University College London Medical School and Camden and Islington Community Health Services (NHS) Trust, and 3Department of Imaging, University College ...
1996, The British Journal of Radiology, 69, 1104-1107
CT appearances of intrathoracic Kaposi's sarcoma in patients with AIDS 1
Z C TRAILL, MRCP, FRCR, 2R F MILLER, FRCP and 3 P J SHAW, MRCP, FRCR
department of Radiology, John Radcliffe Hospital, Oxford, 2Division of Pathology and Infectious Diseases, University College London Medical School and Camden and Islington Community Health Services (NHS) Trust, and 3Department of Imaging, University College London Hospitals (NHS) Trust, London, UK Abstract Although there have been many studies of the plain radiographic appearances of intrathoracic Kaposi's sarcoma in patients with the acquired immunodeficiency syndrome, the computed tomography (CT) findings are less well established. We performed a retrospective review of the thoracic CT findings of 15 patients with tracheobronchial Kaposi's sarcoma diagnosed at bronchoscopy in whom concurrent respiratory infection had been excluded. The commonest CT finding was the presence of ill-defined nodules, seen in all patients. In eight patients more than 20 nodules were seen. Small areas of ground-glass attenuation surrounded one or more nodules in 11 patients. Bilateral perihilar infiltrates were seen in 14 patients, extending into the pulmonary parenchyma along bronchovascular bundles. Interlobular septal thickening was seen in 13 patients and fissural nodularity in 15 patients. To our knowledge this latter finding has not been described before. Discrete areas of ground-glass attenuation were seen in six patients. Small bilateral pleural effusions occurred in six patients; four patients had a pericardial effusion. Shotty mediastinal lymphadenopathy occurred in six patients, and mediastinal nodes greater than 1 cm were present in four patients. None of the patients had CT evidence of chest wall or bone involvement. Although none of these findings are specific, the combination of poorly defined nodules, fissural nodularity and a bronchovascular distribution of perihilar opacities on CT is highly suggestive of pulmonary Kaposi's sarcoma.
Intrathoracic Kaposi's sarcoma is an uncommon but important complication of the acquired immunodeficiency syndrome (AIDS). Although pulmonary Kaposi's sarcoma has been reported in around 10% of patients with AIDS [ 1 ] , this is likely to be an underestimate due to difficulties in antemortem diagnosis. Presenting symptoms are non-specific, concurrent opportunistic respiratory infection is common and tracheobronchial Kaposi's sarcoma may not be seen at bronchoscopy despite the presence of parenchymal tumour [2, 3]. Several previous reports have described the radiographic abnormalities associated with pulmonary Kaposi's sarcoma [1, 3 - 6 ] , but the CT findings are less well established. We undertook a retrospective review of patients with tracheobronchial Kaposi's sarcoma, in whom associated opportunistic infections had been excluded, in order to characterize the CT features of this disease. Patients and methods 17 human immunodeficiency virus-1 (HIV-1) infected patients admitted to the specialist HIV/AIDS inpatient unit at University College London Hospitals between Received 3 April 1996 and in revised form 1 July 1996, accepted 5 August 1996. Address correspondence to Dr Z C Traill, Department of Radiology, John Radcliffe Hospital, Oxford OX3 9DU, UK. 1104
April 1994 and January 1996 had a diagnosis of tracheobronchial Kaposi's sarcoma made at bronchoscopy. All had a thoracic CT scan within 14 days of the bronchoscopy. The clinical records and CT scans of these patients were reviewed retrospectively. Two patients were excluded from the analysis as, although both had pulmonary Kaposi's sarcoma, one had concurrent respiratory infection with Pseudomonas aeruginosa and the other had previously received systemic chemotherapy for extensive cutaneous Kaposi's sarcoma. Of the remaining 15 patients, 14 were men who were all Caucasian and either homosexual or bisexual (one was also an intravenous drug user). The single woman was heterosexual and of African origin. The mean age of the patients was 36 years (range 26-61 years) and they had been HIV-1 antibody positive for between 5 and 84 months (mean 39 months) at the time of diagnosis of tracheobronchial Kaposi's sarcoma. The group was profoundly immunosuppressed with a median CD4 + T lymphocyte count of20x 106 I" 1 (range 10 to 100 x 106 I"1) (normal range 350-2200 x 106 I" 1 ). Five patients had a past history of Pneumocystis carinii pneumonia, none within the preceding 6 weeks. The remaining 10 patients had no past history of respiratory disease. 11 patients were cigarette smokers. 12 patients had concomitant cutaneous Kaposi's sarcoma, 11 of whom also had palatal Kaposi's sarcoma; two other patients had palatal but no cutaneous Kaposi's The British Journal of Radiology, December 1996
