Evolution of Classification of Thymic Epithelial Tumors in the Era of Dr Thomas V. Colby

The classification of TET has been challenging because of (1) the presence of various combinations of epithelial cells and lymphocytes that compose these tumors; (2) the paucity of these neoplasms, making large studies difficult; (3) an uncertainty about the neoplastic component(s) of TET (thymic epithelial cells and/or lymphocytes) in early classifications; (4) controversies about the prognostic significance of classifications and the malignant potential of thymomas; and (5) the diversity of thymic-derived tumors (ie, TETs, lymphomas, germ cell tumors). Many pathologists have struggled with the classification of TET. For instance, Ewing⁷  noted in 1916 that “no group of tumors has more successfully resisted attempts at interpretation and classification than those of the thymus.” Lowenhaupt⁸  stated in 1948 that “thymic tumors, because of their rarity and diversity, have long resisted attempts to reach valid conclusions from [the] variously described and variously interpreted material available from the literature.” In 1961, Bernatz et al⁹  began their article on classification of thymoma with the remarks that “a study of thymic tumors must enjoy a prominent but unenviable position in any list of frustrating experiences inherent in the pursuit of medical knowledge” and that “remarkable as is this dissidence all reports overwhelmingly agree that there is much confusion as well as controversy about this group of tumors.”

In the early years of histomorphologic classification of TETs, it was not clear whether the epithelial cells and/or the lymphocytes were the neoplastic component. In 1917, Bell¹⁰  noted, in a case report of a thymoma, that the tumor was “likely of epithelial origin.” Ewing¹¹  mentioned in 1928 that “the chief source of [lymphosarcoma or thymoma] is probably the reticulum cell, but lymphocytes are often present in abundance.” Although it was still debated, in 1948, Lowenhaupt⁸  remarked on “the epithelial derivation of this entire group of neoplasms,” seeming to conclude that the epithelial cells were the neoplastic cells. In 1972, Zeiller and Dolan¹²  showed in experiments in mice and rats that thymocytes are not derived from the thymus but rather originate from the bone marrow and migrate to the thymus. These findings suggested that although they are present in close vicinity in the same organ, the embryologic origin of thymic epithelial cells differs from that of thymocytes. Rosai and Levine¹³  concluded in 1976 that a simultaneous neoplastic transformation of 2 cellular components, epithelial and lymphocytic, of different embryogenesis within the same organ would be unlikely. Rather, they suggested that the epithelial cells were the neoplastic cells of thymomas because (1) the epithelial cells have morphologic and biologic properties of neoplastic cells, (2) the epithelial cells are present in the invasive tumor and metastases, (3) some thymomas are only composed of epithelial cells, and (4) “in cytologically malignant thymomas the anaplastic features are restricted to [the epithelial cells].” The authors considered the lymphocytes of thymomas to be benign because (1) it had been shown that thymic lymphocytes are derived from the bone marrow, (2) light microscopic and ultrastructural features of lymphocytes in thymoma are not different from those of normal “resting” or stimulated lymphocytes, and (3) the lymphocytes lack biologic properties of neoplastic cells. Today the neoplastic nature of the epithelial cells and the benign/reactive properties of lymphocytes in TET have been well accepted.

