampling site. Patients who underwent allogeneic bone marrow transplantation were excluded from analysis. Patients were further evaluated for presence and duration offever (temperature greater than38.3°C) and neutropenia preceding BAL, the chest x-ray film pattern and its chronologie change, the number of days from the x-ray detection of pulmonary infiltrates to BAL and from BAL to autopsy results of cultures from the two weeks preceding BAL until death, and antileukemic and antimicrobial therapy before BAL. Changes of therapeutic regimens as a consequence of the results obtained by BAL were also analyzed. Bilateral pulmonary infiltrates on chest x-ray film were categorized as diffuse involvement, while all other infiltrates were considered to be localized disease. BAL was performed in patients who were considered to have extensive pulmonary infiltrates or who developed new pulmonary infiltrates while receiving antibacterial antibiotics. Bronchoscopy was performed with an Olympus BF-B3 brochoscope (Olympus Corporation of America), The bronchoscope was passed transnasally or transorally When patients were intubated, bronchoscopy was performed through an endotracheal tube. The tracheobronchial tree was routinely examined, and then the bronchoscope was advanced further and wedged into a subsegmental bronchus of an involved area. Lavage was performed using 60 to 200 ml of sterile normal saline solution in 20- to 3O-ml aliquots through a bronchoscopic channel. Six patients had bronchial brushings done after the lavage procedure. A portion of BAL 8uid was submitted to the clinical microbiology laboratory for bacterial testing, including Legionella sp, fungal, and mycobacterial cultures. Gram stain and acid-fast stain were also done at the same laboratory Viral cultures and direct immunofluorescent staining for Legionella sp were performed, and the remainder of the lavage 8uid was centrifuged at 500 g for 10 minutes and the sediment smeared and stained by the Papanicolaou, MayGrunwald-Giemsa, and Gomori methenamine silver methods. Brushing samples were cultured for bacterial, fungal, and mycobacterial cultures and stained using the same methods as BAL 8uid for cytologic examination. Sections of lung specimens from postmortem examination were examined using hematoxylin and eosin stain, Gomori methenamine silver stain, and Gram stain. All slides submitted for cytologic and histopathologic examination were reviewed by one of us (H.D. ~ Fungal and bacterial pneumonia were diagnosed by the presence of tissue invasion, whether or not this was confirmed by culture results. When the destruction of alveolar structures was present without any identifiable microorganism, the diagnosis of nonspecific pneumonitis was made. RESULTS
The clinical characteristics of the 22 patients who met our inclusion criteria are shown in Table 1. There were ten men and 12 women, who had a median age of55 years (range, 20 to 81 years). Underlying diseases were acute nonlymphocytic leukemia (ten patients), acute lymphocytic leukemia (seven patients), and CML in blastic transformation (four patients) and in accelerated phase (one patient). Four patients had received antileukemic chemotherapy and total body irradiation followed by autologous bone marrow transplantation, eight were receiving their first induction chemotherapy two were in remission following therap~ and 12 were in leukemic relapse. All but one patient was severely neutropenic (a neutrophil count 748
less than IOO/I-LI), and all had a platelet count of less than 50 X I()3/I-LI (less than 20 x l()3/I-LI in 14 patients). The pulmonary infiltrates were diffuse in 12 patients and localized in ten. Arterial blood gases were obtained in 17 patients, 11 of whom were receiving supplemental oxygen. The Pa02 was less than 60 mm Hg in six patients, between 60 and 80 mm Hg in six, and more than 80 mm Hg in five. In the two weeks preceding BAL, 21 patients received one or more broad-spectrum Ji-Iactam antibiotics with or without an aminoglycoside. 'Irimethoprim-sulfamethoxazole was also given to 11 patients and Acyclovir to four patients. Amphotericin B was given to 15 patients for a median dose of 360 mg (range, 10 to 1271 mg). The median interval between the detection of pulmonary infiltrates and BAL was seven days (range, 0 to 23 days); it was nine days between BAL and autopsy (range, 1 to 20 days). Autopsy was performed within 72 hours after BAL in seven patients. Results of BAL and autopsy examinations are summarized in Table 1. Correlation between the BAL cytology and autopsy findings for Candida pneumonia was considered present only when all fields of a slide examined for cytology showed a very large number of yeasts and pseudohyphae. BAL was diagnostic in only three patients, all of whom had histologic documentation of Candida pneumonia on autopsy In the remaining four patients without autopsy evidence of Candida pneumonia, the submitted BAL specimens showed scattered fungal elements, predominantly yeast cells with or without pseudohyphae, and all four were considered negative. The BAL cultures from 21 patients yielded different bacterial, fungal, mycobacterial, and viral organisms but did not correlate with the histopathologic findings at autopsy except in four cases of Candida pneumonia. Bronchial brushing was done in six patients but was nondiagnostic. Results of BAL in seven patients in whom autopsy was performed within 72 hours correlated with the pathologic findings in only one, who had Candida pneumonia. However, a concurrent bacterial pneumonia in the same patient was missed by BAL. Except for the above-mentioned patient with Candida pneumonia, there was no positive correlation between BAL results and autopsy findings in five patients who underwent BAL less than 24 hours after the pulmonary infiltrates were first detected. The overall diagnostic yield of BAL was 15 percent (three of 20 cases of disease). With respect to Candida pneumonia, BAL cytologic study yielded a sensitivity of 75 percent (3/4), a specificity of 100 percent (3/3), and a positive predictive value of 100 percent (3/3). The BAL did not result in change in the therapeutic regimen in any of the patients. All three patients whose results of BAL cytology were positive for Candida pneumonia had been receiving amphotericin SALin Diagnosis of Pulmonary InfiltratesIn Leukemia (Saitost 8/)
Table I-Patient Characteriatict and Beaulta o/BAL and Autopay* TIme Interval (Days) Between
Amphotericin B(mg); Amt. Patient Age, yrl Leukemia PI and BAL and Before BAU No. Sex Type BAL Autopsy Total Amt.
