CT Features of Peripheral Pulmonary Carcinoid Tumors

Cardiopulmonary Imaging Original Research Meisinger et al. CT of Pulmonary Carcinoid Tumors Cardiopulmonary Imaging Original Research Quinn Colin Meisinger 1 Jeffrey S. Klein 1 Kelly J. Butnor 2 George Gentchos 1 Bruce J. Leavitt 3 Meisinger QC, Klein JS, Butnor KJ, Gentchos G, Leavitt BJ Keywords: MDCT, octreotide, pulmonary carcinoid tumor, solitary pulmonary nodule, transthoracic needle biopsy DOI:10.2214/AJR.10.5954 Received October 15, 2010; accepted after revision April 7, 2011. 1 Department of Radiology, University of Vermont College of Medicine, 111 Colchester Ave, Burlington, VT 05401. Address correspondence to J. S. Klein (Jsklein1@mac.com). 2 Department of Pathology, University of Vermont College of Medicine, Burlington, VT. 3 Department of Surgery, University of Vermont College of Medicine, Burlington, VT. CME This article is available for CME credit. See www.arrs.org for more information. AJR 2011; 197:1073 1080 0361 803X/11/1975 1073 American Roentgen Ray Society CT Features of Peripheral Pulmonary Carcinoid Tumors OBJECTIVE. Pulmonary carcinoid tumors are low-grade malignant neoplasms thought to arise primarily within the central airways in 85% of cases. The CT features of pulmonary carcinoid tumors that arise as solitary pulmonary nodules (SPNs) have not been well elucidated. We reviewed our experience with primary pulmonary carcinoid tumors to determine the distribution of lesions within the lung at diagnosis and to identify CT features that might aid in distinguishing these neoplasms from benign pulmonary nodules. MATERIALS AND METHODS. CT scans, if available, of all patients with a primary pulmonary carcinoid tumor diagnosed by biopsy or surgical resection over the previous 15 years were reviewed. The CT scans were reviewed for the following features: lesion location; order of bronchus involved; lesion size, contour, and density; contrast enhancement; and the presence of peripheral atelectasis, hyperlucency, and bronchiectasis. We defined central lesions as those involved with a segmental or larger bronchus. Subsegmental bronchial involvement and tumors surrounded by lung parenchyma without direct airway involvement were defined as peripheral lesions. The final pathologic diagnosis for all cases was confirmed by review of cytologic or histologic specimens. RESULTS. Twenty-eight carcinoid tumors were identified in 28 patients: 24 typical carcinoids and four atypical carcinoids. The study group was composed of 23 females and five males with a mean age of 52.4 years (range, 14 83 years). Twelve of the 28 lesions (43%) were central (i.e., involved a segmental or larger bronchus), and the remaining 16 lesions (57%) were peripheral. The mean tumor diameter for the 16 peripheral tumors was 14 mm (range, 9 28 mm); the majority (14/16, 88%) had a lobulated contour. Of six peripheral lesions with unenhanced and contrast-enhanced CT nodule enhancement studies, the mean maximal enhancement was 55.2 HU (range, 34 73 HU). Thirteen of the 16 peripheral carcinoid tumors (81%) involved a subsegmental bronchus, with 10 (63%) showing peripheral hyperlucency, bronchiectasis, or atelectasis. CONCLUSION. In our series, primary pulmonary carcinoid tumors presenting as peripheral SPNs were more common than central endobronchial lesions in contrast to the published literature. The CT features of peripheral carcinoid tumors presenting as SPNs that suggest the diagnosis include lobulated nodules of high attenuation on contrast-enhanced CT; nodules that densely enhance with contrast administration; the presence of calcification; subsegmental airway involvement on thin-section analysis; and nodules associated with distal hyperlucency, bronchiectasis, or atelectasis. A carcinoid is an uncommon malignancy that can arise in a wide range of tissues that harbor neuroendocrine cells. After the gastrointestinal system, the bronchopulmonary system is the most common location for primary carcinoid tumor. Carcinoid tumors arising from all locations are increasing in incidence at an average rate of 6% per year [1]. More specifically, the incidence of lung neuroendocrine tumors including pulmo- nary carcinoid tumors is increasing [2]. Bronchopulmonary carcinoid tumors had an incidence of 1.57 cases per 100,000 patients in 2003 in the United States. Although rare when compared with non small cell lung cancer, carcinoid tumors are the second most common primary malignant pulmonary neoplasm in adults. Carcinoid tumors possess an indolent clinical course with infrequent metastases, but they are true malignancies with the potential AJR:197, November 2011 1073