At least 24 histomorphologic classifications of TET have been proposed during the last century (Table 1). Eighty-eight years ago, in 1928, Ewing¹¹  categorized thymic tumors into 2 main groups, the “lymphosarcoma or thymoma” and the “carcinoma” (Table 2). A third category, “spindle-cell or myxosarcoma,” comprised rare tumors that were thought to derive from the stroma. Although Ewing still questioned the cell of origin of the “lymphosarcoma or thymoma,” he concluded from his own case studies that the reticulum [epithelial] cells are the main or sole source of the tumor and the lymphocytes are largely passive but “are often present in abundance.” In 1948, Lowenhaupt⁸  classified thymic tumors according to the stage of embryonic development of the normal thymus (Table 2). Although this classification was not used in subsequent classifications, Lowenhaupt⁸  introduced some fundamental thoughts that were later confirmed in other studies, such as that TETs are of epithelial origin and that “early surgical excision, if possible, is the treatment of choice.” In 1955, Castleman¹⁴  defined thymomas as tumors composed of lymphocytes and epithelial cells. Castleman did not divide thymomas into subtypes because he argued that “there are so many variations within a given tumor that there are no foolproof criteria differentiating the thymoma in a patient with myasthenia gravis from one without myasthenia gravis, and that there are no microscopic differences between the thymoma that is limited to the thymic gland and the one that extends beyond its confines. . . .” He was not entirely convinced that thymomas are malignant, even though he described that in a quarter of thymomas the tumor replaces the entire thymic gland and invades into and through the pleura, pericardium, nerves, and/or blood vessels. In addition, he described the discrete nodules in the pleura and pericardium that are not connected with the primary tumor as implants rather than true metastases. However, although Castleman acknowledged lymph node metastases, he avoided terms such as malignant thymoma or carcinoma of the thymic gland because (1) he did not observe any thymomas with “embolic metastases” and (2) he noted that encapsulated thymomas were morphologically similar to thymomas that spread within the mediastinum. In his opinion, tumors that were previously classified as “malignant thymoma” with metastases to distant organs did not have typical morphologic features of thymomas but were usually lymphomas, teratomas, or primary tumors of the lung, pancreas, or stomach that presented with large metastases to mediastinal lymph nodes. Moreover, he suggested that although tumors with implants and/or extension, invasion, or penetration of neighboring structures might be considered malignant, surgical removal of these tumors before implants have occurred or before the tumor extended within the mediastinum might have cured the patients. Recurrences, he felt, occurred likely because of incomplete resection.

In 1961, “with the hope that some prognostic significance could be gained through study of cell types,” Bernatz et al⁹  classified thymomas based on the predominant cell type (Table 2). Bernatz et al⁹  reported on 138 patients who underwent surgery for thymoma at the Mayo Clinic Rochester. Based on the proposed morphologic classification, they found that patients with predominant lymphocytic or spindle cell thymomas have a better prognosis than patients with mixed cell type or predominant epithelial cell thymoma. Furthermore, predominant lymphocytic or spindle cell thymomas were more often encapsulated than thymomas with other predominant cell types. Moreover, based on the data from this study, Bernatz et al⁹  concluded that lack of invasion is of more prognostic significance than cell type, although statistical analysis was not performed at that time. They realized, however, that the classification of thymomas would likely need to be further developed and concluded their article with the words from Effler and McCormack¹⁵ “that the intriguing subject of thymic neoplasms still requires new thought and clarification.” The study by Bernatz et al⁹  also showed the previously disputed benefit of thymectomy for at least some patients with myasthenia gravis. Eighteen of 64 patients (28.1%) with thymoma in the setting of myasthenia gravis who underwent thymectomy were well either without (n = 11) or with (n = 7) medication after a mean follow-up of 8 years. In a subsequent study, Bernatz et al¹⁶  applied statistical analysis to outcome data of 80 patients with thymoma and myasthenia gravis (64 of which were also included in the original study of Bernatz et al⁹  from 1961). The authors confirmed that myasthenia gravis did not affect survival of invasive thymomas. However, in noninvasive thymomas the survival of patients with myasthenia gravis was worse than for patients without that disease. Wilkins and Castleman¹⁷  in 1979 did not find myasthenia gravis to be an adverse prognostic factor (anymore), possibly because of improved treatment. This was also confirmed in 1981 by Masaoka et al,¹⁸  who showed that the 5-year survival of patients with thymoma and myasthenia gravis was similar to patients with thymoma without myasthenia gravis. However, the 10-year survival of patients with myasthenia gravis was worse in that study, which might have been because many deaths due to myasthenia gravis occurred during the early years of the study and, therefore, influenced the 10-year survival. A recent analysis of the retrospective Japanese database of TETs that included 598 TETs in patients with myasthenia gravis did not reveal any difference in 5- and 10-year overall or recurrence-free survival.¹⁹  A recent multicenter study from Italy including 375 patients with thymoma showed a slight protective effect of the presence of myasthenia gravis on overall survival, although that was not confirmed in multivariate analysis.²⁰  These findings suggest that today the outcome of patients with thymoma and myasthenia gravis is at least as good as the outcome of patients with thymoma without myasthenia gravis.

