CT findings of minimally invasive adenocarcinoma (MIA) of the lung and comparison of solid portion measurement methods at CT in 52 patients Sang Min Lee & Jin Mo Goo & Kyung Hee Lee & Doo Hyun Chung & Jaemoon Koh & Chang Min Park
Received: 7 August 2014 / Revised: 5 November 2014 / Accepted: 20 January 2015 # European Society of Radiology 2015
Abstract Objectives We aimed to retrospectively investigate CT findings of minimally invasive adenocarcinoma (MIA) and to determine the appropriate method for measurement of solid portions in MIAs at CT. Methods From May 2012 to April 2014, 55 pulmonary nodules in 52 patients were pathologically confirmed as MIAs and were included in this study. CT findings of MIAs and measurements of solid portions at CT were evaluated by two independent radiologists. Results Mean size of MIAs was 10.5 mm±4.8 (range, 4– 28 mm). Fifty-two MIAs manifested as 28 pure ground glass nodules (GGNs) (53.8 %%), 22 part-solid GGNs (42.3 %%), and 2 two solid nodules (3.8 %%) at CT. Lobulated border, bubble lucency, and pleural retraction were frequently found in both observers (26.9–42.3 %%). Differences according to window settings between solid portion size and invasive component size were not significantly different in both observers (p>0.05). As for interobserver agreement, 95 % CIs for solid portion size in the mediastinal window setting (-2.2 to 3.4; mean, 0.6) were slightly narrower than those in the lung window setting (-2.6 to 3.1; mean, 0.3).
S. M. Lee : J. M. Goo (*) : K. H. Lee : C. M. Park Department of Radiology, Seoul National University College of Medicine, and Institute of Radiation Medicine, Seoul National University Medical Research Center, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Korea e-mail: [email protected] J. M. Goo Cancer Research Institute, Seoul National University College of Medicine, Seoul, Korea D. H. Chung : J. Koh Department of Pathology, Seoul National University College of Medicine, Seoul, Korea
Conclusions Nearly all MIAs appear as pure and part-solid GGNs. Mediastinal and lung window settings can be applied for measurement of solid portions at CT without a significant difference. Key Points • Nearly all MIAs appear as pure and part-solid GGNs. • MIAs show frequent interval growth at follow-up. • MIAs with solid portion ≥5 mm ranged from 7.7 % to 19.2 %. • Mediastinal and lung window settings can be applied for solid portion measurement. Keywords Minimally invasive adenocarcinoma . CT . Solid portion measurement . Invasive component . Window settings Abbreviations MIA minimally invasive adenocarcinoma GGN ground glass nodule IASLC International Association for the Study of Lung Cancer ATS American Thoracic Society ERS European Respiratory Society AAH atypical adenomatous hyperplasia AIS adenocarcinoma in situ
Introduction Minimally invasive adenocarcinoma (MIA) is a recently introduced term in the new classification of lung adenocarcinoma, proposed by the International Association for the Study of Lung Cancer, American Thoracic Society, and European Respiratory Society (IASLC/ATS/ERS) . In this classification, MIA is defined as a small, solitary adenocarcinoma (≤3 cm)
with a predominantly lepidic pattern with invasion (≤5 mm) in its greatest dimension . In terms of computed tomography (CT) findings of MIA, MIA can present as a part-solid ground-glass nodule (GGN) consisting of predominantly ground-glass opacity (GGO) and a small solid portion measuring 5 mm or less at CT [2, 3]. These CT findings are based on CT-pathologic correlation of GGNs, in which GGOs and solid portions of GGNs at CT represent lepidic growth and invasive components of MIA on pathology, respectively . However, as the solid portion can also represent areas of fibroblastic proliferation, collapsed alveolar space, and mucin, as well as invasive components of the tumour , CT findings at present are limited to describe the various CT findings of MIAs in detail. Although Zhang et al.’s study  dealt with CT findings of MIA, there is a limitation in generalizing their results, as they included only MIAs manifesting as GGNs that were smaller than 2 cm. Therefore, there is a need to address the various CT findings of MIAs in a large study population including all lesions pathologically confirmed as MIAs. Another issue when dealing with MIAs is the precise measurement of the solid portion at CT as well as the prediction of MIAs. These issues are of clinical relevance for appropriate selection of a treatment plan in patients with GGNs, as MIAs are considered to be candidates for sublobar resection [7–10]. Regarding the measurement of the solid portion of GGNs, although the Fleischner Society recommends the mediastinal window setting , it remains a source of controversy. In small lesions, the size of the tumours measured in the lung window setting has been shown to correlate better with that at pathologic examination than in the mediastinal window setting . However, to our knowledge, there have been no studies comparing the measurement of solid portion size between lung and mediastinal window settings in GGNs at CT using the invasive component size on pathology as a reference standard. Therefore, the purpose of our study was to retrospectively investigate the CT findings of MIA and to determine the appropriate method for measurement of solid portions in MIAs at CT.
