TNM classifications have been established for various

Lymphogenous and Hematogenous Metastasis of Thymic Epithelial Tumors Kazuya Kondo, MD, PhD, and Yasumasa Monden, MD, PhD Department of Oncological and Regenerative Surgery, School of Medicine, University of Tokushima, Tokushima, Japan GENERAL THORACIC Background. A TNM classification has been established for various tumors. However, the TNM classification of thymic epithelial tumor has not been established yet. Methods. We received replies to a questionnaire on thymic epithelial tumors from 115 institutes. We compiled a database of 1,320 patients with thymic epithelial tumor (1,093 thymomas, 186 thymic carcinomas, and 41 thymic carcinoids) who were treated between 1990 and 1994. We used a tentative TNM classification of thymoma presented by Yamakawa and associates in 1991. The regional lymph nodes of the thymus were classified into three groups: anterior mediastinal lymph nodes (N1), intrathoracic lymph nodes (N2), and extrathoracic lymph nodes (N3). Results. The rate of lymphogenous metastasis in thymoma, thymic carcinoma, and thymic carcinoid was 1.8%, 27%, and 28%, respectively. Most tumors with lymph node metastasis metastasized to N1 (thymoma, 90%; thymic carcinoma, 69%; thymic carcinoid, 91%). The 5-year survival rates of N0, N1, and N2 thymoma were 96%, 62%, and 20%, respectively. The 5-year survival rates of N0, N1, N2, and N3 thymic carcinoma were 56%, 42%, 29%, and 19%, respectively. The 5-year survival rates of M0 and M1 thymoma were 95% and 57%. The 5-year survival rates of M0 and M1 thymic carcinoma were 51% and 35%. Multivariate analysis demonstrated that survival of patients with thymoma was dependent on the clinical stage of Masaoka and complete resection. In thymic carcinoma, survival was dependent on lymph node metastasis and complete resection. Conclusions. The N factor was one of the predictors of survival in thymoma and thymic carcinoma. However, M factor showed less influence on survival than T or N factors. (Ann Thorac Surg 2003;76:1859 65) 2003 by The Society of Thoracic Surgeons TNM classifications have been established for various tumors, and we can predict the prognosis of a patient with a malignant tumor and choose the therapy by the clinical stage on the basis of the classification. However, the TNM classification of thymic epithelial tumor has not been established yet. We think that the main reason for this is that the influence of the N or M factor on prognosis in thymic epithelial tumors is not clear. The clinical staging of thymoma, which is one of the thymic epithelial tumors, has been revised and updated during the years [1 3]. The idea of clinical staging of thymoma was introduced by Bergh and colleagues [1] and later modified by Wilkins and Castleman [2], and was almost established by Masaoka and associates in 1981 [3]. The staging system of Masaoka and associates is the most popular system for thymoma now. In this system, the state of local invasion by thymoma (T factor) is strongly emphasized in comparing lymphogenous and hematogenous metastasis (N and M factors) because of the rarity of lymphogenous or hematogenous metastasis in thymoma. Although the T factor is classified into four categories (stages I, II, III, and IVA) according to the Accepted for publication June 3, 2003. Address reprint requests to Dr Kondo, Department of Oncological and Regenerative Surgery, School of Medicine, University of Tokushima, Kuramoto-cho, Tokushima 770-8503, Japan; e-mail: kondo@clin.med. tokushima-u.ac.jp. degree of local invasion, both lymphogenous and hematogenous metastasis are classified as stage IVB. Suster and Moran [4] emphasized that a system for pathologic and clinical staging should be adapted to the entire spectrum of thymic epithelial tumors. Thymic carcinoma or carcinoid frequently metastasizes lymphogenously or hematogenously [5 7]. It is necessary to analyze the pattern of lymphogenous or hematogenous metastasis and determine whether or not these metastases influence