Epidermal Growth Factor Receptor Mutation in Lung Adenocarcinomas: Relationship with CT Characteristics and Histologic Subtypes 1

Transcription

1 Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at Original Research n Thoracic Imaging Epidermal Growth Factor Receptor Mutation in Lung Adenocarcinomas: Relationship with CT Characteristics and Histologic Subtypes 1 Hyun-Ju Lee, MD Young Tae Kim, MD Chang Hyun Kang, MD Binsheng Zhao, DSc Yongqiang Tan, PhD Lawrence H. Schwartz, MD Thorsten Persigehl, MD 2 Yoon Kyung Jeon, MD Doo Hyun Chung, MD 1 From the Departments of Radiology (H.J.L.) and Pathology (Y.K.J., D.H.C.), Seoul National University Hospital, 28 Yeongeon-dong, Chongno-gu, Seoul , Republic of Korea; Department of Thoracic and Cardiovascular Surgery, Seoul National University Hospital, Cancer Research Institute, Xenotransplantation Research Center, Clinical Research Center, Seoul National University College of Medicine, Seoul, Republic of Korea (Y.T.K., C.H.K.); Department of Clinical Radiology, University Hospital Münster, Münster, Germany (T.P.); and Department of Radiology, Columbia University Medical Center, New York, NY (B.Z., Y.T., L.H.S., T.P.). Received November 28, 2011; revision requested February 14, 2012; revision received July 29; accepted September 10; final version accepted January 3, Supported by Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology (2012R1A1A ). Address correspondence to H.J.L. ( leehyunju00@gmail.com). 2 Current address: Department of Radiology, University Hospital Cologne, Cologne, Germany. q RSNA, 2013 Purpose: Materials and Methods: Results: Conclusion: To retrospectively identify quantitative computed tomographic (CT) features that correlate with epidermal growth factor receptor (EGFR) mutation in surgically resected lung adenocarcinomas stratified by the International Association for the Study of Lung Cancer (IASLC), American Thoracic Society (ATS), and European Respiratory Society (ERS) classification in an East Asian cohort of patients known to have a high prevalence of EGFR mutations. An institutional review board approved this study and waived informed consent. In 153 surgically resected lung adenocarcinomas, EGFR mutation was determined by direct DNA sequencing. Histologic subtype was classified according to IASLC/ATS/ERS classification of lung adenocarcinoma. At preoperative chest CT, the percentage of ground-glass opacity (GGO) volume and total tumor volume of each tumor were measured by using a semiautomated algorithm. Distribution of EGFR mutation according to histologic subtype, percentage of GGO volume, and total tumor volume was evaluated by using the Fisher exact test, the Student t test, trend analysis, and multiple logistic regression analysis. Exon 21 missense mutation was more frequent in lepidic predominant adenocarcinomas than in other histologic subtypes (odds ratio, 3.44; 95% confidence interval: 1.53, 7.74; P =.003). GGO volume percentage in tumors with exon 21 missense mutation (61.7% [standard deviation]) was significantly higher than that in EGFR wild-type tumors (30.0% ) (P =.0001) and exon 19 mutated tumors (28.9% ) (P =.0006). A significant trend of prevalence of exon 21 missense mutation increasing along with increasing GGO volume (P =.0008) was found. GGO volume percentage in tumors with exon 21 missense mutation was significantly higher than that in tumors with other EGFR mutation status. This can be related to the fact that exon 21 missense mutation was significantly more frequent in lepidic predominant adenocarcinomas, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma, according to IASLE/ATS/ERS classification. q RSNA, 2013 Supplemental material: /suppl/doi: /radiol /-/dc1 254 radiology.rsna.org n Radiology: Volume 268: Number 1 July 2013

2 Epidermal growth factor receptor (EGFR) mutations are correlated with specific characteristics, such as nonsmoking status, adenocarcinoma histologic findings, female sex, and East Asian ethnicity (1,2). In turn, EGFR mutations are closely associated with a high response rate to treatment with EGFR tyrosine kinase inhibitors (2 4). Recently, the question of whether exon 19 deletion and exon 21 L858R missense mutation, the two most common types of EGFR mutation, exhibit differences in tumor characteristics has been raised (5). On the other hand, EGFR amplification has been reported to occur invariably on mutated and not wild-type allele and to be associated with solid histologic findings Advances in Knowledge nn Exon 21 missense mutation was more frequent in lepidic predominant adenocarcinomas, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma (41.4%, 24 of 58), than in other subtypes of dominant histologic findings (acinar, papillary, micropapillary, and solid predominant as well as invasive mucinous adenocarcinoma) (17.5%, 14 of 58) (odds ratio, 3.44; 95% confidence interval: 1.53, 7.74; P =.003) when adjustment was performed for sex, smoking, and epidermal growth factor receptor (EGFR) amplification. nn Ground-glass opacity (GGO) volume percentage in tumors with exon 21 missense (61.7% ) was significantly higher than that in EGFR wild-type tumors (30.0% ) (P =.0001) and exon 19 mutated tumors (28.9% ) (P =.0006). nn A significant trend of prevalence of exon 21 mutation increasing along with increasing GGO volume percentage (P =.0008) was found. of adenocarcinoma, advanced clinical stage, and significantly worse diseasefree survival (6,7). Lung adenocarcinoma is the most common histologic subtype of lung cancer in most countries (8). In the past decade, many advances have taken place in oncology, molecular biology, pathologic examination, radiology, and surgery of lung adenocarcinoma. With this background, an international multidisciplinary classification system sponsored by the International