Meisinger et al. to cause significant morbidity and mortality [3]. Although the CT findings of central endobronchial pulmonary carcinoid tumors have been detailed, descriptions of carcinoid tumors presenting as incidental pulmonary nodules are limited, with little detail about features that might distinguish these slowgrowing malignant neoplasms from benign lung nodules. Pulmonary carcinoid tumors are classified into one of two categories on the basis of cytologic criteria: atypical or typical. Typical carcinoids represent 80 90% of pulmonary carcinoids and are most frequently diagnosed in the fifth and sixth decades of life but can occur at any age. Atypical carcinoid neoplasms possess more malignant potential than typical carcinoids, with a greater propensity to nodal and distant metastases [3]. Eighty-five percent of carcinoid tumors are located centrally within the lung: 10% arise from the mainstem bronchi, and 75% from the lobar bronchi [1, 4 6]. However, the larger studies describing this central predilection of pulmonary carcinoid tumors pre-date the widespread availability of MDCT [1]. The majority of patients with pulmonary carcinoid tumors (58%) are symptomatic at presentation [3]. The most common symptoms include cough, hemoptysis, wheezing, and pneumonia as a result of central airway involvement. Rare associations of pulmonary carcinoid tumors include carcinoid syndrome, Cushing disease from ectopic adrenocorticotropic hormone (ACTH) production, and acromegaly from ectopic production of growth hormone. Because symptomatic patients are more likely to undergo diagnostic evaluation, a central predilection for pulmonary carcinoid tumors predominates in published series. However, there are likely a significant number of carcinoid tumors, particularly the more indolent typical carcinoid tumors, that remain undiagnosed during a patient s life. This hypothesis is supported by the finding in a populationbased study that 24% of typical carcinoids and 7% of atypical carcinoids are among the tumors incidentally identified on autopsy [2]. As a result of the more widespread availability and use of MDCT, incidental solitary pulmonary nodules (SPNs) are being detected with increasing frequency [7]. The CT features of pulmonary carcinoid tumors that arise peripherally as SPNs, however, have not been well elucidated to date. Therefore, the primary objective of our study was to determine whether the location and appearance of pulmonary carcinoid tumors in our patient population differs from those previously described in the literature. An evaluation of the features of this tumor as seen on thin-section CT including lesion location, size, density, and contour and the presence of calcification and associated changes in adjacent tissues might lead to better identification of these neoplasms. This knowledge, in turn, would lead to more prompt diagnosis and effective treatment of these lowgrade malignant neoplasms. Materials and Methods Patients This retrospective study received approval from the institutional review board of Fletcher Allen Health Care. The informed consent requirement was waived. A search of the Fletcher Allen Health Care s laboratory system (CoPath, Cerner Corporation) was used to obtain the pathology records of patients with a final diagnosis of typical or atypical primary pulmonary carcinoid tumor from 1996 until April 2010. Forty patients matching the search criteria were identified. Each of the 40 patients had only a solitary pulmonary carcinoid tumor according to the pathology records. Of these 40 patients, pathology slides were available for 34 patients and were retrieved from storage for review. A fellowship-trained pulmonary pathologist reviewed each case to confirm the diagnosis of typical or atypical carcinoid tumor. The diagnoses of typical and atypical carcinoid tumors were based on the criteria established by the World Health Organization [8]. Tumors were characterized by bland polygonal or spindled cells with scant cytoplasm and finely granular chromatin arranged in organoid nests, trabeculae, or cords. In typical carcinoid tumors, necrosis was absent and the mitotic activity was low, not exceeding 2 mitoses per 2 mm 2. Atypical carcinoid tumors showed similar morphologic features but can be distinguished from typical carcinoid tumors by the presence of foci of necrosis, increased mitoses (2 10 mitoses per 2 mm 2 ), or both. A total of 34 patients had a diagnosis of atypical or typical carcinoid tumors confirmed by a pathologist. Of the 34 lesions reviewed, 30 lesions were labeled as typical tumors and four lesions as atypical carcinoid tumors. According to the pathology records, the six patients without locatable slides had a diagnosis of typical carcinoid tumors, and this was accepted as the final pathologic diagnosis. The location of the tumor within the lung was then documented in the pathology records. The type of procedure used for initial pathologic diagnosis was also documented. Three different procedures were used: transbronchial needle aspiration using bronchoscopy, CT-guided transthoracic fine-needle aspiration biopsy, and surgical lobectomy. Transbronchial needle aspiration was performed with a 13-mm, 22-gauge needle using a standard flexible videobronchoscope. During the procedure, specimens were obtained by transbronchial needle aspiration with or without