Sputumt Culture
BAL Cytology
Culture
1
39/M
ALL
1
1
II/51
2
561F
ALL ANLL CML
8 10 17
15 8 17
1361498
61/F 37/F
411205 127112171
Yeast
5
58/M
ANLL
2
211101
Yeast
6
69/F
ANLL
7
5
361/611
C albicans
7
831M
CML
7
1
3011441
Gram-negative rods
Pseudomonas
8
20/F
ALL
23
19
40611041
Paeroginosa
P aeroginosa
9
311M
ANLL
o
010
Yeast
10
81/F
ANLL
8
7
0/226
11
431F
ALL
7
16
720/1460
12
681M
CML
6
3
0/0
13
67/M
ANLL
2
3
010
14
231F
CML
10
10
4501825
15
261F
ANLL
18
12
58111421
16
47/M
ALL
4
12
1140/2010
Yeast
Yeast
17
49/F
ANLL
5
12
1361451
C albicaf&8
Yeast
18
651M
ANLL
21
3
401/401
19
59/M
ALL
o
4
010
ND
Yeast, Mycobacterium avium-intraceUulare Yeast
20
77/F
ALL
12
15
om6
ND
Yeast
21
541M
CML
10
0/440
Yeast
22
381F
ANLL
20
36011240
3 4
9
Candida Candida albictJf&8
Yeast
Klebsiella pneumonia Acinetobacter sp
Enterococcus
Aspergillus
flavus
Candida Yeast, Streptococcus Candida C albictJf&8, Adenovirus Yeast
C albictJns C albictJns Candida spp
maltophiliD
Yeast
.Staphylococcus epitlermidis, yeast Yeast
Enterococcus
Penicillium
S epidemidis
Yeast, Enterococcus
Autopsy (Lung) Candida pneumonia, bacterial pneumonia; grampositive cocci Candida pneumonia Candida pneumonia Nonspecific pneumonia Candida pneumonia Bacterial pneumonia Aspergillus pneumonia Nonspeciflc pneumonia Aspergillus pneumonia Diffuse alveolar damage Aspergillus pneumonia Aspergillus pneumonia Cunninghamella pneumonia Cytomegalovirus pneumonia Aspergillus pneumonia Nonspecific pneumonitis Aspergillus pneumonia S epidermidis pneumonia Diffuse alveolar damage P maltophilia pneumonia Nonspecific pneumonitis Diffuse alveolar damage Aspergillus pneumonia Aspergillus pneumonia Aspergillus pneumonia
*BAL = bronchoalveolar lavage; PI = pulmonary infiltrates; ALL = acute lymphocytic leukemia; ANLL = acute nonlymphocytic leukemia; CML = chronic myelogenous leukemia; ND=not done. tSputum culture done within two weeks before BAL.
B before BAL. Four patients who had Aspergillus pneumonia had not received amphotericin B before BAL; yet the lavage procedure failed to yield the diagnosis. BAL was also nondiagnostic in the five patients with Aspergillus pneumonia who were receiving amphotericin B before BAL (a median dose of
406 mg: range, 360 to 720 mg). Because of worsening pulmonary status in four patients, empiric therapy with amphotericin B was instituted on the same day BAL was performed (patients 10, 11, 20 and 21). In all four, BAL results were nondiagnostic. Of 14 patients who had fungal pneumonia, 12 had CHEST I 84 I 4 100000ER. 1988
747
disseminated fungal infection at autopsy Two(one with Candida pneumonia and one with Aspergillus pneumonia) had disease limited to the lungs. The pulmonary disease was the direct cause of death in nine patients and contributed significantly to death in 12 other patients. The cause of death was attributed to extrapulmonary pathology in the one patient with the localized, nonspecific pneumonia. Despite the fact that the overwhelming majority of patients were found to have residual leukemia at autopsy no patient had pulmonary parenchymal involvement by leukemic cells. No major complications occurred in any of our patients. DISCUSSION
In our study BAL had a low diagnostic yield (15 percent, 3/20 specific diseases) for pulmonary infiltrates in a subset of adults with acute leukemia, fever, profound neutropenia, and pulmonary infiltrates suggestive of multiple processes. This low yield was particularly striking among patients with pulmonary aspergillosis. These results contrast with findings of previous studies, 18-21 which reported higher diagnostic yields of BAL in immunocompromised patients without AIDS. There are, however, significant differences in patients' characteristics between our study and those of others. While other studies included heterogeneous groups of patients having different underlying diseases, receiving various immunosuppresive drug regimens or undergoing allogeneic bone marrow transplantation, the present study included only patients with acute leukemia receiving antineoplastic chemotherapy This point is important to emphasize, since certain underlying diseases are more frequently associated with specific pathogens than others. In contrast to patients undergoing allogenenic bone marrow transplantation, patients with acute leukemia rarely develop Pneumocqstis carinii or CMV pneumonia. 1.2 Yet, the majority of patients included in the studies showing a high diagnostic yield of BAL had one or the other of these two diagnoses. Because infections from these organisms have a significant alveolar component, a high diagnostic yield of BAL, which samples a large area of alveolar space, can be expected. 17•25,26 On the other hand, infection with Aspergillus sp, which constituted the majority of the specific diagnoses obtained only at autopsy in our' study is frequently associated with vascular invasion and subsequent hemorrhagic infarction of the pulmonary parenchyma, without significant infectious process in the alveolar space. 27•28 Therefore, BAL would be less successful in establishing the diagnosis of Aspergillus pneumonia. In the study by Kahn et al,22 nine of 17 patients with invasive pulmonary aspergillosis had Aspergillus hyphae in the BAL sample. However, this study included 748
