Clinical Radiology (1998) 53, 255-257

Invasive Aspergillosis in the Immunocompromised Host: Utility of Computed Tomography and Bronchoalveolar Lavage

M. J. BROWN, S. A. WORTHY, J. D. A. FLINT* and N. L. MLILLER

Departments of Radiology and *Pathology, University of British Columbia and Vancouver Hospital and Health Sciences Centre, Vancouver, BC, Canada

Objective: Bronchoalveolar lavage is performed almost routinely in immunocompromised patients with suspected pneumonia, but it has a low yield in the diagnosis of pulmonary aspergillosis. The aim of this study was to determine whether computed tomography (CT) is helpful in determining the likelihood of a positive bronchoalveolar lavage by allowing distinction of patients with angioinvasive aspergillosis from those with Aspergillus bronchopneumonia.

Methods and Results: A retrospective study was performed including consecutive immunocompromised patients with suspected pneumonia who underwent CT scanning of the chest and bronchoalveolar lavage and who had definite diagnosis of pulmonary aspergillosis. The CT scans were reviewed by two chest radiologists and classified as showing features consistent with angioinvasive or airway invasive aspergiilosis. Twenty- one patients met the inclusion criteria. Bronchoaiveolar lavage was positive for fungi in two of 11 patients with CT findings consistent with angioinvasive aspergillosis and eight of 10 patients with CT scans consistent with Aspergillus bronchopneumonia (P<0.01, chi-squared test). CT findings of angioinvasive aspergillosis included nodules measuring 1-3.5 cm in diameter in six, segmental consolidation in three, and both nodules and segmental consolidation in two patients. CT findings ofAspergillus bronchopneumonia including peribronchial consolidation in five, small centrilobular micronodules in one, and both in four patients.

Conclusions: Chest CT is helpful in determining the likelihood of successful diagnosis of pulmonary aspergillosis by bronchoalveolar iavage. Brown, M. J., Worthy, S. A., Flint, J. D. A. & Mtiller, N. L. (1998). Clinical Radiology 53, 255-257. Invasive Aspergillosis in the Irmnunocompromised Host: Utility of Computed Tomography and Bronchoalveolar Lavage

Accepted for Publication 18 July 1997

Invasive pulmonary aspergillosis is an important cause of morbidity and mortality in the immunocompromised patient. Survival is dependent on early diagnosis and prompt institution of therapy. Bronchoalveolar lavage (BAL) is performed almost routinely in immunocompro- mised patients with suspected pneumonia. However, the majority of studies have reported a lower than 50% sensi- tivity of BAL in diagnosing pulmonary aspergillosis in these patients [1-4].

The two main pulmonary complications of aspergillosis in immunocompromised patients are angioinvasion result- ing in haemorrhagic infarcts and airway invasion resulting in bronchiolitis and bronchopneumonia [5-7]. It has been shown that angioinvasive and airway invasive aspergillosis result in different computed tomography (CT) findings [5-7]. We hypothesized that patients with Aspergillus bronchopneumonia would be more likely to yield a positive BAL than patients with angioinvasive aspergillosis.

The aim of this study was to determine whether CT findings suggestive of Aspergillus bronchiolitis or broncho- pneumonia were more likely to yield a positive BAL diagnosis than patients with CT findings suggestive of angioinvasive aspergillosis.

Correspondence to: Dr N. L. MOiler, Department of Radiology, Van- couver Hospital and Health Sciences Centre, 855 W. 12th Ave, Vancouver, BC, Canada V5Z 1M9.

�9 1998 The Royal College of Radiologists.