Association for the Study of Lung Cancer (IASLC), American Thoracic Society (ATS), and European Respiratory Society (ERS) was proposed (9). In this new classification, there are several changes for surgically resected tumors: (a) The term bronchioloalveolar carcinoma is no longer used, and it is now to be referred to as lepidic pattern. (b) The term adenocarcinoma in situ is proposed for small ( 3 cm) solitary adenocarcinomas with pure lepidic growth without invasion. (c) Minimally invasive adenocarcinoma is proposed for small ( 3 cm) lepidic predominant tumors with invasion of 0.5 cm or less. (d) Invasive adenocarcinomas should be classified according to the predominant subtype after semiquantitative assessment, such as acinar predominant, papillary predominant, micropapillary predominant, solid predominant, and lepidic predominant invasive adenocarcinoma. (e) Micropapillary predominant invasive adenocarcinoma is to be added as a major subtype. (f) Former mucinous bronchioloalveolar carcinomas are to be now classified as invasive mucinous adenocarcinomas. To our knowledge, only a few studies have attempted to correlate imaging features with molecular findings. Identification of the relationship between Implication for Patient Care nn Identification of the relationship between CT imaging traits and EGFR molecular status can help to define categories of lung adenocarcinoma that have distinct clinical, radiologic, molecular, and pathologic characteristics. computed tomographic (CT) imaging traits and EGFR molecular status can help to define categories of lung adenocarcinoma that have distinct clinical, radiologic, molecular, and pathologic characteristics. Therefore, the purpose of this study was to retrospectively identify quantitative CT features that correlate with EGFR mutation status in surgically resected lung adenocarcinomas stratified by IASLC/ATS/ERS classification in an East Asian cohort of patients known to have a high prevalence of EGFR mutations. Materials and Methods This study was approved by the Institutional Review Board of Seoul National University Hospital. Informed consents for tissue collection and gene analyses for research purposes were obtained from individual patients before surgery according to the policy of the Lung Cancer Tissue Bank of the Cancer Research Institute of Seoul National University. Patient Selection Among 161 consecutive patients who underwent surgical resection for primary lung adenocarcinoma at Seoul National University Hospital in Korea from Published online before print /radiol Content code: Radiology 2013; 268: Abbreviations: ATS = American Thoracic Society EGFR = epidermal growth factor receptor ERS = European Respiratory Society GGO = ground-glass opacity IASLC = International Association for the Study of Lung Cancer Author contributions: Guarantor of integrity of entire study, H.J.L.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, H.J.L., T.P.; clinical studies, H.J.L., Y.T.K., C.H.K., B.Z., Y.T., L.H.S., T.P., Y.K.J., D.H.C.; statistical analysis, H.J.L.; and manuscript editing, H.J.L., C.H.K., B.Z., L.H.S., T.P. Conflicts of interest are listed at the end of this article. See also the editorial by Galvin and Franks in this issue. Radiology: Volume 268: Number 1 July 2013 n radiology.rsna.org 255

3 October 2007 to October 2008, 153 patients (mean age, 63 years 6 10 [standard deviation]; age range, years; 73 men [mean age, 62 years 6 10; age range, years], 80 women [mean age, 63 years 6 10; age range, years]) were included in this study on the basis of the availability of pathologic records for EGFR mutation and EGFR gene copy number in resected specimens. Lobectomy was performed in 143 patients, pneumonectomy in one, segmentectomy in one, and wedge resection in eight. The final pathologic lung cancer stages according to the seventh edition Union for International Cancer Control and American Joint Committee on Cancer TNM classification (10,11) were stage IA in 91 patients (59.5%), stage IB in 21 (13.7%), stage IIA in eight (5.3%), stage IIB in six (3.9%), and stage IIIA in 27 (17.6%). Histologic Evaluation and Molecular Analysis All resected specimens were formalin fixed and stained with hematoxylin-eosin in accordance with the routine regulations of our hospital. One board-certified pathologist (D.H.C., with 20 years of experience performing pathologic diagnosis of lung cancer) reviewed pathologic specimens and recorded the pathologic subtype of each tumor according to the IASLC/ATS/ERS classification of lung adenocarcinomas (9). Mutation status of EGFR exons 18, 19, 20, and 21 was examined with a polymerase chain reaction based assay and confirmed through direct sequencing as has been previously described (1,12). EGFR gene copy number was analyzed by using fluorescence in situ hybridization (13,14) and was classified into six categories according to the number of EGFR gene copies and chromosome 7 centromeres by using the classification scheme developed at the University of Colorado (6): disomy, low trisomy, high trisomy, low polysomy, high polysomy, and gene amplification. CT Imaging All patients included in this study had undergone chest CT for preoperative staging, which had been conducted within 1 month prior to surgery. CT imaging was performed by using one of four CT systems (LightSpeed Ultra, GE Medical Systems, Milwaukee, Wis; Sensation 16, Siemens Medical Systems, Erlangen, Germany; Brilliance 64 and MX8000, Philips Medical Systems, Best, the Netherlands). CT parameters were as follows: detector collimation, mm; beam pitch, ; rotation time, second; tube voltage, 120 kvp; tube current, ma; and a reconstruction kernel with a high-frequency algorithm. Reconstruction thicknesses and intervals were 1.0 and 1.0 mm in 77 patients, 1.25 and 1.25 mm in 37, 2.5 and 2.5 mm in two, and 