a heterogeneous group of patients, and the majority of 65 patients who were considered not to have Aspergillus pneumonia did not have tissue confirmation. Thus, the diagnostic yield of BAL for Aspergillus pneumonia remains to be determined in larger prospective studies. The diagnostic yield could be improved by assay for Aspergillus antigen or oxalic acid level in the BAL sample. 29,30 An additional inherent source of bias in our study is that it was limited to autopsy subjects. It is possible that a correct diagnosis might have been obtained in patients who recovered and were therefore not included in this stud)! Such a correct diagnosis may have significantly contributed to the management and successful outcome of these patients. This is the first report to our knowledge of the BAL diagnosis of Candida pneumonia confirmed by concurrent tissue examination in acute leukemia. Of four patients who had documented Candida pneumonia at autopsy three had numerous yeasts and pseudohyphae ill all examined fields in the BAL cyctologic study Linder et al" reported that greater than ten organisms (yeast or pseudohyphae) per cytocentrifuge slide of BAL specimen were clinically significant in Candida pneumonia, although there was no correlation with histolopathology When their criteria are used, three additional specimens in our study would be categorized as positive (actually false positive). Therefore, the specificity falls from 100 to 83 percent, and the positive predictive value from 100 to 50 percent, while the sensitivity remains the same (75 percent). There was a significant number of false positive results with BAL. Of 27 organisms isolated from a BAL specimen, only four (all Candida sp) were considered pathogenic at autopsy examination. Although quantitative cultures were not done in this study two recent reports have emphasized their utility in detecting causative pathogens in acute bacterial pneumonia. 32.33 Thorpe et al32 found a high correlation between a positive (~10' cfulml) BAL semiquantitative culture and clinical findings in acute aerobic bacterial pneumonia. On the other hand, Kahn et al33 showed that even when quantitative culture yielded more than lOS cfu/ml, the presence of more than 1 percent of squamous epithelial cells in the BAL sample accurately predicts the presence of heavy contamination of the sample by oropharyngeal flora, Additional studies to validate either observation would be valuable in the diagnosis of acute bacterial pneumonia. Nevertheless, it is important to remember that patients with acute leukemia who develop fever and pulmonary infection while neutropenic receive broad-spectrum antibacterial agents, usually in combination.P' Viral pneumonia is also rare in this patient population. 1.2 Hence, fungal pneumonia is our concern, especially since it constitutes the majority of fatal pulmonary infiltrates in BAL In Diagnosisof PulmonaryInfiltratesIn Leukemia (S8Ito et III)
patients with acute leukemia. 35.36 BAL appears to have a good sensitivity and specificity for Candida pneumonia but a low diagnostic yield for pulmonary aspergillosis in patients with acute leukemia, profound neutropenia, and pulmonary infiltrates suggestive of multiple processes. It is important to remember, however, that even lung biopsy results can be misleading in a high proportion of cases, is not safe for many leukemia patients, and has not improved outcome in this patient population. Because of its availability and safet}; BAL remains an important procedure for the evaluation of pulmonary infiltrates in immunocompromised patients. Larger prospective studies of the BAL procedure, including assays for fungal antigens and metabolic products, as well as cytologic and microbiologic studies of the BAL specimen, are needed to improve our understanding and management of these seriously ill patients. REFERENCES 1 Bodey G~ Powell RD, Hersh EM, Yeterian A, Freireich EJ. Pulmonary complications of acute leukemia. Cancer 1966; 19:781-93 2 Tenholder MF, Hooper LB. Pulmonary infiltrates in leukemia. Chest 1980; 78:468-73. 3 Sickles EA, Young VM, Greene WH, Wiernik PH. Pneumonia in acute leukemia. Ann Intern Med 1973; 79:528-34 4 Chang H~ Rodriguez ~ Narboni G, Bodey G~ Luna MA, Freireich EJ. Causes of death in adults with acute leukemia. Medicine 1976; 55:259-68 5; McCabe RE, Brooks RG, Mark JD, Remington JS. Open lung biopsy in patients with acute leukemia. Am J Med 1985; 78:60916 6 HiatlJR, Gong H, Mulder DG, Ramming KE The value of open lung biopsy in the immunosuppressed patient Surgery 1982; 92: 285-91 7 McKenna RJ. Mountain CF, McMurtrey MJ. Open lung biopsy in immunocompromised patients. Chest 1984; 86:671-74 8 Pennington JE, Feldman NI Pulmonary infiltrates and fever in patients with hematologic malignancy: assessment of transbronchial biops}t Am J Med 1977; 62:581-87 9 Nishio IN, Lynch JI? Fiberoptic bronchoscopy in the immunocompromised host: the significance of a "nonspecific" transbronchial biops}t Am Rev Respir Dis 1980; 121:3C11-12 10 PuksaS, Hutcheon MA, HylandRH. UsefuInessoftransbroncial biopsy in immunosuppressed patients with pulmonary infiltrates. Thorax 1983; 38:146-50 11 Albelda SM, 11llbot GH, Genon SL, Miller WI: Cassileth PA. Role of fiberoptic bronchoscopy in the diagnosis of invasive pulmonary aspergillosis in patients with acute leukemia. Am J Med 1984; 76:1027-34 12 Castellino RA, Blank N. Etiologic diagnosis of focal pulmonary infection in immunocompromised patients by 8uoroscopica1ly guided percutaneous needle aspiration. Radiology 1979; 132:
563-67