In 1976, Rosai and Levine¹³  defined thymoma simply as an epithelial tumor originating from the thymic gland. The authors proposed to restrict the designation of thymoma to neoplasms of thymic epithelial cells regardless of the proportion of accompanying lymphocytes. The authors suggested that terms such as seminomatous thymoma, round cell thymoma, or granulomatous thymoma should not be used anymore; these tumors should be designated according to the cell of origin, that is, germinoma (seminoma) of thymus, thymic carcinoid, or Hodgkin disease, respectively. Rosai and Levine supported their belief that “once the term thymoma is restricted to the tumor of epithelial thymic cells, with or without a lymphocytic component, all further subdivisions are artificial” based on the review of 164 cases of thymoma. In a subsequent review, Levine and Rosai²¹  added a component of thymic carcinoma to the classification (Table 2).

Subsequently, TETs were classified in more detail and classifications focused on the relationship between neoplastic cells and their possible origin from the cortex and/or medulla of the benign thymic gland. Evidence had shown that in the benign thymic gland medullary epithelial cells are distinct from cortical epithelial cells based on morphology, ultrastructure, and immunology and therefore provide different microenvironments.²²  Furthermore, it was suggested that differences in nuclear and cytoplasmic features allowed the distinction between medullary and cortical thymic epithelial cells by microscopy (Table 3; Figure 1, A and B). Therefore, a classification into medullary predominant, cortical predominant, and mixed cell types was appealing. In 1985, Marino and Mueller-Hermelink²²  first proposed a classification of TET based on the compartment of the normal thymic gland from which the neoplastic epithelial cells likely arose (Table 2). In addition, thymic carcinomas were defined by the presence of invasive growth and almost pure composition of epithelial cells with cytologic criteria of malignancy. They applied this classification to a series of 58 thymomas and 13 thymic carcinomas and found that cortical thymomas had a tendency to occur in younger patients than other thymomas.²²  Furthermore, myasthenia gravis was only identified in patients with cortical thymoma. The more aggressive nature of cortical thymomas was reflected by their tendency for invasive growth and their clinical behavior, with 14 of 25 cases (56%) being locally invasive and/or having intrathoracic (13; 52%) or extrathoracic (1; 4%) metastases. A single case recurred. In addition, the authors noted differences in the lymphoid component, with medium- and large-sized lymphocytes more commonly seen in pure or predominantly cortical thymoma whereas small pleomorphic lymphocytes were mostly seen in medullary or mixed thymoma with medullary predominance. Moreover, the number of lymphocytes was high in cortical and mixed thymomas, whereas pure medullary thymomas had only rare lymphocytes. The concept by Marino and Müller-Hermelink²²  of classifying TET according to the structural components of the normal thymus was further developed by Kirchner et al²³  in 1989 based on a study of 95 TETs, patient's outcome, and proliferation assays of 12 TETs (Table 2). Kirchner et al²³  introduced a category of well-differentiated thymic carcinoma, defined as TET with incomplete loss of organotypic differentiation, tightly packed epithelial cells with slight to moderate atypia, and some immature CD1-positive T cells. Small areas of cortical differentiation could be present in well-differentiated thymic carcinoma. In contrast to well-differentiated thymic carcinoma, other thymic carcinomas were defined by lack of organotypic features. Based on outcome studies, Kirchner et al²³  considered medullary and mixed thymomas as benign; predominantly cortical thymomas, cortical thymomas, and well-differentiated thymic carcinomas as low-grade malignant tumors with increasing invasiveness and metastatic capacity; and thymic carcinoma as overtly malignant tumors. In vitro studies confirmed that the proliferative rate of neoplastic epithelial cells correlates with the different tumor types and their growth behavior in vivo. For instance, cells from mixed thymomas (n = 3) did not grow in vitro or ceased to proliferate after less than 3 passages, whereas cells from well-differentiated thymic carcinomas (n = 3) had the highest growth rate among the TETs tested, could be passaged for a maximum of 35 times, and were kept in culture for up to 8 months. The proliferative rate of epithelial cells of cortical (n = 3) and predominantly cortical thymomas (n = 3) was between that of mixed thymomas and well-differentiated thymic carcinomas.

In 1999, Suster and Moran^24,25  proposed a 3-tiered classification that was based upon morphologic features of differentiation (Table 2). This classification subtyped TETs according to their degree of cytologic atypia, presence of organotypic features of thymic differentiation, and “closeness to benign thymus.” Organotypic features of differentiation were defined as features that “most closely resemble the normal appearance of the thymus in its mature or involuted state” and are summarized in Table 4. Areas of medullary differentiation were described as well-circumscribed foci containing plump epithelial cells with scant lymphocytes within an otherwise cortical-appearing neoplastic population of cells, with or without Hassall corpuscles. According to this classification, thymomas exhibit all or most organotypic features, thymic carcinomas do not have any or have only minimal organotypic features, and atypical thymomas show intermediate morphologic features. The classification is illustrated in Figure 2, A through D. In addition, the authors considered all thymomas to potentially be malignant.

