| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Departments of Thoracic/Head and Neck Medical Oncology (A.O., R.S.H., T.I.), Radiology (D.H.C., R.F.M., M.T.T., E.M.M.), Thoracic and Cardiovascular Surgery (A.M.C., A.A.V.), and Pathology (M.Z.G.), University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd, Unit 57, Houston, TX 77030. From the 2003 RSNA Annual Meeting. Received September 29, 2004; revision requested December 2; revision received December 15; accepted January 18, 2005. Address correspondence to E.M.M. (e-mail: emarom{at}di.mdacc.tmc.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: This study was approved by the institutional review board, with waiver of informed consent, and was in compliance with HIPAA regulations. Findings of chest computed tomography (CT) were retrospectively evaluated in 72 patients who underwent resection of pathologic stage I NSCLC; tumor diameter, presence of calcifications, type of contour, type of margins, attenuation of the nodule, presence of a halo, presence of cavitation, and tumor location were documented. Immunohistochemical studies were performed in surgical specimens. Imaging and molecular data were correlated with patient outcome. Cox proportional hazards regression models were used to correlate biologic and radiographic variates with clinical outcome.
RESULTS: There were 38 men (53%) and 34 women (47%) (median age, 65.5 years). Median follow-up was 56.3 months; median overall survival, 76.3 months. A strong correlation was found between tumor diameter measured by radiologists and that measured by pathologists (P < .001; Pearson correlation coefficient, 0.81). EGFR overexpression was found in 48 (67%) tumors; significantly more was found in squamous cell carcinomas than was found in adenocarcinomas (P = .028), and more was found in T2 tumors than was found in T1 tumors (P = .001). HER2 overexpression was found in 13 (18%) tumors; cavitation, in 16 (22%) tumors. Cavitary lesions were significantly more common in squamous cell carcinomas than were in adenocarcinomas (P = .013) and in EGFR-overexpressing tumors (P = .012) than in tumors that did not overexpress EGFR. Cavitary lesions were significantly associated with shorter disease-free survival time (P = .01) and shorter overall survival time (P < .007).
CONCLUSION: Patients who have stage I NSCLC with cavitary lesions have an adverse prognosis and are likely to have tumor EGFR overexpression.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
A new understanding of the genetic background of lung cancer has emerged from research advances in the field of molecular biology (3). The ErbB family of receptors seems to show promise in this field because they play a pivotal role in tumor cell proliferation, survival, adhesion, migration, and differentiation and in tumor angiogenesis (4). The ErbB family comprises four structurally related receptors: ErbB1, more commonly known as epidermal growth factor receptor (EGFR) and also called HER1; ErbB2, or HER2; ErbB3, or HER3; and ErbB4, or HER4. Inhibition of the ErbB1 and ErbB2 receptors has been studied extensively, and several newly developed compounds have shown promising results in preclinical and clinical trials in the inhibition of ErbB-related pathways (5). In 2003, on the basis of results from phase II trials that involved patients with NSCLC, gefitinib (Iressa; AstraZeneca Pharmaceuticals, Wilmington, Del) became the first approved EGFR-targeted agent for use in patients with previously treated advanced NSCLC (6) (Fig 1), yet only 10% of patients benefit from this therapy. Factors that may be used to predict response or resistance to gefitinib and other EGFR inhibitors are currently being intensively explored.
|
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Clinical end points for the study were overall survival time (from surgery to death) and disease-free survival time (from surgery to diagnosis of local or distant recurrence). Our institutional review board reviewed our proposal and approved our retrospective study, with a waiver of informed consent, and this study was in compliance with Health Insurance Portability and Accountability Act regulations.
CT Imaging and Evaluation
All patients included in the study had undergone chest CT, which had been conducted within the month before surgery. CT examinations were performed at multiple institutions with a variety of helical and nonhelical scanners. The section thicknesses used were 10 mm (n = 64), 8 mm (n = 3), 7 mm (n = 1), 5 mm (n = 3), and 4 mm (n = 1).
All hard copies of the chest CT scans were retrospectively interpreted by two thoracic radiologists (D.H.C. and E.M.M., who had 10 and 7 years of experience, respectively), and any differences were resolved with discussion and consensus. At the time of interpretation, both radiologists were informed that the involved patients had early-stage surgically treated NSCLC, but they were blinded to patients' clinical data (ie, pathology reports, immunohistochemical test results, and clinical outcomes).
