Eur. Radiol. 10, 1318±1322 (2000) Ó Springer-Verlag 2000
European Radiology
Pediatric radiology Original article
Spiral CT of the lung in children with malignant extra-thoracic tumors: distribution of benign vs malignant pulmonary nodules S. Grampp1, A. A. Bankier1, A. Zoubek2, P. Wiesbauer2, B. Schroth2, C. B. Henk1, N. Grois2, G. H. Mostbeck1, 3 1
Universitaetsklinik für Radiodiagnostik, Waehringer Guertel 18±20, A-1090 Vienna, Austria St. Anna Kinderspital, Kinderspitalgasse 6, A-1090 Vienna, Austria 3 Krankenhaus der Barmherzigen Brueder, Abt. Roentgendiagnostik, Esterhazystrasse 26, A-7000 Eisenstadt, Austria 2
Received: 9 November 1999; Revised: 31 January 2000; Accepted: 4 February 2000
Abstract. The purpose of this paper is to clarify the distribution of benign vs malignant pulmonary nodules which are seen on spiral CT in children with malignant extra-thoracic solid tumors. Seventy-four children with known solid, extra-thoracic tumors underwent spiral CT of the chest. According to the initial and follow-up (interval 9.2 4.7 months) findings, the children were graded into four groups: I = normal; II = solitary nodule unchanged at follow-up; III = multiple nodules with one or more than one unchanged at follow-up; and IV = solitary or multiple nodules all changed at follow-up. Nodules without change at follow-up were regarded as benign. Fortynine children did present with normal pulmonary CT exams. In 7 cases solitary pulmonary nodules were found unchanged (group II) at follow-up and in 2 cases (group III) some of the nodules were stationary. Thus, 12 % (9 of 74) presented with at least one pulmonary nodule that did not change at follow-up. Solitary nodules (in groups II and IV) with a diameter < 5 mm were in 70 % (7 of 10) unchanged at follow-up and regarded as benign. In children with known solid extra-thoracic tumors at initial presentation, 70 % of solitary nodules ( < 5 mm) may be benign. To avoid overstaging, smaller solitary nodules must not automatically be regarded as metastases. Key words: Children ± Respiratory system ± CT ± Secondary lung neoplasms
Introduction The diagnosis of pulmonary metastatic disease is one of the clinical problems a radiologist has to face on a routine basis. When malignant disease is present a pulmonary nodule will be regarded by many pediatric radiologists as metastatic disease [1]. However, metastatic disCorrespondence to: S. Grampp
ease can be mimicked by a variety of conditions including granuloma, sarcoidosis, infection, inflammatory pseudotumor, benign tumor, and atelectasis [2, 3]. The superior ability of CT in the detection of pulmonary nodules over chest radiographs and conventional tomography was found in both adults and children [4, 5, 6]. With the advent of spiral applications [7, 8] the sensitivity of pulmonary CT could be further improved. For the planning of further therapeutic interventions the question as to whether a nodule appears to be benign or malignant is crucial. This dilemma is even more pronounced if the nodule is solitary. In pediatric as well as adult tumors accurate staging is essential to determine appropriate treatment [3], since pulmonary metastases are common in numerous solid tumors [9, 10]. It has, however, been addressed in a preliminary report that the scan criteria for the differentiation of benign and metastatic pulmonary nodules in adults, such as CT number, size, and number of lesions, may not be successfully applied in children [3]. There is also the possibility of development of benign pulmonary nodules in children receiving cytotoxic chemotherapy [11]. Therefore, in spiral-CT exams of children with malignant tumors a considerable number of pulmonary nodules may be of benign nature. Incorrect classification of benign pulmonary nodules as malignant may alter the long-term prognosis or affect therapeutic decision making [9]. The aim of this study was to establish the distribution of benign vs malignant pulmonary nodules which are seen on spiral CT in children with malignant tumors, and to investigate if there are differences in size, morphology, or location which can be helpful in establishing a differential diagnosis. Materials and methods Seventy-four children and adolescents (32 females and 42 males) with histologically proven, solid malignant tumors (Table 1) were examined by spiral-CT of the chest.