other conventional diagnostic procedures such as bronchoalveolar lavage, bronchial brushings, endobronchial forceps biopsy, and transbronchial lung biopsy. CT-guided transthoracic biopsy was performed using either conventional CT guidance or CT fluoroscopy when available. Image Technique A subsequent search of our institution s PACS and film library was performed to determine which of the 40 patients with a pathologic diagnosis of carcinoid tumor had a CT examination before diagnosis that was available for review. Of the 40 patients, 28 had CT scans available either electronically or on film; the 12 patients with a diagnosis of carcinoid tumor but without available CT scans were excluded from further evaluation. Four scans obtained before 2000 were obtained on single-detector CT scanners, whereas the remaining 24 patients underwent MDCT on one of five scanners with 4 64 detectors. The CT slice thickness used for the analysis of lesion features ranged from 0.9 to 2.0 mm. Seven of the 28 patients had more than one CT examination before diagnosis; in those seven patients, the study obtained closest to the time of diagnosis was used for analysis. Image Analysis The indication for each CT examination and the patient s presenting symptoms were documented. Each CT scan was reviewed by two thoracic radiologists for specific features that were then recorded on a spreadsheet. The two radiologists had 21 and 10 years of experience at the time of the study. The presence or absence of IV contrast material was recorded, and those patients with both unenhanced and contrast-enhanced studies or dedicated nodule enhancement studies were noted. The lesion features evaluated included the lobe of lung involved, the presence or absence of airway involvement with order of bronchus involved (main, lobar, segmental, or subsegmental), lesion attenuation, degree of contrast enhancement for those with both unenhanced and contrast-enhanced or dedicated enhancement studies, maximum lesion diameter in the axial plane, contour (smooth, lobulated, or spiculated), presence of calcification, and presence of abnormalities distal to the lesion (i.e., peripheral bronchiectasis, peripheral hyperlucency, and peripheral opacity). A mean attenuation measurement (in Hounsfield units) was obtained for each lesion by use of a region-of-interest 1074 AJR:197, November 2011

CT of Pulmonary Carcinoid Tumors tool sized to encompass approximately two thirds of the long and short axes of the nodule and positioned over the central aspect of the nodule as seen on axial images displayed at mediastinal windows to minimize partial volume effect. Attenuation was measured at the time of imaging review by the authors of this study. We defined central lesions as those involving segmental or larger bronchus; subsegmental bronchial involvement and tumors surrounded by lung parenchyma without direct airway involvement were defined as peripheral lesions [9]. Airway involvement was defined as present if a CT scan showed a component of the lesion either within the lumen of or obstructing the lumen of a visible airway. All data were then organized into a data sheet for subsequent analysis. Interobserver analysis was performed on the CT assessment of airway involvement of peripheral tumors using the Cohen kappa coefficient. Patients who underwent preoperative nuclear imaging, including whole-body 111 In-octreotide scanning and thoracic FDG PET, were noted. The results of those studies were noted as either positive or negative within the primary lesion after review of the study report generated at the time of the examination. If the attending nuclear radiologist stated that the study was positive in the impression section of the dictation, it was counted as positive. There were no indeterminate studies, only negative and positive. Any presenting symptoms that led to an imaging evaluation and the subsequent diagnosis of carcinoid tumor were documented from a review of the patients medical records. Results Pathology Data Twenty-eight carcinoid tumors in 28 patients met the selection criteria (Table 1): 24 tumors (86%) were typical carcinoid and four tumors (14%) were atypical carcinoid. The study group was composed of 23 females (82%) and five males (18%) with a mean age of 52.4 years and median age of 52 years (range, 14 83 years). The initial diagnosis of carcinoid tumor was made in 14 tumors (50%) by CT-guided fine-needle aspiration, in 11 tumors (39%) by surgical lobectomy, and in three tumors (11%) by transbronchial fine-needle aspiration under bronchoscopy. Of the 14 patients with the initial diagnosis based on CT-guided biopsy, eight had subsequent lobectomy and six were lost to follow-up without histologic confirmation. Of TABLE 1: Patient Characteristics and Pathologic Data Patient No. Age (y) Sex Diagnosis Method of Diagnosis Anatomic Location Symptom 1 58 F Carcinoid Lobectomy RLL Recurrent pneumonia 2 83 F Atypical CT-guided FNA LLL None 3 50 F Carcinoid CT-guided FNA RML None 4 78 F Carcinoid CT-guided