MATERIALS AND METHODS

A retrospective review of the records of the Departments of Pathology and Radiology, University of British Columbia and Vancouver Hospital and Health Sciences Center, between September 1988 and December 1995 yielded 21 immunocompromised patients with suspected pulmonary infection who had undergone both BAL and CT, and who had a definite diagnosis of pulmonary aspergillosis. The patients included 15 men and six women, and had a mean age of 41 years (range 14-68). The median interval between CT scan and diagnosis of pulmonary aspergillosis was 4 days (range 1-16). The median time interval between the CT and BAL was 2 days (range 0-5) and the median time interval between BAL and definitive diagnosis other than by BAL was 3 days (range 1-12 days) The patients were immunocompromised on the basis of bone marrow trans- plant for haematological malignancy in 12, chemotherapy for haematological malignancy in six, solid organ malig- nancy in one, lung transplant in one, and corticosteroid treatment for systemic lupus erythematosus in one patient.

CT scans were performed using 1-1.5-mm collimation sections reconstructed with a high-spatial frequency algo- rithm (high-resolution CT) at 10-mm intervals through the chest in 20 patients and using contiguous 10-mm collima- tion sections in one patient. Scans were viewed at standard lung windows (window level - 700 Hounsfield units (HU),

256 CLINICAL RADIOLOGY

width 1000-1500 HU). The CT scans were assessed by two radiologists (MJB, SAW) who were not aware of the BAL or pathological findings. CT features considered consistent with angioinvasive aspergillosis included nodules measur- ing 1 cm or more in diameter, nodules with halo of ground- glass attenuation, nodules with air crescent sign, or areas of segmental consolidation [5,6,8,9] (Fig. 1). Findings considered consistent with Aspergillus bronchiolitis or bronchopneumonia included small centrilobular nodules, peribronchial consolidation, or lobar consolidation [7] (Fig. 2). The two observers assessed the findings indepen- dently. Degree of agreement between the observers was assessed using Kappa statistics. Following initial assess- ment, the observers reviewed the findings and reached a final decision by consensus.

In 10 patients, the diagnosis was made based on the identification of fungal hyphae or a positive culture from BAL specimens. In patients in whom the BAL was negative for fungi, the diagnosis of invasive pulmonary aspergillosis was made by open-lung or video-assisted thoracoscopic biopsy in five, autopsy in two, transbronchial biopsy in two, and fine needle aspiration biopsy in two. Comparison between the likelihood of positive broncho- alveolar lavage in patients with CT findings consistent with angioinvasive aspergillosis and findings consistent with bronchopneumonia was made using the chi-squared test with Yate's correction.

RESULTS

Twenty-one patients met the inclusion criteria for the

Fig. 2 - - Aspergillus bronchopneumonia. High-resolution CT scan shows focal areas of peribronchial consolidation (arrow). The patient was a 36- year-old man who had recently undergone bone marrow transplant.

study. Of these, 11 demonstrated CT findings consistent with angioinvasive aspergillosis, and 10 demonstrated fea- tures of airway invasive aspergillosis. BAL results were positive for fungal hyphae in two of 11 patients with CT features of angioinvasive aspergillosis and in eight of 10 with airway invasive aspergillosis (P<0.01, chi-squared test).

CT findings considered consistent with angioinvasive aspergillosis included multiple nodules ( n= 2), nodules with air crescent (n = 4), segmental consolidation (n = 3), or a combination of nodules and segmental consolidation (n = 2). The nodules in patients with angioinvasive asper- gillosis ranged from 1 to 3.5 cm in diameter. CT findings considered consistent with airway invasive aspergillosis included peribronchial consolidation (n = 5), peribronchial consolidation plus centrilobular nodules (n = 4), and centri- lobular nodules (n = 1). The centrilobular nodules measured 2-5 mm in diameter.

The BAL specimen was negative for fungal hyphae in 11 cases. Open-lung or video-assisted thoracoscopic biopsy confirmed the diagnosis in five, demonstrating a vaso- invasive fungal infarct (n=2) , Aspergillus broncho- pneumonia with tissue invasive deep to the basement membrane (n = 1), a fungal abscess (n = 1), and broncho- centric granulomatosis due to Aspergillus (n = 1). The two cases diagnosed at autopsy demonstrated a fungal infarct. In two cases the diagnosis was made by transbronchial biopsy, and in two by fine needle aspiration biopsy.

