Age‐ and gender‐specific disease distribution and the diagnostic accuracy of CT for resected anterior mediastinal lesions

Background Anterior mediastinal lesions account for approximately half of all mediastinal masses and computed tomography (CT) is known to exhibit limited differentiating performance. Our aim was to evaluate the age‐ and gender‐specific distribution of anterior mediastinal lesions and the diagnostic accuracy of multi‐detector CT (MDCT). Methods This retrospective study included 549 consecutive patients with proven anterior mediastinal lesions and diagnostic MDCT scans. The age‐ and gender‐specific distribution of proven diagnoses and diagnostic accuracy were reviewed. The CT features of malignant and benign diseases having the lowest accuracy were compared with those of the most commonly misdiagnosed diseases. Results The proportion of malignancy showed a V‐shape relationship with age (lowest, 52.7% [50s]). The most prevalent malignancies were lymphoma (20s), lymphoma/thymoma (30s), thymoma (40s–50s), and thymoma/thymic carcinoma (≥60s). The most prevalent benign diseases were thymic remnant/hyperplasia (20s–30s), and thymic bed cyst (≥40s). The first‐choice diagnostic accuracy of MDCT decreased with age regardless of gender: 75.4% (20s), 75.0% (30s), 67.8% (40s), 58.5% (50s), and 53.4% (≥60s), primarily due to incorrect diagnoses of thymic bed cyst and thymic carcinoma (accuracy, 42.3% and 30.5%), which were prevalent in older patients and mostly misdiagnosed as thymoma. The most powerful differentiating MDCT features were water attenuation (≤20 HU) (OR, 42.7 [95%CI, 8.8–‐208.3], P < 0.001) for thymic bed cyst and mediastinal lymphadenopathy (6.8 [1.7–27.2], P = 0.006) for thymic carcinoma, but both showed low sensitivity (34.5% and 18.6%, respectively). Conclusions MDCT accuracy depended on age, owing to the age‐specific distribution of thymic carcinoma and thymic bed cyst, which frequently lacks distinguishable CT features from thymoma.


Introduction
Anterior mediastinal lesions account for approximately half of all mediastinal masses, 1,2 which include various disease entities from benign cysts to cancers. 3,4 Computed-tomography (CT) has been conventionally used for primary evaluation of anterior mediastinal lesions, but is known to exhibit limited differentiating performance. The previous study reported that single-channel chest CT offers modest diagnostic accuracy, with a correct first-choice diagnosis of 61% (95% CI, 52-69%; 78/126). 2 As a result, some anterior mediastinal pathologies still receive incorrect imaging diagnoses, leading to nontherapeutic thymectomy in 22% to 44% of cases. 5,6 The distribution of anterior mediastinal lesions depends on age and gender 3 and chest CT scan is primarily obtained on multi-detector CT (MDCT) scanners in current routine practice. Nevertheless, how the diagnostic performance of chest MDCT for anterior mediastinal lesions is affected by age and gender in a sufficient number of cases has been limitedly evaluated. Thus, the purpose of this study was to evaluate the age-and sex-specific distribution of anterior mediastinal lesions and MDCT diagnostic accuracy.

Methods Patients
The Institutional Review Board of our hospital approved this retrospective study and the requirement for informed consent was waived. According to a pathologic database search performed by the study coordinator (J.G.N.), patients who met the following inclusion criteria were included (Fig 1): (i) pathologically proven anterior mediastinal lesion between January 2006 and November 2014; (ii) MDCT images obtained with the standardized protocol for chest CT; (iii) a mean interval between pathologic diagnosis and MDCT of no longer than 1.5 months; and (iv) no history of previous treatment such as chemotherapy. For some lymphoma patients (34/78), a pathologic confirmation was obtained from a body part other than the anterior mediastinum, and all were cervical lymph nodes. Patients with poor CT image quality and indeterminate pathologic diagnosis were excluded. Finally, 549 consecutive patients with a mean age (AE SD) of 50. 6

MDCT image acquisition
Because of the rapid development of MDCT technology, CT scans were performed using various MDCT scanners with 4-(n = 63), 8-(n = 79), 16-(n = 96), 64-(n = 282), and 320-rows (n = 29). The scanning parameters were a tube voltage of 120 kVp with automatic exposure control of the tube current, slice thickness of 1.0-3.0 mm, and a reconstruction interval of 0.6-3.0 mm in most of CT scans, along with 3 mm coronal multi-planar reformatted images. Patients were scanned craniocaudally from the lung apex to the costophrenic angle in the supine position at full inspiration during a single breath-hold. A total of 90 mL of iodinated contrast media was injected into an antecubital vein at 3.0 mL/second, followed by a saline chase of 30 mL at the same rate. The postcontrast scan was usually taken 60 seconds after the contrast media injection.