FNA LUL None 5 35 F Carcinoid Lobectomy LUL Hemoptysis 6 37 M Carcinoid Lobectomy RML Cough 7 31 F Carcinoid Transbronchial FNA Left main bronchus Hemoptysis 8 58 F Atypical CT-guided FNA RML None 9 51 F Carcinoid Transbronchial FNA LUL, LLL Wheezing 10 49 F Carcinoid Transbronchial FNA Left main bronchus Wheezing 11 35 M Carcinoid CT-guided FNA Left lingula (LUL) Cough 12 67 F Carcinoid CT-guided FNA RLL None 13 55 F Carcinoid CT-guided FNA RLL None 14 28 M Carcinoid Lobectomy RLL Cough 15 43 F Carcinoid Lobectomy RLL Cough 16 61 F Carcinoid CT-guided FNA LLL None 17 56 F Carcinoid CT-guided FNA RML None 18 72 F Carcinoid Lobectomy RLL None 19 78 F Carcinoid CT-guided FNA RML Cushing disease 20 74 F Carcinoid Lobectomy LLL Chest pain 21 54 F Atypical CT-guided FNA RUL None 22 53 M Carcinoid Lobectomy LLL None 23 35 M Atypical Lobectomy Left main bronchus Unknown 24 14 F Carcinoid Lobectomy LUL Recurrent pneumonia 25 34 F Carcinoid Lobectomy LLL Cough 26 50 F Carcinoid CT-guided FNA LLL Chest pain 27 49 F Carcinoid CT-guided FNA LLL None 28 80 F Carcinoid CT-guided FNA RML None Note RLL = right lower lobe, FNA = fine-needle aspiration, LLL = left Lower Lobe, RML = right middle lobe, LUL = left upper lobe, RUL = right upper lobe. AJR:197, November 2011 1075

Meisinger et al. the three patients that were diagnosed on the basis of transbronchial fine-needle aspiration findings, two patients had subsequent lobectomy and one patient underwent bronchoscopic resection. The laterality of tumor origin was nearly evenly distributed between the left and right lungs, with 15 tumors (54%) arising in the left lung and 13 (46%) in the right lung. Of the 28 lung tumors, 7 (25%) were located in the left lower lobe, 4 (14%) were located in the left upper lobe, 1 (4%) was located in both left upper and lower lobes, and 3 (11%) were located in the left main bronchus. Six (21%) were located in the right middle lobe, 6 (21%) were located in the right lower lobe, and 1 (4%) was located in the right upper lobe. CT Image Analysis Of the 28 tumors, three tumors (11%) arose in the left main bronchus; six (21%) in a lobar bronchus; three (11%) in a segmental bronchus; and 13 (46%) in a subsegmental bronchus. The remaining three tumors (11%) had no discernable airway involvement. Based on our definition of peripheral tumors as being those arising from a subsegmental or more distal bronchus, 12 of the 28 tumors (43%) were central carcinoid tumors and 16 (57%) were peripheral. The mean maximal lesion diameter of the 28 tumors was 17.7 mm (range, 9 51 mm) (Table 2). The mean lesion diameter of the 16 peripheral tumors was 14 mm (range, 9 28 mm). CT analysis of lesion contour was possible for 26 of 28 tumors; two central lesions were not adequately circumscribed to allow evaluation of the outer contour. Twenty-two of 26 lesions (85%) had a lobulated contour and four of 26 (15%) showed a smooth contour. Of the 16 peripheral tumors, 14 (88%) had a lobulated contour. There was calcification in three of 28 tumors (11%). One of the three calcified tumors was peripheral. Of the 28 lesions, lesion attenuation was evaluated in 26. Seventeen lesions had a single contrast-enhanced CT study. The mean lesion attenuation was 67 HU (range, 16 100 HU; median, 70 HU; SD, 21.4). Four patients underwent CT without IV contrast material. The mean lesion attenuation was 74 HU (range, 46 116 HU; median, 60 HU; SD, 37.0). Five patients had dedicated CT nodule enhancement studies, and two additional patients had separate unenhanced and contrastenhanced CT scans on different dates. In these seven patients, the mean unenhanced attenuation was 14.9 HU, the mean contrastenhanced attenuation was 68.4 HU, and the mean increase in enhancement was 53.4 HU (range, 34 73 HU). For the 16 peripheral tumors, the mean of all Hounsfield measurements on CT, both contrast-enhanced and unenhanced combined, was 58.8 HU (range, 40 96 HU; median, 60 HU; SD, 17.9). Of the six peripheral lesions with unenhanced and contrast-enhanced CT nodule enhancement studies, the mean maximum enhancement value was 55.2 HU (range, 34 73 HU) (Fig. 1). Of the 28 lesions described, 25 lesions (89%) had direct airway involvement. Thirteen of 16 peripheral tumors (81%) had direct airway involvement. Indirect findings of airway involvement seen on thin-section CT in the study population included peripheral bronchiectasis in nine of 28 patients (32%) (Fig. 2), peripheral pulmonary hyperlucency in 11 of 28 (39%) (Fig. 3), and peripheral pulmonary opacity or atelectasis in seven of 28 (25%). Twenty-three of 28 patients (82%) had one or more findings of peripheral bronchiectasis, hyperlucency, or atelectasis. Of the 16 peripheral tumors, four (25%) exhibited peripheral bronchiectasis, five (31%) showed peripheral hyperlucency, and three (19%) showed peripheral atelectasis. Of the 16 patients with peripheral tumors, 10 patients (63%) exhibited at least one indirect finding of airway involvement: bronchiectasis, hyperlucency, or atelectasis. The