the prognosis in thymic carcinoma and carcinoid. However, because of the rarity of the thymic carcinoma and carcinoid, most published series involve only small numbers of patients. In our previous study, we compiled a database of 1,093 thymomas, 186 thymic carcinomas, and 41 thymic carcinoids treated between 1990 and 1994 at 115 institutes in Japan [8]. We demonstrated that the clinical stage of Masaoka and coworkers [3] is an excellent system for predicting prognoses not only in thymoma but also in thymic carcinoma, and that radical surgery is the most important factor in survival rate from thymic epithelial tumor [8]. However, it is unknown whether the addition of radiation or chemotherapy to the surgery can prolong survival and prevent the recurrence in invasive thymoma and thymic carcinoma [8]. In the present study, we analyzed the pattern of lymphogenous or hematogenous metastasis of thymic epithelial tumor, and determined whether N or M factor can be predictors of survival. We would hope to contribute to the estab- 2003 by The Society of Thoracic Surgeons 0003-4975/03/$30.00 Published by Elsevier Inc doi:10.1016/s0003-4975(03)01017-8

GENERAL THORACIC 1860 KONDO AND MONDEN Ann Thorac Surg METASTASIS OF THYMIC EPITHELIAL TUMORS 2003;76:1859 65 Table 1. TNM Classification of Thymic Epithelial Tumors T factor T1: Macroscopically completely encapsulated and microscopically no capsular invasion T2: Macroscopically adhesion or invasion into surrounding fatty tissue or mediastinal pleura, or microscopic invasion into capsule T3: Invasion into neighboring organs, such as pericardium, great vessels, and lung T4: Pleural or pericardial dissemination N factor N0: No lymph node metastasis N1: Metastasis to anterior mediastinal lymph nodes N2: Metastasis to intrathoracic lymph nodes except anterior mediastinal lymph nodes N3: Metastasis to extrathoracic lymph nodes M factor M0: No hematogenous metastasis M1: Hematogenous metastasis lishment of a TNM classification for thymic epithelial tumor. Material and Methods Patients We sent a questionnaire on thymic epithelial tumors to 185 institutes certified by the Japanese Association for Chest Surgery, and received replies from 115 institutes (62%). We compiled a database of 1,320 patients with thymic epithelial tumor who were treated between 1990 and 1994. These patients included 1,093 thymomas (82.8%), 186 thymic carcinomas (14.1%), and 41 thymic carcinoids (3.1%). Final pathologic staging was decided by a tentative TNM classification of thymoma (Table 1) presented by Yamakawa and associates in 1991 [9]. The pleural spread of thymoma was considered T4. The regional lymph nodes of the thymus were classified into three groups: anterior mediastinal lymph nodes (N1), intrathoracic lymph nodes except anterior mediastinal lymph nodes (N2), and extrathoracic lymph nodes (N3). When the lymph node metastasis to N1, N2, or N3 was positive, we denoted N1, N2, or N3, respectively. When the lymph node metastasis to N1, N2, or N3 was negative, we denoted N1,N2,orN3, respectively. When any of N1, N2, or N3 was positive, we denoted N. And when all of N1, N2 and N3 were negative, we denoted N. We decided that the postoperative pathologic findings judged the lymph node metastasis for operable cases, and that computed tomography or magnetic resonance imaging judged it for inoperable cases. Statistical Analysis Survival analysis was performed by the method of Kaplan-Meier, and comparisons of survival by the logrank test were made. Multivariate analysis by the Cox proportional hazards model using SPSS for Windows (version 11.0.1; SPSS Inc, Chicago, IL) was performed. Possible factors for inclusion in the model consisted of sex, age, histologic type (spindle cell type versus not spindle cell type in thymoma, squamous cell carcinoma versus not squamous cell carcinoma in thymic carcinoma), complications (myasthenia gravis, pure red cell aplasia, and hypergammaglobulinemia or hypogammaglobulinemia), clinical stage by Masaoka and coworkers [3] (I and