Thymic neoplasms are the most common tumors of

Thymic Carcinoma: A Multivariate Analysis of Factors Predictive of Survival in 290 Patients Benny Weksler, MD, Rajeev Dhupar, MD, MBA, Vishal Parikh, BS, Katie S. Nason, MD, MPH, Arjun Pennathur, MD, and Peter F. Ferson, MD Department of Cardiothoracic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania Background. Thymic carcinoma is a rare, aggressive disease with low 5-year survivals. We undertook this study to identify factors that impact prognosis and to better define the relationship between survival and surgical intervention. Methods. We queried the Surveillance, Epidemiology, and End Results cancer database and identified patients with thymic carcinoma. We performed univariate and multivariate analyses to identify factors prognostic for survival, focusing on demographic, tumor, and treatment variables. Results. For 290 patients with thymic carcinoma, the median survival was 48 months with 5-year survival of 30%. In multivariate analysis, type of surgical therapy (none, incomplete excision, complete thymic excision, debulking), Masaoka stage, and sex were important determinants of survival. Patients who underwent complete thymic excision had a significantly longer median survival than those who did not receive surgical therapy (105 versus 29 months; p < 0.001). In patients who underwent complete thymic excision, Masaoka stage and race were important determinants of survival in multivariate analysis. Conclusions. Complete thymic excision is the preferred primary treatment for thymic carcinoma. Masaoka stage has significant prognostic implications for all patients, including those who undergo complete thymic excision. (Ann Thorac Surg 2013;95:299 304) 2013 by The Society of Thoracic Surgeons Thymic neoplasms are the most common tumors of the anterior mediastinum and generally run an indolent course, with most patients surviving for many years [1]. However, a small percentage of these tumors are thymic carcinoma, which typically presents as an aggressive, late-stage disease, with 5-year survival in 28% to 67% of patients [2, 3]. Although the classification of thymic carcinomas has remained a controversial topic, in the 1970s and 1980s distinct histologic criteria and clinical staging for thymic carcinoma emerged, which remain widely applied today [4 6]. The gold standard treatment for resectable thymic carcinoma is surgery, with chemotherapy and radiation frequently used as adjunct therapies. Although efforts have been made to study the natural history of thymic carcinoma, the rarity of this disease has limited the ability to do so. The largest study, to our knowledge, reported a cohort of 186 patients with thymic carcinoma as part of a larger cohort of patients with thymic epithelial malignancies [3]. A number of reports have identified factors that may be prognostic in thymic carcinoma, such as resectability, histology, stage, and tumor size [1, 2, 7 12]. However, the roles that surgery, radiation, or multimodal therapy play in survival are not well established. Accepted for publication Sept 4, 2012. Address correspondence to Dr Weksler, Department of Cardiothoracic Surgery, C-800, UPMC Presbyterian Hospital, 200 Lothrop St, Pittsburgh, PA 15213; e-mail: wekslerb@upmc.edu. Using the Surveillance, Epidemiology, and End Results (SEER) database, we were able to identify a cohort of 290 patients with thymic carcinoma. Although chemotherapy data are lacking in the SEER database, a number of factors believed to be important in the prognosis of thymic carcinoma were extracted. We undertook this study to (1) identify factors that impact prognosis, and (2) better define the relationship between survival and surgical intervention in patients with thymic carcinoma. Material and Methods The SEER database is sponsored by the National Cancer Institute and has been used to track cancer incidence and patient survival since 1973. The SEER database currently covers approximately 28% of the US population and captures 98% of all cancer cases within the surveyed geographic areas. We used the SEER 17 Registry including the Hurricane Katrina Impacted Louisiana Cases for this analysis (SEER Program, www.seer.cancer.gov; SEER*Stat Database: Incidence SEER 17 Regs Research Data Hurricane Katrina Impacted Louisiana Cases, Nov 2009 Sub [1973 2007 varying], National Cancer Institute, Division of Cancer Control & Population Sciences, Surveillance Research Program, Cancer Statistics Branch, released April 2010, based on the November 2009 submission). This SEER database was queried for all cases of thymic carcinoma from January 1, 1973, to December 31, 2008, using the ICD-03 codes for thymic carcinoma (8586, 8588, and 8589). The first case of thymic 2013 by The Society of Thoracic Surgeons 0003-4975/$36.00 Published by Elsevier Inc http://dx.doi.org/10.1016/j.athoracsur.2012.09.006