Cohen kappa coefficient for interobserver agreement of airway involvement by peripheral tumors was very good at 1.0. From these data, the CT finding seen with the highest frequency in carcinoid tumors was nodule enhancement of greater than 15 HU. Nuclear Medicine Fifteen patients underwent 16 nuclear imaging studies before resection, with one patient having two studies. Twelve studies were 111 In-octreotide scans and four studies were FDG PET/CT scans. Two of the four FDG PET scans (50%) showed positive findings according to finalized nuclear reports. The average lesion size based on the PET studies was 16.75 mm (range, 12 25 mm). The mean lesion diameter for the two positive PET scans was 18.5 mm, whereas the mean lesion size for the two negative PET scans was 15 mm. Five of 12 octreotide scans (42%) were positive. The mean lesion diameter for all the octreotide scans was 20.6 mm (range, 10 51 mm). The mean lesion diameter for the positive octreotide scans was 24.8 mm (range, 12 51 mm), whereas the mean diameter for negative octreotide scans was 17.7 mm (range, 10 28 mm). Six octreotide scans and three PET scans were obtained of peripheral tumors. Two of six octreotide scans (33%) and one of three PET scans (33%) were positive. One PET scan and two octreotide scans were obtained of three peripheral atypical carcinoids. All three studies were negative. There were no nuclear medicine studies performed of central atypical carcinoids. Presentation Clinical history to assess for the presence of symptoms at presentation was available for 27 patients; one patient s presentation was not documented. Fourteen of the 27 patients (52%) presented with symptoms, whereas 13 (48%) were asymptomatic. Of the 14 symptomatic patients, five (36%) presented with an acute or chronic cough, two (14%) with recurrent pneumonia, two (14%) with hemoptysis, two (14%) with wheezing, two (14%) with pleuritic chest pain, and one (7%) with Cushing disease subsequently proved to be secondary to ectopic ACTH production by a middle lobe carcinoid tumor. Of the asymptomatic group, seven patients had an incidental SPN found on chest radiography, three had nodules incidentally discovered on CT scans being obtained for other indications, and three had nodules discovered on pulmonary CT angiography performed for evaluation of acute symptoms thought to be unrelated to their lung nodules subsequently shown to reflect carcinoid tumor. One of 11 central tumors (9%) presented asymptomatically. Thirteen of 16 peripheral tumors (81%) presented asymptomatically. Symptoms in the three patients with peripheral tumors were cough, Cushing disease, and pleuritic chest pain in one patient each. Discussion Most of the larger studies of pulmonary carcinoid tumors have found that 85% of lesions arise within the central airways as endobronchial masses [1, 4 6]. For this reason, the most common presenting symptoms include cough, shortness of breath, wheezing, hemoptysis, and postobstructive pneumonitis. Given that these tumors typically come to clinical attention because of symptoms associated with central airway obstruction, it is not surprising that the majority of published series have found a marked central predilection for these tumors. Our study found a majority 1076 AJR:197, November 2011

CT of Pulmonary Carcinoid Tumors TABLE 2: CT Characteristics Patient No. Bronchus Affected (57%) of pulmonary carcinoid tumors arising distal to a central airway (i.e., associated with or distal to a subsegmental airway) on thin-section CT analysis. Aside from an article by Magid et al. [10] in 1989 describing a small cohort of five central and seven peripheral carcinoid tumors, articles in the literature about carcinoid tumors exhibiting a peripheral tendency are scarce. The peripheral distribution of the 16 tumors in this study likely accounts for our finding that only 14 of 27 patients (52%) had symptoms directly related to the presence of the tumor. Of the patients with peripheral lesions, Lesion Size (mm) Contour Calcification Density (HU) a Enhancement (HU) Bronchiectasis Hyperlucency Opacity 1 Lobar 17 Macrolobulated Unknown Yes 2 None b 23 Microlobulated 80 3 Subsegmental b 15 Macrolobulated 40 Yes 4 Subsegmental b 19 Macrolobulated 50 5 Lobar 17 Macrolobulated 84 Yes Yes 6 Lobar 29 Smooth Calcification 68 Yes 7 Left main 17 Macrolobulated 100 Yes 8 Subsegmental b 17 Macrolobulated 46 Yes 9 Lobar 51 Unknown 98 Yes Yes 10 Left main 10 Macrolobulated 75 Yes 11 Subsegmental b 12 Macrolobulated 76 68 (8) Yes 12 Subsegmental b 24 Macrolobulated Calcification Unknown c Yes 13 Subsegmental b 12 Macrolobulated 78 73 (5) 14 Lobar 16 Macrolobulated 71 Yes 15 Segmental 25 Macrolobulated Calcification 60 43, 17 d Yes Yes 16 Subsegmental b 12 Macrolobulated 46 43, 3 d Yes 17 None b 15 Macrolobulated 46 c 18 Subsegmental b 13 Unknown 60 c Yes Yes 19 Subsegmental b 15 Macrolobulated 70 Yes 20 Segmental 25 Macrolobulated 54 Yes 21 Subsegmental b 28 Macrolobulated 87 22 Subsegmental b 11 Macrolobulated 41 Yes Yes 23 Left main 13 Smooth 16 Yes 24 Segmental 13 Smooth 116 c Yes 25 Lobar 21 Macrolobulated 90 Yes 26 Subsegmental b 9 Smooth 57 34 (23) 27 Subsegmental b 14 Macrolobulated 67 45 (22) Yes 28 None b 14 Macrolobulated 96 70 (26) Yes Note Dash ( ) indicates not present. Data in parentheses are unenhanced values in HU. a Density measurements were obtained after IV contrast administration unless noted otherwise. b Peripheral carcinoid tumor. c Density without IV contrast material (unenhanced studies). d Enhancement determined from two separate CT scans. 