II versus III and IV), lymph node metastasis, distant metastasis, and type of surgery (complete resection versus subtotal resection and inoperable). All clinical data refer to our previous reports [8]. Significance was defined as a p value less than 0.05. Deaths as a result of complications (eg, myasthenia gravis and pure red cell aplasia) or unrelated disease were excluded. Results Lymph Node Metastasis of Thymic Epithelial Tumors Figure 1 demonstrates the lymph node metastasis of thymoma, thymic carcinoma, and thymic carcinoid. Only 19 cases (1.8%) of 1,064 thymomas metastasized to lymph nodes. N thymoma consisted of 14 (N1 and N2 ) thymomas, 3 (N1 and N2 ) ones, and 2 (N1 and N2 ) ones. None of the thymomas was N3 (Fig 1A). Forty-nine cases (26.8%) of 183 thymic carcinomas metastasized to lymph nodes. N carcinoma consisted of 34 N1 (69.4%), 17 N2 (34.7%), and 15 N3 (30.6%) carcinomas. Nineteen cases (55.9%) of N1 carcinoma metastasized to N1 alone (N1 and N2 and N3 ). Thirteen cases (38.2%) of N1 carcinoma metastasized to either N2 or N3 (N1 and N2 and N3, 10 cases [29.4%]; N1 and N2 and N3, 3 cases [8.8%]). Two cases (5.9%) of N1 carcinoma metastasized to N2 and N3 (N1 and N2 and N3 ). Five cases (29.4%) of N2 carcinoma metastasized to N2 alone (N1 and N2 and N3 ). Ten cases (66.7%) of N3 carcinoma metastasized to N3 alone (N1 and N2 and N3 ). There was no case with (N1 and N2 and N3 ) carcinoma (Fig 1B). Eleven (27.5%) of 40 thymic carcinoids were N.N carcinoids consisted of 10 N1 (90.9%), 5 N2 (45.5%), and3n3 (27.3%) carcinoids. Three cases (30%) of N1 carcinoids metastasized to N1 alone (N1 and N2 and N3 ). Five cases (50%) of N1 carcinoids metastasized to N2 (N1 and N2 and N3 ). Two cases (20%) of N1 carcinoids metastasized to N3 (N1 and N2 and N3 ). Single metastasis of N2 or N3 was 0 and 1, respectively (Fig 1C). Lymph Node Metastasis and Invasion to Neighboring Organs Sixteen (5.6%) of 284 thymomas with invasion to neighboring organs were N. On the other hand, only 3 (0.003%) of 771 thymomas without invasion to neighboring organs were N. Forty-four (28.8%) of 153 thymic carcinomas with invasion to neighboring organs were N. Only 4 (14.8%) of 27 thymic carcinomas without invasion to neighboring organs were N. Nine (37.5%) of 24 thymic carcinoids with invasion of the neighboring organs were N, whereas only 2 (12.5%) of 16 thymic carcinoids without invasion of the neighboring organs were N.

Ann Thorac Surg KONDO AND MONDEN 2003;76:1859 65 METASTASIS OF THYMIC EPITHELIAL TUMORS 1861 GENERAL THORACIC Fig 2. Survival curve of thymic epithelial tumors according to N factor. (A) Survival curve of thymoma according to N factor (thin straight line) thymoma with N0 (dark dotted line) thymoma with N1 (thick line) thymoma with N2. (B) Survival curve of thymic carcinoma according to N factor (thin straight line) carcinoma with N0 (dark dotted line) carcinoma with N1 (thick line) carcinoma with N2 (light dotted line) carcinoma with N3. Fig 1. Lymph node metastasis of thymic epithelial tumors The number in each circle is the number of the case. (LN lymph node.) (A) Pattern of lymph node metastasis of thymoma. (B) Pattern of lymph node metastasis of thymic carcinoma. (C) Pattern of lymph node metastasis of thymic carcinoid. Survival Curve of Thymic Epithelial Tumors According to N Factor Figure 2 shows the survival curves according to N factor of the thymoma and thymic carcinoma. The 5-year survival rates of N (n 895), N1 and N2 (n 13), and N2 (n 5) thymoma were 95.6%, 61.5%, and 20%, respectively (Fig 2A). There was a significant difference in survival rate between thymoma without lymph node metastasis and thymoma with lymph node metastasis (p 0.0001). Although the survival rate of patients with N2 thymoma was worse than that of patients with (N1 and N2 ) thymoma, no significant difference between them was observed. The 5-year survival rates of N (n 104), N1 and N2 and N3 (n 20), N2 and N3 (n 15), and N3 (n 11) thymic carcinoma were 56.0%, 42.1%, 29.3%, and 18.8%, respectively (Fig 2B). A significant difference in survival rate was observed between thymic carcinoma without lymph node metastasis (N ) and thymic carcinoma with lymph node metastasis (N ; p 0.0028). There was a significant difference of survival rate between patients with N carcinoma and N2 and N3 carcinoma (p 0.0111), and between N1 and N2 and N3 carcinoma and N3 carcinoma (p 0.0167). There was a tendency for survival rate of patients with thymic

