Rapidly growing small peripheral lung cancers ... - BIR Publications

17 downloads 52 Views 497KB Size Report
Abstract. 12 peripheral small lung cancers (,20 mm) of rapid growth (volume doubling time. ,150 days), detected by repeated low dose CT screening, were ...
The British Journal of Radiology, 73 (2000), 930±937

E

2000 The British Institute of Radiology

Rapidly growing small peripheral lung cancers detected by screening CT: correlation between radiological appearance and pathological features 1

J-C WANG, MD, 1S SONE, PhD, MD, 1L FENG, PhD, MD, 1Z-G YANG, MD, 1 S TAKASHIMA, PhD, MD, 1Y MARUYAMA, PhD, MD, 1M HASEGAWA, MD, 1S KAWAKAMI, MD, 2 T HONDA, PhD, MD and 3T YAMANDA, PhD, MD Departments of 1Radiology, 2Laboratory Medicine and 3Surgery, Shinshu University School of Medicine, Asahi, Matsumoto 390-8621, Japan

Abstract. 12 peripheral small lung cancers (,20 mm) of rapid growth (volume doubling time ,150 days), detected by repeated low dose CT screening, were evaluated to examine their CT features and to correlate such features with histopathological ®ndings. Each patient's CT images, including follow-up and thin section CT images, were studied retrospectively to determine tumour growth rate and CT morphological features. Nine of the tumours exhibited a solid tumour growth pattern: seven of these showed a well de®ned, homogeneous, soft tissue density with spicular or lobulated margin. These seven tumours included small cell lung cancer (n53), moderately differentiated adenocarcinoma (n52), poorly differentiated adenocarcinoma (n51) and squamous cell carcinoma (n51). The other two tumours, a moderately differentiated adenocarcinoma and a well differentiated adenocarcinoma, appeared as irregular, soft tissue density nodules with poorly de®ned margins. The latter exhibited an air bronchogram pattern and a small cavity. The remaining three tumours exhibited a lepidic tumour growth pattern. They showed ground glass opacity or ground glass opacity with a higher density central zone on CT images and were well differentiated adenocarcinomas. In conclusion, most peripheral small lung cancers of rapid growth were adenocarcinomas. They also included small cell lung cancer and squamous cell carcinoma. The majority showed solid tumour growth pattern and lacked an air bronchogram and/or small air spaces in the nodule. Some well differentiated adenocarcinomas with lepidic tumour growth pattern also showed rapid growth. The prognosis of lung cancer correlates well with the tumour volume doubling time (VDT) [1, 2] and the latter in turn correlates with histopathological type [2±4]. Furthermore, some studies have indicated that the prognosis of surgically resectable tumour is better than that of non-resectable tumour. The essential issue for a successful surgical excision is detection of early stage lung cancers. Thus, accurate characterization of the CT features of rapidly growing small lung cancers is crucial for early diagnosis and treatment to improve patient prognosis. Low dose spiral CT has recently been introduced to detect lung cancers in a population-based screening study. As many lesions detected in low dose CT images are small and do not qualify for immediate histological evaluation, they are often Received 25 January 2000 and in revised form 28 April 2000, accepted 4 May 2000. Address correspondence to Shusuke Sone, PhD, MD, Department of Radiology, Shinshu University School of Medicine, 3-1-1 Asahi, Matsumoto, Nagano 3908621, Japan. 930

re-examined after an interval. At the later examination, a change in morphological appearances is considered highly signi®cant for making a ®nal diagnosis since most benign and malignant nodules change in size, shape and density in different ways. This enables an evaluation of the natural history of rapidly growing small lung cancers (,20 mm) on CT images. To our knowledge, no studies have previously characterized the natural history of rapidly growing small lung cancers based on CT images, and a correlation between the CT characteristics of such tumours on thin section CT image and histopathological features is not available. Our goal was to determine the growth rate and CT features of rapidly growing small lung cancers and correlate these to histopathological ®ndings.