13 of 16 lesions (82%) were detected incidentally as SPNs in patients without symptoms related to the tumor itself. Indeed, three of the patients in this series with SPNs eventually found to be carcinoid tumor had undergone pulmonary CT angiography for acute chest symptoms that resolved spontaneously. These findings are almost certainly the result of the widespread availability and use of MDCT for the evaluation of chest disease, which is more sensitive for the detection of pulmonary nodules than conventional or single-detector CT. The margin of an SPN is one of numerous features that is predictive of benign or malignant characteristics. A lobulated or spiculated margin of an SPN is associated with an increased likelihood of malignancy. For this reason, the lobulated or spiculated margin of a solitary nodule as seen on CT has been incorporated into a number of artificial neural networks that attempt to use thin-section CT characteristics to improve radiologists abilities to distinguish benign from malignant nodules [11]. The retrospective thin-section dataset available from all volumetric thoracic studies performed on MDCT scanners with 16 or more detectors provides high-resolution evaluation of the margins of lesions and AJR:197, November 2011 1077

Meisinger et al. A Fig. 1 55-year-old woman with peripheral typical carcinoid tumor at lobectomy. CT contrast enhancement of right lower lobe nodule. A, Unenhanced CT scan shows density of nodule (arrow) is 5 HU. B, CT scan obtained 60 seconds after contrast administration shows density of nodule is 78 HU. A Fig. 2 78-year-old woman with carcinoid tumor and associated Cushing disease. A, Sagittal CT scan obtained at lung windows through level of medial segment middle lobe bronchus shows lobulated nodule is obstructing subsegmental bronchus (arrow). B, Sagittal CT scan obtained at lung windows medial to A shows dilated mucus-filled bronchus (arrow) distal to lesion. B B their relationship to the small bronchi, thereby helping to distinguish endobronchial lesions from benign SPNs, such as granulomas and hamartomas, that typically bear no relationship to visible airways. In our study, the majority of patients (22/28 of all tumors [79%], 14/16 of peripheral tumors [88%]) were found to have lesions with a lobulated margin. Although benign lesions in particular, hamartomas can show a lobulated contour, recognition of a lobulated margin of an SPN should raise the possibility of a carcinoid tumor especially if the lesion has high attenuation, enhances significantly on contrastenhanced CT, and shows direct or indirect airway involvement on thin-section analysis. Contrast-enhanced MDCT can provide information about the attenuation of solitary lung lesions. Studies performed as dedicated dynamic nodule enhancement examinations or using contrast-enhanced dual-energy CT techniques can assess the enhancement of SPNs; information about enhancement can prove useful in distinguishing benign from potentially malignant SPNs [12, 13]. In our study, most patients (24/28, [86%]) underwent either contrast-enhanced CT or both unenhanced and contrast-enhanced CT studies before diagnosis. The mean attenuation value of the 24 lesions as measured on contrast-enhanced CT was 64.6 HU. This value clearly reflects the hypervascularity seen pathologically in these lesions, even those presenting as incidental pulmonary nodules. For the seven patients that had lesion attenuation measured on both unenhanced and contrastenhanced CT studies, including five patients with dedicated CT nodule enhancement studies, the mean enhancement from baseline was 53.4 HU. The mean enhancement for the six patients who had peripheral lesions was 56 HU. These attenuation values confirm that these lesions show intense enhancement in distinction to benign SPNs, which show no or minimal enhancement on contrast CT densitometry [12]. Therefore, the results of our study indicate that the presence of a high-attenuation or densely enhancing lobulated lesion should suggest the diagnosis of carcinoid tumor and, when recognized, should prompt further evaluation. The presence of calcification was noted in only three of 28 patients (11%), which is lower than the published incidence of calcification in carcinoid tumors [9]; this discrepancy is likely explained by the small mean size (17.7 mm) and peripheral distribution of the majority of the lesions in this study. The endobronchial origin of central (i.e., main, lobar, or