GENERAL THORACIC 1862 KONDO AND MONDEN Ann Thorac Surg METASTASIS OF THYMIC EPITHELIAL TUMORS 2003;76:1859 65 rate between M0 carcinoma and M1 carcinoma was observed. Multivariate Analysis of Survival in Thymoma and Thymic Carcinoma Survival of patients with thymoma was dependent on the clinical stage of Masaoka [3] (p 0.001) and complete resection (p 0.001; Table 2). Sex, age, histologic type, complications, lymph node metastasis, and distant metastasis were not factors predictive of survival. In thymic carcinoma, survival was dependent on lymph node metastasis (p 0.001) and complete resection (p 0.001). Sex, age histologic type, clinical stage of Masaoka [3], and distant metastasis were not factors predictive of survival. Fig 3. Survival curve of thymic epithelial tumors according to M factor. (A) Survival curve of thymoma according to M factor (straight line) thymoma without distant metastasis (dotted line) thymoma with distant metastasis. (B) Survival curve of thymic carcinoma according to M factor (straight line) carcinoma without distant metastasis (dotted line) carcinoma with distant metastasis. carcinoma to become worse according to progression of N factor. M Factor of Thymic Epithelial Tumors Only 13 (1.2%) of 1,073 thymomas had distant metastasis. Most of the metastatic sites in thymoma were lung (11 of 13, 84.6%). Twenty-two (12.0%) of 183 thymic carcinomas had distant metastasis. The metastatic sites in thymic carcinoma were lung (13 of 22, 59%), bone (5 of 22, 22.7%), and liver (5 of 22, 22.7%). Only 1 (2.5%) of 40 thymic carcinoids had distant metastasis. Figure 3 shows the survival curve according to M factor of the thymoma and thymic carcinoma. The 5-year survival rates of M0 (n 903) and M1 (n 10) thymoma were 95.0% and 57.1%, respectively. A significant difference in survival rate was observed between M0 thymoma and M1 thymoma (p 0.0001; Fig 3A). The 5-year survival rates of M0 (n 133) and M1 (n 16) thymic carcinoma were 51.3% and 34.7%, respectively (Fig 3B). No significant difference of survival Comment Several recent studies have reported prognostic factors for thymoma and thymic carcinoma using multivariate analysis [6, 10 20]. Clinical stage of Masaoka [10 12, 14 16], complete resection of the tumor [11, 13, 14], histologic classification (including that of Bernaz and associates [12, 14], one by Muller-Hermelink and coworkers [11, 15], and the World Health Organization classification [10]), Karnofsky performance status [12], and tumor size [14] are considered to be prognostic factors for thymoma. The current study also recognized clinical stage of Masaoka and complete resection of the tumor as prognostic factors of thymoma. In thymic carcinoma, clinical stage of Masaoka [17, 19], complete resection of the tumor [18, 19], histologic classification (squamous cell carcinoma versus other cancer) [6, 18, 19], Karnofsky performance status [18], and innominate vessel invasion [20] are considered to be prognostic factors. We recognized complete resection of the tumor as a prognostic factor for thymic carcinoma. Prognostic factors of multivariate analysis in the previous studies did not include N or M factors because of the rarity of thymoma with lymph node or distant metastasis and thymic carcinoma. The staging system of Masaoka [3] from 1981 is the most popular clinical staging system of thymoma. Our previous study demonstrated that this clinical staging system is an excellent system for predicting prognoses not only in thymoma but also in thymic carcinoma [8]. In the study, the 5-year survival rates