Materials and methods Between May 1996 and May 1999, annual screening for lung cancer was conducted in Nagano Prefecture of Japan using low dose The British Journal of Radiology, September 2000

CT features of rapidly growing small lung cancers

spiral CT. A total of 81 lung cancers was found in this study; 34 of these lung cancers received initial and annual repeat low dose CT screenings, permitting us to observe the interval change of the lesions. We reviewed the repeat CT of these patients to determine the VDTs, and we evaluated diagnostic CT images to assess the morphological features of rapidly growing peripheral small lung cancers. We de®ned a rapidly growing peripheral small lung cancer as a tumour smaller than 2 cm in maximum diameter, a solitary parenchymal nodule in the lung, with a VDT,150 days. The following inclusion criteria were selected: (1) the nodule was found incidentally on annual rescreening CT, and follow-up examinations were performed by diagnostic CT, including thin section CT; (2) tumours were peripheral, i.e. located within the peripheral two-thirds of the lung on CT images, without contact with lobar or segmental bronchi; (3) the period between initial screening CT and ®nal diagnostic CT was .150 days; (4) the VDT of the tumour was ,150 days; (5) targeted, thin section (high resolution) CT (HRCT) images with collimation of 1 mm taken within 1 month prior to surgery were available for review; and (6) the lesion was subsequently treated by surgery. 12 patients (10 male and 2 female), ranging in age from 56 years to 76 years (mean 70 years) ful®lled the inclusion criteria. Initial screening CT examination was performed using a low dose spiral CT (model CT-W950SR, Hitachi Medical Co., Japan). The technical scan parameters were: 120 kVp, 50 mA (11 cases) or 25 mA (1 case), 10 mm s21 table speed, 2 s per rotation of the Xray tube, 10 mm collimation, 10 mm reconstruction interval with a standard reconstruction algorithm and data processing algorithm of 180 Ê linear interpolation. A state-of-the-art CT scanner (Hi-speed Advantage, GE Medical System, Milwaukee, WI) was used in the diagnostic CT examination. CT images were obtained from the lung apices to the lung bases with 10 mm collimation. Technical parameters were: 120 kVp, 200 mA, 1 s scan time, 10 mm collimation and 320 mm ®eld of view. One additional targeted spiral CT sequence was performed through the nodule with 1 mm collimation in each patient and CT images were reconstructed with a bone algorithm, 20 cm ®eld of view and 0.5 mm reconstruction interval. All CT images were obtained during breath-holding at mid inspiration. The lung tumour VDT was calculated using the Schwartz formula [5]. Initial tumour size was measured on CT images shown on the cathode ray tube monitor, using a window width of 1000 HU and window level of2700 HU. Lesions in ®ve subjects were invisible or ,3 mm in The British Journal of Radiology, September 2000

diameter on the initial screening CT images. Taking partial volume effect into account, we de®ned the size of each of these ®ve tumours to be 3 mm in diameter. Because intervals between patients' repeated low dose CT screenings and ®nal thin section CT imaging varied, and the longest interval between initial and ®nal CT images will yield a more accurate VDT, ®nal tumour size was measured on ®nal thin section CT images. Two observers (J-CW, SS), blinded to the pathological ®ndings, independently reviewed each CT image. They recorded the CT features of each tumour, including density, margin and internal texture. Discrepancies in interpretation between the observers were resolved by consensus. The medical records were examined for history of exposure to well documented pulmonary carcinogens such as cigarette smoke and asbestos. All surgical specimens were ®xed in an in¯ated state by transbronchial infusion of formalin liquid and were sliced transversely at the centre of the tumour to provide optimal correlation with the HRCT image. The gross appearance of the nodule and the microscopic ®ndings on the hematoxylin eosin stained pathological material was examined in each case. The margin and internal texture of each tumour were assessed on thin section CT images and then correlated with pathological ®ndings.