segmental bronchial) carcinoid tumors is readily evident on CT and fiberoptic bronchoscopy. In addition to visualization of the endoluminal component of these lesions, radiographic and CT findings that are associated with central airway obstruction as a result of carcinoid tumor include bronchiectasis peripheral to the lesion, hyperlucency of distal lung, and distal atelectasis or consolidation. Involvement of a subsegmental or smaller airway by a peripheral carcinoid tumor is difficult to detect radiologically. The thin-section technique inherent to MDCT of the chest allows a detailed analysis of the smaller airways, thereby allowing detection of both direct airway involvement and of indirect findings of peripheral airway obstruction. The presence of first-order subsegmental airway endoluminal involvement was evident on thin-section analysis in 13 of 16 patients (81%) with peripheral lesions. Additionally, we found significant percentages of peripheral bronchiectasis (25%), distal hyperlucency (31%), distal opacity or atelectasis (19%), or some combination of these findings in 63% of patients with peripheral tumors. Thus, the majority of patients with peripheral carcinoid tumors will have direct and indirect findings of airway involvement on thin-section analy- 1078 AJR:197, November 2011

CT of Pulmonary Carcinoid Tumors A B Fig. 3 53-year-old man with subsegmental bronchus carcinoid tumor and peripheral hyperlucency. A, Axial CT scan obtained at lung windows shows lobulated nodule (straight arrow) is obstructing subsegmental branch of anteromedial basal segmental bronchus (curved arrow). Hyperlucency (asterisk) is seen peripheral to lesion. B, Coronal CT reconstruction at lung windows shows tumor (arrow) with peripheral subsegmental hyperlucency. Lobectomy showed 14 8 8 mm typical carcinoid tumor. sis that should aid in their diagnosis and distinction from benign SPNs. Furthermore, the measured interobserver agreement for the presence of CT findings of direct or indirect airway involvement for the 16 patients in this study was perfect, suggesting that CT detection of airway involvement is reproducible. Another conclusion of our study is that CTguided transthoracic needle biopsy is useful for diagnosing carcinoid tumor. The use of transthoracic needle biopsy for diagnosis in our series was the direct result of the relatively large percentage of peripheral lesions identified on CT that showed features suggestive of malignancy. This accounts for the large percentage of patients (50%, 14/28) for whom CT-guided transthoracic needle biopsy was used as the diagnostic technique to provide cytologic evidence of carcinoid or atypical carcinoid tumors. The same features used to diagnose carcinoid and atypical carcinoid tumors in biopsy and resection specimens are also applied to cytologic specimens. In cytologic specimens, atypical carcinoid tumors sometimes exhibit slightly larger cells with somewhat larger and more hyperchromatic nuclei than typical carcinoid tumors. However, it is the presence of foci of necrosis or more than 2 mitoses per 10 high-power field coupled with the absence of features of small cell carcinoma, including marked pleomorphism, prominent nuclear molding, and numerous mitoses, that is most helpful in the diagnosis of atypical carcinoid tumors in cytologic specimens [14]. Although surgical resection remains the treatment of choice for patients with carcinoid tumors, accurate preoperative diagnosis is important in guiding appropriate staging studies and determining the surgical approach to treatment. Given that carcinoid tumor was the primary diagnostic consideration after CT, a relatively large percentage of patients in our study (43%) underwent octreotide scanning, either for lesion characterization or preoperative nodal staging in those with biopsy-proven carcinoid tumors. Of these patients, 42% had positive findings on In 111 -pentetreotide studies (OctreoScan, Covidien) localized to the lung lesion under evaluation, which is considerably lower than the positivity rate in the literature, perhaps as a result of the relatively small size of the lesions in our study (mean diameter = 20.6 mm). The finding of 50% FDG PET positivity in a small subgroup of patients with carcinoid tumors presenting as SPNs is consistent with the relatively higher false-negative rate of FDG PET for carcinoid tumors (~ 25%) [15] than for malignant SPNs resulting from bronchogenic carcinoma. The conclusions drawn from our study are limited in several respects. The patient population size is small. The relative preponderance of peripheral carcinoid tumors presenting as SPNs is most certainly a result of the detection of small lung lesions by MDCT, which is now widely used for chest evaluation. In support of this explanation is the finding that half of the tumors in our study were detected in patients without symptoms referable to their lesions. Because this study was a retrospective study of patients diagnosed with typical or atypical carcinoid