Prognostic factors
Table 3 illustrates the analysis of predictors (univariable and multivariable analysis) using 3 end-points: OS, CSS and FFR.
Table 3 Univariable and multivariable analysis
Univariable Multivariable
95% CI 95% CI
HR P-value Lower Upper HR P-value Lower Upper
Overall survival
 Sex (male) 1.508 0.322 0.669 3.397 2.438 0.092 0.865 6.871
 Age (continuous) 1.026 0.161 0.990 1.063 1.017 0.395 0.978 1.058
 Myasthenia gravis (yes) 0.234 0.019 0.069 0.790 0.551 0.459 0.114 2.666
 TC vs thymoma 6.315 0.000 2.652 15.040 6.506 0.002 1.956 21.646
 Incomplete resection 3.916 0.001 1.727 8.882 6.696 0.003 1.944 23.060
 Pleural surgery
  Primary pleural surgery vs pleural surgery for recurrencea 4.392 0.017 1.302 14.813 1.493 0.603 0.330 6.757
 Type of pleural surgery
  EPP vs LP + TP 2.351 0.040 1.041 5.312 2.491 0.122 0.784 7.913
 Preoperative therapy (yes)b 3.086 0.013 1.271 7.463 1.495 0.427 0.554 4.032
 Postoperative therapy (yes)c 2.558 0.030 1.093 5.988 1.733 0.293 0.622 4.831
Cause-specific survival
 Sex (male) 1.740 0.331 0.569 5.325 3.206 0.128 0.715 14.372
 Age (continuous) 1.041 0.108 0.991 1.093 1.039 0.197 0.980 1.101
 Myasthenia gravis (yes) 0.022 0.110 0.000 2.387 0.000 0.952 0.000 171.39
 TC vs thymoma 12.129 0.000 3.867 38.044 13.144 0.001 2.726 63.370
 Incomplete resection 2.004 0.255 0.606 6.628 2.053 0.387 0.402 10.487
 Pleural surgery
  Primary pleural surgery vs pleural surgery for recurrencea 2.198 0.239 0.593 8.130 1.213 0.836 0.194 7.576
 Type of pleural surgery
  EPP vs LP + TP 2.301 0.155 0.729 7.265 1.063 0.934 0.252 4.488
 Preoperative therapy (yes)b 2.179 0.180 0.697 6.803 1.072 0.914 0.593 3.788
 Postoperative therapy (yes)c 2.066 0.203 0.675 6.329 1.020 0.976 0.288 3.613
Freedom from recurrenced
 Sex (male) 1.541 0.096 0.926 2.566 1.800 0.033 1.050 3.086
 Age (continuous) 1.007 0.523 0.985 1.031 1.009 0.504 0.983 1.036
 Myasthenia gravis (yes) 0.623 0.090 0.361 1.077 0.661 0.208 0.346 1.260
 TC vs thymoma 2.748 0.006 1.330 5.676 2.442 0.027 1.106 5.388
 Pleural surgery
  Primary pleural surgery vs pleural surgery for recurrencea 1.138 0.629 0.002 1.919 1.131 0.676 0.633 2.024
 Type of pleural Surgery
  EPP vs LP + TP 1.491 0.165 0.849 2.618 1.467 0.241 0.773 2.784
 Preoperative therapy (yes)b 0.852 0.542 0.508 1.427 0.577 0.068 0.319 1.042
 Postoperative therapy (yes)c 0.750 0.270 0.450 1.250 0.849 0.578 0.476 1.513
TC: thymic carcinoma; LP: local pleurectomy; TP: total pleurectomy; EPP: extrapleural pneumonectomy.
a Analysis of primary pleural surgery versus pleural surgery for recurrence.
b Patients received (pseudo-)neoadjuvant therapy, including chemotherapy ± radiotherapy.
c Patients received (pseudo-)adjuvant therapy, including chemotherapy ± radiotherapy.
d Freedom from recurrence was analysed only for patients after R0 resection.
At univariable analysis, the presence of MG was beneficial at increasing OS [hazard ratio (HR) 0.234; P = 0.019], while TC, incomplete resection, TC vs thymomas, primary pleural surgery vs surgery for pleural recurrence, type of pleural surgery (EPP vs TP/LP), preoperative and postoperative therapy had a negative impact on OS.
TCs in comparison with thymomas at multivariable analysis had a negative impact on OS [HR 6.506; P = 0.002], CSS [HR 13.144; P = 0.001] and FFR [HR 2.442; P = 0.027], respectively. Further, incomplete resections predicted worse OS [HR 6.696; P = 0.003] and male sex predicted worse FFR [HR 1.800; P = 0.033] at multivariable analysis. In order to eliminate the potential strong bias of incomplete resection on other predictors, we also performed multivariable analysis only on patients after complete resection (R0): negative impact on OS male sex [HR 3.176; P = 0.025], TC [HR 3.988; P = 0.013], primary pleural surgery compared with surgery for pleural recurrence [HR 4.132; P = 0.040]; no effect was found for MG, age, type of surgery, preoperative or postoperative treatments. Analysis of multimodality therapy (all patients with [pseudo-]neo- and [pseudo-]adjuvant therapy combined: n = 126) vs surgery alone (n = 26) revealed no statistically significant differences.
Univariable analysis for the number of resected nodules (n = 122) with a cut-off value of 4 (median number of resected nodules) for dichotomizing groups revealed no statistically significant differences.