Diagnostic Criteria for Proliferative Thyroid Lesions in Bony Fishes

Abstract

Thyroid proliferative lesions are rather common in bony fishes but disagreement exists in the fish pathology community concerning diagnostic criteria for hyperplastic versus neoplastic lesions. To simplify the diagnosis of proliferative thyroid lesions and to reduce confusion regarding lesion interpretation, we propose specific criteria for distinguishing hyperplastic from neoplastic lesions. Development of these criteria was based on the examination of a large series of proliferative lesions from Japanese medaka (Oryzias latipes), lesions from other small fish species, and a reexamination of the 97 cases of proliferative thyroid lesions from bony fishes deposited in the Registry of Tumors in Lower Animals. Specific diagnostic criteria are provided for all lesion categories including follicular cell hyperplasia (simple, nodular, or ectopic), adenoma (papillary or solid), and carcinoma (well- or poorly differentiated). These criteria should assist fish pathologists in describing and categorizing naturally occurring proliferative lesions from wild fishes, lesions that develop in laboratory fishes due to suboptimal culture practices or water quality, those in fishes used in toxicological assays, and captive aquarium fishes.

Introduction

Criteria for distinguishing hyperplastic thyroid lesions from thyroid neoplasia in bony fishes have been debated by fish pathologists for approximately 100 years (Hoover, 1984). Confusion exists mainly because thyroid tissue in teleosts is unencapsulated, capable of widespread ectopic growth, and frequently predisposed to hyperplastic proliferation. In some cases, hyperplasia can be extensive and follicles can extend into normal tissues (Harshbarger, 1984). Factors that may cause proliferation of fish thyroid tissue include iodine deficiency (Baker et al., 1955) and other nutritional imbalances, poor water quality (Nigrelli and Ruggieri, 1974), genetic susceptibility, seasonal factors, and exposure to goitrogenic substances (Hoover, 1984). Thyroid tissue in teleosts mainly occurs in the pharyngeal region but is often widely distributed anatomically (Baker, 1958; Leatherland, 1994). Because of this diverse anatomic distribution, normal-appearing thyroid follicles in nonpharyngeal (ectopic) sites have sometimes been diagnosed as thyroid neoplasms (Nigrelli, 1952; Berg et al., 1953; Mawdesley-Thomas, 1975; Blasiola et al., 1981; RTLA, 2005). Although thyroid neoplasms have been documented to occur in fish (Park et al., 1993; Chen et al., 1996; Spitsbergen et al., 2000), we contend that many of the reports are actually cases involving nonneoplastic ectopic follicles or lesions that are hyperplastic rather than neoplastic.

Leatherland and Down (2001) reviewed tumors and related lesions of the endocrine system of fishes and included a section on thyroid lesions. They agreed that the terminology utilized by different investigators to categorize thyroid lesions is confusing. The authors also retained the terms “simple hyperplasia,” “adenoma,” and “carcinoma” recognizing that there are potentially many morphological variants within these main categories. They indicated that the majority of pathological thyroid conditions in fishes appear to be simple hyperplasia or, more rarely, adenomas.

Because of the increased use of fishes in aquatic toxicology, carcinogenicity testing, environmental monitoring, and biomedical research, the development of meaningful diagnostic criteria and standardized nomenclature for proliferative lesions in fishes is critical. Criteria have already been established for nonneoplastic and neoplastic liver lesions in medaka (Boorman et al., 1997), proliferative swimbladder lesions in guppy and medaka (Fournie et al., 1999), and exocrine pancreatic lesions in the guppy (Fournie et al., 1987; Fournie and Hawkins, 2002) and mummichog (Fournie and Vogelbein, 1994). Stringent diagnostic criteria are still currently unavailable for many tissues and organ systems in fish including thyroid. Well-defined criteria are an essential aid to pathologists who evaluate fish studies because they promote diagnostic uniformity, allow for the meaningful communication of results, permit comparisons of findings among studies, and encourage the formation of a useful historic database. These are all benefits that will advance the field of fish pathology.

In this report, we propose specific criteria for distinguishing hyperplastic from neoplastic lesions in teleosts to simplify the diagnosis of proliferative thyroid lesions and to clarify confusion regarding lesion interpretation. We propose as primary lesion categories the following: follicular cell hyperplasia (simple, nodular, or ectopic), adenoma (papillary or solid), and carcinoma (well- or poorly differentiated). Key diagnostic features are presented for each lesion type. We also provide and discuss some reported cases for which we believe lesions may have been misinterpreted, and we will discuss specifics of the scheme proposed by Leatherland and Down (2001) in terms of our proposed diagnostic criteria.