of T1 N0 M0, T2 N0 M0, T3 N0 M0, and T4 N0 M0 thymoma were 100%, 98.3%, 89.2%, and 73.1%, respectively. A significant difference in survival rate was observed between T2 N0 M0 and T3 N0 M0 (p 0.0001) and between T3 N0 M0 and T4 N0 M0 (p 0.0015) [8]. Moreover, multivariate analysis in the current study demonstrated that survival of patients with thymoma was significantly dependent on the clinical stage of Masaoka (p 0.001). The 5-year survival rates of T1 plus T2 N0 M0, T3 N0 M0, and T4 N0 M0 thymic carcinomas were 88.2%, 54.5%, and 41.9%, respectively. A significant difference in survival rate was observed between the T1 plus T2 N0 M0 group and the T3 plus T4 N0 M0 (p 0.012) [8]. In patients with thymoma and thymic carcinoma without lymphogenous or hematogenous me-

Ann Thorac Surg KONDO AND MONDEN 1863 2003;76:1859 65 METASTASIS OF THYMIC EPITHELIAL TUMORS Table 2. Multivariate Analysis of Survival in Thymoma and Thymic Carcinoma Thymoma Thymic Carcinoma Feature Hazard Ratio p Value 95% CI Hazard Ratio p Value 95% CI GENERAL THORACIC Sex 0.001 0.999 0.552 1.809 2.468 0.116 0.346 1.124 Age 0.082 0.774 0.974 1.019 3.293 0.070 0.998 1.049 Histologic type a 0.417 0.518 0.067 3.892 1.087 0.297 0.371 1.353 Complications b 2.481 0.115 0.148 1.231 Stage c 17.978 0.001 4.976 78.562 1.187 0.276 0.580 6.713 Lymph node metastasis 3.758 0.053 0.991 4.540 10.239 0.001 1.452 4.724 Distant metastasis 0.108 0.743 0.399 3.624 0.052 0.819 0.466 2.620 Complete resection 13.615 0.001 1.764 6.386 11.421 0.001 1.495 4.541 a Spindle type versus others (thymoma), squamous cell carcinoma versus others (thymic carcinoma). hypergammaglobulinemia or hypogammaglobulinemia. c Stage I and II versus III and IV. CI confidence interval. b Myasthenia gravis, pure red cell aplasia, tastasis, the T factor was a valuable item that can predict the prognosis. Thymic carcinoma and carcinoid frequently show lymphogenous metastasis but thymoma does not. Lymphogenous metastasis was observed in 0% to 12% (mean, 2.6%) of patients with thymoma [9]. Sixteen percent to 35% of patients with thymic carcinoma and approximately 30% of patients with thymic carcinoid show lymphogenous or distant metastasis [6, 21, 22]. The current study demonstrated that the rates of lymphogenous metastasis in thymoma, thymic carcinoma, and thymic carcinoid were 1.8%, 26.8%, and 27.5%, respectively. Yamakawa and associates [9] in 1991 defined the status of lymphogenous metastasis as follows: metastasis to anterior mediastinal lymph nodes around the thymus as N1, metastasis to intrathoracic lymph nodes other than anterior mediastinal lymph nodes as N2, and metastasis to extrathoracic lymph nodes as N3. They regarded N1 as the primary lymph nodes of thymoma. We examined the pattern of lymphogenous metastasis in thymic epithelial tumor according to this N factor classification. Most tumors with lymph node metastasis metastasized to N1 (thymoma; 90%, thymic carcinoma; 69%, thymic carcinoid; 91%). Eighteen percent, 44%, and 60% of N1 tumors in thymoma, thymic carcinoma, and thymic carcinoid, respectively, metastasized to N2 or N3. Thymoma (11%) and thymic carcinoid (9%) infrequently had a skip metastasis to N2 or N3, whereas one third of thymic carcinomas had skip metastasis to N2 or N3. From these data we think that it is reasonable to consider N1 to be the primary lymph node in thymoma and thymic carcinoid. However, because thymic carcinoma easily undergoes skip metastasis, it may be not proper to consider N1 to be a primary lymph node in thymic carcinoma. Thymic epithelial tumor with invasion to neighboring organs easily metastasizes to lymph nodes compared with tumors without invasion (thymoma, 5.6% versus 0.003%; thymic carcinoma, 28.8% versus 14.8%; thymic carcinoid, 37.5% versus 12.5%). We conclude that it is necessary to dissect or sample lymph nodes of N1 in thymic carcinoma and thymic carcinoid, as well as in thymoma with invasion to neighboring organs. A significant difference in survival rate was observed between N and N in