|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 1999 American Society for Clinical Oncology Pneumonia in Febrile Neutropenic Patients and in Bone Marrow and Blood Stem-Cell Transplant Recipients: Use of High-Resolution Computed TomographyFrom the Departments of Radiology and Internal Medicine III, Divisions of Hematology and Pneumology, Johannes Gutenberg-University, Mainz, Germany. Address reprint requests to Dr Claus Peter Heussel, Department of Radiology, Johannes Gutenberg-University Mainz, Langenbeckstrasse 1, D-55131 Mainz, Germany; email heussel{at}mail.uni-mainz.de
PURPOSE: To obtain statistical data on the use of high-resolution computed tomography (HRCT) for early detection of pneumonia in febrile neutropenic patients with unknown focus of infection. MATERIALS AND METHODS: One hundred eighty-eight HRCT studies were performed prospectively in 112 neutropenic patients with fever of unknown origin persisting for more than 48 hours despite empiric antibiotic treatment. Fifty-four of these studies were performed in transplant recipients. All patients had normal chest roentgenograms. If pneumonia was detected by HRCT, guided bronchoalveolar lavage was recommended. Evidence of pneumonia on chest roentgenograms during follow-up and micro-organisms detected during follow-up were regarded as documentation of pneumonia. RESULTS: Of the 188 HRCT studies, 112 (60%) showed pneumonia and 76 were normal. Documentation of pneumonia was possible in 61 cases by chest roentgenography or micro-organism detection (54%) (P < 10-6). Sensitivity of HRCT was 87% (88% in transplant recipients), specificity was 57% (67%), and the negative predictive value was 88% (97%). A time gain of 5 days was achieved by the additional use of HRCT compared to an exclusive use of chest roentgenography. CONCLUSION: The high frequency of inflammatory pulmonary disease after a suspicious HRCT scan (> 50%) proves that pneumonia is not excluded by a normal chest roentgenogram. Given the significantly longer duration of febrile episodes in transplant recipients, HRCT findings are particularly relevant in this subgroup. Patients with normal HRCT scans, particularly transplant recipients, have a low risk of pneumonia during follow-up. All neutropenic patients with fever of unknown origin and normal chest roentgenograms should undergo HRCT.
INFECTIOUS COMPLICATIONS are the major risk during neutropenia induced by high-dose chemotherapy, and the lung is the most frequently involved organ.1-7 Localization of the focus and identification of underlying micro-organisms are essential for appropriate therapy.8 The use of computed tomography (CT), especially high-resolution computed tomography (HRCT), for investigation of the lungs in bone marrow transplant (BMT) recipients has been investigated previously. Retrospective reviews, follow-ups, and comparisons with infiltrations seen on chest roentgenograms have been performed.3,6,9-12 Prospective studies have also been performed using HRCT and chest roentgenography at standardized positions within a diagnostic algorithm to search for a focus in cases of fever of unknown origin7,13-15 (Fig 1). The use of HRCT with subsequent guided bronchoalveolar lavage (BAL) has been recommended as the most sensitive technique for the detection of pneumonia.2,4,6,9-11,13,16,17
The frequency of particular HRCT patterns and their associations with certain micro-organisms, especially Aspergillus species and Pneumocystis carinii, in immunocompromised hosts have been described.6,9,11,12,18,19 Few statistical data exist on further follow-up of HRCT patterns or on HRCT itself (eg, specificity or predictive values) for the early detection of pneumonia when chest roentgenograms are normal. Empiric antibiotic therapy for fever of unknown origin is necessary but is costly and has life-threatening adverse effects.1,2,4,8,20 Studies on the reduction of empiric therapy on the basis of HRCT findings are warranted. Determining the benefit of routine HRCT and the implications of particular patterns represents the first step in this approach. To our knowledge, no prospective study involving a large patient population and transplant recipients has been previously undertaken. In this study, we evaluated special patterns detected with HRCT alone and the use of the diagnostic algorithm based on recognition of pneumonia by HRCT and documentation during follow-up, using pneumonic infiltrates on chest roentgenograms or evidence of micro-organisms as relevant clinical end points.