13 Cunningham JH, Zavala DC, Corry RJ, Keim LW 'Irephtne air drill, bronchial brush, and fiberoptic transbronchiallung biopsies in immunosuppressed patients. Am Rev Respir Dis 1977; 115:213-20 14 Webster J, Clarke J. A doctors dilemma-l980 [Editorial]. Chest 1980; 78:417-18 15 Wardman AC, Cooke NJ. Pulmonary infiltrates in adult acute leukaemia: empirical treatment or lung biopsy [editorial]. Thorax 1984; 39:647-50
16 Robin ED, Burke CM. Lung biopsy in immunosuppressed patients. Chest 1986; 89:276-78 17 Kelley J, Landis IN, Davis GS, Uainer rn, Jakab GJ, Green GM. Diagnosis of pneumonia due to pneumocystis by subsegmental pulmonary lavage via the fiberoptic bronchoscope. Chest 1978; 74:24-28 18 Young}A, Hopkin JM, Cuthbertson WE Pulmonary infiltrates in immunocompromised patients: diagnosis by cytological examination of bronchoalveolar lavage fluid. J Clio Pathol 1984; 37:390-97 19 Stover DE, Zaman MB, Hajdu S1: Lange M, Gold J, Armstrong D. Bronchoalveolar lavage in the diagnosis of diffuse pulmonary infiltrates in the immunosuppressed host. Ann Intern Med 1984; 101:1-7 20 Cordonnier C, Bernaudin JF, Fleury J, Feuilhade M, Haioun C, Payen 0, et al. Diagnostic yield ofbronchoalveolar lavage in pneumonitis occurring after allogeneic bone marrow transplantation. Am Rev Respir Dis 1985; 132:1118-23 21 Springmeyer SC, Hackman RC, Holle R, Greenberg GM, Weems CE, Myerson D. et al. Use ofbronchoalveolar lavage to diagnose acute diffuse pneumonia in the immunocompromised host. J Infect Dis 1986; 154:604-10 22 Kahn ~ Jones JM, England OM. The role ofbronchoalveolar lavage in the diagnosis of invasive pulmonary aspergillosis. Am J Clio Patholl986; 86:518-23 23 Broaddus C, Dake MD, Stulbarg MS, Blumenfeld ~ Haldley WK, Golden}A, et ale Bronchoalveolar lavage and transbronchial biopsy for the diagnosis of pulmonary infections in the acquired immunodeficiency syndrome. Ann Intern Med 1985; 102:74752 24 Martin WJ, Smith TF, Sanderson DR, Brutinel WM. Cockerill FR, Douglas WW Role of bronchoalveolar lavage in the assessment of opportunistic pulmonary infections: utility and complications. Mayo Clio Proc 1987; 62:549-57 25 Price RA, Hughes WI: Histopathology of PneumocyBtis carini. infestation and infection in malignant disease in childhood. Hum Patholl974; 5:737-52 26 Craighead JE. Pulmonary cytomegalovirus infection in the adult Am J Patholl971; 63:487-99 27 Orr D~ Myerowitz RL, Dubois PJ. Patho-radiologic correlation of invasive pulmonary aspergillosis in the compromised host. Cancer 1978; 41:2028-39 28 Young RC, Bennett JE, \\)gel CL, Carbone P~ DeVita VI: Aspergillosis: the spectrum of the disease at 98 patients. Medicine 1970; 49:147-73 29 Andrews C£ Weiner MH. Aspergillus antigen detection in bronchoalveolar lavage fluid from patients with invasive aspergillosis and aspergillomas. Am J Med 1982; 73:372-80 30 Benoit G, Chauvin MF, Cordonnier C, Astier A, Bernaudin JF: Oxalic acid level in bronchoalveolar lavage fluid from patients with invasive pulmonary aspergillosis. Am Rev Respir Dis 1985; 132:748-51 31 Linder J, Vaughan W£ Armitage JO, Ghafouri MA, Hurkman 0, Mroczek EC, et al. Cytopathology of opportunistic infection in bronchoalveolar lavage. Am J Clin Patholl987; 88:421-28 32 Thorpe JE, Baughman R£ Frame YI: Wesseler TA, Staneck JL. Bronchoalveolar lavage for diagnosing acute bacterial pneumonia. J Infect Dis 1987; 155:855-61 33 Kahn ~ Jones JM. Diagnosing bacterial respiratory infection by bronchoalveolar lavage. J Infect Dis 1987; 155:862-69 34 Bodey GE Antibiotics in patients with neutropenia. Arch Intern Med 1984; 144:1845-51 35 Bodey GE Fungal infections complicating acute leukemia. J Chronic Dis 1966; 19:667-87 36 Degregorio M~ 'Lee WF, Linker CA, Jacobs RA, Ries CA. Fungal infections in patients with acute leukemia. Am J Med 1982; 73:543-48 CHEST I 94 I .. I OClOBER. 1988
748
speci6c diseases); all three were Candida pneumonia. The sensitivity of HAL was 75 percent and its speci6city 100 percent, for Candida pneumonia. HAL did not result in a speci6c diagnosis for the 17 remaining diseases, nine of which were Aspergillus pneumonia. In seven patients in whom autopsy was performed within 72 hours of BAL, lavage results correlated with those of autopsy in only one who had Candida pneumonia. All HAL cultures were falsely positive, except in four cases of Candida pneumonia. The therapeutic regimen was not m0di6ed according to the HAL results in any of the 22 patients. There were no major complications associated with the procedure. (Cheat 1988; 94:74&-49)
The diagnosis and management of pulmonary infiltrates in patients with acute leukemia is a challenge to clinicians caring for these patients. Pulmonary infiltrates in this population can have a variety of infectious and noninfectious causes and are often fatal. l -4 Many studies have reported on the utility of various diagnostic procedures in this clinical setting, including open lung biops~5-7 bronchoscopic exami-
serious complications.P'" As a result, BAL has been advocated as a safe and expeditious way to obtain a diagnosis in immunocompromised patients with pulmonary infiltrates. However, with the exception of two studies on patients undergoing bone marrow transplantation,20·21 the reported series included heterogeneous groups of patients with different underlying diseases and immunosuppressive drug regimens, 17-19.22 and a final diagnostic confirmation by histopathology was not available for most patients. Histopathologic examination obtained by open lung biopsy has been misleading in a high proportion of patients with acute leukemia." As a result, autopsy examination remains the most definitive end point for assessing the value of any diagnostic procedure, and even then a significant number will not have a specific diagnosis. The utility of BAL for the diagnosis of pulmonary infiltrates in a homogeneous group of patients with acute leukemia has not been determined. We attempt to do so, reporting our experience with BAL in 22 consecutive adults with acute leukemia and pulmonary infiltrates, all of whom had histopathologic confirmation at autopsy
