Clinical Data and CT Image Features
Among all of the 194 patients, 32 patients were confirmed of low-risk thymomas (including 7 type A, 17 type AB and 8 type B1), 52 patients were confirmed of high-risk thymomas (including 24 type B2, 11 mixed type B2 and B3, 17 type B3), and 110 patients were confirmed of thymic carcinomas. The longest diameter (mean ± SD) between thymic carcinomas (7.84 ± 2.37 cm), high-risk thymomas (7.67 ± 3.88 cm) and low-risk thymomas (6.52 ± 3.47 cm) has no statistical significance (F = 2.409, P = 0.093), while the longest diameter in thymic carcinoma was larger than that of low-risk thymomas with statistical significance (t=-2.031, P = 0.049). The average age of patients with thymic carcinomas (56.8 ± 11.8 years), high-risk thymomas (50.4 ± 13.3 years) and low-risk thymomas (56.4 ± 11.7) has statistical significance (F = 3.909, P = 0.022).
The presence of irregular shape, heterogeneous enhancement, necrotic or cystic change, mediastinal invasion, lung invasion, vessel invasion, lymphadenopathy, pericardial mass, pericardial effusion, pleural effusion, pleural metastasis, lung metastasis and distant organ metastasis were more frequently seen in thymic carcinomas than that of high-risk thymomas and low-risk thymomas (P < 0.05). The main CT features of the 194 patients are summarized in Table 1.
Table 1
CT features of 194 patients with thymic epithelial tumors
Characteristics | No. of patients (%) |
Thymic carcinomas | High-risk thymomoas | Low-risk thymomas |
Tumor size (cm) | 7.84 ± 2.37 | 7.67 ± 3.88 | 6.52 ± 3.47 |
Tumor shape | | | |
Regular | 5 (4.5) | 15 (28.8) | 21 (65.6) |
Irregular | 105 (95.5) | 37 (71.2) | 11 (34.4) |
Enhancement pattern | | | |
Homogenous | 23 (20.9) | 20 (38.5) | 15 (46.9) |
Heterogeneous | 87 (79.1) | 32 (61.5) | 17 (53.1) |
Calcification | | | |
Present | 44 (40.0) | 21 (40.4) | 9 (28.1) |
Absent | 66 (60.0) | 31 (59.6) | 23 (71.9) |
Necrosis | | | |
Present | 83 (75.5) | 27 (51.9) | 14 (43.8) |
Absent | 27 (24.5) | 25 (48.1) | 18 (56.2) |
Mediastinal invasion | | | |
Present | 106 (96.4) | 46 (88.5) | 4 (12.5) |
Absent | 4 (3.6) | 6 (11.5) | 28 (87.5) |
Lung invasion | | | |
Present | 46 (41.8) | 19 (36.5) | 0 (0.0) |
Absent | 64 (58.2) | 33 (63.5) | 32 (100.0) |
Vessel invasion | | | |
Present | 39 (35.5) | 7 (13.5) | 0 (0.0) |
Absent | 71 (64.5) | 45 (86.5) | 32 (100.0) |
Lymphadenopathy | | | |
Present | 67 (60.9) | 4 (7.7) | 1 (3.1) |
Absent | 43 (39.1) | 48 (92.3) | 31 (96.9) |
Pericardial mass | | | |
Present | 19 (17.3) | 9 (17.3) | 0 (0.0) |
Absent | 91 (82.7) | 43 (82.7) | 32 (100.0) |
Pleural metastasis | | | |
Present | 27 (24.5) | 12 (23.1) | 32 (100.0) |
Absent | 83 (75.5) | 40 (76.9) | 0 (0.0) |
Lung metastasis | | | |
Present | 30 (27.3) | 7 (13.5) | 0 (0.0) |
Absent | 80 (72.7) | 45 (86.5) | 32 (100.0) |
Distant organ metastasis | | | |
Present | 20 (18.2) | 1 (1.9) | 0 (0.0) |
Absent | 90 (81.8) | 51 (98.1) | 32 (100.0) |
Pleural effusion | | | |
Present | 40 (36.4) | 15 (28.8) | 0 (0.0) |
Absent | 70 (63.6) | 37 (71.2) | 32 (100.0) |
Pericardial effusion | | | |
Present | 49 (44.5) | 8 (15.4) | 1 (3.1) |
Absent | 61 (55.5) | 44 (84.6) | 31 (96.9) |
* Mean ± SD |
Prognostic Analysis
In the group of thymic carcinomas, 20 patients had tumor progress, 12 patients had local relapse and 14 patients had metastases occurred within 3 years since first CT examination. These 46 patients (41.8%) were categorized as the poor outcome group, within which 32 patients died. Sixty-four patients (58.2%) survived without any evidence of tumor progress, relapse or metastases since first CT examination, and were categorized as the good outcome group. In univariate analysis, radiologic features including tumor size, enhancement pattern, necrosis or cystic change, lung invasion, great vessel invasion, pericardium mass, pleural metastases, lung metastases, lymphadenopathy, pleural effusion, pericardial effusion and distant organ metastasis were associated with poor clinical outcomes (P < 0.05). Multivariate logistical regression analysis showed that only vessel invasion and pericardium mass remained the significant independent predictor (Table 2, Fig. 1). The presence of vessel invasion and pericardium mass was significantly correlated with poor clinical outcome, with an OR of 18.61 (p = 0.000) and 8.50 (p = 0.029). Further ROC curve analysis showed that the area under the curve (AUC) was 0.867 (Fig. 2A), suggesting the multivariate logistic regression model is a reasonable predictor in this study.