Materials and Methods

Many of the thyroid lesions described and illustrated in this manuscript occurred in small fish species such as Japanese medaka (Oryzias latipes), zebrafish (Danio rerio), and sheepshead minnow (Cyprinodon variegatus). These fishes were either maintained as stock or used in toxicological studies (primarily carcinogenicity assays) at the Gulf Coast Research Laboratory (Ocean Springs, MS), and the U.S. Environmental Protection Agency (USEPA) Mid-Continent Ecology Division (Duluth, MN) or Gulf Ecology Division (Gulf Breeze, FL). Zebrafish that were evaluated were from toxicological studies or submitted as cases to the Diagnostic Service at the Zebrafish International Resource Center (Eugene, OR) or Oregon State University (Corvallis, OR). Also, proliferative thyroid lesions in bony fishes from the National Cancer Institute’s Registry of Tumors in Lower Animals (RTLA) (Experimental Pathology Laboratories, Inc., Sterling, VA) were reexamined, and selected cases are included. The Registry is a collection of over 7,500 pathologic specimens of cold-blooded vertebrates and invertebrates with historical data and pertinent literature.

The standard procedure for obtaining diagnostic microscopic slide sections from small fish in toxicological studies was as follows. Each fish was euthanized by overdose exposure to the anesthetic MS-222 (tricaine methanesulfonate). The abdominal cavity was opened to enhance internal fixation, and the entire fish was placed into Lillie’s fixative, Bouin’s fixative, or 10% neutral buffered formalin (Fournie et al., 2000). Some specimens were decalcified in a commercial decalcifying solution for 8 to 24 hours. Specimens were rinsed in water, dehydrated in ethanol, cleared in xylene or a commercial xylene substitute, and embedded in paraffin. Sections were taken through the whole fish along parasagittal and sagittal planes, mounted on glass slides, and stained with Harris’ hematoxylin and eosin.

Slides were examined with a Nikon Eclipse E600 microscope with Plan Apochromatic lenses and a Spot RT Slider (Model 2.3.1) high resolution digital camera system (Diagnostic Instruments, Inc., Sterling Heights, MI). Images were captured from the system using Spot (version 3.3 for Windows) capture software at 1600 × 1200 dpi resolution. Grayscale conversion, brightness/contrast adjustments, and plate production were performed in Adobe Photoshop 6.0 for Windows (Adobe Systems, Inc., San Jose, CA).

Results

Normal Anatomy and Histology

In contrast to mammals, the thyroid tissue of most bony fishes lacks a discrete fibrous capsule. Even in the pink salmon (Oncorhynchus gorbuscha), in which the follicles are gathered together as a highly vascularized gland, the follicles are not contained within a connective tissue capsule (Leatherland, 1994). Thyroid follicles are scattered predominantly throughout the connective tissue of the pharyngeal region near the ventral aorta or wall of the posterior branchial cavity and are often numerous proximate to the first and second branchial arteries. Adjacent muscle, cartilage, and bone may also contain random follicles (Gudernatsch, 1911). Possibly due to the lack of a connective tissue barrier, thyroid follicles can have widespread ectopic distribution; normal-appearing thyroid follicles can be found in the ocular choroid, kidney, spleen, intestine, liver, heart, and other tissues. Nonproliferative ectopic thyroid tissue has been reported in several fish species including goldfish, swordtails, trout, and barbs (Baker, 1958). Histologically, individual thyroid follicles of most fishes resemble those of other vertebrates. The follicles are usually round to oval and are lined by a single-layered epithelium that in most species is squamous or cuboidal depending on the level of metabolic activity. Laboratory-reared male medaka are a notable exception; their follicles are lined by cuboidal to columnar epithelial cells. These epithelial cells have microvilli that project into the follicular lumen (Ferguson, 1989). The center of each follicle is filled with colloid that is comprised of the protein-bound form of thyroid hormone. The amount of colloid varies according to the metabolic activity of the follicles and the extent of thyroid stimulating hormone production. The loss of luminal colloid begins at the periphery of the follicle (see Figure 4) with the appearance of clear vesicles (Leatherland, 1994). Parafollicular cells (clear cells or C-cells) are not present in the piscine thyroid. Figures 1 and 2 show the typical location, quantity, and appearance of thyroid tissue in Japanese medaka.