Patients Patients with fever (temperature > 38.3°C) of unknown origin during neutropenia (neutrophil granulocyte count < 500/µL) caused by chemotherapy underwent chest roentgenography. When chest roentgenograms were normal, HRCT was performed the same day or by 8 AM the following day. Patients with known focus of infection and those with chest roentgenograms indicating pneumonia were excluded. Chest roentgenograms were initially read by the radiologist on duty. Chest roentgenograms and HRCT scans were evaluated, and recommendations were given, prospectively and follow-up chest roentgenograms were read retrospectively by two experienced chest radiologists (C.P.H. and H.-U.K.), who were blinded to clinical information and were required to reach consensus. Twenty initial chest roentgenography and HRCT studies were excluded later because initial chest roentgenography demonstrated faint but suspicious opacifications. Thus, we prospectively performed 188 HRCT studies in a search for a focus of infection in 112 patients with completely normal chest roentgenograms (57 women, 55 men; median age, 50 years; range, 20 to 82 years). Seventy-six studies were performed during different episodes of neutropenia in the same patients or after initially normal HRCT scans when persisting or recurrent fever was present. Fifty-four of the 188 studies were performed in transplant recipients (eight allogeneic BMT recipients, two autologous BMT recipients, one allogeneic peripheral-blood stem-cell transplant [PSCT] recipient, and 34 autologous PSCT recipients). Chest roentgenograms obtained during follow-up were included as further radiologic data. From this, we assessed the interval between HRCT scans suggestive of pneumonia and chest roentgenograms indicating pneumonia. Additionally, we documented the interval between HRCT and detection of a micro-organism. Patients underwent chemotherapy for treatment of acute myelogenous leukemia (n = 49), non-Hodgkin's lymphoma (n = 26), acute lymphatic leukemia (n = 11), chronic myelogenous leukemia (n = 10), breast cancer (n = 8), multiple myeloma (n = 3), aplastic anemia (n = 2), chronic lymphatic leukemia (n = 1), Hodgkin's disease (n = 1), or histiocytosis (n = 1). The median duration of neutropenia was 17 days (range, 4 to 67 days) in the whole sample and 15 days (range, 6 to 66 days) in BMT and PSCT recipients. The median duration of fever was 9 days (range, 2 to 66 days) in the whole sample and 7 days (range, 3 to 66 days) in BMT and PSCT recipients.
Technique HRCT patterns were documented separately for each lobe (lingula categorized as left middle lobe). Ground-glass opacities, ill-defined nodules, consolidation, cavitations, and poorly defined linear opacities are considered highly indicative of inflammation or pneumonia.9,15,18,21-24 If at least one of these findings was present, the study was classified as suggestive of pneumonia. Ground-glass opacities were additionally classified by the extent within each lobe (< 25%, 25% to 50%, or > 50% of the lobar area imaged). Performance of specific BAL or other invasive diagnostic procedures (surgery, CT-guided biopsy) was suggested to the referring physician. BAL was performed in 49 cases after recommendation and in four additional cases with normal HRCT and therefore without radiologic recommendation. In the remaining 63 cases, the patient did not consent to BAL, fever or neutropenia disappeared the day before BAL was to be performed, the different focus of or reason for fever was detected, or thrombocytopenia was present and thus the risk of massive bleeding caused by BAL was too high. Treatment was influenced by microbiologic findings only. Evidence of Aspergillus species in sputum specimens was taken into account. All other detected micro-organisms must have been evident in BAL samples. To reduce the possibility of contamination of the bronchoscope during intubation, Candida species were taken into account only when the concentration was greater than 100,000/mL. Additionally, all findings were classified as relevant or irrelevant by both the hematologist and microbiologist given the location from which the specimen had been taken and the patient's profile at bedside.