Results Among the ten male patients, eight were heavy smokers (>30 pack-years), one was a mild smoker (,30 pack-years) and the other one was a nonsmoker with a family history of malignancy. Of the two female patients, one was a passive smoker while the other was a non-smoker with a family history of malignancy. The mean size of the 12 lesions on initial screening CT images was 4.7 mm (range 3.0±8.5 mm), and 11.5 mm (range 6.0±14.5 mm) on the ®nal, thin section CT images. The tumour VDT ranged from 54 days to 132 days (Table 1). Among the 12 cases of rapidly growing lung cancer, four of eight adenocarcinomas were well differentiated, three were moderately differentiated and one was a poorly differentiated adenocarcinoma. The remaining four cases included one poorly differentiated squamous cell carcinoma (SCC) and three small cell lung cancers (SCLCs) in the lung periphery. 9 of the 12 cancers were located in the right lung and 3 were in the left lung. The proportion of rapidly growing cancers among the 34 small peripheral cancers that were detected by annual repeat CT was 28% (8/29) of adenocarcinomas, 50% (1/2) of SCCs and 100% (3/3) of SCLCs. 931

J-C Wang, S Sone, L Feng et al Table 1. Data for 12 patients with rapidly growing small lung cancer No.

Age (years) /sex

Smoking history (pack-years)

Diameter on initial CT (mm)

Time to detection (days)

Diameter at diagnosis (mm)

VDT (days)

Location

Cell type

1 2 3 4 5 6 7 8 9 10 11 12

67/M 75/M 68/M 73/M 76/M 70/F 64/M 56/M 69/M 70/F 74/M 71/M

34 25 72 40 40 Passive smoker 30 30 34 0 30 0

3Z3 3Z3 3Z3 3Z3 3Z6 3Z3 5Z4 5Z8 3Z5 8Z5 10Z7 6Z8

366 299 372 357 396 299 363 380 376 361 159 386

15Z14 17Z7 11Z11 13Z8 14Z10 6Z6 13Z9 15Z13 7Z7 17Z9 14Z9 18Z11

54 60 66 72 84 100 120 127 128 130 131 132

RLL RUL LLL RLL LUL RUL RML RUL RUL RLL RLL LUL

SCLC SCLC PD SCC MD Adeno WD Adeno WD Adeno MD Adeno PD Adeno WD Adeno WD Adeno MD Adeno SCLC

VDT, tumour volume doubling time. RLL, right lower lobe; RUL, right upper lobe; LLL, left lower lobe; LUL, left upper lobe; RML, right middle lobe. SCLC, small cell lung carcinoma; PD SCC, poorly differentiated squamous cell carcinoma; MD Adeno, moderately differentiated adenocarcinoma; WD Adeno, well differentiated adenocarcinoma; PD Adeno, poorly differentiated adenocarcinoma.

All cancers were treated by surgical resection. The pathological and clinical stages were T1N0M0 in eight patients, T1N1M0 in two, T1N2M0 in one and T2N1M0 in one patient. The patient with a moderately differentiated adenocarcinoma (T2N1M0) had invasion of the visceral pleura by the tumour and died 8 months after operation owing to pulmonary metastases. The remaining 11 patients were still alive at 15±24 months after surgery. Most of the rapidly growing lung cancers appeared as soft tissue density nodules on CT images (n59) (Figure 1); however, two nodules exhibited ground glass opacities (GGOs) (Figure 2) and one nodule showed GGO with a higher density central zone. Tumours with homogeneous attenuation and well de®ned, smooth margins (n57) were more common than those with heterogeneous attenuation and ill de®ned, irregular margins (n55). Half of the tumours had spiculation and few of them showed a pleural tag (n53) (Figure 3), convergence of pulmonary vessels towards the tumour (n51) or halo sign (n51). Air bronchogram and small air spaces were found in only one tumour (Figure 4; Table 2). Regarding the CT manifestations according to histology, well differentiated adenocarcinomas tended to show GGO (n53), although one tumour showed soft tissue density. GGO in the adenocarcinoma was based on lepidic tumour growth (alveolar lining tumour cell growth) histologically (Figure 2c). Moderately differentiated adenocarcinomas were more likely to 932