There was good interobserver agreement in the interpret- ation of the CT findings and in the distinction of airway invasive from angioinvasive aspergillosis on CT (Kappa statistics = 0.62).

Fig. 1 - - Angioinvasive pulmonary aspergillosis. High-resolution CT scan demonstrates a nodule with characteristic halo of ground-glass attenuation. The patient was a 50-year-old woman undergoing chemotherapy for acute myelogenous leukaemia.

DISCUSSION

Invasive aspergillosis is a relatively common pulmonary complication in immunocompromised patients. Early diag- nosis and prompt institution of therapy is essential in determining survival. However, the yield of non-invasive diagnostic tests is limited. Although fibreoptic broncho- scopy may provide an alternative to open lung biopsy, in

�9 1998 The Royal College of Radiologists, Clinical Radiology, 53, 255-257.

INVASIVE ASPERGILLOSIS IN THE IMMUNOCOMPROMISED HOST 257

some patients coagulopathy precludes the safe use of transbronchial biopsy. BAL is usually safe and simple to use and has a high specificity but a relatively low sensitivity in the diagnosis of pulmonary aspergillosis in immuno- compromised patients [1-4]. In one study, Aspergillus hyphae were identified in nine of 17 BAL samples from patients with invasive pulmonary aspergillosis (sensitivity 53%) and from three of the remaining 65 study patients without this diagnosis (specificity 97%) [1]. The small percentage of false-positive diagnosis on BAL may be further decreased by the assessment of fungal antigens [10]. Because of its high specificity identification of Asper- gillus hyphae in BAL fluid in immunocompromised patients has been considered diagnostic of Aspergillus infection [11-13]. Saito et aL [3] assessed the utility of BAL in the diagnosis of pulmonary infiltrates in 22 patients with acute leukaemia. BAL had a diagnostic yield of only 15% (three of 20) specific diseases; all three were Candida pneumonia. While the sensitivity of BAL was 75% and its specificity 100% for Candida pneumonia, BAL did not result in a specific diagnosis for the remaining 17 diseases, nine of which were pulmonary aspergillosis [3]. Levy et al. assessed the value of bronchoalveolar lavage and bronchial washings in the diagnosis of invasive pulmonary asper- gillosis in 300 consecutive patients [4]. In this study a diagnosis of invasive pulmonary aspergillosis was made in a total of 21 patients who underwent 25 bronchoscopies; 14 of the 21 patients (67%) with invasive aspergillosis were detected by either BAL cytology or culture.

The wide range of sensitivity of BAL in the diagnosis of invasive aspergillosis, ranging from 0% [3] up to 67% [4] may be due to the timing of the BAL [4] or be due to different types of invasive aspergillosis. Our study demon- strates that BAL is positive in 80% of patients with CT findings consistent with airway invasive aspergillosis com- pared to only 18% of patients with CT findings consistent with angioinvasive aspergillosis. Airway invasive asper- gillosis is characterized by the presence of bronchitis, bronchiolitis, or bronchopneumonia [14] and is therefore likely to yield a positive BAL. Angioinvasive aspergillosis is characterized by the presence of haemorrhagic infarcts and large nodules, and therefore less likely to yield a positive BAL. The CT findings of both forms of pulmonary aspergillosis have been previously shown to correlate with those seen in pathology specimens [5,9]. The CT findings of angioinvasive aspergillosis consist of nodules, frequently with a halo of ground-glass attenuation, mass-like infil- trates, and areas of segmental consolidation due to haemor- rhagic infarcts [5,6,9]. The CT findings of airway invasive aspergillosis consist of small centrilobular nodules which reflect the presence of bronchiolitis and peribronchial con- solidation due to bronchopnenmonia [7]. It should be noted that the CT findings of airway invasive aspergillosis are non-specific, being similar to those seen in patients with bacterial, mycoplasma or viral bronchiolitis, or

bronchopneumonia [7]. The CT findings of angioinvasive aspergillosis, however, are more specific. It has been sug- gested that in the appropriate clinical setting the presence of nodules with a halo of ground-glass attenuation allow a confident diagnosis of angioinvasive aspergillosis [5].

