Lung Cancer in Chronic Interstitial Pneumonia: Early Manifestation From Serial CT Observations

Cardiopulmonary Imaging Original Research Yoshida et al. CT of Lung Cancer Cardiopulmonary Imaging Original Research Rika Yoshida 1 Hiroaki rakawa Yasushi Kaji Yoshida R, rakawa H, Kaji Y Keywords: CT, high-resolution CT, interstitial lung disease, lung cancer DOI:10.2214/JR.11.7516 Received July 13, 2011; accepted after revision October 12, 2011. 1 ll authors: Department of Radiology, Dokkyo Medical University, School of Medicine, 880 Kita-kobayashi, Mibu-Machi, Shimotsuga-gun, Tochigi, 321-0293, Japan. ddress correspondence to H. rakawa (arakawa@dokkyomed.ac.jp). JR 2012; 199:85 90 0361 803X/12/1991 85 merican Roentgen Ray Society Lung Cancer in Chronic Interstitial Pneumonia: Early Manifestation From Serial CT Observations OJECTIVE. The purpose of this study was to use serial CT observations to characterize early-stage lung cancer in patients with chronic interstitial pneumonia. MTERILS ND METHODS. We found 23 lung cancers in 22 patients during routine follow-up of chronic interstitial pneumonia between 1999 and 2010. Patients with lung cancer found at initial CT were excluded. Two radiologists independently reviewed serial CT scans, determined the earliest scan showing lung cancer, and evaluated the tumor shape, size, density, and location. Delay in diagnosis was measured from the time of the earliest scan showing lung cancer and the subsequent clinical diagnosis. RESULTS. During the mean follow-up period of 4.1 years, CT scans were obtained eight times on average. The median tumor size at presentation was 11 mm, and at clinical diagnosis was 22 mm. The median delay in diagnosis was 409 days. Fifteen tumors (65.2%) were in the interface between normal and fibrotic lung cysts (honeycomb cysts, paraseptal emphysema, and traction bronchiolectasis), four were in the area of ground-glass opacity, and one was in the midst of honeycomb cysts. Twelve tumors were round or oval, eight tumors had an ill-defined stellate shape, and two had a bandlike shape. One tumor appeared as an area of illdefined increased lung attenuation. CONCLUSION. Nearly one half of the tumors had a stellate or bandlike shape and were difficult to recognize as tumors initially. Most of the tumors were located at the interface between normal lung and fibrotic cysts; only rarely were tumors located in the midst of honeycomb cysts. T he incidence of idiopathic interstitial pneumonia mostly idiopathic pulmonary fibrosis has been reported to have increased [1 3]. It has been suggested [4 7] that diffuse interstitial fibrosis, including usual interstitial pneumonia, which is the pathologic counterpart of idiopathic pulmonary fibrosis, is a risk factor for lung cancer, although this theory is not widely accepted. ecause patients with lung cancer often present with few clinical symptoms, and because lung cancer can be incidentally found on CT scans obtained during follow-up of chronic interstitial pneumonia, radiologists must correctly identify tumors at an early stage. However, identifying a small tumor against a background of interstitial opacities is often difficult in patients with chronic interstitial pneumonia. Therefore, awareness of the characteristic CT manifestations of lung cancer associated with chronic interstitial pneumonia in its early stage is important. The aim of this study was to identify the CT characteristics of early lung cancer in relation to chronic interstitial pneumonia on serial images obtained from the time of no identifiable tumor to clinical diagnosis of overt lung cancer. Materials and Methods Patients We retrospectively searched the radiology information system in our department for the cases of patients with diagnoses of chronic interstitial pneumonia (including idiopathic pulmonary fibrosis and nonspecific interstitial pneumonia and interstitial pneumonia in collagen vascular disease) and lung cancer who had undergone serial chest CT between 1999 and 2010. We excluded patients whose initial CT report described obvious tumor opacity. We found the cases of 22 patients (18 men, four women) with chronic interstitial pneumonia who had lung cancer that developed during the follow-up period and who underwent diagnostic intervention. In no patient was a lung tumor identified on the initial CT images. In one JR:199, July 2012 85