show homogeneous soft tissue density with a well de®ned margin (Figure 3), which corresponded well to the histology showing solid tumour growth. Poorly differentiated adenocarcinomas, SCC and parenchymal SCLC appeared as homogeneous, soft tissue density nodules with well de®ned, lobulated margins and ®ne spiculation, which correlated with solid tumour growth on the histological specimen (Figure 1). Small cell lung cancers characteristically showed a well de®ned, distinctly lobulated appearance.

Discussion The VDT of lung nodules has been widely accepted as an index of tumour growth rate. Steele and Buell [6] suggested that a VDT of 30±490 days represents a malignant zone indicative of malignancy, while a VDT outside the above range was referred to as benign zone. The VDTs of lung cancer were noted to have a wide range among the same histological type [1, 4] and, furthermore, several studies have suggested varying VDTs at different stages of lung cancers [1, 7]. The currently available data on VDT of lung cancer were based on chest radiographs [3, 7], which usually do not allow detection of small lung cancer (,10 mm). There is therefore a need to de®ne the growth patterns of smaller lung cancers such as those detected initially on CT scan. There is currently no consensus on the de®nition of rapidly growing lung cancer. According to Usuda et al [1] and Hayabuchi et al [8], who de®ned the cut-off VDTs between rapidly and The British Journal of Radiology, September 2000

CT features of rapidly growing small lung cancers

(a)

(b)

(c)

Figure 1. Case 1: a small cell carcinoma in the right lower lobe in a 67-year-old male patient. (a) Initial screening CT (left) suggests a small nodule (arrow) (,3 mm) with an ill de®ned margin in the right lower lobe. On the re-screening CT image (right), 1 year later, the nodule had increased in size and presented as a soft tissue density nodule with a lobulated margin. This interval change strongly suggested that the nodule needed further thin section CT examination to rule out cancer. (b) Pre-operative thin section CT image shows a homogeneous soft tissue density nodule with a well de®ned lobulated margin (arrows). The tumour size is 14 mm Z 15 mm on the thin section CT. (c) Pathological specimen (Z1.25) shows small cell carcinoma (intermediate cell type) with smooth margin and homogeneous internal texture.

Table 2. Thin section CT ®ndings according to histopathological type in rapidly growing small peripheral lung cancer Thin section CT ®ndings

n Nodule density Soft tissue density GGO Nodule margin Well de®ned Poorly de®ned Smooth Irregular Spiculation Lobulation Pleural tag Convergence of vessels and bronchi Halo sign Internal features Homogeneous Heterogeneous Air bronchogram Small air space Calci®cation

Histological type WD Adeno

MD Adeno

PD Adeno

PD SCC

SCLC

Total

4

3

1

1

3

12

1 3

3 0

1 0

1 0

3 0

9 3

1 3 0 4 1 0 0 0 1

1 2 2 1 1 2 3 1 0

1 0 1 0 1 1 0 0 0

1 0 1 0 1 1 0 0 0

3 0 3 0 2 3 0 0 0

7 5 7 5 6 7 3 1 1

0 4 1 1 0

2 1 0 0 1

1 0 0 0 0

1 0 0 0 0

3 0 0 0 0

7 5 1 1 1

GGO, ground glass opacity; WD Adeno, well differentiated adenocarcinoma; MD Adeno, moderately differentiated adenocarcinoma; PD Adeno, poorly differentiated adenocarcinoma; PD SCC, poorly differentiated squamous cell carcinoma; SCLC, small cell lung carcinoma.