In 1989, a “WHO [World Health Organization] committee for the histologic classification of tumours of the thymus” was organized by Dr Juan Rosai. Subsequently, pathologists from 8 countries debated during 1 decade whether a histologic subclassification of thymomas is possible and useful. Although some pathologists favored a classification that just distinguished noninvasive from invasive thymomas, others thought that histologic types of thymoma correlated with their aggressiveness and clinical behavior and therefore further morphologic subclassification would be useful. Subsequently, a “compromised” histologic classification was published.²⁶  Because the association between thymoma subtype and compartment of normal thymic gland was still controversial, the terminology chosen by the WHO was “noncommittal” and was designated by letters and numbers (Table 2). All thymomas composed of bland-appearing spindle and/or oval cells were classified as type A; thymomas that contained dendritic or plump (“epithelioid”) cells were categorized as type B. Type B thymomas were further subdivided based on the proportional increase in tumor cells in relation to thymocytes, increasing cytologic atypia, and presence or absence of medullary differentiation. Figure 3, A through F, illustrates the morphologic features of the WHO classification. In addition to these “classic” thymomas, uncommon thymomas were described, including micronodular thymoma (Figure 4, A), microscopic thymoma (Figure 4, B), and metaplastic thymoma (Figure 4, C and D). Thymic neuroendocrine tumors were defined identically to neuroendocrine tumors in the lung, including typical carcinoid tumor, atypical carcinoid tumor (defined as 2–10 mitoses per 10 high-power fields and/or necrosis), large cell neuroendocrine carcinoma, and small cell carcinoma. The 1999 WHO classification²⁶  emphasized that the classification of cytoarchitectural features of thymoma should occur independent of staging because the classification “based on invasive/metastasizing properties of the tumour relates more closely to recurrence and outcome than the one based on cytoarchitectural features.” The WHO classification was revised in 2004 to include descriptions of clinical symptoms, macroscopic findings, immunohistochemical characteristics, genetic features, and prognostic data.²⁷  Type C thymoma (1999 WHO classification²⁶) was now defined as thymic carcinoma; the morphologic subtypes of thymomas remained unchanged. The most recent WHO classification,²⁸  published in 2015, advocated a similar histomorphologic classification that had retained the designation of the tumors by letters and numbers. The 2015 WHO classification used an interdisciplinary approach to the diagnosis of TET and included contributions from radiologists, thoracic surgeons, and oncologists. Epidemiologic and prognostic data for the WHO classification were derived, in part, from the worldwide retrospective database of the International Thymic Malignancy Interest Group (ITMIG), which comprises more than 6000 cases. Findings on imaging studies, specifically computer tomography/positron emission tomography and cytologic features, were also included. Some histomorphologic features and immunohistochemical criteria were refined in an attempt to enhance the reproducibility of subtyping of thymomas and to facilitate the distinction between thymomas and thymic carcinomas.²⁹  For instance, to help with the distinction between type B1 and B2 thymoma, the definition of type B1 thymomas included features such as thymuslike architecture and cytology, abundance of immature T cells, areas of medullary differentiation (medullary islands), and paucity of polygonal or dendritic epithelial cells without clustering (<3 contiguous epithelial cells) (Figure 5, A and B). Hassall corpuscles or perivascular spaces were considered optional. In addition, the distinction between type A and type AB thymomas was discussed in more detail. Although both types contain bland-appearing spindle/oval cells, they should be distinguished from each other by a low (“easily countable,” type A) or high (TdT-positive, “T-cells impossible to count,” and/or “moderate infiltrate of TdT+ lymphocytes [difficult to count] in > 10% of the tumour areas,” type AB) content of lymphocytes. Furthermore, an atypical type A thymoma variant was introduced. This TET is characterized by features of conventional type A thymoma with bland-appearing spindle/oval cells and atypical findings including hypercellularity, increased mitotic activity, and focal necrosis. This variant was included in the WHO classification because studies have shown an association between necrosis in type A thymomas and recurrence and extrathoracic metastasis.³⁰  Moreover, in a study of type A and AB thymomas, necrosis was associated with stage in univariate and multivariate analysis.³¹  However, neither study showed an association between mitotic activity and stage or outcome in type A³⁰  or type A and AB thymomas.³¹  The 2015 WHO classification recommended reporting of all thymoma subtypes in 10% increments (except type AB thymomas) if more than one subtype is identified in a resection specimen. Furthermore, all subtypes of thymoma were considered malignant because they can show an aggressive behavior. Exceptions were micronodular and microscopic thymoma, in which no fatal outcome has been reported. It was also suggested that immunohistochemistry might be used for the diagnosis and subtyping of thymomas with ambiguous histology. Thymic carcinoma subtypes have been broadened and now include the NUT carcinoma (Table 5).