Histopathology

Follicular Cell Hyperplasia

Follicular cell hyperplasia is characterized by an increase in the number of follicles that usually contain colloid. The hyperplasia may be categorized as simple, nodular, or ectopic. Simple follicular cell hyperplasia is a proliferation of thyroid tissue at or near the anatomic site in which it normally occurs. For most fish species, this is the pharyngeal region. Usually, hyperplastic follicles are lined by variably basophilic cuboidal to columnar epithelial cells (Figures 3 and 4), although the epithelial cells may be flattened in follicles that contain excessive amounts of colloid. Some simple hyperplastic lesions appear especially basophilic at low magnification because the follicles are smaller and contain a reduced amount of colloid (Figure 5). Regardless of the amount of colloid, hyperplastic follicles are very well-differentiated and do not form discrete nodules or masses. Occasionally, hyperplastic follicles surround tissues such as skeletal muscle, bone, or nerve (Figure 5). The presence of nonthyroid tissues within hyperplastic lesions should not be misinterpreted as neoplastic invasion.

Nodular follicular cell hyperplasia, commonly referred to as goiter (e.g., Wolf et al., 1998), is represented by an increase in the number of thyroid follicles within a discretely circumscribed area. Nodular hyperplasia presents either as a single discrete mass as illustrated in a zebrafish lesion (Figure 6), or as a multinodular mass as seen in a rainbow trout lesion (Figure 7). Microscopically, these lesions are similar to simple follicular cell hyperplasia in that they are comprised of well-differentiated, colloid-containing follicles of various shapes and sizes (Figures 8 and 9). In many cases, surrounding tissues such as gills may contain large numbers of follicles.

Ectopic follicular cell hyperplasia is diagnosed when hyperplastic follicles occur in a variety of tissues that are distant from the primary site of the normal thyroid follicles. These lesions are comprised of well-differentiated tissue, and the proliferation of follicles may be extensive. In some cases, thyroid follicles may occupy as many as 9 different tissues, and large areas of an organ may be replaced with follicles (Figure 10). The follicles are usually lined by a cuboidal to columnar epithelium and often contain abundant colloid. In some fish species, ectopic hyperplasia may occur in anatomic locations in which preexisting ectopic follicles are unlikely to be present. Presumably, in such locations the hyperplastic tissue may arise from “rests” of primordial cells that can differentiate into thyroid tissue. Figures 10 and 11 show extensive ectopic hyperplasia in the liver, spleen, and gonad of a Japanese medaka. In these lesions, there is no evidence of abnormal cellular features that characterize neoplasia. This thyroid tissue is considered hyperplastic (versus normal ectopic) because the follicles are large, crowded, irregularly shaped, and have basophilic cuboidal to columnar epithelium.

Follicular Cell Adenoma

Follicular cell adenomas are neoplasms that are distinguished from nodular hyperplasia by their increased structural complexity and particularly by the presence of cellular atypia. Adenomas occur as single or multiple masses within normal thyroid tissue or in areas of follicular cell hyperplasia. Adenomas are usually well-demarcated masses that often compress the adjacent tissue (Figure 12). A large adenoma may be surrounded by a “pseudocapsule” of compressed stroma and collapsed follicles from the neighboring nonneoplastic thyroid tissue (Figure 13). Unlike normal or hyperplastic thyroids, adenomas may display papillary or solid growth patterns. Papillary follicular cell adenomas are characterized by follicles of varying size with internal papillary projections that protrude into the follicular lumen (Figures 14 and 15). Follicular cell adenomas with a solid growth pattern are characterized by an absence of obvious follicular structure. Where solid adenomas occur in areas of follicular hyperplasia, there may be a distinct tinctorial difference between the neoplastic and hyperplastic cells (Figure 16). In some large solid adenomas, the neoplastic cells may be arranged in cords that are separated by a fibrovascular stroma. In all adenomas, the cells exhibit a greater degree of atypia compared to hyperplastic epithelial cells. Mitotic figures sometimes occur in adenomas, but the number per microscopic field (mitotic index) is generally low.