The British Journal of Radiology, September 2000

933

J-C Wang, S Sone, L Feng et al

(a)

(b)

Figure 2. Case 5: a well differentiated adenocarcinoma in the left upper lobe in a 76-year-old male patient. (a) Initial screening CT (left) shows a 3 mm Z 6 mm ground glass opacity nodule (arrow) in the left upper lobe identi®ed retrospectively. The rescreening CT image 1 year later (right) shows that the size of the ground glass opacity has enlarged at the same position (arrow). (b) Pre-operative thin section CT image shows a ground glass opacity nodule with an ill de®ned margin. The nodule is 14 mm Z 10 mm in size. (c) Pathological specimen (Z1.25) shows well differentiated adenocarcinoma with alveolar lining tumour growth and mild thickening of the alveolar septa (type A of Noguchi's classi®cation). (c)

slowly growing lung cancers at 113 days and 150 days, respectively, and taking into account the slower growth rate shown by small lung cancers compared with large ones [1], in this study we de®ned lung cancers with a VDT of less than 150 days as being rapidly growing. Peripheral lung cancers arise from the small airways of the lung. Their growth pattern has traditionally been classi®ed into two types: hilic (solid) and lepidic growth [9]. In the hilic growth pattern, the tumour develops as a solid mass, displacing the surrounding lung. The tumour may be smooth or irregular in margin, round or lobulated in con®guration, and shows mostly homogeneous soft tissue density on CT images [10]. In this study, most (n59) of the rapidly growing lung cancers showed solid tumour growth pattern in pathological specimens; three were SCLCs, three moderately differentiated adenocarcinomas, one well differentiated adenocarcinoma, one poorly differentiated adenocarcinoma and one SCC. Seven of these showed a well de®ned homogeneous soft tissue density. 934

However, the remaining two, one well differentiated and one moderately differentiated adenocarcinoma, showed soft tissue density nodules accompanied by ill de®ned margins, the latter correlated with lepidic growth pattern in a small portion of the tumour periphery on pathological specimens. On the other hand, a lepidic growth pattern is characterized by tumour cell growth replacing normal alveolar lining cells [9]. A heterogeneous hazy density with ill de®ned opacity may be seen on HRCT and is characteristic of well to moderately differentiated adenocarcinoma [11, 12]. This type of lung cancer was regarded as very slow growing [12] and has the most favorable prognosis among adenocarcinomas in general [11]. According to Noguchi et al [11], the 5-year survival rate of 28 cases of this type of cancer was 100%. It should be noted that three localized bronchioloalveolar carcinoma in this study presented as rapidly growing. Our present study indicated that this type of adenocarcinoma might The British Journal of Radiology, September 2000

CT features of rapidly growing small lung cancers

(a)

(b)

(c)

Figure 3. Case 11: a moderately differentiated adenocarcinoma in the right lower lobe in a 74-year-old male patient. (a) Initial screening CT (left) shows a 7 mm Z 10 mm heterogeneously low density nodule (arrow). The re-screening CT image 6 months later (right) demonstrates a soft tissue density nodule with an increasing size and density. Cancer was highly suspected. (b) Pre-operative thin section CT image shows a soft tissue density nodule with irregular shape, ®ne spiculation and a pleural tag (arrow). The tumour size of 9 mm Z 14 mm was measured on thin section CT image. (c) Pathological specimen (Z1.25) shows moderately differentiated adenocarcinoma with pleural reactive thickening and irregular margin. No air bronchogram pattern or small air spaces are found in the tumour.

grow rapidly, with a possibility of unfavorable outcome. Kuriyama et al [13] reported HRCT ®ndings of 20 small lung cancers (,20 mm); 13 (72%) of 18 adenocarcinomas were reported to exhibit an air bronchogram pattern on CT images. They suggested that the air bronchogram pattern in a lung nodule was helpful in discriminating adenocarcinomas from benign lesions. Other groups found that an air bronchogram also appeared on CT images in 28.7±66.7% of lung cancers of different sizes and pathological types [14, 15]. In eight cases of adenocarcinoma in this study, we found only one adenocarcinoma with an air bronchogram on CT images. We also found only one case having small air spaces in rapidly growing lung cancers, although in one study [16] this pattern was reported in 39% of CT images of small lung cancers. Our study demonstrated that most of the rapidly growing small lung cancers lacked an air bronchogram or small air spaces in the tumour on CT images. Thus, lung nodules

