Apparent Alveolar Bronchiolar Tumors Arising in the Mediastinum of F344 Rats

Abstract

Rare tumors were observed in chronic studies in F-344 rats that were purely or largely free in the mediastinal cavity, yet had the histological architecture of alveolar bronchiolar tumors. They had originally been diagnosed as either pulmonary alveolar bronchiolar tumors, mediastinal mesotheliomas, or thymomas. The authors described these tumors, estimated the fraction of thoracic tumors that they represented, and carried out a preliminary immunohistochemical investigation of whether they were of pulmonary or mesothelial origin. Sections of 715 thoracic tumors originally diagnosed as alveolar bronchiolar tumors, mesotheliomas, or thymomas from control or treated F-344 rats in NTP two-year studies were reviewed. Thirty (4%) were found to be purely or largely mediastinal, yet to have an alveolar bronchiolar histological pattern. A subset of these tumors and some typical intrapulmonary alveolar bronchiolar carcinomas and pleural mesotheliomas were immunostained for Clara cell secretory protein (CCSP), β-tubulin IV, and Wilm’s tumor 1 susceptibility gene products (WT1). The tumors with the histological architecture of alveolar bronchiolar tumors immunostained positive for CCSP and negative for WT1, implying they might have been of alveolar bronchiolar origin, despite their predominantly mediastinal location, although more certain identification would require the use of a larger panel of antibodies.

Introduction

Rare tumors were noted on review of chronic studies in F-344 rats that were purely or largely free in the mediastinal cavity, yet had the histological architecture of alveolar bronch-iolar tumors (that would be expected to be purely or largely within the lung tissue). They had originally been diagnosed as either pulmonary alveolar bronchiolar carcinomas, thymo-mas, or thoracic mesotheliomas. Alveolar bronchiolar tumors with a predominantly mediastinal location have not been described previously in rodents. The aim of this study was to describe these tumors and to investigate whether they were of pulmonary or mesothelial origin. Accurate diagnosis of tumors in carcinogenesis bioassay studies is essential for identification of differences in incidence between control and treated groups, especially for tumors that occur infrequently.

Mediastinal tumors are infrequent in F-344 rats. Thymomas (benign or malignat) occurred in about 0.5% vehicle control F344 rats from National Toxicology Program (NTP) carcinogenicity studies that started between 1998 and 2002 (NTP 2007). Pulmonary alveolar bronchiolar carcinomas of the lung occurred in 21/2,298 (0.9%) vehicle control rats (NTP 2007) and have potential to metastasize to the mediastinum (Boorman and Eustis 1990). The thoracic mesothelium is rarely affected by mesothelioma in F-344 rats, although this tumor is frequent in the peritoneal cavity (Hall 1990). In addition to the presence of these tumor types, mononuclear cell leukemia may affect the mediastinal lymph nodes, thymus and lungs.

Thymomas, tumors of thymic epithelial cells (Stefanski, Elwell, and Stromberg 1990), typically occur in the anterior mediastinum and are usually benign. Malignant thymomas invade mediastinal tissues or nearby organs. Both benign and malignant thymomas may be partially or wholly composed of spindle cells and frequently contain variable numbers of tubules and cords of epithelial cells, sometimes with areas of squamous differentiation. Thymomas may contain variable numbers of lymphocytes.

Alveolar bronchiolar tumors in rats are believed to arise from alveolar type II or Clara cells (Boorman and Eustis 1990). Alveolar bronchiolar carcinomas (Boorman and Eustis 1990; Rittinghausen et al. 1992; Boorman and Herbert 1996) are described as arising in the lung parenchyma; spontaneous carcinomas tend to occur in the peripheral regions of the lung (Mohr 1992). They may invade airways, interstitial tissue, the pleura, or blood vessels. They may have a papillary, alveolar, tubular, and/or solid pattern and, especially in some induced carcinomas, a prominent connective tissue component or scirrhous reaction. Areas of squamous metaplasia may also be present. The epithelial cells are usually cuboidal to columnar but may be spindle-shaped in less-differentiated carcinomas. Non-neoplastic cells resembling macrophages or sloughed type II alveolar cells are often present in the lumens of papillary, alveolar, or glandular areas.

Mesotheliomas are tumors of pluripotential mesothelial stem cells that can differentiate into epithelial or mesenchymal cells (Hoogsteden et al. 1997). They characteristically grow along serosal surfaces and rarely infiltrate the underlying lung or embolize into the lung (Hall 1990). All mesotheliomas are considered malignant. They may have an epithelial, mixed, or mesenchymal histological pattern.