CT appearances of intrathoracic Kaposi's sarcoma in AIDS patients
sarcoma. Only one patient had no cutaneous or palatal tumour. One patient with cutaneous and palatal Kaposi's sarcoma also had biopsy confirmed gastric lesions. All fibreoptic bronchoscopies were performed by RFM. At bronchoscopy, following inspection of the tracheobronchial tree, bronchoalveolar lavage was performed using a previously described technique [ 7 ] . Bronchoalveolar lavage fluid was stained and cultured for bacteria, mycobacteria, fungi and viruses, as previously described [7]. In the 10 most recent patients an aliquot of lavage fluid was subjected to polymerase chain reaction (PCR) amplification to detect human herpes virus-8/Kaposi's sarcoma associated herpes virus (HHV8/KSHV) DNA [8, 9]. The diagnosis of tracheobronchial Kaposi's sarcoma was made on the basis of visual identification of characteristic red or violaceous flat or raised plaques in all patients [10]. Tracheobronchial Kaposi's sarcoma was categorized as localized if lesions were only in the segmental bronchi of a single lobe or on the tracheal wall and widespread if there was involvement of the trachea and a single lobe, or if the segmental bronchi of two or more lobes were affected [11]. The diagnosis was confirmed in 10 patients by detection of HHV8/KSHV DNA in bronchoalveolar lavage fluid. In one of these patients, and in two others, histological confirmation of the diagnosis was made at autopsy 4, 8 and 11 weeks after the bronchoscopic diagnosis. The CT scans were performed on a GE Prospeed CT scanner (Slough, Berkshire, UK). High resolution CT (HRCT) scans (2 mm thick sections at l c m intervals using an edge-enhancing algorithm) were obtained in 14 patients and a conventional 10 mm collimation scan in one. Images were viewed at window settings appropriate for lung and mediastinum. The scans were reviewed as a batch by two radiologists (ZCT and PJS) who were aware of the clinical and bronchoscopic diagnosis. Predetermined criteria were used: note was specifically made of the presence or absence of parenchymal abnormalities including nodules, perihilar infiltrates, alveolar opacities, cystic air spaces and interlobular septal thickening; pleural and pericardial disease, mediastinal and axillary lymphadenopathy, and chest wall and bone involvement. Lymph nodes were considered enlarged if greater than 1 cm in short-axis diameter and "shotty" if increased in number but less than 1 cm in diameter.
Figure 1. HRCT scan showing a poorly defined nodule with an irregular, incomplete rim of ground-glass attenuation. Interlobular septal thickening is also seen.
Perihilar infiltrates were present in 14 patients and had a broncho vascular distribution (Figure 2). The infiltrate was predominantly linear in one patient, predominantly nodular in one and was mixed in 12 patients. Fissural nodularity was seen in all 15 patients (Figure 3) and fissural distortion in 14 patients. Interlobular septal thickening was present in 13 patients (Figure 4). Alveolar opacities were not seen. Areas of ground-glass attenuation, separate from the peri-nodular haze described earlier, occurred in six patients (Figure 5). Pleural effusions were present in six patients being
Results
The commonest abnormality on CT was the presence of parenchymal nodules. These were present in all 15 patients and were always poorly defined. Between 10 and 20 nodules were seen in three patients and in excess of 20 nodules in eight patients. In nine patients the predominant nodule size was between 1 cm and 3 cm, and in the other six most nodules were less than 1 cm in size. There was no zonal predilection. Sub-pleural nodules occurred in 13 patients. In 11 patients small areas of ground-glass attenuation surrounded one or more of the nodules, resulting in a peri-nodular "haze" (Figure 1). Cavitation of a single nodule was seen in one patient.
Vol. 69, No. 828
Figure 2. HRCT scan showing bilateral linear and nodular perihilar infiltrates in a peribronchovascular distribution.
1105
Z C Traill, R F Miller and P J Shaw
Figure 3. In addition to poorly defined nodules this HRCT scan shows nodularity and distortion of the right major fissure.
Figure 4. HRCT scan of the right lower lobe showing marked interlobular septal thickening which clearly defines a secondary pulmonary lobule (arrow).
Figure 5. HRCT scan showing areas of ground-glass attenuation separate from parenchymal nodules.