5.0 and 5.0 mm in 37. Ninety-eight lesions were examined by using CT after enhancement with intravenous contrast medium. Visual Interpretation of CT Images Two radiologists (H.J.L. and T.P., with 14 and 10 years of experience, respectively, performing chest image interpretations) independently interpreted CT images. Both radiologists were aware that patients had surgically resected lung adenocarcinomas but were unaware of the pathologic reports as well as the results of EGFR status. In terms of morphologic characteristics, the presence or absence of air bronchogram, bubblelike lucency, cavitation, lobulated border, notch, and roundness were assessed. Bubblelike lucency was defined as small spots of air attenuation within lesions. A lobulated border was defined when a portion of the surface of a lesion showed a shallow wavy configuration, except regions abutting the pleura (15,16). A notch was defined as V-shaped indentation of the border deeper than 3 mm (17). The round tumor was defined as the tumor with almost identical maximum and perpendicular diameter without a notch. All tumors were categorized into pure solid nodules or ground-glass opacity (GGO) containing nodules. Differences between two readers in the interpretation of the nodule categorization (eight of 153 nodules, 5.2%), the presence of air bronchogram (six nodules, 3.9%), bubblelike lucency (15 nodules, 9.8%), lobulated border (two nodules, 1.3%), and notch (seven nodules, 4.6%) were resolved by discussion until consensus was reached. Computer-aided Volumetric Measurement Computer-aided volume measurement was performed in two steps for GGOcontaining nodules. For the first step of the volume measurement, the entire tumor mass was separated from surrounding anatomic structures by using a semiautomated segmentation algorithm developed in the Laboratory for Computational Image Analysis in the Department of Radiology, Columbia University Medical Center (18). This algorithm combined the image analysis techniques of watershed transformation and active contours and was devised to automatically extract the bronchial and cavitary structures in the mass. Computer-generated tumor boundaries were then visually inspected by a radiologist (H.J.L.) for correctness and consistency. If any segmentation results were considered suboptimal, tumor contours that were superimposed on the original images were edited by the same radiologist (H.J.L.) with an image-viewing system developed by using Interface Description Language (IDL Solutions, Germantown, Wis), which facilitated both manual and computer-aided measurements. As the second step, segmentation of the inner solid portion was performed inside the tumor boundary obtained at the first step for the entire tumor mass (Fig 1). A Gaussian mixture model was used to estimate the opacity distributions of solid and GGO portions. The solid portion was then segmented by using a Markov random field model on the basis of the opacity distribution. Once the segmentation and manual correction (if needed) were completed, the volumes of the entire mass and GGO portion were automatically calculated by using the computer algorithm. Tumor volume was defined as the sum of all tumor voxels, including voxels on and inside the tumor boundary, multiplied by the image resolutions in the x, y (in-plane), and z directions. In pure solid nodules, only the first step was 256 radiology.rsna.org n Radiology: Volume 268: Number 1 July 2013

4 Figure 1 Figure 1: EGFR mutation status, EGFR copy number abnormalities, and histologic subtypes. Each column is a lung adenocarcinoma sample. EGFR mutations were categorized as exon 19 deletion, exon 21 missense mutation, exon 18 mutation, exon 20 mutation, or EGFR wild type (no mutation). EGFR copy number was categorized into disomy, trisomy, low polysomy, high polysomy, and gene amplification. Histologic subtypes were classified according to IASLC/ATS/ERS classification. ADC = adenocarcinoma, AIS = adenocarcinoma in situ, LPIA = lepidic predominant invasive adenocarcinoma, MIA = minimally invasive adenocarcinoma, pred = predominant. performed. Estimated tumor diameter was calculated under the assumption that each tumor was of spherical shape (V = 4/3p [d/2] 3 ), where V is volume and d is diameter (19). To study the intra- and interobserver agreements, two radiologists (H.J.L. and T.P.) performed the same measurement on 47 randomly selected tumors. Statistical Analysis Inter- and intraobserver agreements were assessed by using 95% Bland-Altman limits of agreement and intraclass correlation coefficient. An intraclass correlation coefficient greater than 0.75 was considered to represent good agreement. The prevalence of nominal variables was compared by using the Fisher exact test. Difference of mean values between continuous variables was compared by using the Student t test. Trend analysis was performed by using the two-sided x 2 trend test. Multiple logistic regression analysis was performed to test the association between EGFR mutation and histologic subtypes with adjustment of other factors. All statistical analyses were performed by using a commercial software package (SPSS, version 13, SPSS, Chicago, Ill; MedCalc, MedCalc Software, Mariakerke, Belgium). A P value of less than.05 was considered to indicate a significant difference. Results CT Features according to IASLC/ATS/ERS Classification CT features according to IASLC/ATS/ ERS histologic subtypes are presented in Table 1. When tumors were divided into lepidic predominant adenocarcinomas, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma, versus other subtypes of dominant histologic findings (acinar, papillary, micropapillary, and solid predominant as well as invasive mucinous adenocarcinoma), GGO volume percentages were significantly higher in lepidic predominant