that do not exhibit an air bronchogram or small air spaces are likely to be rapidly growing tumours and this indicates the need for early diagnosis and treatment. The CT features of small nodules of SCLCs have not been adequately described. In this study, we encountered three cases of small nodule SCLC and we were able to study their interval changes. They appeared as well de®ned, homogeneous, soft tissue density nodules with smooth margins and a lobulated con®guration on CT images. The VDTs of these tumours were 54, 60 and 132 days, respectively (Table 1). According to Quoix et al [17], less than 5% of cases of SCLC present as solitary pulmonary nodules. This type of SCLC has been regarded as a biologically distinct subset characterized by relatively slow growth and which is potentially curable by surgery [17, 18]. According to our calculation, using the formula of Schwartz [5], the VDTs of three cases of peripheral SCLC reported by Urschel [18] were 180, 210 and 360 days. Our cases had a noticeably

The British Journal of Radiology, September 2000

935

J-C Wang, S Sone, L Feng et al

(a)

(c)

shorter VDTs than those in previous studies, and were similar to those of the central SCLCs [19]. There are several limitations in this study. First, the study included a relatively small sample size. Thus, further studies using a larger number of cases are necessary to de®ne the spectrum of CT and pathological ®ndings of rapidly growing lung cancers. The other limitation was measurement of initial tumour size on low dose spiral CT images rather than on HRCT images. To avoid an overestimation of tumour VDT, we de®ned the minimal size of small lung cancer on screening CT images as 3 mm when they were invisible or smaller than 3 mm in the initial screening CT image. However, we believe that the measurements performed in this study were more accurate than those reported in previous studies in which tumour VDT was measured on conventional chest radiographs. 936

(b)

Figure 4. Case 10: a well differentiated adenocarcinoma in the right lower lobe in a 70-year-old female patient. (a) Initial screening CT scan (left) shows a 5 mm Z 8 mm irregular opacity in the right lower lobe. Annual re-screening CT image 1 year later (right) shows an increase in size of the same nodule, with heterogeneous density. (b) Pre-operative thin section CT shows a lobulated, soft tissue density nodule with an air bronchogram pattern. The margin of the nodule is irregular and partly ill de®ned. The tumour size is measured as 9 mm Z 17 mm. (c) Pathological specimen (Z1.25) shows that the lesion is well differentiated adenocarcinoma with an irregular margin and air bronchogram pattern. Tumour shows replacement of the growth pattern with active ®broblastic proliferation. Some remnant small air spaces are found in the tumour.

In conclusion, in the present study we described the characteristics of rapidly growing small lung cancers as they appear on thin section CT images. Peripheral small lung cancers of rapid growth mostly were adenocarcinomas; they also included SCLCs and SCCs. However, as most peripheral lung cancers were adenocarcinomas in our repeat CT screening study, the proportion of rapidly growing lung cancer among adenocarcinomas is lower than that of SCLC and SCC. CT features of these included a soft tissue density tumour that lacked an air bronchogram pattern or small air spaces in the nodules. We also showed the rare occurrence of rapid growth in cases of well differentiated adenocarcinomas, which are characterized by lepidic tumour growth with GGO on CT images and which have otherwise been reported in the past to show a slow growth. A long-term study of patients treated surgically is needed to correlate the growth rate of small lung The British Journal of Radiology, September 2000

CT features of rapidly growing small lung cancers

cancers with prognosis and to establish the effects of early diagnosis and surgical treatment of lung cancer on prognosis.

Acknowledgments We thank Kazuhisa Hanamura, BS, and Kazuhiro Asakura, EE, from the Telecommunications Advancement Organization of Japan Matsumoto Research Center for their contribution to this study.