S. Grampp et al.: Spiral CT of lung in children with malignant extra-thoracic tumors
1319
Image analysis
Table 1. Distribution of solid malignant tumors Solid malignant tumor
No. of patients
Hodgkin's disease Osteosarcoma Ewing's sarcoma Neuroblastoma Rhabdomyosarcoma Wilm's tumor Others
14 13 11 8 7 7 14
Table 2. Number of CT scans performed No. of CT exams
No. of patients
%
1 2 3 4 5 6 7
26 23 13 8 1 2 1
35.1 31.1 17.6 10.7 1.4 2.7 1.4
The mean age was 9.4 years (SD 6.3 years, range 1 month to 25.5 years). A total of 167 CT exams were performed with each patient receiving an average of 2.3 1.3 exams (Table 2). Image acquisition All chest exams were acquired with a Philips SR 7000 scanner (Philips, Best, The Netherlands). The collimation was 10 mm in 58 and 5 mm in 16 of the patients. At the time of the study, no collimation between 5 and 10 mm was available due to the scanner configuration. Other scanning parameters included 120 kVp, 1 s per rotation, 125±200 mA, and a pitch of 1.0±2.0, selected to image from the thoracic inlet through the lung bases in 80 % of the examinations. In 20 % of the examinations the protocol was adjusted for imaging of the chest and abdomen with one spiral-CT acquisition. Whenever possible, the scans were obtained during suspended ventilation in deep inspiration. The decision to use either 5or 10-mm collimation depended on the size of the children. Twenty-one patients were examined without intravenous contrast agent, the remaining 53 patients received non-ionic intravenous contrast (1±2 ml/kg body weight, 300 mg/ml; Omnipaque, Schering, Berlin, Germany). The contrast agent was injected at a rate of 0.2±2.0 ml/s depending on body weight and size. Scanning commenced 25±40 s after the initiation of injection of the contrast agent. The contrast agent was usually administered into an antecubital vein. Images were all reconstructed with a lung-specific kernel, a 250- to 360mm field of view adjusted according to thoracic size. A 360 linear interpolation algorithm and 1- to 2-mm overlapping reconstructions were used in all patients. Images were photographed with a laser imager by using lung-specific window setting (level ±400 HU, width 1500 HU).
Spiral-CT exams were retrospectively interpreted by two blinded radiologists (G. H. M. and A. A. B.), initially by means of individual reading and finally by means of consensus reading. All examinations were graded for the presence of pulmonary nodules. If one or more lesions were present, they were described for their number (1, 2±3, 4±5, > 5). Their size was defined as the largest diameter along the major axis of the largest lesion. In addition, the location of each lesion within the lung was classified as being peripheral (subpleural), or central, and the side and lobe were recorded [12]. In adults, peripheral lung represented all lung within 2 cm of the pleural surfaces (subpleural), including the fissures. In our study this anatomical definition was individually adapted by visual estimation according to the smaller size of the children (0.8- to 1.6-cm rim). Other parameters assessed were the margin of the nodules (sharp/unsharp) and the presence of calcifications, which was assessed on a subjective basis The relationship with reference to the bronchovascular bundle was also defined. According to the study design developed by Pass et al. [6], nodules were classified by the dynamics of their radiological properties. A nodule could increase in size or remain stable; Figs. 1, 2). Since the tendency of rapid growth in pulmonary metastases in children is well known [6, 7, 8, 9], pulmonary nodules without change at follow-up may likely be regarded as benign lesions. In addition, nodules with a decrease in size at follow-up in patients receiving cytotoxic chemotherapy were classified as metastatic lesions if other causes for such a decrease were absent (antimicrobial therapy). The patients were graded into four distinct groups: 1. Group I: Patients without pulmonary nodules at initial or follow-up spiral CT 2. Group II: Patients with a solitary pulmonary nodule, which was unchanged at follow-up 3. Group III: Patients with multiple pulmonary nodules with one or more than one which were unchanged at follow-up 4. Group IV: Patients with solitary/multiple pulmonary nodules that were all changed at follow-up. The means and standard deviations of all demographic and radiological parameters of the patient groups were calculated. Results In 15 of the 167 spiral-CT examinations no consensus between both observers was found after reading of the hard copies. In these cases the examinations were additionally reviewed by both observers in a joined session using on-screen the cine-mode capabilities of the spiral CT. Following this procedure, consensus was found in all cases. As shown in Table 3, the majority of the patients (66 %; n = 49) had no pulmonary nodules at the time of