adenocarcinomas (67.3% ) than those in other subtypes of predominant histologic findings (22.7% ) (P,.0001). Total tumor volume presented as estimated tumor diameters showed no significant difference between two categories of histologic subtypes (P =.145). Air bronchogram was more frequently found in lepidic predominant adenocarcinomas (P,.0001). Round shape was more frequent in lepidic predominant adenocarcinomas (P =.039). EGFR Mutation and Demographics The results of EGFR mutation are summarized in Figure 1. Eighty-three of 153 (54.2%) adenocarcinomas showed EGFR mutation: exon 21 missense in 40 (26.1%), exon 19 deletion in 38 (24.8%), exon 18 deletion or missense mutation in three (2.0%), and exon 20 insertion mutation in two (1.3%). Exon 21 missense mutation was significantly more frequent in women (P =.028) (Table 2). Exon 21 missense mutation was also significantly more frequent in never smokers (P =.010) (Table 3). No difference was found between the ages of patients with exon 21 missense mutation (62.6 years ), exon 19 deletion mutation (61.8 years ), and EGFR wild type (62.9 years ) (P =.522). EGFR Mutation and Histologic Subtypes EGFR gene status according to IASLC/ ATS/ERS histologic subtypes is presented in Table 4. Exon 21 missense Radiology: Volume 268: Number 1 July 2013 n radiology.rsna.org 257

5 Table 1 CT Features according to IASLC/ATS/ERS Histologic Subtypes in Adenocarcinomas Invasive Mucinous AD P Value Solid Predominant Micropapillary Predominant Papillary Predominant Acinar Predominant CT Feature AIS MIA LPIA Proportion of GGO (%)* ,.0001 Estimated tumor diameter (cm)* Morphologic CT feature Air bronchogram 3 (37.5) 8 (72.7) 29 (74.4) 20 (35.1) 4 (50.0) 0 (0.0) 0 (0.0) 4 (100.0),.0001 Bubblelike lucency 2 (25.0) 1 (9.1) 11 (28.2) 14 (24.6) 2 (25.0) 0 (0.0) 3 (33.3) 2 (50.0).469 Cavity 0 (0.0) 0 (0.0) 1 (2.6) 2 (3.5) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0).619 Notch 1 (12.5) 2 (18.2) 26 (66.7) 30 (52.6) 3 (37.5) 1 (50.0) 4 (44.4) 2 (50.0).568 Lobulated border 6 (75.0) 11 (100.0) 39 (100.0) 52 (91.2) 7 (87.5) 1 (50.0) 9 (100.0) 4 (100.0).187 Round 5 (62.5) 2 (18.2) 3 (7.7) 3 (5.3) 1 (12.5) 0 (0.0) 1 (11.1) 0 (0.0).039 Total (n = 138) 8 (100.0) 11 (100.0) 39 (100.0) 57 (100.0) 8 (100.0) 2 (100.0) 9 (100.0) 4 (100.0) Note. Unless otherwise indicated, data are numbers, with percentages in parentheses. Statistical analysis for the proportion of GGO and estimated tumor diameter was performed by using the Student t test. Statistical analysis for the morphologic CT features was performed by using the Fisher exact test. All statistical comparisons were performed under the condition that tumors were divided into lepidic predominant adenocarcinomas (adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma) versus other subtypes of dominant histologic findings (acinar, papillary, micropapillary, and solid predominant invasive adenocarcinomas as well as invasive mucinous adenocarcinoma). AD = adenocarcinoma, AIS = adenocarcinoma in situ, LPIA = lepidic predominant invasive adenocarcinoma, MIA = minimally invasive adenocarcinoma. * Data are means 6 standard deviations. Estimated tumor diameter was calculated from the measured total tumor volume under the assumption that each tumor is of spherical shape (V = 4/3p [d/2] 3 ). In 15 of 153 adenocarcinomas, subtype could not be evaluated because the pathologic slides were not available; therefore, subtypes in 138 tumors are presented. mutation was more frequent in lepidic predominant adenocarcinomas (odds ratio, 3.44; 95% confidence interval: 1.53, 7.74; P =.003) when adjustment was performed for sex, smoking, and EGFR amplification (Table 5). Agreement in Measurement of GGO Volume Percentage and Total Tumor Volume The 95% limits of inter- and intraobserver agreements obtained by using Bland-Altman analysis in the measurement of GGO volume percentage were 216.9% to 20.2% and 225.1% to 23.6%, with intraclass correlation coefficients of (95% confidence interval: 0.962, 0.992) and (95% confidence interval: 0.881, 0.974), respectively (P,.001). The 95% limits of inter- and intraobserver agreements in the measurement of total tumor volume were 221.4% to 23.5% and 215.4% to 17.3%, with intraclass correlation coefficients of (95% confidence interval: 0.995, 0.998) and (95% confidence interval: 0.996, 0.999), respectively (P,.001). In all nodules, the computerized segmentation was successful (Fig 2). In 68 of 153 nodules, tumor contours were partially edited. EGFR Mutation and GGO Volume Percentage GGO volume percentage in tumors with exon 21 missense mutation (61.7% ) was significantly higher than that in EGFR wild-type tumors (30.0% ) (P =.0001) and exon 19 mutated tumors (28.9% ) (P =.0006). The frequencies of GGO greater than 90%, GGO greater than 80% and less than or equal to 90%, GGO greater than 60% and less than or equal to 80%, GGO less than or equal to 60%, and pure solid presentation were 20.0%, 20.0%, 25.0%, 22.5%, and 12.5%, respectively, in adenocarcinomas with exon 21 missense mutation. The frequencies were 5.3%, 7.9%, 21.1%, 5.3%, and 60.5%, respectively, in adenocarcinomas with exon 19 deletion. The frequencies were 8.6%, 7.1%, 20.0%, 4.3%, and 60.0%, respectively, in EGFR wild-type tumors (Fig 3a). A significant trend of prevalence of exon 21 mutation increasing 258 radiology.rsna.org n Radiology: Volume 268: Number 1 July 2013

