Thymic carcinomas are rare neoplasms. Accordingly, no robust staging system exists for these tumors. Herein 33 cases of thymic carcinoma are presented with emphasis on staging, follow-up, and usefulness of prior staging systems. All of the cases shared similar clinical, surgical, and pathologic features, which makes them suitable for appropriate staging. Based on our experience, a novel 3-tier staging schema is presented. According to our proposed schema, 7 patients had stage I tumors, 11 had stage II tumors, and 15 had stage III tumors at the time of resection. Follow-up revealed that 19 patients were alive (mean follow-up, 57 months) and 14 patients had died (mean survival, 44 months). The survival curves obtained using our proposed staging system showed significant differences between the individual stages contrary to those obtained using prior staging systems. Our proposed schema seems to correlate well with survival and seems to be more practical than prior systems.
Thymic epithelial neoplasms are rare tumors that comprise thymomas and thymic carcinomas. Histologically, thymomas recapitulate organotypic (or thymus-like) features and have at least low-grade malignant potential. Thymic carcinomas, on the other hand, are outright malignant tumors with an invasive growth pattern, clear-cut cytologic atypia, and largely absent organotypic features. Thymic carcinomas in particular can show diverse differentiation resembling various types of carcinoma found outside the thymic gland. Thymic carcinoma has an even lower incidence than thymoma, and its rarity has so far hampered definitive conclusions as to its biologic behavior, prognosis, and best treatment modalities.
The small number of cases has also impeded the implementation of a reliable staging system specifically designed for these tumors. To date, the Masaoka system,1 a staging system originally developed for thymoma, is the most popular one for staging not only thymomas, but also thymic carcinomas. However, several previous reports2–4 and our own observations (see part 1 in this issue of the Journal) have questioned the validity of this system when applied to thymic carcinoma by demonstrating that survival of the patients is not predicted by stage as defined by Masaoka et al.1 In 1994, a Japanese group proposed a system specifically designed for the staging of thymic carcinomas.5 However, this system was based on a series including only 16 cases, and statistical significance of the survival curves could not be demonstrated, limiting its prognostic value.
We described our experience with a large series of thymic carcinomas (see part 1) and demonstrated that the Masaoka system is not suitable for predicting outcome in this patient group. Furthermore, we showed that lymph node status is a significant predictor of survival. Based on these results, we introduce a 3-tiered TNM-derived staging system that offers a more practical approach to staging these tumors and that correlates well with clinical outcome.
Materials and Methods
We identified 65 cases of primary thymic carcinoma from the surgical pathology files of The MD Anderson Cancer Center, Houston, TX, and the personal consultation files of one of us (C.A.M.) in a period from 1985 to 2011. In all cases, material from thymectomy specimens was available; biopsy material alone was not included in this study. Of the 65 cases, 33 fulfilled the criteria to be included in the present study. The minimum requirements for inclusion in this study were as follows: (1) diagnosis of primary thymic carcinoma, (2) histopathologic evaluation of tissue derived exclusively from thymectomy, and (3) lymph node resection performed at the time of surgery.
Each of the 33 cases was assigned a specific stage, and comparison was then made by additionally staging the tumors according to the Masaoka Table 21 and Tsuchiya Table 35 staging systems. Clinical and follow-up information was obtained from the patients’ charts or referral information. Survival analysis was performed using the Kaplan-Meier method. Statistical significance was defined as a P values less than .05. All statistical analyses were performed using Statistica software, version 6 (StatSoft, Tulsa, OK).
Among the 33 patients, 23 were men and 10 were women. The age range was from 19 to 81 years with a mean age of 52.8 years. Patient details are summarized in Table 4. Eight patients had a history of a second neoplasm at the time of diagnosis, including breast, prostate, renal cell, colonic, and transitional cell carcinoma and mixed germ cell tumor of the testis. Metastatic spread of these tumors to the thymus was ruled out by thorough clinical, radiologic, and immunohistochemical investigations. None of the patients had an associated paraneoplastic syndrome. Tumor deaths in our patient population were exclusively due to thymic carcinoma.
All tumors were surgically resected, and completeness of resection could be assessed in 30 cases Table 5. In 13 cases (43%), complete surgical resection was achieved. In these cases, 10 patients (77%) were alive 11 to 126 months after diagnosis and 3 (23%) had died 18 to 123 months after diagnosis. In 17 cases, resection was incomplete; 7 (41%) of the patients were alive 3 to 99 months and 10 (59%) had died 3 to 128 months after diagnosis. Completeness of resection could not be assessed in 3 cases.