Yoshida et al. TLE 1: Clinical Findings on 22 Patients With Lung Cancer ssociated With Chronic Interstitial Pneumonia Characteristic woman, two primary lung cancers developed during follow-up. Thus our study group included 23 tumors in 22 patients (median age, 70 years; range, 31 83 years). Clinical information, including smoking history, clinical stage of lung cancer at diagnosis, presence of collagen vascular disease, and histologic or cytologic diagnosis of lung cancer, was evaluated. The clinical stage of lung cancer was classified with the system of the International Union gainst Cancer, sixth edition, on the basis of the CT and PET/CT findings. We based the date of the clinical diagnosis of lung cancer on the date of the tumor biopsy. CT Technique CT was performed with various scanners (Somatom Plus 4, Sensation 40, Sensation 64, all Siemens Healthcare). The CT images of the entire lungs were produced with 8-mm section thickness. dditional high-resolution CT was performed with Value Sex Men 18 (81.8) Women 4 (18.2) ge (y) Median 70 Range 31 83 Smoking habit Present 20 (90.9) Duration (pack-years) Median 47.5 Range 5 137.5 bsent 2 (9.1) Complications Present 5 (22.7) Rheumatoid arthritis 3 men (13.6) Systemic sclerosis 2 women (9.1) bsent 17 (77.3) Clinical stage (no. of lesions, n = 23) I 11 (47.8) I 4 (17.4) II 1 (4.3) II 0 (0) III 4 (17.4) III 1 (4.3) IV 2 (8.7) Note Values are numbers of patients with percentages in parentheses unless otherwise indicated. thin sections (1 or 2 mm) at 10-mm intervals to evaluate the chronic interstitial pneumonia. In selected cases, additional thin-section images of the tumor were obtained without an interval. Lung window images were reconstructed with a high-frequency algorithm (width, 1400 HU; level, 650 HU). Interpretation Two radiologists, one with 20 years experience and the other with 5 years experience in reading chest CT scans, independently reviewed the serial CT images and determined the earliest CT image showing the definite presence of lung cancer. Serial CT images were reviewed with the PCS. On the earliest CT image showing the tumor, we determined the size, location, shape, and type of opacity of each tumor. Disagreement between the two observers was resolved by consensus. The recorded size of the tumor was determined by averaging the two observers measurements. The shape of the tumor was subcategorized as being round or oval, bandlike, stellate, or other. Round or oval was defined as a relatively well-defined sphere or oval with a smooth margin. andlike was defined as elongated linear opacity with the longitudinal diameter more than three times the short diameter. Stellate was defined as having a starlike appearance that radiated from the center and showed irregular morphologic features. The type of opacity of the tumor was subcategorized as solid nodule, part-solid nodule, or ground-glass nodule [8]. In terms of the localization of the tumor, we specifically evaluated the association of the tumor with fibrotic lung cysts such as honeycomb cysts, traction bronchiolectasis, and paraseptal emphysema. The difference in tumor size among the subcategories was tested with analysis of variance, and the Scheffe method was subsequently used (SPSS 11.01 J for Microsoft Windows, SPSS). Results Clinical Findings The clinical findings on 22 patients with chronic interstitial pneumonia are summarized in Table 1. Five of these patients had collagen vascular disease (three, rheumatoid arthritis; two, systemic sclerosis), and the other 17 patients had idiopathic interstitial pneumonia. Twenty patients had a smoking history with a median exposure of 47.5 packyears (range, 5 137.5 pack-years). Two women had never smoked. The distribution of tumors by stage is shown in Table 1. Pathologic Findings The pathologic diagnosis of lung cancer was based on the lobectomy specimen of nine tumors, transbronchial lung biopsy specimen of 12 tumors, and cytologic examination of two tumors. Of the 23 tumors, 11 (47.8%) were adenocarcinoma, eight (34.8%) were squamous cell carcinoma, three (13.0%) were small cell carcinoma, and one (4.3%) was pleomorphic carcinoma. The pathologic diagnosis of interstitial pneumonia was based on examination of the lobectomy specimen (nine lesions) or previous surgical lung biopsy (one lesion). Of these 10 cases of interstitial pneumonia, seven were usual interstitial pneumonia, one was nonspecific interstitial pneumonia, and two were not classified because of atypical histopathologic findings. High-Resolution CT Findings Radiologic follow-up of chronic interstitial pneumonia lasted 1 7.2 years (average, 4.1 years). During the follow-up period, CT was performed 2 23 times for each patient (average, 8 times). The median interval between 86 JR:199, July 2012