Materials and methods This retrospective study was approved by the institutional review board of Seoul National University Hospital, and the requirement for patients’ informed consent was waived. Study population A retrospective search of the electronic medical records of our hospital between May 2012 and April 2014 for patients
diagnosed with MIAs was performed by one chest radiologist (J.M.G. with 21 years of experience in thoracic radiology). Sixty pulmonary nodules in 57 patients were initially found. For accurate evaluation of the CT findings of MIAs, we only included patients who underwent thin-section CT no more than 2 months prior to surgery, with slice thicknesses of 1 mm or 1.25 mm. Among the 60 pulmonary nodules, five nodules in five patients were excluded in our study due to lack of thin-section CT. Finally, 55 pulmonary nodules in 52 patients (mean age, 59.3±9.9; range, 35–78 years) comprised our study population. Of 55 MIAs, solitary MIAs were found in 50 patients, two MIAs were found in one patient, and three MIAs were found in another patient. Among the 52 patients, 15 patients underwent lobectomy, 27 underwent wedge resection, and ten patients underwent segmentectomy. Mean interval between CT and surgery was 19.8 days (range, 1–51 days; median, 19 days). Clinical features of patients One chest radiologist (C.M.P. with 13 years of experience in thoracic radiology) reviewed the medical records of the study population using the electronic medical records system of our hospital. The following clinical features were collected: (a) age, (b) sex, (c) smoking history, and (d) history of previous or concurrent malignancy. CT examinations Chest CT was performed using various scanners: Somatom Definition, Sensation-16 (Siemens Medical Solutions, Forchheim, Germany), Brilliance-64, Ingenuity (Phillips Medical Systems, Netherlands), and Lightspeed Ultra, Discovery HD 750 (GE Medical Systems, Milwaukee, WI) with 120 kVp, 30–200 mAs, pitch of 0.875–1.5, and a collimation of 1 or 1.25 mm. Images were reconstructed using the medium sharp reconstruction algorithm with thicknesses of 1– 1.25 mm. All CT scans were obtained in the supine position at full inspiration. Among the 52 chest CT examinations, 29 used low-dose techniques with mean radiation doses of 1.4 mSv±0.4 and 23 used standard-dose techniques with mean radiation doses of 4.3 mSv± 1.5. In terms of contrast media, 40 were noncontrast-enhanced CTs and 12 were contrast-enhanced CTs. In the case of contrast-enhanced CT, 100 ml of contrast media was injected at a rate of 2 ml/second. Analysis of thin-section CT features Two chest radiologists (S.M.L. and K.H.L. with 7 and 5 years of experience in chest CT, respectively) who were blinded to the pathologic results independently evaluated all CT images. The CT findings were analyzed in the lung window setting
(window level of -700 HU and a width of 1,500 HU) using the Picture Archiving and Communication System (Marotech, Korea). CT findings analyzed for each lesion included: (a) nodule size, (b) nodule type (solid, part-solid, pure ground-glass), (c) solid portion size, (d) margin (spiculated, non-spiculated) (e) border (lobulated, nonlobulated), (f) presence of bubble-lucency, and (g) presence of pleural retraction. To determine the nodule size and solid portion size, the greatest diameters of the entire nodule and of the internal solid portion on axial images were recorded. Solid portion size was reported in millimetres rounded to the nearest whole number. Discrepancies of nodule type between the two observers were resolved by another observer (C.M.P.). In addition, one chest radiologist (J.M.G.) reviewed all CT images to investigate lesion multiplicity (solitary, multiple) and to determine whether there was interval growth. We defined nodule growth as when a nodule or solid portion increased greater than or equal to 2 mm in its longest diameter, or when there was any emerging new solid portion within a GGN. To evaluate interobserver agreement of solid portion measurements at the different window settings, two chest radiologists measured the solid portions in the mediastinal window setting (window level of 30 HU and a width of 400 HU) 4 weeks later (Fig. 1). Pathologic assessment All surgical specimens were fixed in an inflated state