1106
bilateral and small in all. Four patients had a pericardial effusion. Although shotty mediastinal lymphadenopathy was common (six patients), mediastinal nodes greater than 1 cm were present in only four patients. Eight patients had shotty axillary nodes and one patient had bilateral axillary nodes greater than 1 cm. Because the majority of patients had extensive perihilar infiltrates and all but one of the CT scans were performed without intravenous contrast medium, no attempt was made to determine the presence of hilar lymphadenopathy. Cystic air spaces were seen in two patients; one of these, a cigarette smoker, had a past history of Pneumocystis carinii pneumonia as well as CT evidence of emphysema. The second patient had no previous history of respiratory illness but had CT evidence of bronchiectasis. There was no evidence of Kaposi's sarcoma involving chest wall or bone in any patient. The tracheobronchial Kaposi's sarcoma was localized in two patients and widespread in 13 patients. Both of the patients with localized disease at bronchoscopy had extensive parenchymal abnormalities with in excess of 20 poorly defined nodules and perihilar infiltrates in a peribronchovascular distribution. Discussion Respiratory complications are a significant cause of morbidity and mortality in patients with AIDS. Chest radiograph appearances are often non-specific and thoracic CT is increasingly used both to establish disease, and as an aid to specific diagnosis. Previous reports of the CT appearances of intrathoracic Kaposi's sarcoma suggest that these are characteristic with typical findings including perihilar infiltrates, poorly defined nodules, interlobular septal thickening, lymphadenopathy and pleural effusions [12,13]. On the basis of such previously published data, two recent studies found that CT was very accurate (approximately 90%) in the diagnosis of pulmonary Kaposi's sarcoma [14, 15]. In this study we sought to determine the frequency and appearance of peri-nodular ground-glass attenuation in patients with a bronchoscopically confirmed diagnosis of pulmonary Kaposi's sarcoma. This appearance has been briefly commented upon in a previous study of patients with pulmonary Kaposi's sarcoma [14]. We found small areas of ground-glass attenuation forming a rather ill-defined "halo" around one or more nodules in 11 of our patients. This differed from the characteristic halo seen on CT surrounding nodules in invasive aspergillosis in being less regular and often not completely surrounding the nodule [16]. Fissural nodularity was found in all of our patients and fissural distortion in 14 out of 15. To our knowledge this has not previously been described as a CT feature of pulmonary Kaposi's sarcoma and is likely to represent perilymphatic spread of tumour. The presence of poorly defined pulmonary nodules was the commonest finding in our study. Previous studies have reported frequencies of pulmonary nodules of 42% and 85% [12, 14]. Sider et al [13], in a retrospective study of thoracic CT scans in patients with AIDS, found
The British Journal of Radiology, December 1996
CTappearances of intrathoracic Kaposi's sarcoma in AIDS patients
Kaposi's sarcoma to be the commonest cause of single or multiple ill-defined nodules. Other causes included cryptococcal infection, AIDS-related non-Hodgkins lymphoma, Mycobacterium tuberculosis and Mycobacterium avium-intracellulare. Associated findings will often narrow the differential diagnosis. A perihilar infiltrate, usually extending along bronchovascular bundles, is a characteristic finding in patients with pulmonary Kaposi's sarcoma in our study and in others. These appearances were reported by Wolff et al [17] in 13 of 15 patients with AIDS-related intrathoracic Kaposi's sarcoma and in 22 of 24 patients described by Naidich et al [12]. In a retrospective study of 102 patients with thoracic complications of AIDS, Hartman et al [14], found that a predominantly bronchovascular distribution of nodular opacities or consolidation was found almost exclusively in Kaposi's sarcoma. This correlates with the peribronchial and perivascular location of tumour observed at autopsy [5]. Histological examination has also revealed tumour infiltration of interlobular septa [5,12], which would explain the interlobular septal thickening seen in 13 out of 15 of our patients. This finding was observed in only 10 of 26 patients in whom Kaposi's sarcoma was the only thoracic complication of AIDS in the study of Hartman et al [14]. Areas of ground-glass attenuation separate from the localized areas surrounding nodules have not been reported to be a common