6 along with increasing GGO volume percentage (P =.0008) was found. No trend was found for exon 19 deletion (P =.168). EGFR Mutation versus Total Tumor Volume and Morphologic CT Features There was no difference in total tumor volume among tumors with exon 21 missense (10.4 cm ), exon 19 deletion (8.72 cm ), and EGFR wild type (13.9 cm ) (P =.466). The frequencies of estimated tumor diameters 1.0 cm or smaller, greater than 1.0 cm and less than or equal to 1.5 cm, greater than 1.5 cm and less than or equal to 2.0 cm, greater than 2.0 cm and less than or equal to 2.5 cm, greater than 2.5 cm and less than or equal to 3.0 cm, and greater than 3.0 cm were 2.5%, 5.0%, 30.0%, 25.0%, 15.0%, and 22.5%, respectively, in adenocarcinomas with exon 21 missense mutation. The frequencies were 0.0%, 13.2%, 21.1%, 23.7%, 23.7%, and 18.4%, respectively, in adenocarcinomas with exon 19 deletion. The frequencies were 2.9%, 15.7%, 20.0%, 28.6%, 11.4%, and 21.4%, respectively, in EGFR wild-type tumors (Fig 3b). Among the CT morphologic characteristics, air bronchogram was more frequent in exon 21 missense mutation (P =.044) (Table E1 [online]). EGFR Amplification versus Clinical, CT, and Pathologic Characteristics The results of EGFR gene copy number analyses and EGFR amplification status are summarized in Figure 1. The prevalence of EGFR amplification was 14.4% (22 of 153). All EGFR amplifications were only found in EGFR-mutated tumors. The prevalence of other nonamplified EGFR copy number categories were disomy in 22.9% (35 of 153), trisomy in 10.5% (16 of 153), low polysomy in 28.7% (44 of 153), and high polysomy in 23.5% (36 of 153). Predilection of EGFR amplification according to sex (P =.822) and smoking status (P =.643) was not found. In tumors with exon 21 missense, EGFR amplification was significantly less frequent in lepidic predominant adenocarcinomas (9.1%) than in other subtypes Table 2 Sex according to EGFR Mutation Subtype Subtype Total (n = 153) Women (n = 80) Men (n = 73) P Value EGFR wild type (42.5) 36 (49.3).028* Exon 21 missense (33.8) 13 (17.8) Exon 19 deletion (20.0) 22 (30.1) Exon 18 deletion or missense 3 2 (2.5) 1 (1.4) Exon 20 insertion 2 1 (1.2) 1 (1.4) Note. Data are numbers, with percentages in parentheses. Statistical comparison was performed by using a Fisher exact test. * P value was based on comparison between adenocarcinomas with EGFR exon 21 missense mutation and adenocarcinomas with other EGFR mutation status, including EGFR exon 19 deletion, EGFR exon 18 and 20 mutation, and EGFR wild type. Table 3 Smoking Status according to EGFR Mutation Subtype Subtype Total (n = 153) Never smoker (n = 88) Smoker (n = 65) P Value EGFR wild type (38.6) 36 (55.4).010* Exon 21 missense (34.1) 10 (15.4) Exon 19 deletion (23.9) 17 (26.2) Exon 18 deletion or missense 3 2 (2.3) 1 (1.5) Exon 20 insertion 2 1 (1.1) 1 (1.5) Note. Data are numbers, with percentages in parentheses. Statistical comparison was performed by using a Fisher exact test. * P value was based on comparison between adenocarcinomas with EGFR exon 21 missense mutation and adenocarcinomas with other EGFR mutation status, including EGFR exon 19 deletion, EGFR exon 18 and 20 mutation, and EGFR wild type. of predominant histologic findings (42.9%) (P =.034). The prevalence of EGFR amplification showed a significant trend in that EGFR amplification was more often found as GGO volume percentage decreased (P =.023) (Fig 4a) and tumor diameter increased (P =.007) (Fig 4b). Among CT morphologic characteristics, notch was more frequent in tumors with EGFR amplification (P =.012) (Table E1 [online]). Discussion Our study showed that in our selected cohort of patients with primary lung adenocarcinomas GGO volume percentages were significantly higher in lepidic predominant adenocarcinomas, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma, than those in other subtypes of predominant histologic findings. EGFR exon 21 missense mutation was more frequent in lepidic predominant adenocarcinomas, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma. GGO volume percentage in tumors with exon 21 missence mutation was significantly higher than that in EGFR wild-type tumors and exon 19 mutated tumors. The prevalence of EGFR exon 21 missense increased along with increasing GGO volume percentage. Contrarily, EGFR amplification was found less often as GGO volume percentage increased. Our study had several interesting features. First, the demographic characteristics of the study population (East Asian origin) favor the presence of somatic EGFR mutations, and indeed we identified EGFR mutations in 54.2% of cases. Exon 21 L858R missense mutation was significantly frequent in women and in never smokers in this study. As published in another study Radiology: Volume 268: Number 1 July 2013 n radiology.rsna.org 259

7 Table 4 EGFR Mutation and Amplification according to IASLC/ATS/ERS Histologic Subtypes in Adenocarcinomas Amplification AIS MIA LPIA Acinar Predominant Papillary Predominant Micropapillary Predominant Solid Predominant Invasive Mucinous AD EGFR amplification negative Exon 21 missense 3 (37.5) 6 (54.5) 13 (33.3) 7 (12.3) 1 (12.5) 0 (0.0) 0 (0.0) 0 (0.0) Exon 19 deletion 1 (12.5) 3 (27.3) 9 (23.1) 10 (17.5) 2 (25.0) 1 (50.0) 0 (0.0) 0 (0.0) Exon 18 deletion or missense 0 (0.0) 0 (0.0) 0 (0.0) 1 (1.8) 1 (12.5) 0 (0.0) 0 (0.0) 0 (0.0) Exon 20 insertion 0 (0.0) 0 (0.0) 0 (0.0) 1 (1.8) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) EGFR wild type 4 (50.0) 2 (18.2) 10 (25.6) 27 (47.3) 4 (50.0) 1 (50.0) 8 (88.9) 4 (100) EGFR amplification positive Exon 21 missense 0 (0.0) 0 (0.0) 2 (5.1) 6 (10.5) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) Exon 19 deletion 0 (0.0) 0 (0.0) 5 (12.8) 4 (7.0) 0 (0.0) 0 (0.0) 1 (11.1) 0 (0.0) Exon 18 deletion or missense 0 (0.0) 0 (0.0) 0 (0.0) 1 (1.8) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) Exon 20 insertion 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) EGFR wild type 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) Total (n = 138)* 8 (100) 11 (100) 39 (100) 57 (100) 8 (100) 2 (100) 9 (100) 4 (100) Note. Data are numbers, with percentages in parentheses. AD = adenocarcinoma, AIS = adenocarcinoma in situ, LPIA = lepidic predominant