by means of transpleural and transbronchial infusion of formalin and embedded in paraffin. The specimens were sliced with a less than 0.3 mm interval and stained with haematoxylin and eosin. For each case, representative histological slides for the largest cross section or representative part of the tumour and the invasive component were selected (D.H.C. with 22 years of experience in pathology). Thereafter, the longest dimensions of tumour and invasive component were measured and reported by two pathologists (D.H.C. and J.K. with 3 years of clinical experience) in consensus. Pathologic stages of lung cancers were also recorded. Statistical analysis In consideration of within-patient correlation, a single nodule was selected randomly prior to nodule evaluation in two patients with more than one MIA. Interobserver agreement for CT findings of MIAs was investigated using weighted κ-statistic. Measurements of solid portions at CT were compared with the size of the invasive components on pathology using the paired t-test. Differences in solid portion measurement
Fig. 1 Measurements of solid portion in the lung and mediastinal window settings at CT. (a) CT scan shows a 13-mm, part-solid GGN in the right middle lobe in a 66-year-old man. The solid portion (arrow) was measured as 6 mm in the lung window setting. (b) In the mediastinal setting, the solid portion (arrow) was measured as 3 mm. (c) Low magnification (haematoxylin and eosin, original magnification 40×) photomicrograph demonstrates invasive component (green arrows) and surrounding lepidic component (black arrows). The size of the invasive component was 4 mm
between the two observers according to window settings were also compared using the paired t-test. Bland-Altman plots were used to assess differences between invasive components and solid portions as well as interobserver agreement between different window settings . Single measure intraclass correlation coefficients (ICCs) were also calculated. Differences between invasive components and solid portions according to window settings and differences in interobserver measurements between the two different window settings were compared using the paired t-test. All statistical analyses were performed using MedCalc, version 12.2.1 (MedCalc Software, Mariakerke, Belgium). A
p value of <0.05 was considered to indicate a significant difference.
Results Clinical and pathological findings of MIAs The 52 patients consisted of 18 men (mean age, 62.2± 8.6; range, 45-77 years) and 34 women (mean age, 57.7 ± 10.3; range, 35-78 years). Among them, 41 patients were never-smokers and 11 were ex-smokers (mean pack-year, 25.1±24.1; range, 1.0–76.5 pack-years). Fifteen patients (28.8 %) had a history of malignancy (concurrent lung cancers in two, previous breast cancers in four, previous hepatocellular carcinomas in three, previous stomach cancers in two, previous thyroid cancers in four, previous pancreatic cancer in one, previous bladder cancer in one, and previous cervical cancer in one). Mean size of MIAs was 10.5 mm±4.8 (range, 4–28 mm). Mean size of invasive components was 2.6 mm±1.5 (range, less than 1 mm to 5 mm). Pathologic stages in the 40 patients who underwent nodal dissections were T1N0. The nodal staging in the remaining 12 patients was not assessed. All MIAs were of the nonmucinous type, except one mucinous type MIA (Fig. 2). CT findings of MIAs MIA manifested as 28 pure GGNs (53.8 %), 22 part-solid GGNs (42.3 %), and two solid nodules (3.8 %). Mean sizes of invasive components in pure, part-solid GGNs, and solid nodules were 2.3 mm±1.6, 2.9 mm±1.5, and 3.0 mm±1.4, respectively. One of the two solid nodules was a mucinous MIA, and the other presented with multifocal invasive components in the background of fibrosis at pathologic examination (Fig. 3). With respect to the internal characteristics of MIAs, lobulated border, bubble lucency, and pleural retraction were frequently found in both observers (26.9 - 42.3 %). The results of interobserver agreement for various CT features of MIAs are summarized in Table 1. Nodule type showed almost perfect agreement (κ=0.832). Other than nodule type, interobserver agreement for CT findings showed moderate to substantial agreement. Fourteen patients had multiple GGNs, including a GGN confirmed as MIA: eight patients had two GGNs (four atypical adenomatous hyperplasias (AAHs), two adenocarcinoma in situ (AIS), one MIA, and one adenocarcinoma); two patients had three GGNs (two MIAs in one patient and one adenocarcinoma and one pathologicallyunproven GGN in one patient); the remaining four