CT abnormality in AIDS patients with pulmonary Kaposi's sarcoma [14]. However, we found areas of ground-glass attenuation separate from nodules in six of our 15 patients. The presence of areas of ground-glass attenuation in patients with characteristic CT features of pulmonary Kaposi's sarcoma should not necessarily suggest a second pathology such as Pneumocystis carinii pneumonia. Previous CT studies have reported lymphadenopathy in 33-53% of patients with pulmonary Kaposi's sarcoma [12, 14, 17]. We found mediastinal lymphadenopathy in four patients and axillary lymphadenopathy in one, a total of 33%. Mediastinal lymph nodes larger than 1.5 cm occurred in only one patient. We therefore concur with Naidich et al [12], that massive coalescent lymphadenopathy is not a feature of pulmonary Kaposi's sarcoma. Marked lymphadenopathy should suggest an alternative diagnosis such as Mycobacterium tuberculosis or Mycobacterium avium-intracellulare. Pleural effusions, seen in six of our 15 patients, have been reported in 35-60% of cases in previous CT studies [12, 14, 17]. In our patients, pleural involvement was always bilateral, similar to the findings reported by Wolff et al [17]. Whereas Wolff et al [17] found chest wall or bone disease in seven of 15 patients, no patient in our study had chest wall or bone involvement. The combination of poorly defined nodules, fissural nodularity and a bronchovascular distribution of perihilar infiltrates on CT is highly suggestive of pulmonary Kaposi's sarcoma. Pulmonary Kaposi's sarcoma may be present in patients without cutaneous or palatal Kaposi's
Vol. 69, No. 828
sarcoma. Although the CT appearances will suggest the correct diagnosis in a high proportion of patients with pulmonary Kaposi's sarcoma, these patients should undergo fibreoptic bronchoscopy to confirm the diagnosis as well as to exclude co-existing opportunistic infections. Fissural nodularity with distortion and localized areas of ground-glass attenuation around poorly defined nodules are entirely in keeping with the diagnosis. References 1. ZIBRAK, J D, SILVESTRI, R C, COSTELLO, P ET AL, Bronchoscopic and radiologic features of Kaposi's sarcoma involving the respiratory system, Chest, 90, 476-479 (1986). 2. OGNIBENE, F P, and SHELHAMER, J H, Kaposi's sarcoma, Clin. Chest Med., 9, 459-465 (1988). 3. GARAY, S M, BELENKO, M, FAZZINI, E and SCHINELLA, R, Pulmonary manifestations of Kaposi's sarcoma, Chest, 91, 39-43 (1987). 4. DAVIS, S D, HENSCHKE, C I, CHAMIDES, B K and WESTCOTT, J L, Intrathoracic Kaposi's sarcoma in AIDS patients: radiographic pathologic correlation, Radiology, 765,495-500(1987). 5. SIVIT, C J, SCHWARTZ, A M and ROCKOFF, S D, Kaposi's sarcoma of the lung in AIDS: radiologicpathologic analysis, A JR., 148, 25-28 (1987). 6. GRUDEN, J F, HUANG, L, WEBB, W R ET AL, AIDSrelated Kaposi sarcoma of the lung: radiographic findings and staging system with bronchoscopic correlation, Radiology, 195, 545-552 (1995). 7. MILLER, R F, KOCJAN, G, BUCKLAND, J ET AL, Sputum induction for the diagnosis of respiratory disease in HIV positive patients, J. Infect., 23, 5-16 (1991). 8. WHITBY, D, HOWARD, M R, TENANT-FLOWERS, M ET AL, Detection of Kaposi's sarcoma associated herpes virus in peripheral blood of HIV infected individuals and progression to Kaposi's sarcoma, Lancet, 346, 799-802 (1995). 9. HOWARD, M, BRINK, N S, MILLER, R and TEDDER, R S, Association of human herpes virus with pulmonary Kaposi's sarcoma, Lancet, 346, 712 (1995). 10. HUGHES-DAVIES, L, KOCJAN, G, SPITTLE, M F and MILLER, R F, Occult alveolar haemorrhage in bronchopulmonary Kaposi's sarcoma, J. Clin. Pathoi, 45, 536-537 (1992). 11. MILLER, R F, TOMLINSON, M C, COTTRILL, C P ET AL, Bronchopulmonary Kaposi's sarcoma in patients with AIDS, Thorax, 41, 721-725 (1992). 12. NAIDICH, D P, TARRAS, M, GARAY, S ET AL, Kaposi's sarcoma. CT-radiographic correlation, Chest, 96, 723-728 (1989). 13. SIDER, L, GABRIEL, H, CURRY, D R and PHAM, M S, Pattern recognition of the pulmonary manifestations of AIDS on CT scans, Radiographics, 13, 771-784 (1993). 14. HARTMAN, T E, PRIMACK, S L, MULLER, N L and STAPLES, C A, Diagnosis of thoracic complications in AIDS: accuracy of CT, AJR, 162, 547-553 (1994). 15. KANG, E-Y, STAPLES, C A, MCGUINNESS, G ET AL, Detection and differential diagnosis of pulmonary infections and tumors in patients with AIDS: value of chest radiography versus CT, AJR, 166, 15-19 (1996). 16. GEFTER, W B, The spectrum of pulmonary aspergillosis, J. Thorac. Imaging, 7, 56-74 (1992). 17. WOLFF, S D, KUHLMAN, J E and FISHMAN, E K, Thoracic Kaposi's sarcoma in AIDS: CT findings, JCAT, 77,60-62(1993).
1107