Staging classifications of TET in general are based upon invasion, implants, lymph node involvement, and/or distant metastases. During the past half-century, at least 14 staging systems have been proposed, even though many of these were not outright designated as “staging systems” (Table 1). This review will focus on those that introduced new and interesting concepts of classification. In 1961, Bernatz et al⁹  divided thymomas into noninvasive and invasive thymomas (Table 6).⁹  This categorization was based on their study that showed that only 6 of 27 patients (22.2%) with invasive thymoma lived 5 or more years following surgery, in contrast to 49 of 64 patients (76.5%) with noninvasive thymoma, suggesting that patients with invasive thymoma had worse survival than patients with noninvasive thymoma. Similarly, in 1976, Rosai and Levine¹³  suggested that thymomas need only be separated into encapsulated versus invasive tumors for predicting outcome and guiding therapy. Rosai and Levine found that well-encapsulated thymomas without myasthenia gravis remain stable and surgical excision is curative, even though the patient might develop a paraneoplastic syndrome or local recurrence, which can occur either as multiple pleural implants or as localized mediastinal tumor. In contrast, invasive thymomas, although histologically indistinguishable from encapsulated thymoma, have a worse outcome.

In 1978, Bergh et al³²  developed a 3-tiered staging system that further subclassified invasive thymomas (Table 6). When applied to 43 thymomas, that staging system revealed “considerable” differences in survival between stage II and III thymomas, supporting that the extent of invasion has prognostic value. Wilkins and Castleman¹⁷  modified the classification by Bergh et al slightly in 1979, adding invasion into pleura and pericardium to stage II (Table 6). The authors confirmed that patients with invasive thymomas had an overall worse outcome than patients with noninvasive thymomas; however, the authors did not reach any conclusions in regards to outcome of patients with stage II versus stage III thymoma.

In 1981, Masaoka et al¹⁸  pioneered today's most commonly used staging system for thymomas (Table 6). This system was developed for clinical staging of thymomas assuming that all thymomas have a variable degree of malignant potential and prognosis of thymoma is determined from its stage, which is based on invasiveness of the tumor through the capsule into surrounding tissue and organs and metastases. This classification was designed upon their observation that thymomas initially grow locally, subsequently infiltrate or disseminate, and finally metastasize. It was sought to guide therapy, to evaluate the results of surgery, and to define the prognosis of a patient. When this staging system was applied to 96 thymomas, they showed that survival differed based on stage, with 5-year survivals of 92.6% (stage I), 85.7% (stage II), 69.6% (stage III), and 50% (stage IV). In 1994, Koga et al³³  modified the Masaoka staging, changing invasion into the tumor capsule (Masaoka stage II/2) to microscopic invasion through the capsule (transcapsular; Koga stage IIA) to accommodate the most common point of view of pathologists that a tumor is invasive if it grows through the tumor capsule (Table 6). Therefore, thymomas invading into but not through the tumor capsule were now classified as stage I in the Koga staging system. It was confirmed later by Roden et al⁴  that the prognosis is similar for patients with thymomas infiltrating into but not through the tumor capsule when compared with patients with encapsulated thymomas. Applying this system to 79 thymomas, Koga et al found that although the outcome did not differ significantly between stages I and II, there were significant differences between stages II and III and stages II and IV. The Koga staging system is now often referred to as the modified Masaoka or the Masaoka-Koga staging system. This staging system is probably worldwide the most commonly used staging system for thymomas today and is recommended by the WHO.²⁹ 

In 1982, the French Groupe d'Étude des Tumeurs Thymiques proposed a clinical classification that encompassed not only extent of invasion and metastases but also completeness of resection (Table 6).^34,35 