Analysis of MDCT diagnostic accuracy
During the study period, one of four expert thoracic radiologists (J.M.G., C.H.L., H.J.L., and C.M.P., with 12-26 years of clinical experience) primarily read chest CT scans and wrote the formal radiology reports in routine clinical practice. One author (J.G.N.) extracted the first-choice and top three differential diagnoses from the formal reports and assessed the proportion of cases where the radiologic diagnosis was consistent with the pathologic diagnosis. We excluded a total of 17 cases where the radiology report was made after pathologic confirmation.

Analysis of MDCT findings
All MDCT images were retrospectively reviewed by two radiologists (J.G.N. and S.H.Y., with three and 12 years of experience in the interpretation of thoracic imaging) in consensus. CT images were evaluated for morphologic features, ancillary findings, and enhancement features on a picture archiving and communication system workstation (Infinitt, Seoul, Korea) in soft-tissue window settings (width, 400 HU; level, 20 HU). The location of tumors was     categorized as superior or inferior and as median or eccentric according to the relative location of the tumor epicenter and the heart. 7 The lesion was defined to have a superior location if the epicenter of the lesion was located above the heart and to have an eccentric location when the epicenter was located off-midline more than a half-width of the lesion. The size of the tumor was measured bidirectionally and the contour, margin, and shape of the tumor were evaluated. The shape was classified as round if the long-to-short-axis ratio was <1.5, and oval or irregular if the dimension ratio was ≥1.5. 2,7,8 The presence of a focal cystic or necrotic portion, calcification, gross fat, or adjacent organ invasion was also analyzed. 2,7,8 Adjacent organ invasion included invasion of the heart, lung, major vessels, or chest wall. 7,8 The ancillary findings included the presence of pleural effusion, pericardial effusion, mediastinal or hilar lymph node (LN) enlargement (short diameter ≥ 10 mm), distant metastasis, or satellite lesions. 2,8 The CT attenuation value of the lesion was measured using a circular region-of-interest in precontrast (n = 406) and postcontrast (n = 481) axial images. If a postcontrastenhanced scan was performed, the attenuation value was obtained from the area of the lesion that had the greatest attenuation on the postcontrast-enhanced images, excluding the necrotic or cystic portion. Based on the previous cut off, 9 the proportion of lesions with water attenuation of 20 HU or less in pre-and postcontrast images was calculated.

Statistical analysis
The proportions of disease entities according to age and gender were compared using the chi-square test. The chisquare test and Fisher exact test were used for examining statistical differences between the proportion of correct first-choice diagnoses of a particular disease and that of malignant/benign diseases other than that disease. MDCT features were compared between thymic bed cyst and thymoma, and between thymoma and thymic carcinoma, using the Student t-test or the chi-square test, as appropriate. Variables with a P-value <0.10 in the univariate analysis were included in the multivariate logistic regression analysis through the enter method. To examine the impact of differentiating CT features on the MDCT diagnosis, the proportion of correct diagnoses was compared using the Fisher exact test according to the presence of the significant CT features. Statistical analyses were performed using MedCalc version 15.8 (MedCalc, Mariakerke, Belgium). For all tests, P-values <0.05 were considered statistically significant.

Age-and gender-specific distribution of diseases
The comprehensive age-and gender-specific distribution is shown in Figure 2 and Table S1. Malignancy showed a V- The most prevalent benign diseases were thymic hyperplasia or remnant in their 20s-30s and thymic bed cyst in their 40s. Thymic bed cyst is predominant pathology of benign anterior mediastinal lesions in 50s or older, irrespective of gender.
Thymic carcinoma versus thymoma: In the univariate analysis, thymic carcinoma tended to be more ill-defined (P < 0.0001), and to more frequently have focal cystic/necrotic changes (P = 0.002), adjacent organ invasion (P < 0.0001), pleural effusion (P = 0.004), and mediastinal LN enlargement (P < 0.0001). A significantly higher proportion of patients in their 60s or older were found in patients with thymic carcinoma (P = 0.0003).