References 1. Usuda K, Saito Y, Sagawa M, Sato M, Kanma K, Takahashi S, et al. Tumor doubling time and prognostic assessment of patients with primary lung cancer. Cancer 1994;74:2239±44. 2. Arai T, Kuroishi T, Saito Y, Kurita Y, Naruke T, Kaneko M. Tumor doubling time and prognosis in lung cancer patients: evaluation from chest ®lms and clinical follow-up study. Japanese Lung Cancer Screening Research Group. Jpn J Clin Oncol 1994;24:199±204. 3. Geddes DM. The natural history of lung cancer: a review based on rates of tumour growth. Br J Dis Chest 1979;73:1±17. 4. Kerr KM, Lamb D. Actual growth rate and tumour cell proliferation in human pulmonary neoplasms. Br J Cancer 1984;50:343±9. 5. Schwartz M. A biomathematical approach to clinical tumor growth. Cancer 1961;14:1272±94. 6. Steele JD, Buell P. Asymptomatic solitary pulmonary nodules, host survival, tumor size, and growth rate. J Thorac Cardiovasc Surg 1973;65:140±54. 7. Spratt JS, Meyer JS, Spratt JA. Rate of growth of human neoplasms: Part II. J Surg Oncol 1996;61: 68±83. 8. Hayabuchi N, Russell WJ, Murakami J. Slowgrowing lung cancer in a ®xed population sample. Radiological assessments. Cancer 1983;52:1098±104.

The British Journal of Radiology, September 2000

9. Theros EG. Varying manifestations of peripheral pulmonary neoplasms: a radiologic±pathologic correlative study. AJR 1977;128:893±914. 10. Sone S, Sakai F, Takashima S, Honda T, Yamanda T, Kubo K, et al. Factors affecting the radiologic appearance of peripheral bronchogenic carcinomas. J Thorac Imaging 1997;12:159±72. 11. Noguchi H, Morikawa A, Kawasaki M, Matsuno Y, Yamada T, Hirohashi S, et al. Small adenocarcinoma of the lung: histologic characteristics and prognosis. Cancer 1995;75:2844±52. 12. Kuriyama K, Seto M, Kasugai T, Higashiyama M, Kido S, Sawai Y, et al. Ground-glass opacity on thin-section CT: value in differentiating subtypes of adenocarcinoma of the lung. AJR 1999;173: 465±9. 13. Kuriyama K, Tateishi R, Doi O, Higashiyama M, Kodama K, Inoue E, et al. Prevalence of air bronchograms in small peripheral carcinoma of the lung on thin-section CT: comparison with benign tumors. AJR 1991;156:921±4. 14. Gaeta M, Pandolfo I, Volta S, Russi EG, Bartiromo G, Girone G, et al. Bronchus sign on CT in peripheral carcinoma of the lung: value in predicting results of transbronchial biopsy. AJR 1991;157:1181±5. 15. Ma K, Templeton PA, White CS, Cai Z-L, Bai Y-X, Cai Y-Q. Evaluation of the air bronchogram sign on CT in solitary pulmonary lesions. J Comput Assist Tomogr 1996;20:983±6. 16. Kuriyama K, Tateishi R, Doi O, Kodama K, Tatsuta M, Matsuda M, et al. CT±pathological correlation in small peripheral lung cancers. AJR 1987;149:1139±43. 17. Quoix E, Fraser R, Wolkove N, Finkelstein H, Kreisman H. Small cell lung cancer presenting as a solitary pulmonary nodule. Cancer 1990;66:577±82. 18. Urschel JD. Pretreatment natural history of small cell lung cancer presenting as a solitary pulmonary nodule. J Cardiovasc Surg 1994;35:273±5. 19. Brigham BA, Bunn PA Jr, Minna JD, Cohen MH, Ihde DC, Shackney SE. Growth rate of small cell bronchogenic carcinomas. Cancer 1978;42:2880±6.

937