G. Willson, N. Clayton, and G. Pearse (personal communication 2003) found that primary intrathoracic mesotheliomas are very rare in F344 rats from NTP studies, in contrast to mesothe-liomas arising from the testis, epididymis, tunica vaginalis, or abdominal peritoneum, which are relatively frequent spontaneous tumors in this strain. Of the twenty-five cases of primary intrathoracic mesothelioma that they identified, twenty-one arose from the thoracic pleura and featured variable amounts and combinations of epithelial cells in papillae, nodules, trabeculae or sheets, and mesenchymal areas (hyaline and chondroid connective tissue and malignant spindle cells). The remaining four primary intrathoracic mesotheliomas were atriocaval.

A diffuse, serosal, pleurotropic growth pattern is common in human malignant mesothelioma, but other human pleural tumors (most frequently peripheral lung carcinomas and less frequently sarcomas, lymphomas, thymic epithelial tumors, carcinoid tumors, and tumors metastatic to the thorax) may be indistinguishable clinically, radiologically, and at postmortem (Attanoos and Gibbs 2003). The histological appearances of epithelial mesotheliomas can also mimic those of carcinomas. The differentiation is particularly important in humans, for prognostic and legal reasons (Amin et al. 1995). Immunos-taining for Clara cell secretory protein (CCSP) and β-tubulin, an intermediate filament specific to cilia, was used in this study to investigate whether the tumors were of alveolar bronchiolar origin. CCSP, also known as CC-10, or Clara cell antigen, is a low-molecular-weight, phospholipase A₂- inhibitory protein of uncertain function that is present in Clara cells in humans, mice, rats, and other species (Peri et al. 1993; Singh and Katyal 2000; Reynolds et al. 2004). Mason et al. (2000) reported CC-10 mRNA and protein expression in Clara cells but not in normal alveolar cells of mice. Immunostaining for CCSP may not be confined to Clara cells. The amino acid sequence of CCSP is similar to that of rabbit uteroglobin, found in the uterus and the epithelia of several other tissues. Peri et al. (1993) detected CC-10 cDNAs and protein in the lung and various other human tissues.

β-tubulin IV is the major protein of microtubules and is normally found in the cilia of bronchial and bronchiolar cells and in other pulmonary cell types (Sheppard and Thurlow 1992). No published accounts of tubulin immunostaining of the pleura were found.

Immunostaining for the products of Wilm’s tumor 1 susceptibility gene (WT1) is reported to have high sensitivity and selectivity for mesothelioma in tests to differentiate this tumor from lung cancers in humans and rats (Amin et al. 1995; Willson, Clayton, and Pearse, per. comm. 2003). The proteins transcribed from this tumor suppressor gene are found in the nucleus and are synthesized during the transition from mesenchymal to epithelial tissue, as in the developing kidney (Gulyas and Hjerpe 2003). There is no single immunomarker for neoplastic thymic epithelial cells, so we did not investigate by immunohistochemistry the possibility that these mediastinal tumors were of thymic origin.

We reviewed sections of thoracic tumors from NTP chronic studies in F-344 rats, originally diagnosed as alveolar bronch-iolar lung carcinoma, for cases purely or largely mediastinal, without significant pulmonary involvement. Tumors originally diagnosed as thymomas and mesotheliomas were also reviewed for cases where the histological architecture was typical of alveolar bronchiolar carcinomas, despite being purely or largely mediastinal. Immunostaining for CCSP, β-tubulin, and WT1 were used in attempts to identify the nature of these mediastinal tumors.

Materials and Methods

Hematoxylin and eosin (H&E) sections, paraffin blocks, and residual formalin-fixed tissue were obtained from the NTP Archive. In the original studies, full necropsy examinations were done, and lungs and any pleural abnormalities fixed in 10% buffered formalin. All lung lobes were sectioned in the coronal plane, along with masses observed at necropsy. H&E-stained sections were prepared of paraffin embedded tissue. Original diagnoses were recorded by the pathologists responsible for the evaluation of the studies and, after peer review, included in the NTP database of results. Residual wet tissue, blocks, and slides were archived by the NTP. Animal husbandry and experimental procedures complied with the requirements set forth by the Public Health Service’s Guide for the Care and Use of Laboratory Animals.