300 WEKSLER ET AL Ann Thorac Surg ANALYSIS OF 290 THYMIC CARCINOMA PATIENTS 2013;95:299 304 Table 1. Characteristics of 290 Patients With Thymic Carcinoma by Surgical Therapy Full Cohort No Surgery Local Therapy Complete Thymic Excision Debulking n (%) 290 (100) 96 (33.1) 51 (17.6) 121 (41.7) 15 (5.2) Male, n (%) 176 (60.7) 57 (59.4) 28 (54.9) 77 (63.6) 9 (60.0) Mean age (y) 59.1 14.2 58.3 15.1 61.1 16.2 58.8 12.7 61.0 14.7 Caucasian, n (%) 204 (70.3) 67 (69.8) 35 (68.6) 83 (68.7) 13 (86.7) Tumor size (mm) 71.5 33 73.1 26.1 62.0 34.9 72.9 34.3 81.2 39.7 Masaoka stage Stage I 41 (14.1) 8 (8.3) 15 (29.4) 18 (14.9) 0 (0) Stage 2 15 (5.2) 1 (1.0) 3 (5.9) 10 (8.3) 0 (0) Stage 3 129 (44.5) 33 (34.4) 19 (37.3) 65 (53.7) 10 (66.7) Stage 4 74 (25.5) 40 (41.7) 10 (19.6) 19 (15.7) 5 (33.3) Stage unknown 31 (10.7) 14 (14.6) 4 (7.8) 9 (7.4) 0 (0) Radiation therapy, n (%) 154 (53.1) 47 (49.0) 24 (47.1) 69 (57.0) 11 (73.3) carcinoma was not reported in the database until 1986. SEER*Stat software (www.seer.cancer.gov/seerstat) version 6.6.2 was used for data mining. There were 290 patients identified in the SEER database with the diagnosis of thymic carcinoma. Data regarding patient demographics, Masaoka stage, tumor size, tumor grade (grades I to IV), surgical procedure (none; local therapy only including excisional biopsy; thymic excision including simple, total, and radical excision; and debulking), radiation therapy, survival, and cause of death were obtained. Data were incomplete for the requested variables in some patients as follows: radiation therapy, 1.4% (4 of 290); type of surgical procedure, 2.4% (7 of 290); stage, 10.7% (31 of 290); tumor size, 26.6% (77 of 290); and tumor grade, 49.3% (143 of 290). Continuous data variables were analyzed using Student s t test. Nominal data were analyzed utilizing crosstabs and Pearson s 2 test. Kaplan-Meier survival curves were constructed and compared using the log-rank test. A Cox proportional hazard model was used to identify the relevant variables that impact overall survival. Only variables that significantly affected survival in univariate analysis were included in the Cox model. The proportionality of hazards was evaluated using the Cox regression analysis with time-dependent covariates. The assumption of proportionality of hazards was tested and was not broken in any of the Cox regression models. Statistical analysis was performed with SPSS statistical software package version 19.0 (SPSS Inc, Chicago, IL). Significance was set at a probability value less than 0.05. Results All Thymic Carcinoma Patients Of the 290 cases of thymic carcinoma identified in the SEER database, the majority of patients were white men, and most tumors were Masaoka stage III (Table 1). The median survival of all patients with thymic carcinoma was 48 months (95% confidence interval [CI], 38.4 to 57.6), with 5-year survival of 40% and 10-year survival of 40% (Fig 1). In univariate analysis (Table 2), sex (p 0.009), type of surgical intervention (p 0.001), Masaoka stage (p 0.001), and tumor grade (p 0.044) were significant variables affecting survival. Age, tumor size, radiation therapy, and race did not significantly affect survival. In Cox regression analysis, sex, type of surgical procedure, and Masaoka stage remained important determinants of survival (Table 3). Thymic Carcinoma Patients Who Underwent Complete Thymic Excision Next, we analyzed only the patients who underwent an operative intervention to characterize prognostic factors and outcomes after complete thymic excision (excluding partial excision, debulking, and unspecified surgery). One hundred twenty-one of 217 patients (55.8%) underwent simple, total, or radical excision of the thymus (Table 1). Patients who underwent complete thymic excision did significantly better than those who had debulking, local therapy, or no surgical therapy. The median survival of patients who underwent complete thymic excision was 105 months (95% CI, 43.0 to 167.0) with 5-year survival of 58% versus 29 months for median survival (95% CI, 13.9 to 44.1) with 5-year