Fig. 2 Mucinous minimally invasive adenocarcinoma (MIA) appearing as solid nodule. (a) CT scan showed a 9 mm solid nodule (arrow) in the left lower lobe in a 52-year-old woman. This nodule was confirmed as mucinous MIA by lobectomy. (b) Low magnification (hematoxylin and eosin, original magnification 40×) photomicrograph showed multifocal tumors (arrows). (c) On 200× magnification photomicrograph, there were multifocal invasive components (blue arrows) and mucin in the gland epithelium (black arrow). The size of the invasive component was 2 mm
patients had more than ten GGNs. Among the 42 lesions with previous thin-section CT images (interval range, 41– 2,775 days), 29 lesions (69.0 %) showed growth. Among the 32 lesions with more than 1 year of follow-up (median, 1,144 days; range, 411–2,775 days), 28 lesions (87.5 %) showed interval growth. CT measurement of solid portion There were ten GGNs (19.2 %) with solid portions larger than 5 mm in the lung window setting and five (9.6 %) in the mediastinal window setting by observer 1; and eight (15.4 %) in the lung window setting and four (7.7 %) in the mediastinal window setting by observer 2. In comparison with the invasive component on pathology, solid
Eur Radiol Fig. 3 Nonmucinous MIA with a background of severe fibrosis. (a) CT scan showed a 12 mm solid nodule with air-bronchogram (arrow) in the right upper lobe in a 35-year-old woman, who underwent lobectomy. (b) Low magnification (hematoxylin and eosin, original magnification 40×) photomicrograph showed multifocal invasive components (blue arrows) and lepidic growth (black arrows). (c) On 400× magnification photomicrograph, there was dense fibrotic stroma with proliferated vessels (yellow arrow) and an invasive component (blue arrow). (d) On a 400× magnification photomicrograph in another field, there was focal infiltration of inflammatory cells, suggesting severe inflammation (green arrows). The size of the invasive component was 4 mm
portion size in the mediastinal window setting by observer 2 was significantly smaller (p=0.001). Measurements in the lung window setting were significantly larger than those in the mediastinal window setting in both observers (p<0.001). Observer 1 tended to measure solid portion as larger than observer 2 in the mediastinal window setting (p=0.004). Table 2 summarizes the results of the measurement of solid portions at CT of the two observers.
Bland-Altman plots with 95 % limits of agreement between pathology and CT measurements are shown in Fig. 4. 95 % limits of agreements in the mediastinal window setting were slightly narrower than those in the lung window setting in both observers. There were no significant differences between lung
Table 2 Measurements of solid portion at CT by two observers, and invasive component on pathology Characteristics
Fig. 4 Bland-Altman plots showing the relation between CT and pathology and interobserver agreement. (a)–(d), Graphs show differences between the size of the invasive component and the size of the solid portion. X-axes represent invasive component size and Y-axes represent differences in size between invasive component and solid portion. (a) lung window setting by observer 1, (b) mediastinal window
setting by observer 1, (c) lung window setting by observer 2, and (d) mediastinal window setting by observer 2. (e)–(f), Graphs show measurement differences between two observers in the lung (e) and mediastinal (f) window settings. X-axes represent mean measurements of two observers and Y-axes represent measurement differences between two observers
and mediastinal window settings in both observers (p=0.466 and p=0.704).
(mean, 0.3 mm) and from -2.2 mm to 3.4 mm (mean, 0.6 mm), respectively (Fig. 4). Single measure ICCs in the lung and mediastinal window settings were 0.9030 (95 % CI, 0.8368– 0.9431) and 0.8733 (95 % CI, 0.7891–9252). There were no significant differences in interobserver measurement differences between the lung and mediastinal window settings (p=0.558).