Case Selection

H&E-stained sections of tumors from NTP carcinogenesis studies conducted in Fischer 344 rats from the NTP Archive were reexamined. Four hundred sixty tumors originally diagnosed as alveolar bronchiolar carcinoma in the lung were reexamined from the majority of studies that ended between 1976 and 2002 (some studies were not readily available for reevaluation), comprising 88 from vehicle control rats in sixty-one studies; 357 from rats that received test compounds that either clearly induced alveolar bronchiolar carcinomas in one or both sexes (thirteen studies), or where there was equivocal evidence of induction of this tumor in one or both sexes (nine studies); and 15 from rats that received non-lung carcinogens in three studies that ended between 1999 and 2002 (see Table 1 notes). All tumors, in controls and in treated groups, originally diagnosed as mesothelioma in the lung, pleura, or mediastinum (involving the pleural cavity either primarily, or secondarily from the abdominal cavity) and benign or malignant thymomas were also selected from studies that ended between 1976 and 2002. In addition, 45 alveolar bronchiolar adenomas in control F-344 rats were randomly selected, to assess whether any tumors originally diagnosed as being benign also had a predominantly mediastinal location. The numbers of tumors reviewed are summarized in Table 1.

Immunostaining

A subset of tumors representative of the mediastinal tumor types were immunostained for CCSP, β-tubulin, and WT1. These comprised ten tumors that were purely or largely mediastinal, yet had alveolar bronchiolar histological architecture (comprising eight originally diagnosed as alveolar bronchiolar carcinomas, one as alveolar bronchiolar adenoma, and one as thymoma). For comparison, two examples of typical thymo-mas, seven pleural mesotheliomas, and seven alveolar bronch-iolar carcinomas with more typical, significant pulmonary involvement were also immunostained. Immunostaining was done on the original paraffin blocks. Sections (6 μm) were deparaffinized, hydrated, and blocked for endogenous peroxi-dase and biotin staining. Table 2 summarizes the immunohisto-chemical procedures. Detailed epitope retrieval and staining protocols are provided at the NIEHS Laboratory of Experimental Pathology Web site: http://www.niehs.nih.gov/research/atniehs/labs/lep/path-support/immuno/

Results

Of 714 thoracic tumors in F-344 rats reexamined from the NTP database, 30 (4%) were purely or largely mediastinal and had the histological pattern of alveolar bronchiolar tumors (Table 1, Figure 1). The other 684 either were largely within the lung and had the histological pattern of alveolar bronchiolar tumors, or were largely mediastinal with the typical histolgical patterns of mesotheliomas or thymomas. At necropsy, the 30 tumors were described as located in the lung only or in both the lung and mediastinum and were histologically described as purely or largely mediastinal. Histologically, the majority were large tumors, with several or confluent mediastinal foci, occupying a considerable proportion of the mediastinal space. They were largely free in the mediastinum (Figure 1A, C, E, and G), although some had small areas adherent to the visceral pleura (sometimes compressing the underlying lung) or to other mediastinal structures such as the esophagus, trachea, or aorta. These tumors sometimes invaded into the mediastinal fat, lung parenchyma, blood vessels, or interstitial tissue around pulmonary arteries. Some had mesothelial, or thin fibrous, capsules, while in others, papillary, epithelial projections were exposed directly to the mediastinal space. The detailed histolo-gical appearances of these predominantly mediastinal tumors were typical of intrapulmonary alveolar bronchiolar tumors (Figure 1B, D, F, and H). Most had alveolar, glandular, papillary, and solid histological patterns, with round, cuboidal, or columnar cells (sometimes with cytoplasmic vacuoles); and the luminal spaces contained collections of macrophage- or alveolar type II epithelial-like cells. Compared with typical mesothe-liomas, most papillae in these tumors had far less prominent connective tissue cores and the cells bordering the papillae had more cytoplasm; however, some of the tumors included papillae that were similar to those present in mesotheliomas (Figure 1D and F), and in one case, there was mild reactive mesothelial proliferation of the visceral pleura, illustrating the diagnostic difficulties that these tumors can pose. One of the 45 tumors originally diagnosed as an alveolar bronchiolar adenoma, which is pictured in Figure 1G and H, was purely mediastinal and had a benign appearance with little cellular pleomorphism and a relatively regular alveolar and papillary structure, despite occurring in multiple foci. Three of the 29 purely or largely mediastinal, malignant tumors were recorded by the original pathologist as having metastasized outside the thorax, but only 1 of the 88 intrapulmonary alveolar bronchio-lar carcinomas in controls had done so; there were, however, too few extrathoracic metastases to draw any conclusions about whether the largely mediastinal tumors were more prone to extrathoracic metastasis than the purely mediastinal ones. The 29 purely or largely mediastinal tumors originally diagnosed as alveolar bronchiolar tumors or thymomas were unlikely to have been part of abdominal or pericardiac mesotheliomas, because none of these animals also had an original record of mesothe-lioma in any site.