survival of 26% for patients who did not undergo any surgery (Fig 2; p 0.001). In univariate analysis, race (p 0.013) and stage (p 0.001) were the only variables affecting survival (Table 4) of patients who underwent complete thymic excision. Cox regression analysis also revealed that both race and stage affected survival (Table 5). Comment In this study, we analyzed data from a large cohort of patients with thymic carcinoma using multivariate analysis to identify factors that impact prognosis and identify how survival relates to surgical intervention. The main findings were that (1) in patients with thymic carcinoma, there are significant relationships between overall survival and sex, tumor stage, and the type of operative intervention; (2) in patients with thymic carcinoma who undergo complete thymic excision, there is a significant

Ann Thorac Surg WEKSLER ET AL 2013;95:299 304 ANALYSIS OF 290 THYMIC CARCINOMA PATIENTS 301 Fig 1. Overall survival for all patients with thymic carcinoma (median survival, 48 months; 40% 5-year survival). relationship between overall survival and race or tumor stage; and (3) patients who undergo complete thymic excision have a median survival that is significantly greater than those who do not undergo surgical therapy (median survival, 105 versus 29 months; p 0.001). In previous studies, resectability, histology, tumor stage, and tumor size were identified as prognostic factors for patients with thymic carcinoma [1 3, 7 10, 12]. Because thymic carcinoma is a rare disease, many of these studies used small, distinct cohorts and addressed one or just a few of these factors. This study used the SEER database to identify a cohort of 290 patients treated for thymic carcinoma, the largest cohort studied to our knowledge. In all studies of thymic carcinoma to date, surgical treatment has improved survival, underscoring the reason that surgical therapy is the preferred modality of treatment [13, 14]. Consistent with previous studies, we found the same survival benefit of surgical intervention in our cohort. Median survival for patients who underwent complete thymic excision was 105 months with 5-year survival of 58%; median survival for patients who did not undergo surgical therapy was 29 months with 5-year survival of 26%. In comparison, the analysis of survival in 154 patients with thymic carcinoma by Kondo Table 2. Univariate Analysis of Factors Influencing Survival in Patients With Thymic Cancer p Value Sex (male versus female) 0.009 Age 0.339 Tumor size 0.366 Masaoka stage 0.001 Tumor grade 0.044 Radiation therapy (yes versus no) 0.920 Type of surgical intervention 0.001 Race (non-caucasian versus Caucasian) 0.403 and Monden [3] revealed 5-year survivals of 67% after total resection, 30% after subtotal resection, and 24% with no operation. Other studies with fewer patients have found 24- to 167-month median survival and 5-year survival of 27.5% to 61% [7, 11, 14]. The wide range reflects the low number of patients in each study and the inability to get accurate survival information for all patients. In contrast, in our study all patients had complete survival information. Our data support the notion that surgery is the preferred modality of treatment for thymic carcinoma, with complete thymic excision imparting the best median survival. In our study, advanced Masaoka stage was associated with shorter survival. One explanation for this is that advanced stage patients are less likely to receive complete thymic excision, which has been shown by this and other studies to confer the best long-term survival. In fact, although only 26% of the cohort had stage IV disease, this group accounted for 42% of the patients who did not have surgery. Stage has been implicated by many studies to be related to overall survival; it is not surprising that our data support a similar conclusion. Until now, sex has not been described as a significant factor in overall survival for patients with thymic carcinoma. Yano and associates [10] analyzed the relationship between sex and overall survival in a series of 30 patients with thymic carcinoma and found no relationship. Our Table 3. Multivariate Analysis for Factors Important in Overall Survival in Patients With Thymic Carcinoma Hazard Ratio (CI) p Value Type of surgical intervention 0.639 (0.476 0.856) 0.003 Sex (male versus female) 2.152 (1.188 3.901) 0.012 Masaoka stage 1.857 (1.269 2.720) 0.001 Tumor grade 1.086 (0.738 1.600) 0.675 CI confidence interval.