Imaging characteristics of the primary lesions were recorded. These characteristics included diameter of the tumor, presence of calcifications, type of contour (spiculated, smooth, lobular, or mixed [combination of the other types of contour]), type of margins (poorly or well demarcated), attenuation of the nodule (solid, ground-glass, or mixed [a combination of solid and ground-glass]), presence of a halo (ground-glass opacity abutting the solid tumor), relationship of the solid tumor to the halo (central or eccentric), presence of cavitation (defined as the presence of air in the tumor at the time of diagnosis, prior to biopsy or treatment), location of the tumor (center or periphery of the lung [periphery of the lung was defined as within 3 cm of the pleura]), lobe or structure involved, airway order (the largest airway the tumor involves), size of hilar and mediastinal lymph nodes, presence of a pleural effusion, and presence of any adrenal abnormalities.
Fifty-two of the 72 study patients underwent additional imaging studies for staging, and some underwent more than one study. These were retrospectively reviewed (E.M.M. and D.H.C.) to confirm staging and included the following: abdominal CT (n = 44), whole-body bone scanning by using technetium 99m medronate (n = 30), brain CT (n = 35), and brain magnetic resonance (MR) imaging (n = 2).
Immunohistochemical Techniques
The primary antibodies used for immunohistochemical analyses were a mouse monoclonal anti-EGFR (clone 31G7; Zymed, South San Francisco, Calif) and an anti–HER2 test (HercepTest; Dako Cytomation, Carpinteria, Calif). Formalin-fixed paraffin-embedded tissue sections (4–6 µm) were obtained from the Surgical Pathology Laboratory, University of Texas M.D. Anderson Cancer Center. The sections were deparaffinized in xylene, treated with a graded series of alcohol washes (100%, 95%, and 80% ethanol [vol/vol] in double-distilled water), and rehydrated in phosphate-buffered saline (pH 7.5). Immunohistochemical studies were conducted as previously described by using commercially available reagents and well-established staining methods (7,8).
Review and Score Assignment in Tissue Sections
A positive control sample was evaluated with each batch of slides. Two researchers (M.Z.G. and A.O., with 9 and 4 years of experience, respectively) reviewed the stained sections while they were blinded to clinical and imaging data. Expression of EGFR and HER2 was evaluated by means of examination of membrane staining. Each slide was assigned a score by using a semiquantitative four-category system: score 0, no tumor expression; score 1, 1%–10% tumor expression; score 2, greater than 10% to 20% tumor expression; or score 3, greater than 20% tumor expression. Slides with scores of 2 or 3 were considered to overexpress that particular protein.
Statistical Analysis
To compare distributions between demographic, clinical, and biologic factors, 
2 tests were used. The Student t test was used to compare patient age between the sexes. The Kaplan-Meier method (9) was used to estimate the cumulative survival functions and cumulative recurrence hazard functions. The significance level was set at P < .05. The strength of the association between the prognostic effect of the putative covariates and the disease-free survival time and overall survival time was evaluated with hazard ratios by using univariate Cox proportional hazards regression models. Multivariate Cox proportional hazards regression analysis was used in adjustment for the presence of other covariates. Statistical software (SPSS; SPSS, Chicago, Ill) was used to perform the analyses and create graphs.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Fifty-seven (79%) patients underwent removal of the tumor at lobectomy or pneumonectomy; and 15 (21%) patients, at wedge resection or segmentectomy. Sixty-six (92%) patients had undergone mediastinal lymph node sampling, whereas six (8%) patients had undergone only hilar lymph node sampling. Lymph node sampling was performed during surgery in all patients. In addition, three patients underwent mediastinoscopy prior to surgery.
Median follow-up in the 72 patients was 56.3 months, and the median time from surgery to death was 76.3 months. Twenty patients had documented disease recurrence, and the median time from surgery to diagnosis of recurrence was 25.4 months.
Imaging Features
Twenty-eight patients had radiographically identified T1 tumors, and 44 had radiographically identified T2 disease. The median diameter of all measured tumors was 3.5 cm. A strong correlation was found between tumor diameter measured by the pathologists and that measured by the radiologists (P < .001; Pearson correlation coefficient, 0.81).