Immunhistochemistry

Table 3 summarizes the immunohistochemistry findings. The tumors that were purely or largely mediastinal, yet had alveolar bronchiolar histological architecture, all immunostained positive for CCSP and negative or only weakly positive for WT1, including the papillary structures, resembling those of mesothelioma. This suggests that these tumors were in fact of alveolar bronch-iolar origin, despite their mediastinal location.

WT1 was a valuable immunostain in detecting pleural mesotheliomas with typical histological architecture, as these tumors displayed positve (intense and diffuse) nuclear staining (Figure 2A), while typical alveolar bronchiolar carcinomas and thymomas were negative. CCSP was valuable in detecting typical alveolar bronchiolar carcinomas, as these tumors exhibited positive cytoplasmic staining (Figure 2B and C). In contrast, the typical mesotheliomas were negative for CCSP. One of the two typical thymomas was weakly positive for CCSP, but it is not likely that the purely or largely mediastinal tumors were all of thymic origin, because they had alveolar bronchiolar histological architecture, that is not seen in typical thymomas. β-tubulin (although it stained the cilia of normal cells well) was not useful in distinguishing the tumor types, as only one typical alveolar bronchiolar carcinoma was positive.

Discussion

Thirty (4%) of 715 thoracic tumors reviewed in control and compound-exposed F-344 rats had the histological appearance of alveolar bronchiolar tumors, yet were purely or largely mediastinal. Twenty-eight had originally been diagnosed as alveolar bronchiolar tumors of the lung (27 malignant and 1 benign), but in 1 the original diagnosis had been mesothelioma, and in another case it had been thymoma. The immunostaining pattern, positive for CCSP and negative for WT1, confirmed the original diagnoses of alveolar bronchiolar tumors and proved especially useful in the diagnosis of those tumors with a significant papillary component that were particularly difficult to distinguish from mesotheliomas in H&E-stained sections. The histological patterns in H&E-stained sections did not suggest that these tumors arose from the thyroid or parathyroid, although they did not exclude the possibility of thymic origin, since thymic carcinomas of humans are difficult to distinguish from carcinomas of any other organ by light microscopy, because they lack thymic morphologic features; diagnosis relies on exclusion of other possible sites, immuno-histochemistry and electron microscopy (Rosai and Sobin 1999). Due to the possibility that CCSP could occur in tissues other than the lung, our immunostaining results also do not totally rule out a thymic origin for the mediastinal tumors described here.

There is no evidence to support the existence of mediastinal stem cells or alveolar bronchiolar choristomas, which could develop into alveolar bronchiolar tumors. However, the presence of one alveolar bronchiolar adenoma in the mediastinum of a control F-344 rat implies that these tumors can arise de novo in the mediastinum. Alternatively, it is possible that the mediastinal tumor focus, rather than representing a subtype of alveolar bronchiolar tumor, was simply due to invasion from intrapulmonary carcinoma that was not included in the examined sections. The degree of cellular differentiation and architectural organization of the mediastinal tumors was, however, similar to intrapulmonary carcinomas, implying that the mediastinal tumors were not more malignant. Also, the mediastinal foci were usually large, making it unlikely that intrapulmonary foci would have been missed. Finally, the alveolar bronchiolar carcinomas induced by tetranitromethane (Bucher et al. 1991) were included in the 357 cases from studies in which lung tumors were induced. Many of these metastasized outside the thorax, yet the proportion that were purely or largely mediastinal was not greater than in exposed rats in other studies or in controls. These considerations imply that the presence of relatively large amounts of these tumors in the mediastinum was not due simply to a strong tendency to invade from the lung. An alternative possibility is that the tumors arose through metaplasia of mesotheliomas into an alveolar bronchiolar type. There is no evidence for such metaplasia in any species; however, squamous metaplasia of neoplastic and non-neoplastic pleura has been reported in the horse, rabbit, rat, and in man (Topov and Kolev 1987, Matsuo et al. 1993, Baum, Hewicker-Trautwein, and Wohlsein 2004). In none of the 30 cases was there also an extrathoracic malignant tumor that could have metastasized to the thorax. Overall, it is most likely that these tumors arise de novo in the mediastinum.