invasive adenocarcinoma, MIA = minimally invasive adenocarcinoma. * In 15 of 153 adenocarcinomas, subtype was not evaluated because the pathologic slides were not available; therefore, subtypes in 138 tumors are presented. Table 5 Relationship among EGFR Exon 21 Missense Mutation, Histologic Subtypes, Sex, and Smoking Variable Odds Ratio* P Value Lepidic dominant histologic subtype (AIS, MIA, and LPIA) 3.44 (1.53, 7.74).003 Sex (women) 1.19 (0.34, 4.15).788 Smoking (never smoker) 2.12 (0.57, 7.87).259 EGFR amplification 0.45 (0.15, 1.31).143 Note. AIS = adenocarcinoma in situ, LPIA = lepidic predominant invasive adenocarcinoma, MIA = minimally invasive adenocarcinoma. * Data in parentheses are 95% confidence intervals. IASLC/ATS/ERS histologic subtypes were divided into lepidic predominant adenocarcinomas (adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma) versus other subtypes of dominant histologic findings (acinar, papillary, micropapillary, and solid predominant invasive adenocarcinomas as well as invasive mucinous adenocarcinoma) for statistical analysis. with an East Asian population (7), all EGFR amplifications were only found in EGFR-mutated tumors. The ethnic homogeneity of the study population enabled us to identify highly significant associations between genetic, clinical, pathologic, and radiologic features in the absence of a larger population analysis. Second, we studied adenocarcinomas that were candidates for curative surgical resection among patients who did not undergo preoperative neoadjuvant treatment. Although EGFR mutation and EGFR amplification have generated considerable interest because both have been reported to be associated with increased sensitivity to EGFR targeted therapy (1,2,6,20 23), the goal of this study was to identify the relationship between CT imaging traits and EGFR molecular status. Surgically resected adenocarcinomas in preadvanced stages can be more optimal for the investigation of the relationship between EGFR gene status and clinical, radiologic, and pathologic features because the original radiologic features and the original biologic characteristics of the cancer could be preserved. Our study demonstrated that EGFR exon 21 mutations were significantly more common in lepidic predominant adenocarcinomas, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma. This can be supported by the gene expression profiling studies using microarray technologies in which substantially higher frequency of EGFR mutation was observed in the terminal respiratory unit type adenocarcinomas (24). Gene expression profiling revealed two major subtypes consisted of terminal respiratory unit type adenocarcinomas, which are based on lepidic pattern and expression of thyroid transcription factor 1 and surfactant proteins, and non terminal respiratory unit adenocarcinomas that lack these characteristics. Functional annotation showed retention of normal peripheral differentiated lung features in the terminal respiratory unit types, which contrasted with the cell cycling and proliferation-enriched 260 radiology.rsna.org n Radiology: Volume 268: Number 1 July 2013

8 Figure 2 Figure 2: CT images show semiautomated segmentation and quantification of GGO volume percentage in a lepidic predominant invasive adenocarcinoma presented as a GGO-containing nodule in 68-year-old woman. Left: Representative image for the segmentation of solid portion and ground-glass portion. Automatically segmented tumor outer contour (yellow line) and contour of the inner solid portion (red line) are overlapped on each transverse image. Bronchial structures (blue line) were automatically extracted from the calculation of tumor volume. Total tumor volume and GGO volume percentage of this tumor were 7.88 cm 3 and 79.6%, respectively. annotation of genes associated with the non terminal respiratory unit adenocarcinomas (25,26). Terminal respiratory unit type adenocarcinomas can be classified as lepidic predominant adenocarcinomas in this study (9). Several other reports also support the presence of this relationship; one such report from Taiwan (27) indicated that 66% of adenocarcinomas with a lepidic component showed EGFR mutation compared with 21% of adenocarcinomas without a lepidic component showing EGFR mutation. Another report from Japan (28) demonstrated that EGFR mutations were observed in 58% of adenocarcinomas with a lepidic component compared with 37% of adenocarcinomas without a lepidic component. Only few reports have evaluated the association between EGFR mutation status and imaging findings of lung adenocarcinoma before an advanced stage (29,30). Among them, our results are concordant with those of the investigation performed in a Japanese population where GGO was more frequent in tumors with EGFR mutation (73.7%) than in tumors with EGFR wild type (57.1%) (30). Our results differed from the study performed in the United States which concluded that they found no CT characteristics could help suggest the status of EGFR mutation (29). The difference of inclusion criteria may be the origin of the different results; they included only adenocarcinomas with bronchioloalveolar carcinoma features; thus, other subtypes of predominant histologic findings that showed lower frequency of EGFR mutation might have been excluded in their study and the statistical difference reduced (29). In previous studies, investigators had only visually inspected the presence or absence of GGO in the mass and did not perform a more refined quantification of GGO or present the histologic subtypes and EGFR mutation subtypes of their study population. In regard to EGFR amplification, it has already been reported in another study that EGFR amplifications are reversely correlated with the GGO percentage of primary adenocarcinomas (31), and our results confirmed this reversed correlation. Radiology: Volume 268: Number 1 July 2013 n radiology.rsna.org 261