The morphologic characteristics of the tumors are described in Table 1. Tumor cavitation was present in the primary tumor in 16 (22%) patients. Calcifications were present in one patient, and they were undifferentiated T1 NSCLC. Hilar lymph nodes were 5 mm or smaller in diameter in 48 (67%) patients, 6–10 mm in diameter in 15 (21%) patients, and 11–18 mm in diameter in nine (12%) patients. Mediastinal lymph nodes were 5 mm or smaller in diameter in 12 (17%) patients, 6–10 mm in diameter in 43 (60%) patients, and 11–18 mm in diameter in 17 (24%) patients. Pleural effusion was detected in four (6%) patients, and it was minimal in size and occupied less than 10% of the hemithorax. In 44 patients, the adrenal glands were completely visible, and of those, four (9%) patients had an adrenal nodule. Two of the nodules were proved to be benign at fine-needle aspiration biopsy prior to surgery, and one was stable compared with its appearance at abdominal CT performed 12 months earlier. No additional findings suspicious for malignancy were seen on the abdominal CT scans, bone scans, brain CT scans, and MR images that were available for review.
|
Correlation between Radiologic Features and Molecular Study Results
Tumor diameter, as measured by the radiologists, correlated inversely with the patients' outcome. A significant association was found between tumor diameter and disease-free survival time (P = .031); 44 patients with T2 tumors had a shorter disease-free survival time compared with 28 patients with T1 tumors (the median was not reached in both cases). Patients with T2 tumors had a median survival time of 69.1 months, whereas those with T1 tumors had a median survival time of 84.4 months, but this difference did not reach statistical significance (P = .067).
Cavitation was more commonly seen in tumors of the squamous cell carcinoma histologic type (n = 11) than it was in those of the adenocarcinoma histologic type (n = 2) (P = .013). The tumors with cavitation were more commonly T2 (n = 13) tumors than T1 (n = 3) tumors, and more often they were at the lung periphery (n = 13) than they were in the center (n = 3) of the lung. In addition, 13 (81%) of 16 tumors with cavitation were associated with overexpression of EGFR (P = .012) (Fig 2). Overexpression of HER2 was noted in only two tumors with cavitation. Overall, patients with lesions with cavitation had worse prognoses than did the other patients. They had a shorter disease-free survival time (P = .01) (Table 2, Fig 3) and a shorter overall survival time (P < .007) (Table 2, Fig 4). With a multivariate Cox proportional hazards regression model, we confirmed that cavitation is an independent predictor of poor prognosis when adjustment was made for EGFR expression, histologic type, tumor diameter, and tumor location. With multivariate regression analysis, no significant association between EGFR overexpression in patients with tumors with cavitation and histologic type, tumor diameter, and tumor location was found.
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
To our knowledge, our report is the first to focus on the relationship between tumor cavitation and prognosis in patients with stage I disease. In an earlier study, no difference in survival was found between lung cancers with cavitation and those without cavitation (10). That study, however, included a heterogeneous group of patients with all stages and histologic disease types, including small cell lung cancer. For staging purposes, lymph nodes were staged radiographically by using size rather than histologic criteria, and a substantial number of patients were not candidates for surgery. Researchers in a later study (11) of surgically treated patients with stages I to III squamous cell carcinoma found that patients with lung cancer with cavitation had a worse prognosis than did patients with tumors without cavitation. A statistical analysis in which the prognosis for patients with each stage was correlated, however, was not performed.
The pathogenesis of cavitary lesions has not been fully established. It has been suggested that this phenomenon is associated with rapid tumor growth that exceeds the blood supply of the tumor, thus causing central necrosis of the lesion, and this necrosis eventually leads to cavity formation (12). This supposition gains support from findings in a report (13) that tumor-associated vasculature inhibition by bevacizumab, a monoclonal antibody to vascular endothelial growth factor, induced central necrosis and cavity formation in patients with squamous cell carcinoma of the lung. In our study, the tumors with cavitation also were associated with overexpression of EGFR. This receptor is commonly expressed by NSCLC tumors, especially squamous cell carcinomas (7,14,15), but the association with lesions with cavitation has not been shown before. The pathologic role of EGFR is related to enhancement of tumor proliferation and angiogenesis (4). We assume that the high level of EGFR expression in the tumors in our study was associated with the rapid growth of these tumors and the formation of cavitation, as the tumors exceeded their blood supply, and that it was related to the adverse outcome in these patients.