Statistics
Of the 188 HRCT studies, 112 were suggestive of pneumonia (60%). An invasive diagnostic procedure was recommended in these cases. In 39 (35%) of the 112 studies, a new opacification that was typical of pneumonia was detected on the chest roentgenogram during follow-up. The diagnosis of pneumonia was made using the criteria of the German Society of Pneumology25 (major criteria: new infiltrate on chest roentgenograms; minor criteria: fever and neutropenia). In 42 (38%) of the 112 studies, a relevant micro-organism was identified in sputum (n = 5), a BAL specimen (n = 34), a specimen obtained through CT-guided biopsy (n = 2), or a surgical specimen (n = 1). At least one of these documentations of inflammatory lung disease was evident in 61 studies (54%) (  true positive). In 19 patients, follow-up chest roentgenography showed only opacification. In some cases, findings could be confirmed by biopsy or clinical course (graft-versus-host disease [GVHD] in two cases, relapsing leukemia in four cases, overhydration in three cases, and drug toxicity in one case). The remaining 51 cases of pneumonia shown by HRCT (46%) had no evidence of inflammatory lung disease during follow-up (no opacification on chest roentgenograms or evidence of micro-organisms) (Table 1). To help determine possible reasons for the 51 false-positive HRCT findings (46%), we calculated multivariate clinical attributes of patients with pneumonia on HRCT scans with and those without documentation during follow-up. No significant differences were found with regard to age (patients with documentation: mean, 53 years; patients without documentation: mean, 52 years), duration of neutropenia before (12 v 11 days) or after HRCT (10 v 11 days), or duration of fever before (7 v 5 days) or after HRCT (7 v 6 days).
Of the 188 HRCT studies, 76 were normal (40%). In nine (12%) of the 76 cases of initially normal HRCT scans, chest roentgenograms during follow-up were suggestive of pneumonia (n = 7) and/or a relevant micro-organism was detected in lung specimens (n = 4). In the remaining 67 studies with normal HRCT findings (88%), there was no opacification on chest roentgenograms or evidence of micro-organisms during follow-up (Table 1). We calculated multivariate clinical attributes of patients with "false-negative" normal HRCT scans with and those without pulmonary inflammatory episodes during follow-up. Age was found to be significantly different between the two groups (40 v 52 years) (P < .005). Thus, pneumonia develops faster in younger patients. No significant differences were found with regard to duration of neutropenia before (9 v 11 days) or after HRCT (14 v 10 days) or duration of fever before (3 v 6 days) or after HRCT (5 v 7 days). The frequency of suspicious HRCT findings and the frequency of verified inflammatory lung disease during follow-up were lower among BMT and PSCT recipients than in the whole group (Tables 1 and 2). In the nonverified cases, the fever was suspected to have had a noninfectious cause such as GVHD (n = 2), relapse of malignant disease (n = 3), or drug toxicity (n = 1). Additionally, fever persisting for more than 48 hours is frequently caused by a different individual time gain between the start of empiric antibiotic or antifungal treatment and the time when treatment takes effect.
Follow-Up
HRCT used for the early detection of pneumonia in patients with fever of unknown origin and normal chest roentgenograms had a sensitivity of 87% (88% in BMT and PSCT recipients), a specificity of 57% (67%), a positive predictive value of 54% (32%), and a negative predictive value of 88% (97%) (Table 3).
To evaluate the clinical significance of pneumonia detected by HRCT, we calculated the time until normalization of temperature. We excluded patients whose fever ended 1 day before hematologic reconstitution, so that the study population was not affected by disturbances caused by the ending of neutropenia and immunodeficiency. For the whole group of febrile neutropenic patients, the mean time until normalization of fever was 8 days in cases in which pneumonia was detected by HRCT and 7 days in cases of normal HRCT findings (P = not significant [NS]). For the subgroup of BMT and PSCT recipients, the interval was 14 days in cases in which pneumonia was detected by HRCT and 4 days in cases of normal HRCT findings (P < .05) (Fig 3).
To determine the risk of BAL itself, we evaluated the time until normalization of fever for studies associated with (n = 53) and those not associated with BAL (n = 135), independent of HRCT finding. The mean time was 7 days if BAL was performed (range, 0 to 30 days) and 6 days if BAL was not performed (range, 0 to 56 days) (P = NS).