Table 2
Independent predictors by logistical regression analysis
Factors | Category | B value | P value | OR (95% CI) |
Vessel invasion* | Presence | 2.924 | 0.000 | 18.612(6.811, 50.862) |
Pericardium mass* Pericardium mass# | Presence Presence | 2.140 3.530 | 0.029 0.000 | 8.500(1.247, 57.931) 34.125(5.215, 223.280) |
*: thymic carcinomas; #: high-risk thymomas |
In the group of high-risk thymomas, 7 patients had tumor progress, 3 patients had local relapse and 1 patient had hepatic metastases occurred within 3 years since first CT examination. These 11 patients (21.1%) were categorized as the poor outcome group, within which 7 patients died. Forty-one patients (78.9%) survived without any evidence of tumor progress, relapse or metastases since first CT examination and were categorized as the good outcome group. In univariate analysis, radiologic features including lung invasion, pericardium mass, pleural metastases, lung metastases, lymphadenopathy, pericardial effusion and pleural effusion were associated with poor clinical outcomes (P < 0.05). Multivariate logistical regression analysis showed that only pericardium mass remained the significantly independent predictor (Table 2). The presence of pericardium mass was significantly correlated with poor clinical outcome (Fig. 3), with an OR of 34.12 (p = 0.000). Further ROC curve analysis showed that the area under the curve (AUC) was 0.794 (Fig. 2B), suggesting the multivariate logistic regression model is a reasonable predictor in this study.
In the group of low-risk thymomas, no patient was found with any evidence of tumor progress, relapse or metastases since the first CT examination, and the univariate analysis was waived in this group.
Survival Analysis
In the group of thymic carcinomas, 32 patients (29.1%) died during regular follow up process and 78 patients (70.9%) survived by the time of the last follow-up visit. In univariate analysis, radiologic features including tumor size, necrosis or cystic change, lung invasion, great vessel invasion, pericardium mass, pleural metastases, lung metastases, lymphadenopathy, pleural effusion and distant organ metastasis were associated with survival status (P < 0.05). Cox analysis showed lung invasion, great vessel invasion, lung metastases and distant organ metastasis (Fig. 4) remained the significantly independent predictor (Table 3).
Table 3
Independent predictors by Cox regression analysis in thymic carcinomas
Factors | Category | B value | P value | OR (95% CI) |
Lung metastases | Presence | 0.912 | 0.012 | 2.490(1.221,5.080) |
Distant metastasis | Presence | 1.278 | 0.001 | 3.590(1.657,7.778) |
Lung invasion | Presence | 1.439 | 0.000 | 4.217(1.878,9.473) |
Vessel invasion | Presence | 0.868 | 0.025 | 2.383(1.115,5.094) |
In the group of high-risk thymomas, 7 patients (13.5%) died during regular follow up process, and 45 patients (86.5%) survived by the time of the last follow-up visit. In univariate analysis, radiologic features including tumor size, enhancement pattern, lung invasion, pericardium mass, lung metastases, pericardial effusion and pleural effusion were associated with survival status (P < 0.05). Cox analysis showed that only lung invasion and pericardium mass (Fig. 3) remained the significantly independent predictor (Table 4).
Table 4
Independent predictors by Cox regression analysis in high-risk thymomas
Factors | Category | B value | P value | OR (95% CI) |
Pericardium mass | Presence | 4.969 | 0.012 | 143.879(3.029,6833.33) |
Lung invasion | Presence | 3.063 | 0.032 | 21.386(1.307,349.88) |
In the group of low-risk thymomas, no patients died of thymic tumors by the time of the last follow-up visit, thus survival analysis was waived in this group.