We recently found that synchronous overexpression of EGFR and HER2 is associated with poor prognosis in patients with stage I NSCLC (7). Other molecular factors that can be used to predict poor outcome in patients with stage I disease include the presence of the K-ras oncogene (16), a high ratio of type IV collagenases to E-cadherin (17), a low level of bcl-2 protein (18), a high level of retinoic acid receptor-ß protein (19), and the loss of blood group antigen A (20). Tests for these other factors, however, are not routinely performed, and no medical intervention has been developed to target them. Identification of all tumor markers requires tumor sampling and substantial expense. To our knowledge, no radiographic association between these other molecular factors and imaging findings in the tumor has been reported.
A large amount of information about the morphologic appearance of the primary tumor in NSCLC is obtained at cross-sectional imaging. Because there is a subset of patients with poor survival, despite their having only stage I lung cancer, it was our hope to identify them radiographically and to try to find a correlation with receptor expression. Thus, not only would they be identified for prognostic purposes, but also they would be identified in a noninvasive fashion for possible future molecular receptor-specific targeted therapy.
Perhaps a more important role for the association we found between tumor cavitation and expression of EGFR is the potential for therapeutic development. EGFR inhibition currently is used to treat lung cancer: Gefitinib, an EGFR tyrosine kinase inhibitor, was recently approved as a third-line monotherapeutic agent for use in patients with NSCLC (6). This approval followed phase II studies of gefitinib in patients with lung cancer, and these studies demonstrated an approximately 10% response rate and a relatively high safety profile (21,22). To date, the exact profile of patients who may respond to therapy with gefitinib or other EGFR inhibitors has not been fully discovered. These factors had been correlated with response: patients who were female, patients who had adenocarcinoma as the tumor histologic type, patients who had no history of smoking cigarettes, and patients who were from Japan (21,22). Recently, it has been shown that somatic mutations in the EGFR tyrosine kinase site characterize responders to therapy with gefitinib (23,24). Other biologic and clinical factors, such as tumor cavitation, should be studied and correlated with response to EGFR signaling inhibition.
We believe a strength of our study is its long-term follow-up; however, our study had limitations as well. First, its retrospective nature may have introduced selection bias. Second, perhaps our study group was actually heterogeneous, and more rigorous staging might have led to identification of some patients with a higher stage of the disease. We do not think this introduced substantial bias, however, because the survival times we found in our study were those expected for patients with stage I disease, and all patients included had undergone nodal histologic staging. Although whole-body fluorodeoxyglucose positron emission tomography is more accurate for the staging of NSCLC (25,26), it was not available at the time our study population was treated. Third, although no correlation between HER2 overexpression and prognosis was found in our study, this could be reflective of the small number of patients in our series in whom staining was positive for HER2. Thus, studies with larger groups of patients may be necessary to determine the role of HER2 in lung cancer. Finally, the section thickness used was variable and was generally thicker than is commonly used today. Perhaps the use of thin section thickness would have helped in the identification of additional patients with tumors with cavitation. Researchers in future prospective studies may evaluate this point and perhaps quantify the degree of cavitation, but because of the long-term follow-up required for evaluation of survival time in such studies, the imaging parameters used may lag behind technologic advances.
In conclusion, tumor cavitation is associated with EGFR overexpression and is a predictor of poor outcome in patients with stage I NSCLC. The use of findings from imaging and molecular studies will establish a better clinical profile to identify patients at high risk for disease progression who, thus, may benefit from targeted therapy with anti-EGFR agents.
|
FOOTNOTES |
|---|
Abbreviations: EGFR = epidermal growth factor receptor • NSCLC = non–small cell lung cancer
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, A.O., R.S.H., A.M.C., E.M.M.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, A.O., E.M.M.; clinical studies, D.H.C., E.M.M., A.O., T.I., M.Z.G.; statistical analysis, A.M.C.; and manuscript editing, A.O., R.S.H., A.M.C., R.F.M., M.T.T., A.A.V., M.Z.G., E.M.M.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. B. Gadkowski and J. E. Stout Cavitary Pulmonary Disease Clin. Microbiol. Rev., April 1, 2008; 21(2): 305 - 333. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