Immunostaining for CCSP and WT1 allowed the provisional conclusion to be drawn that these primarily mediastinal tumors with alveolar bronchiolar architecture had the same cellular origin as typical, intrapulmonary alveolar bronchiolar carcinomas, as they were positive for CCSP and negative for WT1. Panels of immunomarkers are used to assist in the differential diagnosis of pleural tumors in humans (Attanoos and Gibbs 1997; Attanoos et al. 2001; Gonzales-Lois et al. 2001; Roberts et al. 2001; Ordonez 2003). Although we found that CCSP and WT1 were valuable in suggesting the origin of the tumors, the use of additional markers, such as CK5/6, D2–40, mesothelin, n-cadherin, and calretinin for mesotheliomas; and MOC31, p63, BerEP4, carcinoembryonic antigen, LeuM1 (CD15), and the periodic acid Schiff reaction for neutral mucins for carcinomas would be necessary for precise determination of their origin.

CCSP was specific for bronchial and bronchiolar epithelial cells, and did not stain normal alveolar cells, confirming the observations of Mason et al. (2000) with CC-10 in mice; therefore, our results also imply that typical alveolar bronchiolar carcinomas in F-344 rats originate from bronchiolar epithelium. Normal pleural mesothelium seldom stained for WT1, even when the nuclei of tumor cells were positive, although normal mesothelium does stain in humans (Amin et al. 1995), implying that this antigen is upregulated in neoplastic mesothelial cells in rats.

In preliminary work (results not presented), surfactant apo-protein (SP)-A immunostaining gave results similar to, but less specific than, CCSP, as nuclei of neoplastic cells and of normal bronchiolar, and alveolar epithelial cells were sometimes positive, in addition to their cytoplasm. SP-A staining was also found in normal esophageal epithelial cytoplasm, which was included in the lung sections in six cases, and there was sometimes weak staining in smooth muscle. The possibility that SPA may normally be present in mesothelial and other cells does not support the use of this marker alone for differentiating rat alveolar bronchiolar tumors from pleural mesotheliomas. A more specific marker of alveolar type II cells would be needed to confirm whether the alveolar bronchiolar tumors were of alveolar or bronchiolar origin.

Vimentin is typically used as a marker of cells of mesothelial origin. In preliminary work, we stained several archival blocks and tissue fixed in formalin for extended periods of time for vimentin, and the absence of staining in mesenchymal structures in all samples indicated overfixation of the tissues (Newsholme and Zimmerman 1997); therefore, vimentin was not helpful in distinguishing tumors of mesothelial origin from alveolar bronchiolar tumors in this study. Calretinin is also frequently used in the diagnosis of mesotheliomas in the clinical setting; however, we were not successful in staining a typical mesothe-lioma from a rat with a commercial calretinin antibody.

Ultrastructural features have been used to distinguish adeno-carcinoma from epithelial mesothelioma in humans. Mesothe-lioma cells have longer, more slender and complex microvilli, more prominent intermediate filaments and glycogen foci compared to adenocarcinomas (Attanoos and Gibbs 1997). Electron microscopy was not used in the present study due to a lack of suitable material.

In conclusion, 4% of a sample of thoracic tumors in F-344 rats in NTP studies, originally diagnosed as alveolar bronchio-lar tumors, thoracic mesotheliomas, or thymomas, were purely or largely mediastinal, yet had alveolar bronchiolar histological architecture. The nine immunostained were positive for CCSP and negative or only weakly positive for WT1, implying that they may have been of alveolar bronchiolar origin, despite their predominantly mediastinal location. As a result, the two tumors originally diagnosed as thoracic mesotheliomas or thymomas should probably be reclassified as alveolar bronchiolar tumors.

It is important to be aware that alveolar bronchiolar tumors in F-344 rats can occur in a predominantly mediastinal location. CCSP and WT1 immunostaining are useful in investigating whether tumors with this morphology are of alveolar epithelial or mesothelial origin, although larger panels of antibodies are needed for more precise determination.