Discussion
Our study reported detailed age-and gender-specific distributions of resected malignant and benign anterior mediastinal lesions. The proportion of correct first-choice MDCT diagnoses was still limited, ranging about 62.3% and gradually decreased as age increased regardless of genders from  75.4% in the 20s to 53.4% in ≥60s. This age-dependent decrease of MDCT accuracy resulted from age-dependent increase of thymic bed cyst and thymic carcinoma. The most strongly differentiating CT features for thymic bed cyst and thymic carcinoma were water attenuation and mediastinal LN enlargement, respectively. Those CT features had high specificities (98.4% and 97.3%, respectively), but low sensitivities (34.5% and 18.6%, respectively).
In accordance with a previous study reporting the ageand gender-specific distribution of anterior mediastinal masses 3 in patients <40 years, the prevalent diseases in our study were lymphoma, malignant germ cell tumor, and thymoma in men and lymphoma, thymoma, and benign teratoma in women. However, in patients in their 40s or older, thymoma, thymic bed cyst, and thymic carcinoma accounted for more than three-fourths, unlike the previous study (thymoma and thyroid goiter). Patients with thyroid goiter were not included in this study, as it was typically diagnosed based on CT and rarely required pathologic confirmation. In addition, we separated thymic bed cyst and thymic carcinoma from thymoma.
There is a paucity of literature focused on the agespecific proportion for thymic bed cyst and thymic carcinoma. While thymoma showed a constant proportion of around 30-40% regardless of age in patients ≥40 years, thymic bed cyst had an inverse V-shaped relationship, with a peak in the 50s (42.1%), and thymic carcinoma had a gradual increase with age, with a peak in the 70+ group (30.0%). We confirmed the latter trend by comparing the mean ages of patients with thymoma and thymic carcinoma, as the mean age of patients with thymic carcinoma tended to be higher than those with thymoma, except for type A thymoma (mean age of thymoma, 63 years in type A; 54 years in type AB; 49 years in type B1; 48 years in type B2, 50 years in type B3; 57 years in type C). 10 MDCT provided a proportion of correct first-choice diagnoses of 62.3% (95% CI, 58.  2 More than 80% of resected thymic bed cysts and thymic carcinoma were radiologically misdiagnosed as thymomas resulting in age-dependent decrease of MDCT accuracy. Conventionally, CT differentiation of cysts from tumors is made based on the attenuation value of water (20 HU or less). However, only 34.5% of thymic bed cysts showed water attenuation in our study. Indeed, prior studies including a small number of thymic cysts reported that three-fourths or more of thymic cysts had CT attenuation equal to that of the chest wall muscle, which was attributed to high calcium or protein contents, hemorrhage, or streak artifacts. 11,12 Focal necrosis existed in 33.3% of thymomas in our results, consistent with the results of other studies, where focal necrosis was found in 28-42% of thymomas. In accordance with previous studies, 2,5 thymic bed cysts tended to have a smooth contour and median location, while thymomas tended to be located off-midline.
The differentiating CT features between thymoma and thymic carcinoma have been previously evaluated in a few studies. [13][14][15] In our study, in the multivariate analysis, age older than 60, mediastinal or hilar LN enlargement, and adjacent organ invasion were found to be significant. Likewise, in the literature, [13][14][15] mediastinal LN enlargement and adjacent organ invasion were characteristic CT features of thymic carcinoma, but were only present in 18.6% and 33.9% of cases, showing low sensitivity. Accordingly, even in cases suspected to be thymoma without those CT features, the possibility of thymic carcinoma should be considered, especially for patients older than 60.
Our study has some limitations. First, there was an inevitable selection bias due to its retrospective nature. Second, as we only included only surgically treated cases, the diagnostic accuracy of benign diseases, especially thymic bed cysts or thymic hyperplasia, might have been underestimated. Third, the CT scans were taken with various CT scanners, so the direct comparison of CT attenuation should be carefully interpreted. Finally, we only evaluated the role of conventional CT and did not explore the added role of other imaging techniques, such as dual-energy CT, MRI, or PET.
In conclusion, according to the characteristic age-specific distributions of malignant and benign anterior mediastinal lesions, especially for thymic carcinoma and thymic bed cysts, the MDCT accuracy decreased depending on age regardless of gender. The most strongly differentiating CT features (water attenuation for thymic bed cyst, mediastinal LN enlargement and adjacent organ invasion for thymic carcinoma) had high specificity but low sensitivity. Understanding this age-dependent distribution and MDCT accuracy may potentially help make a correct diagnosis of anterior mediastinal lesions and considering additional noninvasive imaging work-up including MRI.