For editorial comment see page 676 nation with bronchial brushing, washing, and transbronchial biopsy 8-11 and percutaneous needle aspiration or biopsy 12.13 These procedures, however, have been associated with a low diagnostic yield or a number of complications. They also cannot be routinely performed in patients with acute leukemia, because many have coagulation abnormalities or compromised cardiopulmonary systems. 14-16 Recently bronchoalveolar lavage (BAL) has been evaluated for the diagnosis of pulmonary infiltrates in immunocompromised patients, 17-24 where its diagnostic yield in immunocompromised patients without acquired immunodeficiency syndrome (AIDS) has been reported to range from 50 to 93 percent, without ·From the Department of Medical Specialties, and Department of Pathology The University of Texas System Cancer Center, M. D. Anderson Hospital and Tumor Institute, Houston. tSection of Infectious Diseases. § Department of Pathology *Section of Cardiopulmonary Medicine, Department of Pathology Manuscript received February 15, revision accepted March 24 Reprint requests: Dr. Anai8.rie, lnfectiow Diseae&, Box 47, 1515
Holcombe Blvd, Houston 77030
MATERIALS AND METHODS
reviewed the medical records of all adults with acute leukemia cared for at the University of Texas M. D. Anderson Hospital and Tumor Institute between October 1984 and January 1987 and examined by BAL for pulmonary infiltrates. Patients were included only if they met the following criteria: (1) a diagnosis of acute leukemia or chronic myelogenous leukemia (CML) in the terminal phase being treated with intensive chemotherapy; (2) a neutrophil ~
CHEST I 94 I 4 I OClOBER. 1988
745
count of less than I,OOO/fJ.I at the time of BAL; (3) an autopsy examination performed within three weeks of BAL; and (4)a similar location for the disease found at autopsy the pulmonary infiltrates shown on chest x-ray film, and the BAL sampling site. Patients who underwent allogeneic bone marrow transplantation were excluded from analysis. Patients were further evaluated for presence and duration offever (temperature greater than38.3°C) and neutropenia preceding BAL, the chest x-ray film pattern and its chronologie change, the number of days from the x-ray detection of pulmonary infiltrates to BAL and from BAL to autopsy results of cultures from the two weeks preceding BAL until death, and antileukemic and antimicrobial therapy before BAL. Changes of therapeutic regimens as a consequence of the results obtained by BAL were also analyzed. Bilateral pulmonary infiltrates on chest x-ray film were categorized as diffuse involvement, while all other infiltrates were considered to be localized disease. BAL was performed in patients who were considered to have extensive pulmonary infiltrates or who developed new pulmonary infiltrates while receiving antibacterial antibiotics. Bronchoscopy was performed with an Olympus BF-B3 brochoscope (Olympus Corporation of America), The bronchoscope was passed transnasally or transorally When patients were intubated, bronchoscopy was performed through an endotracheal tube. The tracheobronchial tree was routinely examined, and then the bronchoscope was advanced further and wedged into a subsegmental bronchus of an involved area. Lavage was performed using 60 to 200 ml of sterile normal saline solution in 20- to 3O-ml aliquots through a bronchoscopic channel. Six patients had bronchial brushings done after the lavage procedure. A portion of BAL 8uid was submitted to the clinical microbiology laboratory for bacterial testing, including Legionella sp, fungal, and mycobacterial cultures. Gram stain and acid-fast stain were also done at the same laboratory Viral cultures and direct immunofluorescent staining for Legionella sp were performed, and the remainder of the lavage 8uid was centrifuged at 500 g for 10 minutes and the sediment smeared and stained by the Papanicolaou, MayGrunwald-Giemsa, and Gomori methenamine silver methods. Brushing samples were cultured for bacterial, fungal, and mycobacterial cultures and stained using the same methods as BAL 8uid for cytologic examination. Sections of lung specimens from postmortem examination were examined using hematoxylin and eosin stain, Gomori methenamine silver stain, and Gram stain. All slides submitted for cytologic and histopathologic examination were reviewed by one of us (H.D. ~ Fungal and bacterial pneumonia were diagnosed by the presence of tissue invasion, whether or not this was confirmed by culture results. When the destruction of alveolar structures was present without any identifiable microorganism, the diagnosis of nonspecific pneumonitis was made. RESULTS