Patterns and Follow-Up
Patterns and Micro-Organisms
In our series of neutropenic patients with fever of unknown origin and normal chest roentgenograms, 60% of HRCT studies were indicative of pneumonia. In 54% of these studies, either clinical (chest roentgenographic) or microbiologic documentation of pulmonary infiltration was possible. Irrespective of the reason for the inflammatory pulmonary disease that was suspected to be the cause of fever (eg, infection, toxicity, GVHD), micro-organisms in specimens as well as pneumonia on chest roentgenograms were accepted as documentation of pneumonia seen on HRCT scans. More than two thirds of these documentations were obtained within the first 5 days after HRCT. Given the parallel course after day 7 to day 10 (Fig 2), it can be stated that initial HRCT had no causal relation to the late cases of pneumonia. Thus, a time gain of 5 days using HRCT in comparison with the exclusive use of chest roentgenography is obvious. This value of chest roentgenography and HRCT in the early detection of pneumonia has been previously described but with smaller and less clearly defined populations.7,15 In our study, HRCT was used as a screening method for early detection of pneumonia. Therefore, negative predictive value and sensitivity are the most important values to consider. Given HRCT's high sensitivity (87%) and high negative predictive value (88%), this technique is well suited for early detection of pneumonia. The reason the negative predictive value was not 100% is that in some cases, HRCT studies were normal but documentation of pneumonia was obtained during follow-up. In clinical practice, these documentations were obtained more than 5 days after HRCT. This fact lessens the importance of the reduced negative predictive value. Barloon et al13 reported that clinical management was changed in their study's subjects, BMT and PSCT recipients, because of HRCT findings in nine (25%) of 36 episodes. His group performed CT in cases of nonspecific findings on chest roentgenography or by special orders from the clinical staff. The routine use of HRCT after normal roentgenographic findings confirmed inflammatory pulmonary disease in approximately one third of cases. Given HRCT's high sensitivity (88%) and high negative predictive value (97%) in our study's transplant recipients, this technique is well suited for excluding pulmonary inflammatory disease under these clinical conditions. The possibility of escalation, de-escalation, or delay of empiric antibiotic therapy depending on HRCT findings will be evaluated in further studies. In our study's whole population of febrile neutropenic patients as well as in the subgroup of BMT and PSCT recipients, specificity and positive predictive values are moderate in HRCT. These values are of less interest for a screening method (Table 3). Because of the high frequency of false-positive results with HRCT, we also considered the possibility of nonspecific findings. Suspicious HRCT findings might not be related to the cause of fever. To control for a relation with time, we calculated the duration of persisting fever for patients with and those without suggestive findings on HRCT. A slight but nonsignificant difference was found for the whole population (Fig 3). The time of fever after HRCT may have been too short for a significant difference. Even after exclusion of patients 1 day before hematologic reconstitution during the febrile episode, the population was heterogeneous because of different chemotherapy regimens for different malignant diseases. In the transplant subgroup, these differences were smaller, and a significant difference became evident with regard to time until normalization of temperature: fever after HRCT lasted longer if HRCT showed pneumonia (P < .02) (Fig 3). Thus, in transplant recipients, at least a time-based relation of HRCT findings and fever was evident. Therefore, the HRCT findings are connected to the cause of fever. The frequent documentation of inflammatory pulmonary disease until day 5 after pneumonia had been diagnosed by HRCT has high clinical impact (positive predictive value, 54%). In cases involving suspicious findings on HRCT scans, we were interested in characterizing the subgroups of patients who developed pneumonia and those who did not. We previously calculated a slightly longer duration of neutropenia for patients with pneumonia (13 days) compared with those without pneumonia on chest roentgenograms during follow-up (10 days) (P = NS).7 In the present study, a lung-derived micro-organism (or Aspergillus species from sputum) obtained during follow-up was accepted as documentation of pneumonia, in addition to evidence of pneumonia on chest roentgenograms. These criteria correspond to the criteria for pneumonia outlined by the German Society of Pneumology.25 The duration of fever and neutropenia before and after HRCT was calculated separately for patients whose HRCTs demonstrated pneumonia and those whose HRCTs did not. No relevant difference was calculated for these groups of patients. Given the higher rate of BAL in patients with suspicious HRCT scans compared with patients with normal HRCT scans, the number of false-negative results may have been underestimated. We found that patients who developed pneumonia after normal HRCT scans were significantly younger than the patients without verification of pneumonia after HRCT-demonstrated infiltration. Comparable to our previous study,7 we also calculated characterizing data of patients with normal HRCT scans concerning their further outcomes. The level of significance increased from P < .025 to P < .005 after classifying evidence of micro-organisms as