9 Figure 3 Figure 3: Distribution of GGO volume percentages and estimated tumor diameters according to specific types of EGFR mutation. (a) Graph shows distribution of lung adenocarcinomas with different GGO volume percentages (GGO. 90%, 90% GGO. 80%, 80% GGO. 60%, 60% GGO, or a pure solid nodule) according to EGFR mutation status. In adenocarcinomas with the exon 21 missense mutation (E21 mis), GGO-containing nodules were relatively more frequent than in other types of EGFR mutation. The distribution of tumors according to GGO volume percentage for exon 19 deletion (E19 del) was similar to those in tumors with EGFR wild type. (b) Graph shows distribution of estimated tumor diameters according to EGFR mutation status. Among tumors with the two most common types of EGFR mutation, exon 21 missense and exon 19 deletion, and EGFR wild type, frequencies of estimated tumor diameters were not different. D = estimated tumor diameter calculated from measured total tumor volumes under the assumption that each tumor is of spherical shape (V = 4/3p [d/2] 3 ). E18 or E20 mutation = EGFR exon 18 or exon 20 mutation. Figure 4 Figure 4: Distribution of GGO volume percentages and estimated tumor diameters according to EGFR amplification. (a) Graph shows that EGFR amplification is more frequent as GGO volume percentage decreases (P =.023). (b) Graph shows that EGFR amplification is more often found as tumor diameter increases (P =.007). D = estimated tumor diameter calculated from measured total tumor volumes under the assumption that each tumor is of spherical shape. 262 radiology.rsna.org n Radiology: Volume 268: Number 1 July 2013

10 This study also demonstrated the correlation between GGO proportion and exon 21 L858 mutation and the reversed correlation between GGO proportion and EGFR amplification. In the aspect of patient prognosis, this may be meaningful because the high proportion of GGOs in lung adenocarcinomas is well known to be an indicator of better prognosis (32) and EGFR amplification is known to be associated with more aggressive lung carcinomas and worsened prognosis (7,33,34). Moreover, according to a report (5), patients with the exon 21 L858R mutation had better survival than those with EGFR wild type, although the difference was not statistically significant. In this regard, it gives more weight to the belief that radiologic features are closely related to the genetic background and clinical prognosis of lung cancer. The major strength of this study was the calculation of the volume of the GGO component within the tumors. No study has correlated the volumetric proportion occupied by the GGO component with EGFR mutation status. Until now, most studies have used the maximum cross-sectional diameter on CT images to calculate the GGO component (32,35,36). In addition, although many studies were performed by using software for volumetric nodule segmentation, the classic technique for nodule segmentation does not enable the segmentation of tumors that contain a GGO component. In this study, we obtained segmentation results in solid nodules and GGO-containing nodules with good intra- and interobserver agreement. The effect of a single pixel width over the entire surface of a nodule can have a dramatic effect on volume estimates, particularly for smaller nodules and larger pixels (19). In this regard, thin-section CT is highly recommended for volume measurement of lung nodules. Our study had several limitations. First, the sample size was not relatively large and not predetermined in dimension with a power analysis. Second, in this study, the sample size of histologic subtypes was not evenly distributed. Lepidic predominant invasive adenocarcinoma and acinar predominant invasive adenocarcinoma were dominant and other subtypes were few. This may be a limitation in accurately evaluating the correlation between EGFR mutation and each histologic subtype. In conclusion, GGO volume percentage in tumors with exon 21 missense mutation was significantly higher than that in EGFR wild-type tumors and exon 19 mutated tumors. This can be related to the fact that exon 21 missense mutation was significantly more frequent in lepidic predominant adenocarcinomas, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma, according to IASLE/ ATS/ERS classification. Disclosures of Conflicts of Interest: H.J.L. No relevant conflicts of interest to disclose. Y.T.K. No relevant conflicts of interest to disclose. C.H.K. No relevant conflicts of interest to disclose. B.Z. No relevant conflicts of interest to disclose. Y.T. No relevant conflicts of interest to disclose. L.H.S. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: author is consultant for GlaxoSmithKline and Novartis. Other relationships: none to disclose. T.P. No relevant conflicts of interest to disclose. Y.K.J. No relevant conflicts of interest to disclose. D.H.C. No relevant conflicts of interest to disclose. References 1. Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. N Engl J Med 2004;350(21): Paez JG, Jänne PA, Lee JC, et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science 2004;304(5676): Mok TS, Wu YL, Thongprasert S, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361(10): Maemondo M, Inoue A, Kobayashi K, et al. Gefitinib or chemotherapy for non-small-cell lung cancer with mutated EGFR. N Engl J Med 2010;362(25): Shigematsu H, Lin L, Takahashi T, et al. Clinical and biological features associated with epidermal growth factor receptor gene mutations in lung cancers. J Natl Cancer Inst 2005;97(5): Cappuzzo F, Hirsch FR, Rossi E, et al. Epidermal growth factor receptor gene and protein and gefitinib sensitivity in nonsmall-cell lung cancer. J Natl Cancer Inst 2005;97(9): Sholl LM, Yeap BY, Iafrate AJ, et al. Lung adenocarcinoma with EGFR amplification has distinct clinicopathologic and molecular features in never-smokers. Cancer Res 2009;69(21): Curado MP, Edwards B, Shin HR, et al. Cancer incidence in five continents, vol 9. Lyon, France: IARC Scientific Publications, Travis WD, Brambilla E, Noguchi M, et al. International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol 2011;6(2): Goldstraw P. IASLC staging manual in thoracic oncology. International Association for the Study of Lung Cancer. Orange Park, Fla: Editorial Rx Press, Rusch VW, Appleman HD, Blackstone E, et al. Lung. In: Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. AJCC cancer staging manual. American Joint Commission on Cancer. 