In 1991, Yamakawa and colleagues³⁶  proposed a tentative tumor-node-metastasis (TNM) classification that was similar to the Masaoka staging system but distinguished between lymphogenous and hematogenous metastases (Table 6). Lymphogenous metastases occurred in a few patients and were thought to progress from anterior mediastinal lymph nodes to intrathoracic and later to extrathoracic lymph nodes. Hematogenous metastases did not reveal any particular characteristics. The T stage was adopted from the Masaoka classification. In contrast to TNM classifications of other organs, the T stage for TET did not consider size of the tumor because it was thought that the malignant potential of thymoma was due to its invasiveness and its tendency to disseminate rather than its size. The TNM classification was applied to 207 thymomas. When evaluating stage IVB cases (thymomas with lymphogenous and/or hematogenous metastases), no association between stage and prognosis or treatment was identified, possibly because of the small number of cases (8 of 207; 3.9%). The classification was also applied to 13 thymic carcinomas and 6 thymic carcinoid tumors. In this population, there was a much higher frequency of stage IVB cases, and the TNM classification was considered to be more useful in these tumors. Tsuchiya et al³⁷  thought that a pathologic TNM staging system would be applicable to thymic carcinomas and thymic carcinoid tumors but not to thymomas, as Yamakawa et al³⁶  had suggested, and proposed a modification to the TNM staging system in 1994 (Table 6). Stage pT2 now included thymic carcinomas with invasion through the tumor capsule; tumors that invaded through the pleura or the pericardium were classified as stage pT3. The staging groups were also modified. Tumors with N1 disease were now distributed into stages II to IV based upon the T stage because N1 disease was considered resectable and curable. The authors validated the modified TNM staging system on 16 thymic carcinomas. Although there were no stage II thymic carcinomas in this study, survival curves separated them between stages I and III or IV and stages III and IV; the differences were not statistically significant, likely again because of the small number of cases. Moreover, this study predominantly included squamous cell carcinomas and 2 undifferentiated and 1 basaloid carcinoma, and lacked other carcinoma types of high-grade histology such as small cell carcinoma, lymphoepithelioma-like carcinoma, sarcomatoid carcinoma, and clear cell carcinoma. Carcinoid tumors were also not evaluated.

In 1997, Suster and Moran²⁵  developed a simple 3-tiered staging classification that was proposed to be usable to stage both thymic carcinomas and thymomas (Table 7). Weissferdt and Moran³⁸  proposed a 3-tiered staging classification for thymic carcinomas in 2012 (Table 7) with the purpose of simplifying the existing TNM classification. This classification was developed based on outcome data of 33 patients with thymic carcinoma. At the same time, Moran et al³⁹  recommended a staging solely for thymomas (Table 7). The authors introduced a stage 0 or in situ malignancy, which most likely could be controlled by complete resection alone. Only invasive tumors were assigned a stage. However, the authors acknowledged that recurrence has been reported even in encapsulated thymomas. When the authors applied this staging system to 250 patients with thymoma, significant differences in overall and recurrence-free survival were identified when comparing stages 0 and I with stages II and III; no differences were found in outcome between stages 0 and I. Furthermore, there were no differences in outcome between stages IIa, b, and c. The authors considered, however, that distinction was still important because of possible differences in treatment. The prognostic significance of the proposed Moran staging for thymomas was later confirmed by Roden et al,⁴⁰  although in that study, patients with stage 0 disease had a significantly better overall survival than stage I patients.

Currently, neither the American Joint Committee on Cancer nor the Union for International Cancer Control provides a staging system for thymoma or thymic carcinoma. The modified Masaoka staging system (Masaoka-Koga staging system) appears to be the most commonly used staging system for thymomas, whereas the TNM staging as recommended by the WHO (Table 6) might be used for thymic carcinoma.^28,41  To facilitate staging, the 1999 and 2004 WHO included morphologic criteria to facilitate the evaluation of invasion and metastases (Table 8; Figure 6, A through F).^26,27 

