Predicting Role of Preoperative Computed Tomography Features on Invasiveness of Thymic Neoplasms

Objective: The anterior mediastinal masses are composed of neoplasms from many different groups, thymic epithelial tumors being the most common type. Histologic subtype, tumor stage and completeness of resection are considered to be the most important prognostic factors in thymic malignancies. The aim of this study is to evaluate the association between the preoperative Computed Tomography characteristics and the histological types, mostly focusing on thymic neoplasms. Materials and Methods: Preoperative Computed Tomography features and pathologic data of 40 patients who underwent surgery due to an anterior mediastinal mass were retrospectively analyzed. Patients are grouped as thymomas, thymic carcinomas and others. Results: Computed Tomography characteristics such as shape, contour, necrosis, lymph node enlargement, surrounding organization, pathologic invasion pattern, internal density, presence of cystic component, pleural/pericardial effusion and calcification were compared with histologic subtypes. Among these parameters solely pathologic invasion and surrounding organization were statistically different between the groups (p<0.001 and p=0.001, respectively). Correlation between the Computed Tomography and pathologic tumor sizes among histologic subtypes was also analyzed and no statistically significant difference between Computed Tomography and pathologic tumor sizes was detected (p>0,05). Conclusion: According to the results of our study, Computed Tomography imaging has a limited role in predicting stage and malignant potential particularly for thymic neoplasms and anterior mediastinal masses.


INTRODUCTION
The anterior mediastinal masses are composed of neoplasms from many different groups and constitute about 50% of all mediastinal masses [1]. The most common anterior mediastinal mass is thymic epithelial tumors (TET), which constitute approximately 20% of all mediastinal tumors and approximately 47% of anterior mediastinal masses [2]. Histologic subtype, tumor stage and completeness of resection are considered to be the most important prognostic factors in thymic malignancies. Although most of the patients with early stage disease can be cured with complete resection, patients with advanced disease may necessitate neoadjuvant chemotherapy to facilitate complete resection and prolong survival [3][4][5]. Therefore, the ability to detect patients' stage by radiographic predictors is an important consideration to avoid unnecessary or futile surgeries and help select appropriate patients for adjuvant chemotherapy. Computed tomography (CT) is the most commonly used imaging method for preoperative evaluation of patients with anterior mediastinal masses which provides important information about tumor size, shape, content, and its relationship to other mediastinal components or borders [6][7]. The purpose of this study is to evaluate the association between the preoperative CT characteristics and the stage, histological types and malignancy, mostly focusing on thymic neoplasms.

Patient Selection
This retrospective study has been approved by the local ethics committee and conducted in accordance with the Declaration of Helsinki (2000). Informed consent was waived because of the retrospective nature of the study. A total of 40 patients who underwent surgery due to an anterior mediastinal mass between October 2007 and December 2016 were included in the study. Patients whose surgical treatment, radiologic imaging or pathologic examination were not available at our hospital were excluded from the study. Patients are grouped as thymomas, thymic carcinomas and others. Others group included anterior mediastinal masses other than thymic pathologies [Granulomatous inflammation (n=1), neuroendocrine tumor (n=1), pericardial cyst (n=1) and vascular malformation (n=1)]. CT images and medical records of the patients were retrospectively reviewed.

CT technique
Patients underwent thorax CT imaging with a 16-detector CT (Somatom Sensation 16, Siemens Medical Systems, Germany) scanner (tube voltage=130 kV, effective mAs=90, slice thickness 5 mm, collima-tion=2x4 mm, pitch=1.6). One hundred milliliters (ml) of intravenous (IV) contrast material (300 mg/ ml Omnipaque, GE Healthcare, Ireland) at a flow rate of 4 ml/sec was used for CT examinations. The body region between the level of the lung apex and upper pole of kidneys was included in the CT scan area. CT scans were directly sent to the workstation (Centricity Universal Viewer, GE Healthcare, USA).

Image analysis and description of CT findings
CT images were reviewed by two radiologist experienced in thoracic radiology and were blinded to the surgical, clinical and pathology findings, and radiology reports. Imaging characteristics were determined as size in the largest dimension, shape (oval, round or irregular), contour (smooth, lobulated or irregular), internal density, homogenity, calcification, necrosis or cystic appearance, invasion, pleural or pericardial effusion, capsule integrity, lymphadenopathy, central density and ratio of anteroposterior length to sternum-vertebra distance. The morphological features which were ambiguously determined by the reviewers were reassessed together in a second session, and the exact finding was determined by a consensus decision.

Clinical and Pathological Assessment
Clinical symptoms, presence of comorbidities and the duration of hospital stay were recorded for all patients. The tumor subtype, pathological size in the largest dimension, surgical margin positivity and microscopic invasion were also recorded as pathological findings.

Statistical Analysis
Statistical analysis was performed using the SPSS 20.0 for Windows (SPSS, Inc., Chicago, USA) packet program. As a statistical analysis, the categorical variables in the descriptive findings are given in terms of number, percent, and continuous variables are expressed as mean ± standard deviation and median (smallest, largest value). Pearson Chisquare was used to compare categorical variables. The normal distribution of the data was analyzed by Kolmogorov-Smirnov test and Kruskal Wallis, oneway ANOVA and Wilcoxon signed rank tests. The homogeneity of the variances in Anova test was evaluated by Levene Test. In cases where there was a significant difference between the groups, twotailed post-hoc comparisons were made using the Tukey test. Correlation between tumor subtypes and pathologic dimension and correlation between pre and post-operative hemoglobin values was assessed by Pearson or Spearman Correlation tests according to normal distribution fit. The statistical significance level was accepted as p <0.05.

RESULTS
This study consisted of 24 men (64.8%) and 13 women (35.2%) with a median age of 51 years (range, 21 -92 years). Histological subtypes of patients are shown in Table 1. The relationship of the tumor to surrounding tissues was investigated. Eighteen patients (45%) had surrounding tissue invasion (Table  2). CT characteristics such as shape (circle, oval, irregular), contour (lobulated, irregular, smooth), necrosis, lymph node enlargement, surrounding organization, pathologic invasion pattern, internal density (homogeneous, heterogeneous), presence of cystic component, pleural/pericardial effusion and calcification were compared with histologic subtypes (  Correlation between the CT and pathologic tumor sizes among histologic subtypes are summarized in Table 4. There was no statistically significant difference between CT and pathologic tumor sizes among the histologic groups (p>0,05). Correlation analysis also showed a strong correlation between pathologic and CT tumor sizes (correlation coefficient (r): 0.952). The distribution of pathologic and CT sizes among histologic subtypes are summarized in Table 5.

DISCUSSION
The objective of this study was to evaluate which CT characteristics were associated with the pathologic features and malignant potential of anterior mediastinal masses focusing on thymic malignancies. Although we found that some CT parameters were significantly associated with the malignancy of the tumor, most of the features were found to be independent from the malignant potential of thymic tumors. Currently, CT imaging is considered the preferred method for the initial assessment and follow-up for patients with anterior mediastinal masses [8]. In the present study, we primarily focused on the CT features of thymic tumors due to the very limited number of non-thymic anterior mediastinal masses. There are limited studies comparing CT appearance of thymic malignancies with staging of the neoplasm and the results are controversial [9][10][11][12]. Although several studies have indicated that CT imaging is not adequate to differentiate the histological subtypes of thymoma, some studies showed that CT scanning can sufficiently predict histologic classification [6,8,13,14]. Some studies indicated that heterogeneous density of the tumor in CT scan was suggestive of higher stage [10,12]. In Shen's study, there was no correlation between the internal density of the tumor and staging [8]. Results of our