Grossly, the tumors varied from 2 to 14 cm with an average size of 7.0 cm. Histologically, the tumors included 22 squamous cell carcinomas, 2 high-grade undifferentiated carcinomas, 2 lymphoepithelioma-like carcinomas, 2 papillary carcinomas, and 1 each of anaplastic carcinoma, spindle cell carcinoma, rhabdoid carcinoma, mucinous carcinoma, and clear cell carcinoma. Among the squamous cell carcinomas, 16 were poorly differentiated, 5 were moderately differentiated, and 1 was well differentiated. Three of these cases were of the basaloid subtype. Of the cases of thymic carcinoma, 5 showed transition with thymoma and 4 were associated with a multilocular thymic cyst.
In 7 cases, the tumors were confined to the thymic gland with no invasion of the adjacent structures. In 11 cases, there was involvement of pleura, lung, pericardium, great vessels, chest wall, or diaphragm or a combination of these structures. In 30 patients, intrathoracic lymph nodes were sampled at the time of resection. Lymph node metastasis was identified in 11 cases (37%); 19 cases had negative lymph node status (63%). A total of 7 patients had metastatic disease to the lung, liver, bone, brain, adrenal gland, or abdominal lymph nodes at the time of diagnosis. Although in 3 of these cases no lymph node sampling was performed at surgery, the presence of metastatic disease allowed these cases to be staged according to our proposed staging system. Based on these results, cases were stratified into the following groups: 7 in stage I, 11 in stage II, and 15 in stage III Table 6. When applying the Masaoka system, 2 cases were stage I, 5 were stage II, 9 were stage III, 2 were stage IVa, and 15 were stage IVb Table 7. According to the Tsuchiya system, 7 were stage I, 1 was stage II, 14 were stage III, none were IVa, 5 were IVb, and 6 were IVc (Table 7).
Survival curves were drawn for our proposed staging system and for the Tsuchiya and Masaoka systems. Statistical analysis revealed a significant difference (P = .04178) between the stages in our proposed system Figure 4, but not for those in the Masaoka (P = .07762) Figure 5 or Tsuchiya (P = .09081) Figure 6 system.
In addition to surgical resection, 27 of the 33 patients received additional therapy for their tumors (Table 4). Two patients received no adjuvant therapy. In 4 cases, information about additional treatment was not available.
Follow-up information was available for all patients, ranging from 1 to 128 months. Of the patients, 19 were alive 3 to 126 months after diagnosis, and 14 patients had died of their disease 1 to 128 months after diagnosis (Table 4). Of the living patients, 6 had stage I tumors, 7 had stage II, and 6 had stage III. Of the patients who died, 1 had a stage I tumor, 4 had stage II, and 9 had stage III.
The classification and staging of thymic epithelial neoplasms has long been a matter of debate. For thymoma, several staging systems have been proposed over the years,1,6–11 but none of them have shown to be without flaw. Currently, the system proposed by Masaoka et al1 seems to be the most widely accepted one and is commonly used not only for the staging of thymomas, but also for thymic carcinomas. Acknowledging the fact that thymoma and thymic carcinoma are tumors with different biologic behavior warranting separation not only on histologic grounds, but also for the purpose of staging, an attempt was made by Tsuchiya et al5 in 1994 to establish a TNM-based staging system specific for thymic carcinomas; however, in clinical practice, the Masaoka system remains the most popular staging tool.
In our experience with 65 cases of thymic carcinoma (see part 1), we showed that the Masaoka system is not well suited for staging thymic carcinomas and that lymph node status, among other factors, has an important role for the prognosis of these tumors. When the tumors were staged according to the system proposed by Masaoka et al,1 a statistically significant difference in survival rate was not seen between the individual tumor stages but was achieved only when several stages were combined (stages I, II, and III vs stage IV). Similar results were noted by other investigators.2–4 In 1998, Blumberg et al2 retrospectively examined 43 cases of thymic carcinoma and found a lack of correlation between Masaoka stage and survival, refuting the prognostic usefulness of this system for this patient population. Five years later, a Taiwanese study3 confirmed the failure of the Masaoka system to predict outcome for thymic carcinoma by investigating 38 such tumors, and, in the same year, Kondo and Monden4 came to the same conclusion when evaluating a large cohort of 1,320 patients with thymic epithelial neoplasms (186 thymic carcinomas).