302 WEKSLER ET AL Ann Thorac Surg ANALYSIS OF 290 THYMIC CARCINOMA PATIENTS 2013;95:299 304 Fig 2. Overall survival for patients who received surgical therapy. Median survival times: no surgery, 29 months; local therapy, 47 months; complete thymic excision, 105 months; debulking, 41 months (p 0.001). series identified a slight male preponderance with thymic carcinoma (60.7%), which matches most other series [14]. When examining the entire cohort, men had better overall survival in both univariate and multivariate analyses. Interestingly, sex was not associated with differences in survival in patients who underwent complete thymic excision. The reason for this is unclear; perhaps there are unidentified differences in the biology of thymic carcinoma between men and women. In our multivariate analysis, tumor grade was not prognostic for survival. In 1978, Levine and Rosai [5] defined the histologic features of thymic tumors, and subsequent revisions culminated in the widely accepted classification system proposed by the World Health Organization (WHO) in 1999. Several studies have validated that clinically significant differences in outcome are partially based on the WHO histologic classification system [15 17]. Although our study did not confirm this, accurate information on tumor grade was not available for 49% of the patients in the SEER database. Therefore, the findings of this study should not detract from the findings of other studies, but rather should reflect a weakness in this database [1, 9, 11, 12]. Table 4. Univariate Analysis of Factors Affecting Survival in Patients Who Underwent Complete Thymic Excision p Value Sex (male versus female) 0.460 Age 0.410 Tumor size 0.661 Masaoka stage 0.001 Tumor grade 0.708 Radiation therapy (yes versus no) 0.606 Race (non-caucasian versus Caucasian) 0.013 Adjunctive therapies in the treatment of thymic carcinoma remain a controversial issue. There are limited data on the benefit of chemotherapy on survival, with some studies demonstrating a benefit for selected small groups of patients [18 20]. Overall, chemotherapy has not become the modality of choice for the primary treatment of thymic carcinoma. Additionally, several studies have concluded that there is some benefit to radiation therapy for local control and disease-free survival in thymoma and thymic carcinoma patients [21, 22]. Although we did not find a survival benefit in the patients who received radiation therapy, we were unable to analyze diseasefree intervals and recurrence owing to limitations with the SEER database. The true benefit of radiation therapy on survival remains unclear. Additionally, the SEER database does not maintain data on chemotherapy. We believe that in patients with thymic malignancies that are Masaoka stage III or greater, a serious consideration should be given to adjuvant chemotherapy with or without radiation therapy, although it is unclear whether this approach improves long-term survival. There are limitations to consider with this study. Because it is retrospective and is based on a large patient population database, the design is constrained by the retrospective nature of the study, the limited data points available in the database, and possibly inaccuracies in Table 5. Multivariate Analysis for Factors Important in Overall Survival in Patients With Thymic Carcinoma Who Undergo Complete Thymic Resection Hazard Ratio (CI) P Value Masaoka stage 2.692 (1.568 4.621) 0.001 Race (non-caucasian versus 1.625 (1.115 2.369) 0.012 Caucasian) CI confidence interval.

Ann Thorac Surg WEKSLER ET AL 2013;95:299 304 ANALYSIS OF 290 THYMIC CARCINOMA PATIENTS 303 reporting. Also, the diagnosis of thymic carcinoma is difficult and depends on reliable pathologic assessment. The differences between WHO classification B3 and C can be subtle and missed by less experienced pathologists. It is unclear how this affects the study. We attempted to address this concern by identifying all patients with WHO B3 thymomas in the database. We compared the survival between patients classified with WHO B3 thymoma and patients classified with thymic carcinoma and found that patients classified as WHO B3 thymomas had a median survival of 99 months (95% CI, 63.4 to 134.6) compared with 48 months (95% CI, 38.4 to 94.1) in patients with thymic carcinoma (p 0.001 by log rank). Still, the lack of expert pathology review is a significant weakness of this manuscript. Another possible shortcoming of this study is the movement of patients into and out of a SEER geographic location, which may result in occurrences of thymic carcinoma that are not captured by the SEER registry. Additionally, there is a degree of subjective interpretation with reporting in this database for one of our