clinical end points in the present study. This means that younger patients are at a higher risk for developing pneumonia even after normal HRCT scans. Younger patients are frequently treated with more cycles or higher doses of chemotherapy, which may play a role in the faster development of infectious episodes. By using this highly sensitive technique, minimal findings might be irrelevant for the detection of pneumonia. The minimal relevant finding suggestive of pneumonia had to be identified. Therefore, exact evaluation of different patterns detected with HRCT was performed and their statistical implications were calculated (Tables 4 and 5). The minimal specificity was calculated for ground-glass opacification, which is known to be a pattern of low specificity without limitations to a minimal extent.21,23,26 If only a small area of ground-glass opacification exists, a minimal degree of pneumonia without or with low clinical significance may be responsible for this finding. Disturbances of ventilation, temporary overhydration, aspiration, GVHD, or disturbances of perfusion are other possible causes of this pattern. However, the large decrease in sensitivity for ground-glass opacifications of a larger extent within a lobe suggests that even small areas of ground-glass opacifications are important and may have clinical impact. The most useful set of statistical data were obtained for ill-defined nodules. This finding is frequent (evident in 53 studies) and is associated with a well-suited combination of sensitivity (57%), specificity (86%), positive predictive value (66%), and negative predictive value (81%). Better values were calculated for rare patterns (eg, the positive predictive value for cavitations, which were evident in only nine studies, was 78%). However, the combination of statistical data for these rare findings is less useful. We also calculated the frequency of detected HRCT patterns for different groups of micro-organisms. A radiologic finding suggestive of an underlying germ might be useful for instituting the ordering of a microbiologic test. The frequency of underlying micro-organisms is known in this population.2 In general, all groups of micro-organisms were associated with all patterns; only their frequencies were slightly different. The coexistence of ill-defined nodules and fungal pneumonia is known. However, when micro-organisms grow within encapsulated nodules, verification of these organisms is complicated.6,9,11,12,18 Thus, some cases in which pneumonia is not identified may be at least partially caused by fungi that are known to grow within ill-defined nodules and cavitations. Determining pattern distribution, which was not investigated in this study, is also helpful for identifying underlying micro-organisms. Perihilar distribution of ground-glass opacities occurs in P carinii pneumonia.19 This organism is rare in febrile neutropenic patients. In general, assignment of a particular pattern, pattern distribution, or combination of patterns to an underlying micro-organism is difficult or even impossible in early pneumonia. In this phase, only mild, nonspecific findings may be detectable. Most radiologic studies of pneumonia are performed in later phases of the disease. We conclude that radiologic studies can be useful for suggesting pneumonia caused by fungi or P carinii but radiologic determination of underlying micro-organisms is not possible in this early phase. In conclusion, duration of fever was longer in patients with pneumonia on HRCT scans compared with those with normal HRCT scans (P < .02 for transplant recipients). Thus, HRCT findings are associated with the cause of fever. A larger extent of patterns does not increase specificity or positive predictive value. Even minor HRCT findings have a clinical impact and should lead to further diagnostic procedures. In our study, more than 50% of patients with normal chest roentgenograms and pneumonia on HRCT scans had documentation of inflammatory lung disease during follow-up. Most documentations were achieved during the first 5 days after HRCT. When chest roentgenograms were normal, approximately 60% of HRCT scans showed infiltration, and half of these findings were verified during follow-up. Thus, a normal chest roentgenogram does not exclude pneumonia. Patients with normal HRCT scans are at minimal risk for inflammatory lung disease for the following 5 to 10 days. Escalation, de-escalation, or delay of empiric antibiotic therapy may be linked to HRCT findings. Finally, all patients with normal chest roentgenograms should routinely undergo HRCT.
This article contains major parts of the doctoral thesis of M.B.
1. Bodey GP: Empirical antibiotic therapy for fever in neutropenic patients. Clin Infect Dis 17:378S-384S, 1993 (suppl 2) 2. Maschmeyer G, Link H, Hiddemann W, et al: Pulmonary infiltrations in febrile neutropenic patients: Risk factors and outcome under empirical antimicrobial therapy in a randomized multicenter trial. Cancer 73:2296-2304, 1994[Medline]
3.
Mori M, Galvin JR, Barloon TJ, et al: Fungal pulmonary infection after bone marrow transplantation: Evaluation with radiography and CT. Radiology 178:721-726, 1991 4. Chanock S: Evolving risk factors for infectious complications of cancer. Hematol Oncol Clin North Am 7:771-793, 1993[Medline] 5. Fisher BD, Armstrong D, Yu B, et al: Invasive aspergillosis: Progress in early diagnosis and treatment. Am J Med 71:571-577, 1981[Medline] 6. Gartenschläger M, Braunschweig R, Ehninger G, et al: CT and HR-CT in the diagnosis of pulmonary complications of bone marrow transplantation [in German]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 160:303-311, 1994[Medline]
7.