The clinical characteristics of the 22 patients who met our inclusion criteria are shown in Table 1. There were ten men and 12 women, who had a median age of55 years (range, 20 to 81 years). Underlying diseases were acute nonlymphocytic leukemia (ten patients), acute lymphocytic leukemia (seven patients), and CML in blastic transformation (four patients) and in accelerated phase (one patient). Four patients had received antileukemic chemotherapy and total body irradiation followed by autologous bone marrow transplantation, eight were receiving their first induction chemotherapy two were in remission following therap~ and 12 were in leukemic relapse. All but one patient was severely neutropenic (a neutrophil count 748
less than IOO/I-LI), and all had a platelet count of less than 50 X I()3/I-LI (less than 20 x l()3/I-LI in 14 patients). The pulmonary infiltrates were diffuse in 12 patients and localized in ten. Arterial blood gases were obtained in 17 patients, 11 of whom were receiving supplemental oxygen. The Pa02 was less than 60 mm Hg in six patients, between 60 and 80 mm Hg in six, and more than 80 mm Hg in five. In the two weeks preceding BAL, 21 patients received one or more broad-spectrum Ji-Iactam antibiotics with or without an aminoglycoside. 'Irimethoprim-sulfamethoxazole was also given to 11 patients and Acyclovir to four patients. Amphotericin B was given to 15 patients for a median dose of 360 mg (range, 10 to 1271 mg). The median interval between the detection of pulmonary infiltrates and BAL was seven days (range, 0 to 23 days); it was nine days between BAL and autopsy (range, 1 to 20 days). Autopsy was performed within 72 hours after BAL in seven patients. Results of BAL and autopsy examinations are summarized in Table 1. Correlation between the BAL cytology and autopsy findings for Candida pneumonia was considered present only when all fields of a slide examined for cytology showed a very large number of yeasts and pseudohyphae. BAL was diagnostic in only three patients, all of whom had histologic documentation of Candida pneumonia on autopsy In the remaining four patients without autopsy evidence of Candida pneumonia, the submitted BAL specimens showed scattered fungal elements, predominantly yeast cells with or without pseudohyphae, and all four were considered negative. The BAL cultures from 21 patients yielded different bacterial, fungal, mycobacterial, and viral organisms but did not correlate with the histopathologic findings at autopsy except in four cases of Candida pneumonia. Bronchial brushing was done in six patients but was nondiagnostic. Results of BAL in seven patients in whom autopsy was performed within 72 hours correlated with the pathologic findings in only one, who had Candida pneumonia. However, a concurrent bacterial pneumonia in the same patient was missed by BAL. Except for the above-mentioned patient with Candida pneumonia, there was no positive correlation between BAL results and autopsy findings in five patients who underwent BAL less than 24 hours after the pulmonary infiltrates were first detected. The overall diagnostic yield of BAL was 15 percent (three of 20 cases of disease). With respect to Candida pneumonia, BAL cytologic study yielded a sensitivity of 75 percent (3/4), a specificity of 100 percent (3/3), and a positive predictive value of 100 percent (3/3). The BAL did not result in change in the therapeutic regimen in any of the patients. All three patients whose results of BAL cytology were positive for Candida pneumonia had been receiving amphotericin SALin Diagnosis of Pulmonary InfiltratesIn Leukemia (Saitost 8/)
Table I-Patient Characteriatict and Beaulta o/BAL and Autopay* TIme Interval (Days) Between
Amphotericin B(mg); Amt. Patient Age, yrl Leukemia PI and BAL and Before BAU No. Sex Type BAL Autopsy Total Amt.
Sputumt Culture
BAL Cytology
Culture
1
39/M
ALL
1
1
II/51
2
561F
ALL ANLL CML
8 10 17
15 8 17
1361498
61/F 37/F
411205 127112171
Yeast
5
58/M
ANLL
2
211101
Yeast
6
69/F
ANLL
7
5
361/611
C albicans
7
831M
CML
7
1
3011441
Gram-negative rods
Pseudomonas
8
20/F
ALL
23
19
40611041
Paeroginosa
P aeroginosa
9
311M
ANLL
o
010
Yeast
10
81/F
ANLL
8
7
0/226
11
431F
ALL
7
16
720/1460
12
681M
CML
6
3
0/0
13
67/M
ANLL
2
3
010
14
231F
CML
10
10
4501825
15
261F
ANLL
18
12
58111421
16
47/M
ALL
4
12
1140/2010
Yeast
Yeast
17
49/F
ANLL
5
12
1361451
C albicaf&8
Yeast
18
651M
ANLL
21
3
401/401
19
59/M
ALL
o
4
010
ND
Yeast, Mycobacterium avium-intraceUulare Yeast
20
77/F
ALL
12
15
om6
ND
Yeast
21
541M
CML
10
0/440
Yeast
22
381F
ANLL
20
36011240
3 4
9
Candida Candida albictJf&8
Yeast
Klebsiella pneumonia Acinetobacter sp
Enterococcus
Aspergillus
flavus
Candida Yeast, Streptococcus Candida C albictJf&8, Adenovirus Yeast
C albictJns C albictJns Candida spp
maltophiliD
Yeast
.Staphylococcus epitlermidis, yeast Yeast
Enterococcus
Penicillium
S epidemidis
Yeast, Enterococcus
Autopsy (Lung) Candida pneumonia, bacterial pneumonia; grampositive cocci Candida pneumonia Candida pneumonia Nonspecific pneumonia Candida pneumonia Bacterial pneumonia Aspergillus pneumonia Nonspeciflc pneumonia Aspergillus pneumonia Diffuse alveolar damage Aspergillus pneumonia Aspergillus pneumonia Cunninghamella pneumonia Cytomegalovirus pneumonia Aspergillus pneumonia Nonspecific pneumonitis Aspergillus pneumonia S epidermidis pneumonia Diffuse alveolar damage P maltophilia pneumonia Nonspecific pneumonitis Diffuse alveolar damage Aspergillus pneumonia Aspergillus pneumonia Aspergillus pneumonia
*BAL = bronchoalveolar lavage; PI = pulmonary infiltrates; ALL = acute lymphocytic leukemia; ANLL = acute nonlymphocytic leukemia; CML = chronic myelogenous leukemia; ND=not done. tSputum culture done within two weeks before BAL.