CT of Lung Cancer TLE 2: CT Findings of 23 Cancer Nodules ssociated With Chronic Interstitial Pneumonia Finding Value Opacity Solid 13 (56.5) Part solid 6 (26.1) Ground glass 4 (17.4) Size (mm) t clinical diagnosis Median 22 Range 12 60 Smallest identifiable size Median 11 Range 6 25 Shape Round or oval 12 (52.2) Stellate 8 (34.8) andlike 2 (8.7) Other 1 (4.3) Location a Interface of fibrotic cysts and normal lung 15 (65.2) Ground-glass opacity 4 (17.4) Midst of honeycomb cysts 1 (4.3) Normal lung 3 (13.0) Note Values are number of cases with percentages in parentheses unless otherwise indicated. a ll but one of the tumors were in the lung periphery. the time of the earliest tumor presentation and time of clinical diagnosis was 409 days (range, 0 1301 days). The CT findings regarding the 23 lung cancers are shown in Table 2. The median diameter at earliest tumor presentation was 11 mm (range, 6 25 mm), and that at clinical diagnosis was 22 mm (range, 12 60 mm). Of the 23 tumors, 22 (95.7%) were located in the periphery of the lung. The tumor locations are shown in Table 2. Twelve tumors (52.2%) had a well-defined round or oval shape (median diameter, 11 mm), eight (34.8%) had a stellate shape (median diameter, 10 mm [Fig. 1]), and two (8.7%) had a bandlike shape (median diameter, 17 mm [Fig. 2]). One tumor (4.3%) appeared as a peribronchial area of ill-defined ground-glass attenuation (12 mm) and was not subcategorized (Fig. 3). ll but one of the stellate tumors were found in an area of fibrotic cysts and had infiltrated along the walls of the cysts. s the follow-up period progressed, these stellate and bandlike lesions became voluminous and became closer to round (Fig. 4). No statistical association was identified between size and shape of the tumors (p = 0.339). Of the 23 tumors, 13 (56.5%) appeared as solid nodules (median diameter, 13 mm), six (26.1%) as part-solid nodules (median diameter, 7 mm), and four (17.4%) as groundglass nodules (median diameter, 11 mm). lthough solid nodules tended to be larger than part-solid nodules, the difference was not statistically significant (p = 0.052). Discussion In a Japanese study of 3712 consecutive autopsy cases [6], the prevalence of lung cancer was significantly higher among patients with usual interstitial pneumonia than in the age-matched general population without usual interstitial pneumonia (48.2% vs 9.1%; p < 0.001). This finding was confirmed in a population-based cohort study from the United Kingdom [9], the estimated rate ratio of association being 7.31. lthough some studies failed to confirm the association, it is widely expected that radiologists will encounter lung cancer more often in patients with chronic interstitial pneumonia than in those without. Patients with chronic interstitial pneumonia, including idiopathic and associated collagen vascular disease, have a variable life expectancy and undergo repeated chest CT examinations during follow-up. It is important not to miss detection of early-stage lung cancer at CT interpretation. Failure of detection becomes apparent when progression of lung cancer is diagnosed on later CT scans. In this small series of 23 lung cancers associated with chronic interstitial pneumonia, we focused on evaluating the earliest CT findings of lung cancer on serial CT images that initially showed no tumor opacity but later showed growth of lung cancer from minimal opacity to overt tumor opacity. Our results showed that the typical highresolution CT finding of early lung cancer was a tumor 11 mm in median diameter. The typical tumor was solid with a well-defined round shape and abutted fibrotic lung cysts. However, 48% of the tumors had an irregular shape (most often stellate or even bandlike or ill-defined ground glass) and looked atypical for primary lung cancer. One explanation of the irregular shape is that during growth, tumors originating in fibrotic lung may avoid the area of advanced fibrosis, which is usually stiff and irregularly distributed. lthough this explanation cannot be considered the sole cause of a stellate tumor shape, all but one tumor occurred in an area of fibrotic lung cysts. We observed that as the tumor enlarged, these irregularly shaped tumors evolved to take on a round shape, which made them begin to look typical of lung cancer. This observation is important because small lung cancers often appear irregular, especially in