Interobserver agreement in measuring solid portion The 95 % limits of interobserver agreement in the lung and mediastinal window settings ranged from -2.6 mm to 3.1 mm
Discussion In our study, 52 MIAs manifested as 28 pure GGNs (53.8 %), 22 part-solid GGNs (42.3 %), and two solid nodules (3.8 %) at CT; that is, nearly all GGNs. Although 7.7 % to 19.2 % of GGNs in our study had solid portions larger than 5 mm depending on the observers and window settings, this corresponds well with those of previous studies [2, 3, 14, 15]. As nearly all MIAs appeared as GGNs, MIAs can be managed according to subsolid guidelines proposed by the Fleischner Society . Among thin-section findings, lobulated border, bubble lucency, and pleural retraction were frequently found in both observers (26.9–42.3 %). The presence of frequent lobulated margin in MIAs is consistent with the results of a previous study , which suggested invasive adenocarcinomas. However, in contrast with nodule typing (κ=0.832), these CT findings may be limited in wide acceptance for use due to moderate to substantial interobserver agreement (κ=0.420 – 0.689). Interestingly, we found that 28 lesions (87.5 %) showed interval growth among the MIAs with more than 1 year of follow-up. Although there may have been selection bias in that growing GGNs tended to be resected, MIAs may be a practically growing lesion at follow-up, in distinction from AAH or AIS [17, 18]. Even though there were growing GGNs, the pathologic stages of MIAs in all 40 patients who underwent nodal dissections were T1N0. This may be supporting evidence of limited resection for MIAs. In addition, there were 16 MIAs manifesting as pure GGNs less than 10 mm in size in our study, contrary to Lee et al.’s study , and nine MIAs showed interval growth at follow-up. Considering that lesion size was the sole significant differentiator of pre-invasive lesions in pure GGNs in Lee et al.’s study , interval growth can be another useful differentiator between pre-invasive lesion and MIA or invasive adenocarcinoma. In terms of demographics, MIAs were frequently found in women (65.4 %, 34/52) and never-smokers (78.8 %, 41/52). The female predominance seen in our study corresponds well with that of Borczuk et al.’s study . These tendencies in patients with MIAs are consistent with those in patients with AIS (formerly, bronchioloalveolar carcinomas) [19, 20]. In addition, we found one mucinous MIA (1.9 %, 1/52) in our study. Our results are consistent with Yoshizawa et al.’s study , which also revealed that MIA is almost always nonmucinous and that mucinous MIA is rare. With regard to measurement of solid portions at CT, although the mediastinal window setting showed narrower 95 % confidence intervals than the lung window setting when the size of solid portions was compared with that of the invasive component on pathology, there were no significant differences between lung and mediastinal settings in both observers (p=0.466 and p=0.704). This suggests that there are no significant differences in the accuracy of solid portion measurements according to the two window settings.
With respect to interobserver agreement, 95 % limits of confidence intervals in the mediastinal window setting were slightly narrower than those in the lung window setting. However, the differences were minimal, and there were no significant differences in interobserver measurement differences between the lung and mediastinal window settings (p=0.558). Although our results were obtained from a limited number of observers and a relatively small population, we believe that both the mediastinal and lung window settings can be applied for measurement of solid portions at CT without a significant difference. Future prospective trials are warranted to confirm our results. Our study has several limitations. First, our study was of retrospective design. Therefore, there may have been bias. Second, the number of patients is not large. However, we included various CT findings of MIAs, including manifestations of solid nodules and a mucinous MIA. In addition, our study including 52 patients is, in comparison with previous studies [2, 3, 14, 15, 21], the largest study of MIAs, excepting one study . In conclusion, nearly all MIAs appear as pure and partsolid GGNs with frequent interval growth at follow-up, and both mediastinal and lung window settings can be applied for measurement of solid portions at CT without a significant difference. Acknowledgments The scientific guarantor of this publication is Dr. Jin Mo Goo. The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article. This study has received funding by grant no 20130507 from the SK Telecom Research Fund. No complex statistical methods were necessary for this paper. Institutional Review Board approval was obtained. Written informed consent was waived by the Institutional Review Board. Methodology: retrospective, observational, performed at one institution. Conflict of interest All authors have no conflicts to disclose.