B before BAL. Four patients who had Aspergillus pneumonia had not received amphotericin B before BAL; yet the lavage procedure failed to yield the diagnosis. BAL was also nondiagnostic in the five patients with Aspergillus pneumonia who were receiving amphotericin B before BAL (a median dose of
406 mg: range, 360 to 720 mg). Because of worsening pulmonary status in four patients, empiric therapy with amphotericin B was instituted on the same day BAL was performed (patients 10, 11, 20 and 21). In all four, BAL results were nondiagnostic. Of 14 patients who had fungal pneumonia, 12 had CHEST I 84 I 4 100000ER. 1988
747
disseminated fungal infection at autopsy Two(one with Candida pneumonia and one with Aspergillus pneumonia) had disease limited to the lungs. The pulmonary disease was the direct cause of death in nine patients and contributed significantly to death in 12 other patients. The cause of death was attributed to extrapulmonary pathology in the one patient with the localized, nonspecific pneumonia. Despite the fact that the overwhelming majority of patients were found to have residual leukemia at autopsy no patient had pulmonary parenchymal involvement by leukemic cells. No major complications occurred in any of our patients. DISCUSSION
In our study BAL had a low diagnostic yield (15 percent, 3/20 specific diseases) for pulmonary infiltrates in a subset of adults with acute leukemia, fever, profound neutropenia, and pulmonary infiltrates suggestive of multiple processes. This low yield was particularly striking among patients with pulmonary aspergillosis. These results contrast with findings of previous studies, 18-21 which reported higher diagnostic yields of BAL in immunocompromised patients without AIDS. There are, however, significant differences in patients' characteristics between our study and those of others. While other studies included heterogeneous groups of patients having different underlying diseases, receiving various immunosuppresive drug regimens or undergoing allogeneic bone marrow transplantation, the present study included only patients with acute leukemia receiving antineoplastic chemotherapy This point is important to emphasize, since certain underlying diseases are more frequently associated with specific pathogens than others. In contrast to patients undergoing allogenenic bone marrow transplantation, patients with acute leukemia rarely develop Pneumocqstis carinii or CMV pneumonia. 1.2 Yet, the majority of patients included in the studies showing a high diagnostic yield of BAL had one or the other of these two diagnoses. Because infections from these organisms have a significant alveolar component, a high diagnostic yield of BAL, which samples a large area of alveolar space, can be expected. 17•25,26 On the other hand, infection with Aspergillus sp, which constituted the majority of the specific diagnoses obtained only at autopsy in our' study is frequently associated with vascular invasion and subsequent hemorrhagic infarction of the pulmonary parenchyma, without significant infectious process in the alveolar space. 27•28 Therefore, BAL would be less successful in establishing the diagnosis of Aspergillus pneumonia. In the study by Kahn et al,22 nine of 17 patients with invasive pulmonary aspergillosis had Aspergillus hyphae in the BAL sample. However, this study included 748
a heterogeneous group of patients, and the majority of 65 patients who were considered not to have Aspergillus pneumonia did not have tissue confirmation. Thus, the diagnostic yield of BAL for Aspergillus pneumonia remains to be determined in larger prospective studies. The diagnostic yield could be improved by assay for Aspergillus antigen or oxalic acid level in the BAL sample. 29,30 An additional inherent source of bias in our study is that it was limited to autopsy subjects. It is possible that a correct diagnosis might have been obtained in patients who recovered and were therefore not included in this stud)! Such a correct diagnosis may have significantly contributed to the management and successful outcome of these patients. This is the first report to our knowledge of the BAL diagnosis of Candida pneumonia confirmed by concurrent tissue examination in acute leukemia. Of four patients who had documented Candida pneumonia at autopsy three had numerous yeasts and pseudohyphae ill all examined fields in the BAL cyctologic study Linder et al" reported that greater than ten organisms (yeast or pseudohyphae) per cytocentrifuge slide of BAL specimen were clinically significant in Candida pneumonia, although there was no correlation with histolopathology When their criteria are used, three additional specimens in our study would be categorized as positive (actually false positive). Therefore, the specificity falls from 100 to 83 percent, and the positive predictive value from 100 to 50 percent, while the sensitivity remains the same (75 percent). There was a significant number of false positive results with BAL. Of 27 organisms isolated from a BAL specimen, only four (all Candida sp) were considered pathogenic at autopsy examination. Although quantitative cultures were not done in this study two recent reports have emphasized their utility in detecting causative pathogens in acute bacterial pneumonia. 32.33 Thorpe et al32 found a high correlation between a positive (~10' cfulml) BAL semiquantitative culture and clinical findings in acute aerobic bacterial pneumonia. On the other hand, Kahn et al33 showed that even when quantitative culture yielded more than lOS cfu/ml, the presence of more than 1 percent of squamous epithelial cells in the BAL sample accurately predicts the presence of heavy contamination of the sample by oropharyngeal flora, Additional studies to validate either observation would be valuable in the diagnosis of acute bacterial pneumonia. Nevertheless, it is important to remember that patients with acute leukemia who develop fever and pulmonary infection while neutropenic receive broad-spectrum antibacterial agents, usually in combination.P' Viral pneumonia is also rare in this patient population. 1.2 Hence, fungal pneumonia is our concern, especially since it constitutes the majority of fatal pulmonary infiltrates in BAL In Diagnosisof PulmonaryInfiltratesIn Leukemia (S8Ito et III)