There is a need for a staging system that can be applied to all TETs, including thymomas, thymic carcinomas, and thymic neuroendocrine tumors.⁴²  Therefore, the International Association for the Study of Lung Cancer (IASLC) together with ITMIG proposed a staging system in 2014 for the forthcoming (eighth) edition of the TNM classification of malignant tumors (Table 7).⁴²  This classification requires microscopic evidence of findings for pathologic staging. This staging system was validated using a worldwide retrospective database led by ITMIG and supplemented by cases from the Japanese Association for Research on the Thymus and the European Society of Thoracic Surgeons that encompassed more than 10 000 TETs. This proposed TNM classification is different from previous staging classifications of TET in that encapsulated TET and TET invading into the surrounding connective tissue are both classified as T1a because data from the retrospective database did not identify differences in outcome between these tumors. Some aspects of the proposed staging were considered speculative and still need to be validated, such as the distinction between stages IIIa and IIIb, N1 and N2, and M1a and M1b. Overall, subset analysis showed that this staging proposal is applicable to thymomas and thymic carcinomas. The number of thymic neuroendocrine tumors was too low for statistically meaningful conclusions.

Although most studies showed that staging is an important prognostic parameter for TET, there are controversies with regard to significance of the individual elements of staging systems. Some studies did not find any significant differences in outcome between patients with encapsulated thymomas and patients with thymomas that invade into the thymic or perithymic adipose tissue.^33,42,43  In contrast, a recent study by Roden et al⁴⁰  showed significantly worse overall survival for patients with thymomas that invaded through the capsule versus encapsulated thymomas. This controversy might, at least in part, be due to interobserver variability. Therefore, Roden et al⁴⁰  studied the interobserver reproducibility of 3 different staging systems, including the modified Masaoka (Masaoka-Koga), proposed Moran, and proposed IASLC/ITMIG staging system. In that study, the agreement among 3 thoracic pathologists from the same institution was almost perfect for the modified Masaoka staging system (κ = 0.85, n = 315) and the proposed Moran staging system (κ = 0.81, n = 290) and substantial for the proposed IASLC/ITMIG staging system (κ = 0.75, n = 81).⁴  If evaluating only the T component of the proposed IASLC/ITMIG (n = 297), the agreement was also substantial (κ = 0.75). Most common disagreements occurred between encapsulated and minimally invasive tumors, which would correspond to stage I and IIA tumors of the modified Masaoka staging system (15.8% of cases). When studying cases in which all 3 reviewers agreed upon extent of invasion, there was a significant difference in overall survival between modified Masaoka stage I and II patients.⁴⁰ 

Although thymic carcinomas have consistently been shown to have a worse outcome than thymomas,^44–47  the prognostic significance of thymomas has been debated and reports in the literature are conflicting. This debate also gave rise to some histomorphologic classifications. The prognostic significance of the histomorphologic classification of thymomas was especially debated in the years leading up to the first WHO classification. Some pathologists suggested that the classification of thymomas into noninvasive and invasive tumors might be sufficient because there is no additional prognostic significance of more elaborated classifications. Others proposed to subclassify thymomas further because histology might correlate with outcome. Indeed, some evidence suggested that WHO and Bernatz classifications have prognostic value.^46,48–50  For instance, studies have shown significantly worse survival of B3 thymoma versus types A through B2 thymoma,⁵¹  types A and AB thymoma,⁵²  or types A, AB, and B1 thymoma.⁵³  Recently, a study by Weis et al⁵⁴  evaluating 4221 thymomas submitted to the ITMIG retrospective database (including cases between 1983 and 2012) also showed that the WHO classification of thymomas was significantly associated with overall survival in univariate analysis, and although not all thymoma types differed from each other significantly, B3 thymomas had a worse overall survival in R0 resected patients than B1 thymomas. However, after adjusting for age, stage, and resection status, the WHO classification was no longer significant for overall survival. Other studies did not find differences in survival between type B3 thymoma or atypical thymoma and other thymomas^45,55  and failed to show an association of WHO or proposed Suster and Moran classification with outcome.^45,56,57 