In summary, our study demonstrates that CT findings may be helpful in determining the likelihood of a successful diagnosis of pulmonary aspergillosis by BAL. In the appropriate clinical context, namely a patient with severe neutropenia, the presence of nodules with a halo of ground- glass attenuation on CT are highly suggestive of angio- invasive aspergillosis, a condition which is unlikely to be confirmed by BAL. In patients with Aspergillus bronchio- litis or bronchopneumonia, similar to other causes of bronchopneumonia, the diagnosis can usually be made by BAL.

REFERENCES

1 Kahn FW, Jones JM, England DM. The role of bronchoalveolar lavage in the diagnosis of invasive pulmonary aspergillosis. American Journal of Clinical Pathology 1986;86:518-523.

2 Sternberg RI, Baughman RP, Dohn MN, First MR. Utility of broncho- alveolar lavage in assessing pneumonia in immunosuppressed renal transplant recipients. American Journal of Medicine 1993;95:358-364.

3 Saito H, Anaissie EL Mofice RC, Dekmezian R, Bodey GP. Broncho- alveolar lavage in the diagnosis of pulmonary infiltrates in patients with acute leukemia. Chest 1988;94:745-749.

4 Levy H, Horak DA, Tegtmeier BR, Yokota SB, Forman SJ. The value of bronchoalveolar lavage and bronchial washings in the diagnosis of invasive pulmonary aspergillosis. Respiratory Medicine 1992;86: 243-248.

5 Kuhlman JE, Fishman EK, Birch PA, Karp JE, Zerhouni EA, Siegelman SS. CT of invasive pulmonary aspergillosis. American Journal of Roent- genology 1988;150:1015-1020.

6 Gefter WB. The spectrum of pulmonary aspergillus. Journal of Thoracic Imaging 1992;7:56-74.

7 Logan PM, Primack SL, Miller RR, Mtiller NL. Invasive aspergillosis of the airways: radiographic, CT and pathologic findings. Radiology 1994;193:383-388.

8 Curtis AMB, Smith GJW, Ravin CE. Air crescent sign of invasive pulmonary aspergillosis. Radiology 1979;133:17-21.

9 Hruban RH, Meziane MA, Zerbouni EA, Wheeler PS, Dnmler JS, Hutchins GM. Radiologic-pathologic correlation of the halo sign in invasive pulmonary aspergillosis. Journal of Computer Assisted Tomo- graphy 1987;11:534-536.

10 Andrews CP, Weiner MH. Aspergillus antigen detection in broncho- alveolar lavage fluid from patients with invasive aspergillosis and aspergillomas. American Journal of Medicine 1983;73:372-380.

11 Moil M, Galvin JR, Barloon TJ, Gingfich RD, Stanford W. Fungal pulmonary infections after bone marrow transplantation: evaluation with radiography and CT. Radiology 1991;178:721-726.

12 Blum U, Windfuhr M, Buitrago-Tellez C, Sigmund G, Herbst EW, Langer M. Invasive pulmonary aspergillosis: MRI, CT, and plain radiographic findings and their contribution for early diagnosis. Chest 1994; 106:1156-1161.

13 Miller WT Jr, Said GJ, Frank I, Gefter WB, Aronchick JM, Miller WT. Pulmonary aspergillosis in patients with AIDS. Clinical and radio- graphic correlations. Chest 1994;105:37-44.

14 Okudaira M, Schwarz J. Tracheobronchopulmonary mycoses caused by opportunistic fungi, with particular reference to aspergillosis. Laboratory and Investigations 1962;11:1053-1064.

�9 1998 The Royal College of Radiologists, Clinical Radiology, 53, 255-257.