In 1991, a TNM-based system for staging thymic epithelial tumors was introduced by Yamakawa et al8 to account for the cases in which lymphomatous or hematogenous metastasis had taken place. This was done in an attempt to improve the Masaoka system, which, according to the characteristics of thymoma, places more emphasis on regional tumor involvement. Acknowledging the different tumor characteristics of thymoma and thymic carcinoma, Tsuchiya et al5 later modified the TNM system put forward by Yamakawa et al8 according to their own experience with 16 thymic carcinomas. Both of these systems divide the T category into 4 groups (T1–T4) in a similar pattern to the stages of the Masaoka system. In addition, these systems take into account the lymph node status and distant metastasis by adding N and M categories: lymph node metastasis is divided into 3 groups according to location (N1–N3), and hematogenous metastasis is noted as absent or present (M0 or M1). The major difference between the two systems lies in the stage grouping, placing N1 disease into stages II, III, or IV in the Tsuchiya system as opposed to stage IVb in the Yamakawa system based on the belief that N1 disease is resectable and, hence, curable. However, when applied to the cohort studied by Tsuchiya et al,5 the proposed Yamakawa and Tsuchiya systems lacked statistical significance between any of the respective stages, a result attributed to the small number of cases examined.
Aside from the relatively low number of cases examined in that study, a few other points have to be raised. The cases included in the study by Tsuchiya et al5 were represented by 3 histologic variants only (squamous cell carcinoma, lymphoepithelioma-like carcinoma, and basaloid carcinoma), thereby slightly limiting the validity of the study. A few other concerns involve the practicality of the Tsuchiya staging system. First, the division of T3 and T4 tumors (T3, “Tumor breaking through the mediastinal pleura or pericardium, or invading neighboring organs such as great vessels and lung” and T4, “Tumor with pleural or pericardial implantation”) is somewhat vague and may be difficult to use in practice. An even more problematic issue is the division of the lymph node categories. According to Tsuchiya et al,5 the lymph node stage is divided into N0 (no lymph node metastasis), N1 (metastasis to anterior mediastinal lymph nodes), N2 (metastasis to intrathoracic lymph nodes, excluding anterior mediastinal lymph nodes), and N3 (metastasis in extrathoracic lymph nodes). However, in our experience, lymph node dissection takes place in only approximately 40% of patients with thymic carcinoma (see part 1), and systematic nodal sampling (or labeling for that matter) is not performed. This may make elaborate lymph node staging as proposed by Tsuchiya et al5 rather difficult. Last, the stage grouping using 6 categories (I, II, III, IVa, IVb, IVc) seems rather complex for a group of tumors that are rare and for which there seems to be no systematic surgical approach or standardized additional treatment.
Based on this, the aim of the present study was to overcome the problems encountered using the existing staging systems and devise a more practical one specifically aimed at thymic carcinomas. This system is based on data collected from a recent study of 65 thymic carcinomas (see part 1) in which we demonstrated that lymph node involvement is one of the most important prognostic parameters. For this reason, we believe that a TNM-based system is preferable to a non–TNM-based approach.
Contrary to the TNM systems for most other organ systems that use 4 stage categories, our proposed staging system includes only 3 stages. This is the result of our observation that, especially when using the Masaoka and Tsuchiya systems, a statistically significant survival difference is noted only when combining several stage categories (the stages representing local tumor involvement) and comparing with the stages constituting more widespread dissemination. Therefore, in our opinion, it is preferable to condense the 4 categories into 3 (with stage I and II representing regional involvement and stage III representing extrathoracic spread or metastasis) to achieve more prognostically significant results.
Another important issue to address was the categorization of lymph node status. In our experience, lymph node dissection in thymic carcinoma is still performed rather randomly (∼40% of cases) and not in any systematic way. We therefore encountered difficulties staging our cases using the proposed Tsuchiya system, which seems to rely heavily on consistent and standardized lymph node sampling and requires separation of lymph nodes into anterior mediastinal, intrathoracic, and extrathoracic groups. We strongly believe that unless further improvements and standardization of lymph node sampling in these tumors is not achieved, a more simplified approach in terms of lymph node staging (using N0 and N1 categories only) is justified. Contrary to the beliefs of Tsuchiya et al,5 we previously showed (see part 1) that lymph node metastasis is a poor prognostic factor in thymic carcinoma; hence all cases with lymph node involvement were placed in stage III of our proposed system.
According to our observations, our proposed staging system for thymic carcinomas encompasses 3 stage categories based on the extent of regional involvement (stages I and II) and metastatic or distant disease (stage III). Survival analysis of the data derived from our cases applied to the Masaoka, Tsuchiya, and our proposed systems showed that only in our system was the difference of survival between the individual stages statistically significant, implying that our approach to staging seems better suited to this tumor type.
We have devised a new system for the staging of thymic carcinomas based on our experience with 33 cases of this neoplasm (including 9 subtypes) and comparison with the Masaoka and Tsuchiya proposed staging systems. We have demonstrated that for the purpose of staging thymic carcinomas, a simplified approach as represented by our system not only seems more practical but also correlates better with the clinical outcome in this patient group. We propose that until more data become available from larger series of cases and surgery is performed in a more consistent and standardized manner, our system represents the most useful tool for the staging of thymic carcinomas.
We thank Carmen Behrens, MD, for help with statistical analysis.