7th ed. Chicago, Ill: Springer, Yi HG, Kim HJ, Kim YJ, et al. Epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) are effective for leptomeningeal metastasis from non-small cell lung cancer patients with sensitive EGFR mutation or other predictive factors of good response for EGFR TKI. Lung Cancer 2009;65(1): Hirsch FR, Varella-Garcia M, Bunn PA Jr, et al. Epidermal growth factor receptor in non-small-cell lung carcinomas: correlation between gene copy number and protein expression and impact on prognosis. J Clin Oncol 2003;21(20): Jeon YK, Sung SW, Chung JH, et al. Clinicopathologic features and prognostic implications of epidermal growth factor receptor (EGFR) gene copy number and protein expression in non-small cell lung cancer. Lung Cancer 2006;54(3): Ikehara M, Saito H, Kondo T, et al. Comparison of thin-section CT and pathological findings in small solid-density type pulmonary adenocarcinoma: prognostic factors from CT findings. Eur J Radiol 2012;81(1): Hansell DM, Bankier AA, MacMahon H, McLoud TC, Müller NL, Remy J. Fleischner Society: glossary of terms for thoracic imaging. Radiology 2008;246(3): Radiology: Volume 268: Number 1 July 2013 n radiology.rsna.org 263

11 17. Lee HJ, Goo JM, Lee CH, et al. Predictive CT findings of malignancy in ground-glass nodules on thin-section chest CT: the effects on radiologist performance. Eur Radiol 2009;19(3): Zhao B. Automatic segmentation of lung tumors in CT images. Invention Report New York, NY: Columbia University, Zhang L, Yankelevitz DF, Henschke CI, Jirapatnakul AC, Reeves AP, Carter D. Zone of transition: a potential source of error in tumor volume estimation. Radiology 2010;256(2): Hirsch FR, Varella-Garcia M, McCoy J, et al. Increased epidermal growth factor receptor gene copy number detected by fluorescence in situ hybridization associates with increased sensitivity to gefitinib in patients with bronchioloalveolar carcinoma subtypes: a Southwest Oncology Group Study. J Clin Oncol 2005;23(28): Takano T, Ohe Y, Sakamoto H, et al. Epidermal growth factor receptor gene mutations and increased copy numbers predict gefitinib sensitivity in patients with recurrent non-small-cell lung cancer. J Clin Oncol 2005;23(28): Hirsch FR, Varella-Garcia M, Bunn PA Jr, et al. Molecular predictors of outcome with gefitinib in a phase III placebo-controlled study in advanced non-small-cell lung cancer. J Clin Oncol 2006;24(31): Tsao MS, Sakurada A, Cutz JC, et al. Erlotinib in lung cancer: molecular and clinical predictors of outcome. N Engl J Med 2005;353(2): Takeuchi T, Tomida S, Yatabe Y, et al. Expression profile-defined classification of lung adenocarcinoma shows close relationship with underlying major genetic changes and clinicopathologic behaviors. J Clin Oncol 2006;24(11): Yatabe Y, Kosaka T, Takahashi T, Mitsudomi T. EGFR mutation is specific for terminal respiratory unit type adenocarcinoma. Am J Surg Pathol 2005;29(5): Yatabe Y, Mitsudomi T, Takahashi T. TTF-1 expression in pulmonary adenocarcinomas. Am J Surg Pathol 2002;26(6): Hsieh RK, Lim KH, Kuo HT, Tzen CY, Huang MJ. Female sex and bronchioloalveolar pathologic subtype predict EGFR mutations in non-small cell lung cancer. Chest 2005;128(1): Haneda H, Sasaki H, Lindeman N, et al. A correlation between EGFR gene mutation status and bronchioloalveolar carcinoma features in Japanese patients with adenocarcinoma. Jpn J Clin Oncol 2006;36(2): Glynn C, Zakowski MF, Ginsberg MS. Are there imaging characteristics associated with epidermal growth factor receptor and KRAS mutations in patients with adenocarcinoma of the lung with bronchioloalveolar features? J Thorac Oncol 2010;5(3): Yano M, Sasaki H, Kobayashi Y, et al. Epidermal growth factor receptor gene mutation and computed tomographic findings in peripheral pulmonary adenocarcinoma. J Thorac Oncol 2006;1(5): Park EA, Lee HJ, Kim YT, et al. EGFR gene copy number in adenocarcinoma of the lung by FISH analysis: investigation of significantly related factors on CT, FDG- PET, and histopathology. Lung Cancer 2009;64(2): Kim EA, Johkoh T, Lee KS, et al. Quantification of ground-glass opacity on highresolution CT of small peripheral adenocarcinoma of the lung: pathologic and prognostic implications. AJR Am J Roentgenol 2001;177(6): Soh J, Toyooka S, Ichihara S, et al. Sequential molecular changes during multistage pathogenesis of small peripheral adenocarcinomas of the lung. J Thorac Oncol 2008;3(4): Yatabe Y, Takahashi T, Mitsudomi T. Epidermal growth factor receptor gene amplification is acquired in association with tumor progression of EGFR-mutated lung cancer. Cancer Res 2008;68(7): Ohde Y, Nagai K, Yoshida J, et al. The proportion of consolidation to ground-glass opacity on high resolution CT is a good predictor for distinguishing the population of non-invasive peripheral adenocarcinoma. Lung Cancer 2003;42(3): Aoki T, Tomoda Y, Watanabe H, et al. Peripheral lung adenocarcinoma: correlation of thin-section CT findings with histologic prognostic factors and survival. Radiology 2001;220(3): radiology.rsna.org n Radiology: Volume 268: Number 1 July 2013