These varying results might have been, at least in part, due to interobserver variation. Indeed, studies reported an only fair to moderate reproducibility of the 2004 WHO classification of TET, with κ values ranging between 0.39 and 0.53 or an overall concordance rate of 63% to 70%.^58–61  Only if a weighted κ was used, 1 study achieved good agreement for the 1999 WHO and the Bernatz classification (κ = 0.87).⁴⁸  Interobserver agreement for the Mueller-Hermelink classification was reported as 78%.⁶²  Roden et al⁴  compared the reproducibility of the 2004 WHO with the proposed Suster and Moran classification and the classification by Bernatz reviewing 456 TETs. Reproducibility was best for the proposed Suster and Moran classification (κ = 0.74, substantial agreement) followed by the 2004 WHO classification (κ = 0.65, substantial agreement) and the Bernatz classification (κ = 0.52, moderate agreement). The most common disagreements occurred between B1 and B2 thymoma (2004 WHO classification), thymoma and atypical thymoma (proposed Suster and Moran classification), and lymphocyte predominant and mixed lymphocyte and epithelial cell thymoma (Bernatz classification). Moreover, this study showed that the only moderate to substantial reproducibility of histomorphologic classifications indeed plays a role in determining the prognostic significance of TET, with differences in prognosis between the 3 reviewers in multivariate analysis.⁴  Therefore, Roden et al⁴⁰  evaluated the prognostic significance of thymoma using cases in which all 3 thoracic pathologists independently agreed upon a diagnosis. Univariate analysis confirmed that all histomorphologic classifications studied, including Bernatz, Suster and Moran, and WHO (2004), are of prognostic significance for overall and disease-free survival. However, in multivariate analysis, only Bernatz was prognostic for overall survival if adjusted for modified Masaoka staging. On the other hand, modified Masaoka staging and thymoma size were prognostic factors independent of histologic classification. Similarly, in the study by Weis et al⁵⁴  of the ITMIG retrospective database, stage, resection status, and age were independently associated with overall survival.

Taken together, these results suggest that histomorphologic classification is not an independent prognostic marker for thymoma and that staging plays an important role for outcome and therefore for management of thymomas, independent of the histomorphologic classification.

Loss of heterozygosity on chromosome 6q25.2-25.3 is common and has been identified in all types of TET except type B1 thymoma (Table 9).²⁸  A potential target of loss of heterozygosity in 6q25.2-25.3 is the tumor suppressor gene FOXC1.²⁸  In addition, Petrini et al⁶³  showed that frequent copy number loss of FOXC1 is associated with more aggressive tumors and correlates with decreased protein expression. More recently, GTF2I missense mutations have been found in types A and AB thymomas and less commonly in types B1, B2, and B3 thymomas.⁶⁴  These mutations were rather uncommon in thymic carcinoma (Table 9).

Using array comparative genomic hybridization, Petrini et al⁶⁵  identified homozygous 9p21.3 copy number loss in 2 of 7 thymic carcinomas (28.6%) and 2 of 20 type B3 thymomas (10%) but no other thymomas. The copy number loss peak of 9p21.3 included only CDKN2A/B loci. CDKN2A is a known tumor suppressor gene involved in the control of the cell cycle and may be related to uncontrolled tumor cell proliferation. In fact, focal deletion of 9p21.3 is a frequent event in cancer (40% overall and 16% focal copy number loss)⁶⁶  and is associated with poor outcome in patients with lymphoblastic leukemia.⁶⁷  Copy number loss of CDKN2A/B was associated with poor outcome (including disease-related survival and time to progression), suggesting that CDKN2A/B may be of potential value in the prognosis of TET. In addition, copy number loss of CDKN2A correlated with lack of expression of its related protein p16^INK4 (p16) in TET.

In vitro studies showed that siRNA knockdown of antiapoptotic molecules BCL2 and MCL1 leads to reduction of the proliferation of TET cell lines.⁶⁵  Gx15-070 (pan-BCL2 inhibitor) induced necrosis in TET cells and inhibited TET xenograft growth. ABT263 (inhibitor of BCL2/BCL-XL/BCL-W) led to decreased TET cell proliferation when combined with sorafenib, a tyrosine kinase inhibitor.

Overall, biology and genetic findings appear to vary with the histomorphologic type of TET according to the WHO.

A variety and a large number of histomorphologic and staging classifications for TET have been proposed during the era of Dr Thomas V. Colby, reflecting challenges to develop an easy-to-follow, reproducible, and therapeutically and prognostically meaningful classification that can be used for the evaluation of thymomas, thymic carcinomas, and thymic neuroendocrine tumors. The rarity of the disease makes its study even more difficult. Overall, staging is a prognostic marker of TET independent of histomorphologic classification. However, despite many attempts of staging classifications, currently there is no perfect staging system available that is suitable for treatment decisions and prognosis for all TETs. The recently proposed staging classification of IASLC/ITMIG shows promise when applied to a large global retrospective database. Although histomorphologic classifications are not prognostically significant independent of staging, tumor biologic and genetic features appear to differ among histologic subtypes of TET.