patients with acute leukemia. 35.36 BAL appears to have a good sensitivity and specificity for Candida pneumonia but a low diagnostic yield for pulmonary aspergillosis in patients with acute leukemia, profound neutropenia, and pulmonary infiltrates suggestive of multiple processes. It is important to remember, however, that even lung biopsy results can be misleading in a high proportion of cases, is not safe for many leukemia patients, and has not improved outcome in this patient population. Because of its availability and safet}; BAL remains an important procedure for the evaluation of pulmonary infiltrates in immunocompromised patients. Larger prospective studies of the BAL procedure, including assays for fungal antigens and metabolic products, as well as cytologic and microbiologic studies of the BAL specimen, are needed to improve our understanding and management of these seriously ill patients. REFERENCES 1 Bodey G~ Powell RD, Hersh EM, Yeterian A, Freireich EJ. Pulmonary complications of acute leukemia. Cancer 1966; 19:781-93 2 Tenholder MF, Hooper LB. Pulmonary infiltrates in leukemia. Chest 1980; 78:468-73. 3 Sickles EA, Young VM, Greene WH, Wiernik PH. Pneumonia in acute leukemia. Ann Intern Med 1973; 79:528-34 4 Chang H~ Rodriguez ~ Narboni G, Bodey G~ Luna MA, Freireich EJ. Causes of death in adults with acute leukemia. Medicine 1976; 55:259-68 5; McCabe RE, Brooks RG, Mark JD, Remington JS. Open lung biopsy in patients with acute leukemia. Am J Med 1985; 78:60916 6 HiatlJR, Gong H, Mulder DG, Ramming KE The value of open lung biopsy in the immunosuppressed patient Surgery 1982; 92: 285-91 7 McKenna RJ. Mountain CF, McMurtrey MJ. Open lung biopsy in immunocompromised patients. Chest 1984; 86:671-74 8 Pennington JE, Feldman NI Pulmonary infiltrates and fever in patients with hematologic malignancy: assessment of transbronchial biops}t Am J Med 1977; 62:581-87 9 Nishio IN, Lynch JI? Fiberoptic bronchoscopy in the immunocompromised host: the significance of a "nonspecific" transbronchial biops}t Am Rev Respir Dis 1980; 121:3C11-12 10 PuksaS, Hutcheon MA, HylandRH. UsefuInessoftransbroncial biopsy in immunosuppressed patients with pulmonary infiltrates. Thorax 1983; 38:146-50 11 Albelda SM, 11llbot GH, Genon SL, Miller WI: Cassileth PA. Role of fiberoptic bronchoscopy in the diagnosis of invasive pulmonary aspergillosis in patients with acute leukemia. Am J Med 1984; 76:1027-34 12 Castellino RA, Blank N. Etiologic diagnosis of focal pulmonary infection in immunocompromised patients by 8uoroscopica1ly guided percutaneous needle aspiration. Radiology 1979; 132:
563-67
13 Cunningham JH, Zavala DC, Corry RJ, Keim LW 'Irephtne air drill, bronchial brush, and fiberoptic transbronchiallung biopsies in immunosuppressed patients. Am Rev Respir Dis 1977; 115:213-20 14 Webster J, Clarke J. A doctors dilemma-l980 [Editorial]. Chest 1980; 78:417-18 15 Wardman AC, Cooke NJ. Pulmonary infiltrates in adult acute leukaemia: empirical treatment or lung biopsy [editorial]. Thorax 1984; 39:647-50
16 Robin ED, Burke CM. Lung biopsy in immunosuppressed patients. Chest 1986; 89:276-78 17 Kelley J, Landis IN, Davis GS, Uainer rn, Jakab GJ, Green GM. Diagnosis of pneumonia due to pneumocystis by subsegmental pulmonary lavage via the fiberoptic bronchoscope. Chest 1978; 74:24-28 18 Young}A, Hopkin JM, Cuthbertson WE Pulmonary infiltrates in immunocompromised patients: diagnosis by cytological examination of bronchoalveolar lavage fluid. J Clio Pathol 1984; 37:390-97 19 Stover DE, Zaman MB, Hajdu S1: Lange M, Gold J, Armstrong D. Bronchoalveolar lavage in the diagnosis of diffuse pulmonary infiltrates in the immunosuppressed host. Ann Intern Med 1984; 101:1-7 20 Cordonnier C, Bernaudin JF, Fleury J, Feuilhade M, Haioun C, Payen 0, et al. Diagnostic yield ofbronchoalveolar lavage in pneumonitis occurring after allogeneic bone marrow transplantation. Am Rev Respir Dis 1985; 132:1118-23 21 Springmeyer SC, Hackman RC, Holle R, Greenberg GM, Weems CE, Myerson D. et al. Use ofbronchoalveolar lavage to diagnose acute diffuse pneumonia in the immunocompromised host. J Infect Dis 1986; 154:604-10 22 Kahn ~ Jones JM, England OM. The role ofbronchoalveolar lavage in the diagnosis of invasive pulmonary aspergillosis. Am J Clio Patholl986; 86:518-23 23 Broaddus C, Dake MD, Stulbarg MS, Blumenfeld ~ Haldley WK, Golden}A, et ale Bronchoalveolar lavage and transbronchial biopsy for the diagnosis of pulmonary infections in the acquired immunodeficiency syndrome. Ann Intern Med 1985; 102:74752 24 Martin WJ, Smith TF, Sanderson DR, Brutinel WM. Cockerill FR, Douglas WW Role of bronchoalveolar lavage in the assessment of opportunistic pulmonary infections: utility and complications. Mayo Clio Proc 1987; 62:549-57 25 Price RA, Hughes WI: Histopathology of PneumocyBtis carini. infestation and infection in malignant disease in childhood. Hum Patholl974; 5:737-52 26 Craighead JE. Pulmonary cytomegalovirus infection in the adult Am J Patholl971; 63:487-99 27 Orr D~ Myerowitz RL, Dubois PJ. Patho-radiologic correlation of invasive pulmonary aspergillosis in the compromised host. Cancer 1978; 41:2028-39 28 Young RC, Bennett JE, \\)gel CL, Carbone P~ DeVita VI: Aspergillosis: the spectrum of the disease at 98 patients. Medicine 1970; 49:147-73 29 Andrews C£ Weiner MH. Aspergillus antigen detection in bronchoalveolar lavage fluid from patients with invasive aspergillosis and aspergillomas. Am J Med 1982; 73:372-80 30 Benoit G, Chauvin MF, Cordonnier C, Astier A, Bernaudin JF: Oxalic acid level in bronchoalveolar lavage fluid from patients with invasive pulmonary aspergillosis. Am Rev Respir Dis 1985; 132:748-51 31 Linder J, Vaughan W£ Armitage JO, Ghafouri MA, Hurkman 0, Mroczek EC, et al. Cytopathology of opportunistic infection in bronchoalveolar lavage. Am J Clin Patholl987; 88:421-28 32 Thorpe JE, Baughman R£ Frame YI: Wesseler TA, Staneck JL. Bronchoalveolar lavage for diagnosing acute bacterial pneumonia. J Infect Dis 1987; 155:855-61 33 Kahn ~ Jones JM. Diagnosing bacterial respiratory infection by bronchoalveolar lavage. J Infect Dis 1987; 155:862-69 34 Bodey GE Antibiotics in patients with neutropenia. Arch Intern Med 1984; 144:1845-51 35 Bodey GE Fungal infections complicating acute leukemia. J Chronic Dis 1966; 19:667-87 36 Degregorio M~ 'Lee WF, Linker CA, Jacobs RA, Ries CA. Fungal infections in patients with acute leukemia. Am J Med 1982; 73:543-48 CHEST I 94 I .. I OClOBER. 1988
748