1320
S. Grampp et al.: Spiral CT of lung in children with malignant extra-thoracic tumors
a
Fig. 1 a,b. Axial spiral CT of the right lung in a 13year-old girl with Hodgkin's disease (grade IIA). Initial scan (a left) shows a solitary centrally located nodule (arrow) with a diameter of 4 mm in the right upper lobe. Follow-up 6 weeks (a middle), 3 months (a right), 6 months (b left), and 9 months (b right) later does not demonstrate change in the nodule
b the initial spiral CT nor on follow-up exams. Of the remaining 25 patients, 12 presented with solitary and 13 with multiple lesions. In 9 patients, who had pulmonary nodules at first presentation (groups II and III) one or more did not show any tendency of change (group II, n = 7; group III, n = 2; 9 of 25 = 36 %). In 16 patients (group IV) all nodules did change on follow-up exams. Five of these 16 patients had solitary lesions on the initial examination, 2 patients two to three nodules, 2 pa-
tients four to five nodules, and 7 patients more than five nodules. When considering the differences in morphological findings between groups II and IV most obvious differences could be found for nodule size with a mean increase in diameter of 3.2 mm for group II and of 19.0 mm for group IV (Table 3). The delineation of lesions, where 86 % of the benign lesions (group II) presented with an unsharp margin compared with 25 % of
Table 3. Demographic patient data and spiral-CT readings. I normal; II solitary nodule unchanged at follow-up; III multiple nodules with one or more than one unchanged at follow-up; IV solitary or multiple nodules all changed at follow-up Group
I
II
III
IV
Number Mean age (months/SD) Mean no. of spiral CT/SD Mean follow-up (months/SD) Mean nodule size (mm/SD) Peripheral localization (n/%) Side (n; left/right) Lobe (n; upper/middle/lower) Margin unsharp (n/%) Calcification (n/%) Related to bronchovascular bundle (n/%)
49 111/77 1.8/1.1 4.2/6.4 ± ± ± ± ± ± ±
7 128/62 3.6/1.6 7.7/4.3 3.2/1.0 6/86 2/5 4/1/2 6/86 0/0 3/43
2 73/74 5.5/2.1 9.0/1.4 4.0/1.1 1/50 2/1 2/1/2 2/100 0/0 0/0
16 146/68 2.7/0.9 9.9/5.2 19/18.3 13/81 14/12 11/11/10 4/25 1/6 5/31
S. Grampp et al.: Spiral CT of lung in children with malignant extra-thoracic tumors
1321
Fig. 2. Axial spiral CT of the right lung in a 5-year-old girl with Hodgkin's disease (grade IIA). Initial scan (left) shows a solitary subpleural nodule (arrow) with a diameter of 3 mm in the right lower lobe. Follow-up 2 months (middle) and 5 months later (right) does not demonstrate change in the nodule
the malignant lesions (group IV), was an usable parameter to differentiate between both groups. In group IV, of the 5 cases with solitary lesions there was one metastasis of an Ewing's sarcoma (diameter 5 mm), one Hodgkin's disease (diameter 4 mm), and one osteosarcoma (diameter 4 mm), all proven by pathologic work-up after surgical resection. Two solitary lesions on initial spiral CT in a patient with an osteosarcoma (diameter 60 mm) and another with Hodgkin's disease stage IV (diameter 50 mm) were not proven histologically. Altogether there were 10 patients with solitary lesions (n = 7, group II; n = 3, group IV) smaller than 5 mm on initial spiral CT. One of these was proven to be an osteosarcoma, and two disappeared under cytotoxic chemotherapy (patients with Hodgkin's disease stage IV and Ewing's sarcoma). However, seven of ten (70 %) were unchanged over a follow-up period of 7.7 4.3 months (1±12 months) receiving an average of 3.6 spiral-CT exams. The location of the lesion within the lung, the pattern of distribution within the lobes, the presence of calcifications, and the relation of the lesions toward the bronchovascular bundle did not appear to have any relationship to the estimation of malignancy (Table 3). Discussion The discovery of pulmonary nodules in children with solid tumors often means that it is assumed that they have metastatic disease [4, 11], which is not always justified. A study by Cohen et al. [10] reported that one third of the pulmonary lesions in children with cancer seen on conventional radiographs were of a benign nature. The follow-up of lung nodules as performed by Pass et al. [6] showed in a review of bone and soft tissue sarcomas that enlarging or newly developing pulmonary nodules are much more likely to be metastases compared with nodules that are stable. Gross et al. [13] have suggested that nodules, that either decrease in size on anti-neoplastic therapy or enlarge parallel with other proved metastases, are likely to be metastases. Nodules that de-