References 1. Travis WD, Brambilla E, Noguchi M et al (2011) International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol 6: 244–285 2. Travis WD, Garg K, Franklin WA et al (2005) Evolving concepts in the pathology and computed tomography imaging of lung adenocarcinoma and bronchioloalveolar carcinoma. J Clin Oncol 23:3279– 3287 3. Borczuk AC, Qian F, Kazeros A et al (2009) Invasive size is an independent predictor of survival in pulmonary adenocarcinoma. Am J Surg Pathol 33:462–469 4. Goo JM, Park CM, Lee HJ (2011) Ground-glass nodules on chest CT as imaging biomarkers in the management of lung adenocarcinoma. AJR Am J Roentgenol 196:533–543
Eur Radiol 5. Park CM, Goo JM, Lee HJ, Lee CH, Chun EJ, Im JG (2007) Nodular ground-glass opacity at thin-section CT: histologic corr ela ti on and eval ua ti on of change at foll ow -up. Radiographics 27:391–408 6. Zhang Y, Qiang JW, Ye JD, Zhang J (2014) High resolution CT in differentiating minimally invasive component in early lung adenocarcinoma. Lung Cancer 84:236–241 7. El-Sherif A, Gooding WE, Santos R et al (2006) Outcomes of sublobar resection versus lobectomy for stage I non-small cell lung cancer: a 13-year analysis. Ann Thorac Surg 82:408–415, discussion 415-416 8. Nakamura H, Kawasaki N, Taguchi M, Kabasawa K (2005) Survival following lobectomy vs limited resection for stage I lung cancer: a meta-analysis. Br J Cancer 92:1033–1037 9. Okada M, Koike T, Higashiyama M, Yamato Y, Kodama K, Tsubota N (2006) Radical sublobar resection for small-sized non-small cell lung cancer: a multicenter study. J Thorac Cardiovasc Surg 132:769– 775 10. Tsutani Y, Miyata Y, Nakayama H et al (2014) Appropriate sublobar resection choice for ground glass opacity-dominant clinical stage IA lung adenocarcinoma: wedge resection or segmentectomy. Chest 145:66–71 11. Naidich DP, Bankier AA, MacMaho n H et al (20 13) Recommendations for the management of subsolid pulmonary nodules detected at CT: a statement from the Fleischner Society. Radiology 266:304–317 12. Grills IS, Fitch DL, Goldstein NS et al (2007) Clinicopathologic analysis of microscopic extension in lung adenocarcinoma: defining clinical target volume for radiotherapy. Int J Radiat Oncol Biol Phys 69:334–341 13. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307–310
14. Lee KH, Goo JM, Park SJ et al (2014) Correlation between the size of the solid component on thin-section CT and the invasive component on pathology in small lung adenocarcinomas manifesting as ground-glass nodules. J Thorac Oncol 9:74–82 15. Lim HJ, Ahn S, Lee KS et al (2013) Persistent pure ground-glass opacity lung nodules >/= 10 mm in diameter at CT scan: histopathologic comparisons and prognostic implications. Chest 144:1291– 1299 16. Lee SM, Park CM, Goo JM, Lee HJ, Wi JY, Kang CH (2013) Invasive pulmonary adenocarcinomas versus preinvasive lesions appearing as ground-glass nodules: differentiation by using CT features. Radiology 268:265–273 17. Park CM, Goo JM, Lee HJ et al (2006) CT findings of atypical adenomatous hyperplasia in the lung. Korean J Radiol 7: 80–86 18. Lee HJ, Goo JM, Lee CH, Yoo CG, Kim YT, Im JG (2007) Nodular ground-glass opacities on thin-section CT: size change during follow-up and pathological results. Korean J Radiol 8:22–31 19. Gandara DR, Aberle D, Lau D et al (2006) Radiographic imaging of bronchioloalveolar carcinoma: screening, patterns of presentation and response assessment. J Thorac Oncol 1: S20–S26 20. Asamura H (2008) Minimally invasive approach to early, peripheral adenocarcinoma with ground-glass opacity appearance. Ann Thorac Surg 85:S701–S704 21. Yoshizawa A, Motoi N, Riely GJ et al (2011) Impact of proposed IASLC/ATS/ERS classification of lung adenocarcinoma: prognostic subgroups and implications for further revision of staging based on analysis of 514 stage I cases. Mod Pathol 24:653–664