Heussel CP, Kauczor H-U, Heussel G, et al: Early detection of pneumonia in febrile neutropenic patients: Use of thin-section CT. AJR Am J Roentgenol 169:1347-1353, 1997 8. Aisner J, Schimpff SC, Wiernek PH: Treatment of invasive aspergillosis: Relation of early diagnosis and treatment to response. Ann Intern Med 99:539-543, 1983
9.
Blum U, Windfuhr M, Buitrago-Tellez C, et al: Invasive pulmonary aspergillosis: MRI, CT, and plain radiographic findings and their contribution for early diagnosis. Chest 106:1156-1161, 1994
10.
Graham NJ, Müller NL, Miller RR, et al: Intrathoracic complications following allogeneic bone marrow transplantation: CT findings. Radiology 181:153-156, 1991
11.
Kuhlman JE, Fishman EK, Siegelman SS: Invasive pulmonary aspergillosis in acute leukemia: Characteristic findings on CT, the halo sign, and the role of CT in early diagnosis. Radiology 157:611-614, 1985 12. Leutner C, Strunk H, Mueller-Miny H, et al: The typical radiological findings and course of invasive aspergillosis in the immunosuppressed patient [in German]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 167:24-31, 1997[Medline]
13.
Barloon JT, Galvin JR, Mori M, et al: High-resolution ultrafast chest CT in the management of febrile bone marrow transplant patients with normal or nonspecific chest roentgenograms. Chest 99:928-933, 1991 14. Maschmeyer G, Beinert T, Buchheidt D, et al: Diagnostic and therapeutic management of febrile neutropenic patients with lung infiltrates: Standard recommendations of the infectious diseases working party of the German Society of Hematology and Oncology. Dtsch Med Wochenschr (in press) 15. Heussel CP, Kauczor H-U, Matzke G, et al: High resolution computed tomography of the lung in neutropenic patients with fever [in German]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 164:368-375, 1996[Medline] 16. Gerson SL, Talbot GH, Hurwitz S, et al: Prolonged granulocytopenia: The major risk factor for invasive pulmonary aspergillosis in patients with acute leukemia. Ann Intern Med 100:345-351, 1984 17. Cordonnier C, Escudier E, Verra F, et al: Bronchoalveolar lavage during neutropenic episodes: Diagnostic yield and cellular pattern. Eur Respir J 7:114-120, 1994[Abstract]
18.
Brown MJ, Miller RR, Müller NL: Acute lung disease in the immunocompromised host: CT and pathologic examination findings. Radiology 190:247-254, 1994
19.
Chaffey MH, Klein JS, Gamsu G, et al: Radiographic distribution of Pneumocystis carinii pneumonia in patients with AIDS treated with prophylactic inhaled pentamidine. Radiology 175:715-719, 1990 20. Branch RA: Prevention of amphotericin B-induced renal impairment. Arch Intern Med 148:2389-2394, 1988[Abstract] 21. Aberle DR: HRCT in acute diffuse lung disease. J Thorac Imaging 8:200-212, 1993[Medline]
22.
Remy-Jardin M, Remy J, Deffontaines C, et al: Assessment of diffuse infiltrative lung disease: Comparison of conventional CT and high-resolution CT. Radiology 181:157-162, 1991 23. Remy-Jardin M, Remy J, Giraud F, et al: Computed tomography assessment of ground-glass opacity: Semiology and significance. J Thorac Imaging 8:249-264, 1993[Medline]
24.
Herold CJ, Kramer J, Sertl K, et al: Invasive pulmonary aspergillosis: Evaluation with MR imaging. Radiology 173:717-721, 1989 25. Schaberg T, Dalhoff K, Lorenz J, et al: German Society of Pneumology: Recommendations for diagnosis of community-acquired pneumonia. "Diagnosis of Community-Acquired Pneumonia" Working Group [in German]. Pneumologie 51:69-77, 1997[Medline]
26.
Remy-Jardin M, Giraud F, Remy J, et al: Importance of ground-glass attenuation in chronic diffuse infiltrative lung disease: Pathologic-CT correlation. Radiology 189:693-698, 1993 Submitted June 10, 1998; accepted November 9, 1998. This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