thymoma and thymic carcinoma. Multivariate analysis demonstrated that survival was dependent on lymph node metastasis in patients with thymoma (p 0.053) and thymic carcinoma (p 0.001). Moreover, the prognosis tends to become worse according to the progression of N factor. The N factor in thymic epithelial tumor was also one of the indicators predictive of survival, although it weakly influenced the survival in patients with thymoma compared with those with thymic carcinoma. The current study demonstrated the rate of hematogenous metastasis in thymoma. Hematogenous metastasis was observed in 0% to 10% (mean, 5.5%) of patients with thymoma [9]. Frequent metastatic sites for thymic epithelial tumor were the lung (thymoma, 84.6%; thymic carcinoma, 59%), followed by bone and liver. Although M1 thymoma and M1 thymic carcinoma showed worse prognoses compared with M0 thymoma and M0 thymic carcinoma, multivariate analysis demonstrated that survival of patients with thymoma (p 0.743) or thymic carcinoma (p 0.819) is not dependent on distant metastasis. In patients with thymic epithelial tumor, the M factor influences the survival less than T and N factors. In general, M factor strongly influences the survival in most cancers. This complicated relationship between N factor and M factor may be the reason why no TNM classification of thymic epithelial tumor has been established yet. In conclusion, T factor (invasiveness) was an excellent factor that can predict prognosis in not only thymoma but also thymic carcinoma. N factor (lymph node metastasis) is also one of the predictors of survival, although thymoma infrequently metastasizes to the lymph node. It is reasonable to consider the anterior mediastinal lymph node group (N1) to be a primary lymph node of thymic epithelial tumor because of its frequency and pattern of metastasis, anatomic location, and prognosis, although in thymic carcinoma skip metastasis occurs easily. The M

GENERAL THORACIC 1864 KONDO AND MONDEN Ann Thorac Surg METASTASIS OF THYMIC EPITHELIAL TUMORS 2003;76:1859 65 factor (distant metastasis) influences the survival less than T and N factors. References 1. Bergh NP, Gatzinsky P, Larsson S, Lundin P, Ridell B. Tumors of the thymus and thymic region: I. Clinicopathological studies on thymomas. Ann Thorac Surg 1978;25:91 8. 2. Wilkins EW Jr, Castleman B. Thymoma: a continuing survey at the Massachusetts General Hospital. Ann Thorac Surg 1979;28:252 6. 3. Masaoka A, Monden Y, Nakahara K, Tanioka T. Follow-up study of thymomas with special reference to their clinical stages. Cancer 1981;48:2485 92. 4. Suster S, Moran CA. Thymoma, atypical thymoma, and thymic carcinoma. A novel conceptual approach to the classification of thymic epithelial neoplasms. Am J Clin Pathol 1999;111:826 33. 5. Shimosato Y, Mukai K. Tumors of the mediastinum. In: Rosai J, Sobin LH, eds. Atlas of tumor pathology. Third series, Fascicle 21. Washington, D.C.: Armed Forces Institute of Pathology, 1997. 6. Suster S, Rosai J. Thymic carcinoma. A clinicopathologic study of 60 cases. Cancer 1991;67:1025 32. 7. Moran CA, Suster S. Neuroendocrine carcinomas (carcinoid tumor) of the thymus. Am J Clin Pathol 2000;114:100 10. 8. Kondo K, Monden Y. Therapy for thymic epithelial tumors: a clinical study of 1,320 patients from Japan. Ann Thorac Surg 2003;76:878 85. 9. Yamakawa Y, Masaoka A, Hashimoto T, et al. A tentative tumor-node-metastasis classification of thymoma. Cancer 1991;68:1984 7. 10. Okumura M, Ohta M, Tateyama H, et al. The World Health Organization histologic classification system reflects the oncologic behavior of thymoma: a clinical study of 273 patients. Cancer 2002;94:624 32. 11. Lardinois D, Rechsteiner R, Lang RH, et al. Prognostic relevance of Masaoka and Muller-Hermelink classification in patients with thymic tumors. Ann Thorac Surg 2000;69: 1550 5. 12. Gripp S, Hilgers K, Wurm R, Schmitt G. Thymoma: prognostic factors and treatment outcomes. Cancer 1998;83:1495 503. 