the early stage, and are apt to be missed. In our study, delay of clinical diagnosis from earliest presentation of tumor opacity was at most 1301 days (median, 409 days), when the median tumor size became 22 mm, which was twice the diameter at initial presentation. With the exception of a study by Park et al. [10], previous studies have shown that lung cancers are likely to occur in the area of lung fibrosis [4, 6, 11 14], where atypical and dysplastic epithelial changes are often observed [15]. In terms of localization, we confirmed that the tumors preferentially occurred in areas with fibrosis (87.0%), especially in the interface between normal lung and fibrotic lung cysts (65.2%). ecause of these observations, focal increased lung density associated with fibrotic lung cysts should JR:199, July 2012 87

Yoshida et al. Fig. 1 74-year-old man with idiopathic pulmonary fibrosis and usual interstitial pneumonia., Initial high-resolution CT image at level of superior segment of right lower lobe shows peripheral ground-glass opacity, traction bronchiolectasis, and emphysema. No tumor opacity is present., High-resolution CT image 20 months after shows 7-mm stellate part-solid nodule (arrow) abutting small cystic spaces (traction bronchiolectasis) of right lower lobe. Similar opacity (arrowhead) is present at same level, which proved to be interstitial fibrosis at follow-up. C, High-resolution CT image 11 months after shows stellate nodule has grown to 11 mm. ppearance has become suggestive of lung cancer. Small cell lung cancer was diagnosed at transbronchial biopsy. C Fig. 2 74-year-old man with rheumatoid arthritis and surgically confirmed usual interstitial pneumonia., Chest CT image obtained at initial admission for pneumonia shows focal thickening (arrow) of pulmonary vessel in lingula., High-resolution CT image 1 year after shows 13-mm bandlike lesion (arrow) in area of ground-glass opacity that corresponds to lesion in. lthough different section thickness was used, lesion has clearly grown. C, High-resolution CT image 12 months after shows nodule has grown to 23 mm and become plump. denocarcinoma was confirmed after surgical resection of lingular segment. C 88 JR:199, July 2012

CT of Lung Cancer be watched with caution, even if the shape is not necessarily tumorlike. Only one tumor developed in the midst of honeycomb cysts, which are clustered lung cysts with fibrosis, usually with no lung tissue between the cysts [16]. We suppose that lung cancer is unlikely to develop in the cyst wall without adjacent lung tissue, and careful observation of the tumor and surrounding lung condition can add important information to the differential diagnosis. In previous reports, nodules with air bronchograms have most commonly been reported in lung cancer associated with chronic interstitial pneumonia [12, 17]. In our study, however, we classified the nodules as ground Fig. 3 69-year-old man with surgically confirmed idiopathic nonspecific interstitial pneumonia., High-resolution CT image shows small peribronchial area of reticulated opacity (arrow) in central zone of middle right lobe., High-resolution CT image 14 months after shows part-solid ground-glass nodule has grown to 25 mm. Squamous cell carcinoma was confirmed after surgical resection of middle right lobe. Fig. 4 69-year-old man with idiopathic pulmonary fibrosis and usual interstitial pneumonia., High-resolution CT image shows 21-mm bandlike lesion (arrow) amid honeycomb cysts., High-resolution CT image 17 months after shows nodule has grown to 24 mm and become voluminous and rounded. Squamous cell carcinoma was diagnosed at transbronchial biopsy. C, High-resolution CT image cranial to shows top of tumor (arrow) abutting normal lung, confirming tumor is in interface between normal lung and fibrotic cysts. glass (17.4%), part solid (26.1%), or solid (56.5%), the usual approach to describing small lung cancer [8]. In general, the groundglass component of the tumor corresponds to partially aerated lung that bears cells of bronchioloalveolar carcinoma, whereas the solid component corresponds to other adenocarcinomas and carcinomas [8]. Thus the frequency of the nodule type may depend on the frequency of the histopathologic subtype of carcinoma. The relatively lower frequency of ground-glass nodules in our study may reflect the lower frequency of bronchioloalveolar carcinoma. nother explanation is that the ground-glass nodules might not have been detected until a solid component developed from the surrounding ground-glass opacity of interstitial pneumonia. Our study had several limitations. First, because the study was retrospective, the interval between CT examinations was not determined in advance. t our institution, patients with chronic interstitial pneumonia underwent CT scans at intervals of 6 months on average, which is more frequent than the annual screening program for lung cancer. Second, the number of patients enrolled in the study group was small, which limited the statistical analysis. Third, our CT protocol did not always include thin-section scanning of the whole lung volume. Detailed analysis of the tumor might have been impaired when C JR:199, July 2012 89