tumors, the true specificity of the CT findings described in this study in particular, the specificity of thin-section CT for airway involvement by peripheral carcinoid tumors presenting as SPNs cannot be accurately assessed. However, other causes of SPNs such as granulomas or hamartomas have rarely been described to involve the peripheral airways pathologically or radiologically, so it is unlikely that these other lesions would present with CT findings similar to those seen in our subset of patients with peripheral carcinoid tumors [16]. Nevertheless, other lesions that are known to arise within bronchi, such as endobronchial hamartomas, salivary gland tumors, granular cell tumors, non small cell carcinoma, and inflammatory myofibroblastic tumors, could present with similar CT findings. However, these tumors are extremely rare relative to carcinoid tumors and therefore it seems unlikely that they contribute significantly to the total number of endobronchial lesions seen clinically. The high level of CT contrast nodule enhancement seen in our subset of seven patients who had both unenhanced and contrast-enhanced scans involved dedicated nodule enhancement examinations in only five of seven patients, although it is well recognized that carcinoid tumors are vascular lesions that enhance with IV contrast administration, a feature that helps distinguish these lesions, particularly those that are PET negative, from benign SPNs. In summary, our study suggests that peripheral carcinoid tumors presenting as lobulated SPNs are more common than previously thought. The identification of high attenuation on contrast-enhanced CT or nodule enhancement on contrast nodule densitometry in association with direct and indirect findings of subsegmental airway obstruction should suggest the diagnosis of a peripheral carcinoid tumor and help distinguish these lesions from benign SPNs. AJR:197, November 2011 1079

Meisinger et al. Acknowledgments We would like to acknowledge the Fletcher Allen Health Care Department of Radiology CT technologists who assisted in the performance of the diagnostic CT studies and CT-guided transthoracic needle biopsies for the patients in this study. References 1. Modlin IM, Lye KD, Kidd M. A 5-decade analysis of 13,715 carcinoid tumors. Cancer 2003; 97: 934 959 2. Skuladottir H, Hirsch FR, Hansen HH, et al. Pulmonary neuroendocrine tumors: incidence and prognosis of histological subtypes a population-based study in Denmark. Lung Cancer 2002; 37: 127 135 3. Gustafsson BI, Kidd M, Chan A, Malfertheiner MV, Modlin IM. Bronchopulmonary neuroendocrine tumors. Cancer 2008; 113:5 21 4. Davila DG, Dunn WF, Tazelaar HD, et al. Bronchial carcinoid tumors. Mayo Clin Proc 1993; 68: 795 803 5. Forster BB, Müller NL, Miller RR, Nelems B, Evans KG. Neuroendocrine carcinomas of the lung: clinical, radiologic, and pathologic correlation. Radiology 1989; 170:441 445 6. Blondal T, Grimelius L, Nou E, Wilander E, Aberg T. Argyrophil carcinoid tumors of the lung. Chest 1980; 78:840 844 7. Fischbach F, Knollmann F, Griesshaber V, Freund T, Akkol E, Felix R. Detection of pulmonary nodules by multislice computed tomography: improved detection rate with reduced slice thickness. Eur Radiol 2003; 13:2378 2383 8. Travis WD, Brambilla E, Muller-Hermelink HK, Harris CC, eds. Pathology and genetics of tumours of the lung, pleura, thymus and heart. In: Travis WD, Brambilla E, Muller-Hermelink HK, Harris CC, eds. World Health Organization classification of tumours. Lyon, France: IARC Press, 2004:19 20 9. Zwiebel BR, Austin JHM, Grimes MM. Bronchial carcinoid tumors: assessment with CT location and intratumoral calcification in 31 patients. Radiology 1991; 179:483 486 10. Magid D, Siegelman SS, Eggleston JC, Fishman EK, Zerhouni EA. Pulmonary carcinoid tumors: CT assessment. J Comput Assist Tomogr 1989; FOR YOUR INFORMATION This article is available for CME credit. See www.arrs.org for more information. 13:244 247 11. Nakamura K, Yoshida H, Engelmann R, et al. Computerized analysis of the likelihood of malignancy in solitary pulmonary nodules by use of artificial neural networks. Radiology 2000; 214: 823 830 12. Swensen SJ, Viggiano RW, Midthun DE, et al. Lung nodule enhancement at CT: multicenter study. Radiology 2000; 214:73 80 13. Coursey CA, Nelson RC, Boll DT, et al. Dual-energy multidetector CT: how does it work, what can it tell us, and when can we use it in abdominopelvic imaging? RadioGraphics 2010; 30:1037 1055 14. Nicholson SA, Ryan MR. A review of cytologic findings in neuroendocrine carcinomas including carcinoid tumors with histologic correlation. Cancer 2000; 90:148 161 15. Daniels CE, Lowe VJ, Aubry MC, Allen MS, Jett JR. The utility of fluorodeoxyglucose positron emission tomography in the evaluation of carcinoid tumors presenting as pulmonary nodules. Chest 2007; 131:255 260 16. Churg AM, Myers JL, Tazelaar HD, Wright JL. Thurlbeck s pathology of the lung, 3rd ed. New York, NY: Thieme Medical Publishers Inc, 2005:496 1080 AJR:197, November 2011