13. Regnard JF, Magdeleinat P, Dromer C, et al. Prognostic factors and long-term results after thymoma resection: a series of 307 patients. J Thorac Cardiovasc Surg 1996;112: 376 84. 14. Blumberg D, Port JL, Weksler B, et al. Thymoma: a multivariate analysis of factors predicting survival. Ann Thorac Surg 1995;60:908 13. 15. Quintanilla-Martinez L, Wilkins EW Jr, Choi N, Efird J, Hug E, Harris NL. Thymoma. Histologic subclassification is an independent prognostic factor. Cancer 1994;74:606 17. 16. Pescarmona E, Rendina EA, Venuta F, et al. Analysis of prognostic factors and clinicopathological staging of thymoma. Ann Thorac Surg 1990;50:534 8. 17. Hsu HC, Huang EY, Wang CJ, Sun LM, Chen HC. Postoperative radiotherapy in thymic carcinoma: treatment results and prognostic factors. Int J Radiat Oncol Biol Phys 2002;52: 801 5. 18. Ogawa K, Toita T, Uno T, et al. Treatment and prognosis of thymic carcinoma: a retrospective analysis of 40 cases. Cancer 2002;94:3115 9. 19. Liu HC, Hsu WH, Chen YJ, et al. Primary thymic carcinoma. Ann Thorac Surg 2002;73:1076 81. 20. Blumberg D, Burt ME, Bains MS, et al. Thymic carcinoma: current staging does not predict prognosis. J Thorac Cardiovasc Surg 1998;115:303 8. 21. Hsu CP, Chen CY, Chen CL, et al. Thymic carcinoma. Ten years experience in twenty patients. J Thorac Cardiovasc Surg 1994;107:615 20. 22. Economopoulos GC, Lewis JW Jr, Lee MW, Silverman NA. Carcinoid tumors of the thymus. Ann Thorac Surg 1990;50: 58 61. INVITED COMMENTARY Doctors Kondo and Monden have organized a large contemporary database to see if the TNM staging system would fit thymic epithelial tumors. After a careful review of their data I believe the answer to be no for thymoma and maybe for thymic carcinoma. A staging system is valuable to estimate prognosis, decide on therapy and to compare results. Most would agree that simplicity and clinical validity are the two cardinal features of a successful staging system. Unnecessary subdivisions of tumor characteristics result in survival curves that do not separate well (and hence reduce the statistical difference between groups of patients) and make the system needlessly complex and difficult to remember. Clinical validity relates to the most important tumor variables (proven prognostic factors derived from large high quality databases utilizing multivariate analysis) for that specific tumor. For example, in soft tissue sarcoma, the histologic grade of the tumor is by far the most important prognostic variable while the lymph node status is not really relevant. The art of designing and comparing tumor staging schemes is complex and not standardized. A recent paper examining this problem in detail using thymoma as a model is worth reviewing to understand the uncertainties even our biostatistician colleagues have in this very complex area [1]. Less than 2% of patients with thymomas have nodal metastases. Only 1% of patients with thymoma have distant metastases. The rarity of this biologic behavior indicates that the N factor would not be a good candidate for stratifying survival in a staging system. This is emphasized by the p value in the multivariate analysis indicating a modification of the Masaoka system (stage 1, 2 vs 3, 4) was far superior (p 0.001) to nodal metastases (p 0.053) for predicting survival. While the Masaoka staging system is widely used, most agree it could be improved upon as there is no difference between stage 1 and 2 in almost all modern reports, stage 3 is heterogeneous and the new WHO histologic grading system is not used (most reports indicate the histologic type of thymoma stratifies survival, as well as the Masaoka stage, and is independently predictive of survival). I do not think the TNM system would be appropriate for thymoma given the information in this report. In thymic carcinoma lymph node metastases are relatively common (27% of cases) and there is a difference in survival between node negative and positive patients. 2003 by The Society of Thoracic Surgeons 0003-4975/03/$30.00 Published by Elsevier Inc doi:10.1016/s0003-4975(03)01316-x