Yoshida et al. a small tumor fell within the scan interval in thin-section CT acquisitions. Fortunately, our review showed that most of the tumors were scanned with thin-section CT even in the earliest stage of presentation. In summary, lung cancer associated with chronic interstitial pneumonia initially appears as a solid nodule or ground-glass opacity approximately 11 mm in diameter preferentially in the periphery of the lungs and often is associated with fibrotic lung cysts (i.e., honeycomb cysts, traction bronchiolectasis, and paraseptal emphysema). One half of the tumors had a stellate shape or even an elongated bandlike shape, often intermingled with fibrotic lung opacities and lung cysts, giving an appearance atypical of lung cancer. However, the tumors became rounded as they grew. These observations may be enlightening for radiologists interpreting the CT images of patients with chronic interstitial pneumonia, giving hints to avoid missing lung cancer. References 1. Coultas D, Zumwalt RE, lack WC, Sobonya RE. The epidemiology of interstitial lung diseases. m J Respir Crit Care Med 1994; 150:967 972 2. Navaratnam V, Fleming KM, West J, et al. The rising incidence of idiopathic pulmonary fibrosis in the UK. Thorax 2011; 66:462 467 3. Gribbin J, Hubbard R, Le Jeune I, Smith CJ, West J, Tata LJ. Incidence and mortality of idiopathic pulmonary fibrosis and sarcoidosis in the UK. Thorax 2006; 61:980 985 4. Kawai T, Yakumaru K, Suzuki M, Kageyama K. Diffuse interstitial pulmonary fibrosis and lung cancer. cta Pathol Jpn 1987; 37:11 19 5. Nagai, Chiyotani, Nakadate T, Konno K. Lung cancer in patients with idiopathic pulmonary fibrosis. Tohoku J Exp Med 1992; 167:231 237 6. Matsushita H, Tanaka S, Saiki Y, et al. Lung cancer associated with usual interstitial pneumonia. Pathol Int 1995; 45:925 932 7. Raghu G, Nyberg F, Morgan G. The epidemiology of interstitial lung disease and its association with lung cancer. r J Cancer 2004; 91(suppl 2):S3 S10 8. Henschke CI, Yankelevitz DF, Mirtcheva R, Mc- Guinness G, McCauley D, Miettinen OS. CT screening for lung cancer: frequency and significance of part-solid and nonsolid nodules. JR 2002; 178:1053 1057 9. Hubbard R, Venn, Lewis S, ritton J. Lung cancer and cryptogenic fibrosing alveolitis: a population-based cohort study. m J Respir Crit Care Med 2000; 161:5 8 10. Park J, Kim DS, Shim TS, et al. Lung cancer in patients with idiopathic pulmonary fibrosis. Eur Respir J 2001; 17:1216 1219 11. Kishi K, Homma S, Kurosaki, Motoi N, Yoshimura K. High-resolution computed tomography findings of lung cancer associated with idiopathic pulmonary fibrosis. J Comput ssist Tomogr 2006; 30:95 99 12. ubry MC, Myers JL, Douglas WW, et al. Primary pulmonary carcinoma in patients with idiopathic pulmonary fibrosis. Mayo Clin Proc 2002; 77:763 770 13. Martinod E, zorin JF, Sadoun D, et al. Surgical resection of lung cancer in patients with underlying interstitial lung disease. nn Thorac Surg 2002; 74:1004 1007 14. Sakai S, Ono M, Nishio T, Kawarada Y, Nagashima, Toyoshima S. Lung cancer associated with diffuse pulmonary fibrosis: CT-pathologic correlation. J Thorac Imaging 2003; 18:67 71 15. Meyer EC, Liebow. Relationship of interstitial pneumonia honeycombing and atypical epithelial proliferation to cancer of the lung. Cancer 1965; 18:322 351 16. Hansell DM, ankier, MacMahon H, McLoud TC, Muller NL, Remy J. Fleischner Society: glossary of terms for thoracic imaging. Radiology 2008; 246:697 722 17. Lee HJ, Im JG, hn JM, Yeon KM. Lung cancer in patients with idiopathic pulmonary fibrosis: CT findings. J Comput ssist Tomogr 1996; 20:979 982 FOR YOUR INFORMTION The comprehensive book based on the RRS 2012 annual meeting categorical course on Pitfalls in Clinical Imaging is now available! For more information or to purchase a copy, see www.arrs.org. 90 JR:199, July 2012