major measures, complete thymic excision. Complete thymic excision as defined by the database does not mean R0 resection. We do not have information on the completeness of resection, and it is possible that some patients who had complete thymic resection did not have a complete resection of the tumor. In summary, thymic carcinoma is a rare disease that is best treated with surgery as a primary therapy. Advanced Masaoka stage is prognostic for shorter survival in all patients, including those who undergo complete thymic excision. We also found that male sex is prognostic for longer survival, but the reasons for this are unclear. Further studies are needed to establish additional prognostic criteria and to better understand the implications of primary and adjunctive treatments on overall survival in patients with thymic carcinoma. The authors would like to thank Shannon Wyszomierski for her expert editorial review of the manuscript. References 1. Blumberg D, Port JL, Weksler B, et al. Thymoma: a multivariate analysis of factors predicting survival. Ann Thorac Surg 1995;60:908 14. 2. Tseng YL, Wang ST, Wu MH, Lin MY, Lai WW, Cheng FF. Thymic carcinoma: involvement of great vessels indicates poor prognosis. Ann Thorac Surg 2003;76:1041 5. 3. 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. 4. Snover DC, Levine GD, Rosai J. Thymic carcinoma. Five distinctive histological variants. Am J Surg Pathol 1982;6: 451 70. 5. Levine GD, Rosai J. Thymic hyperplasia and neoplasia: a review of current concepts. Hum Pathol 1978;9:495 515. 6. 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. 7. Liu HC, Hsu WH, Chen YJ, et al. Primary thymic carcinoma. Ann Thorac Surg 2002;73:1076 81. 8. Maruyama R, Suemitsu R, Okamoto T, et al. Persistent and aggressive treatment for thymic carcinoma. Results of a single-institute experience with 25 patients. Oncology 2006; 70:325 9. 9. Suster S, Rosai J. Thymic carcinoma. A clinicopathologic study of 60 cases. Cancer 1991;67:1025 32. 10. Yano M, Sasaki H, Yokoyama T, et al. Thymic carcinoma: 30 cases at a single institution. J Thorac Oncol 2008;3:265 9. 11. Hosaka Y, Tsuchida M, Toyabe S, Umezu H, Eimoto T, Hayashi J. Masaoka stage and histologic grade predict prognosis in patients with thymic carcinoma. Ann Thorac Surg 2010;89:912 7. 12. 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. 13. Venuta F, Anile M, Diso D, et al. Thymoma and thymic carcinoma. Eur J Cardiothorac Surg 2010;37:13 25. 14. Chung DA. Thymic carcinoma analysis of nineteen clinicopathological studies. Thorac Cardiovasc Surg 2000;48: 114 9. 15. Okumura M, Ohta M, Miyoshi S, et al. Oncological significance of WHO histological thymoma classification. A clinical study based on 286 patients. Jpn J Thorac Cardiovasc Surg 2002;50:189 94. 16. Kondo K. Tumor-node metastasis staging system for thymic epithelial tumors. J Thorac Oncol 2010;5(10 Suppl 4):S352 6. 17. Detterbeck FC. Clinical value of the WHO classification system of thymoma. Ann Thorac Surg 2006;81:2328 34. 18. Yano T, Hara N, Ichinose Y, Asoh H, Yokoyama H, Ohta M. Treatment and prognosis of primary thymic carcinoma. J Surg Oncol 1993;52:255 8. 19. Weide LG, Ulbright TM, Loehrer PJ Sr, Williams SD. Thymic carcinoma. A distinct clinical entity responsive to chemotherapy. Cancer 1993;71:1219 23. 20. Carlson RW, Dorfman RF, Sikic BI. Successful treatment of metastatic thymic carcinoma with cisplatin, vinblastine, bleomycin, and etoposide chemotherapy. Cancer 1990;66: 2092 4. 21. Masaoka A. Staging system of thymoma. J Thorac Oncol 2010;5(10 Suppl 4):S304 12. 22. Onuki T, Ishikawa S, Yamamoto T, et al. Pathologic radioresponse of preoperatively irradiated invasive thymomas. J Thorac Oncol 2008;3:270 6. INVITED COMMENTARY Weksler and colleagues report a multivariate analysis of predictive factors for survival based on 290 patients from the SEER (Surveillance Epidemiology and End Results) database [1]. The strengths of this study include the large number of patients (the largest reported in the literature), the use of the SEER database that allows analysis of a large number of patients with a rare disease, a thorough and sophisticated data analysis, and thoughtful discussion. Not surprisingly, performing a complete resection and Masaoka stage were associated with improved survival. When a complete resection was performed, stage remained as an important prognostic factor. More surprising was the finding that men had improved survival compared with women in the whole cohort. This finding is novel and has gone unexplained. Unlike patients with thymoma, who almost always have resection as their 2013 by The Society of Thoracic Surgeons 0003-4975/$36.00 Published by Elsevier Inc http://dx.doi.org/10.1016/j.athoracsur.2012.09.067