crease in size without anti-neoplastic therapy or remain stable during follow-up of at least 12 months are likely to be benign. In our study, of all children with pulmonary nodules, approximately one third did not show an increase in size on follow-up exams. Our findings agree with the results of Rosenfield et al. [3] who found that the accurate prediction of the nature of a pulmonary nodule in a child with malignancy is difficult. They used a conventional CT technique with 10-mm-thick contiguous slices. We were able to apply more advanced spiral techniques [7, 8] with more complete coverage of the lung volume, improved resolution, and overall image quality [14]. This allowed us to establish morphologic criteria which may support the radiologist in a reasonable prediction of the proposed outcome. As demonstrated in adults, the sharp delineation of a pulmonary lesion is a morphologic criterion, which may point to the diagnosis of malignancy [5, 15]. Another finding in our study was that solitary nodules, at least when they are small, have a reasonable propensity to be benign. This finding complements previous results [13] where multiple pulmonary nodules are more likely to represent metastatic disease in younger patients than in older patients. One limitation of our study is the fact that only a minority of pulmonary nodules were assessed histologically after surgical resection. However, for clinical reasons and due to different therapy protocols, histologic workup was not possible in all cases. Our findings can be summarized in that solitary pulmonary lesions smaller than 5 mm, especially with unsharp margins, have a significant tendency to remain stable (70 %). Lesions over 5 mm, especially when they are multiple and do present with sharp margins, are almost always malignant. These facts have to be kept in mind when reading spiral-CT exams of children with a history of malignant disease to avoid overstaging. Acknowledgement. This work was supported by the ªLudwigBoltzmann Institut für klinische und experimentelle Radiologie,º Vienna, Austria.
1322
S. Grampp et al.: Spiral CT of lung in children with malignant extra-thoracic tumors
References 1. van Kaick G (1994) Zur Differentialdiagnostik der Lungenmetastasen. Radiologe 34: 561 2. Cahan WG, Shah JP, Castro EB (1978) Benign solitary lung lesions in patients with cancer. Ann Surg 187: 241±244 3. Rosenfield NS, Keller MS, Markowitz RI, Touloukian R, Seashore J (1992) CT differentiation of benign and malignant lung nodules in children. J Pediatr Surg 27: 459±461 4. Wiliams JA, Douglass EC, Magill HL, Fitch S, Hustu HO (1988) Significance of pulmonary computed tomography at diagnosis in Wilms' tumor. J Clin Oncol 6: 1144±1146 5. Davis SD (1991) CT evaluation for pulmonary metastases in patients with extrathoracic malignancy. Radiology 180: 1±12 6. Pass HI, Dwyer A, Makuch R, Roth JA (1985) Detection of pulmonary metastases in patients with osteogenic and soft-tissue sarcomas: the superiority of CT scans compared with conventional linear tomograms using dynamic analysis. J Clin Oncol 3: 1261±1265 7. Kalender WA, Polacin A, Süss C (1994) A comparison of conventional and spiral CT: an experimental study on the detection of spherical lesions. J Comput Assist Tomogr 18: 167±176
8. Friese SA, Rieber A, Fleiter T, Brambs HJ, Claussen CD (1994) Pulmonary nodules in spiral volumetric and single slice computed tomography. Eur J Radiol 18: 48±51 9. Cushing B, Slovis T (1992) Imaging of Wilms' tumor: What is important? Urol Radiol 14: 241±251 10. Cohen M, Smith WL, Weetman R, Provisor A (1981) Pulmonary pseudometastases in childen with malignant tumors. Pediatr Radiol 141: 371±374 11. Hidalgo H, Korobkin M, Kinney TR, Falletta J, Heaston DH, Kirks DR (1983) The problem of benign pulmonary nodules in children receiving cytotoxic chemotherapy. AJR 140: 21±24 12. Rusinek H, Naidich DP, McGuiness G et al. (1998) Pulmonary nodule detection: low-dose versus conventional CT. Radiology 209: 243±249 13. Gross BH, Glazer GM, Bookstein FL (1985) Multiple pulmonary nodules detected by computed tomography: diagnostic implications. J Comput Assist Tomogr 9: 880±885 14. Paranjpe DV, Bergin C (1994) Spiral CT of the lungs: optimal technique and resolution compared with conventional CT. AJR 162: 561±567 15. Zerhouni E (1990) CT and MRI of focal lung disease. In: Zerhouni E (ed) CT and MR of the thorax. Churchill Livingstone, New York, pp 102±124