Follicular Cell Carcinoma

Follicular cell carcinomas exhibit anaplastic features such as cellular pleomorphism, nuclear atypia, and/or a high mitotic index. For example, a follicular cell carcinoma in a flagfish (Jordanella floridae) features moderate nuclear atypia and a very high mitotic index, with 15–20 mitotic figures per 40× field (Figure 17). Some carcinomas present as discrete masses (Figure 18) but the majority have indistinct boundaries and show evidence of local invasion (Figure 19) or distant metastasis. Follicular cell carcinomas may be characterized as well- or poorly differentiated. Well-differentiated follicular cell carcinomas have densely packed follicles of varying shapes and sizes that are usually smaller than normal follicles and contain only small amounts of colloid. The follicular architecture is apparent in a well-differentiated carcinoma from a medaka (Figures 20 and 21); however, there is very little colloid present. In addition to anaplastic features that are characteristic of carcinomas (e.g., increased nuclear to cytoplasmic ratio, cellular and nuclear pleomorphism, nuclear hyperchromasia, empty-appearing nuclei, coarsely-clumped chromatin, and prominent nucleoli), atypical cellular features in piscine follicular cell carcinomas can include piling up of multiple cell layers creating the appearance of stacked nuclei (Figure 22) as illustrated in the carcinoma from the golden trout (Oncorhynchus aguabonita).

Some neoplasms that we have identified as well-differentiated carcinomas are characterized by heterogeneous growth patterns. For example, the rainbow trout (O. mykiss) follicular cell carcinoma had areas that consisted of variably shaped, colloid-containing follicles, and other more cellular areas that contained only a few follicles (Figure 23). Some of these less-differentiated areas of the tumor were barely recognizable as thyroid tissue (Figure 24). On the other hand, poorly differentiated follicular cell carcinomas bear little resemblance to normal follicular structures and usually exhibit a solid growth pattern in which the tumor cells are often deeply basophilic and have small irregular nuclei. The zebrafish lesion illustrated in Figure 25 is a large, poorly differentiated follicular cell carcinoma. Although this densely cellular neoplasm is well circumscribed, it is a carcinoma rather than an adenoma because the tumor cells have many anaplastic features (Figure 26). In some cases, carcinomas arise within the pharyngeal thyroid tissue and metastasize to the liver and kidney. Other nonpharyngeal carcinomas, however, arise de novo from ectopic tissue in the kidney or spleen. Figure 27 shows a well-differentiated follicular cell carcinoma derived from ectopic thyroid tissue in the kidney of a medaka. This tumor displayed a disorganized growth pattern and anaplastic cellular features (Figure 28). The pharyngeal thyroid tissue in this specimen did not exhibit evidence of neoplasia in the examined sections.

Proliferative Thyroid Lesions

At this writing, there are 97 cases of thyroid lesions in the RTLA from 33 teleost species representing 17 families and 8 orders and including both hyperplastic and neoplastic lesions. Of the 97 cases, a total of 27 different diagnostic categories or individual terms were generated by the scientists who originally examined these lesions (Table 1). We reevaluated these cases according to our standardized criteria, and Table 2 presents both a shortened version of each original diagnosis and the final revised diagnosis. Two cases that were originally diagnosed as adenocarcinoma were subsequently diagnosed as nodular hyperplasia. However, there were several cases that had an initial diagnosis of “hyperplasia versus carcinoma” or “adenoma versus goiter” that were assigned a diagnosis of nodular hyperplasia upon review. A single case that was initially diagnosed as a goiter was changed to follicular cell carcinoma

We also examined a series of Japanese medaka with thyroid lesions from a carcinogenicity bioassay conducted by one of the authors. A total of 402 fish were examined including 60 control and 342 experimental animals (Table 3). The total thyroid lesion prevalence in this study was 80%. The lesions were more abundant in males and were not treatment related. Although the underlying cause of the thyroid hyperplasia in this study was not specifically identified, we do consider iodine deficiency to be a possible factor. Proliferative thyroid lesions in this group of medaka included follicular cell hyperplasias, adenomas, and carcinomas.

Criteria used to distinguish hyperplastic and neoplastic thyroid lesions in teleosts are summarized in Table 4. Details concerning growth pattern, cellular arrangement, and cytological features essential for diagnosis were presented in the previous section. Table 5 expands our diagnostic criteria in terms of pharyngeal and nonpharyngeal sites.

Discussion

Care must be taken in distinguishing true neoplasms from hyperplastic lesions. In both man and animals, neoplasia is inherently far more significant than hyperplasia in terms of treatment, prognosis, and outcome. This is because the hallmark of hyperplasia is its potential for reversibility, whereas neoplasia is irreversible and involves one or more permanent cellular alterations that result in autonomous tissue proliferation. Due to unique features of the normal and proliferating teleost thyroid, extra caution must be exercised before rendering a diagnosis of thyroid neoplasia in bony fishes. It is important to remember that whether it is ectopic or pharyngeal, follicular cell hyperplasia in fishes can be extensive.

The scientific literature contains numerous reports of thyroid neoplasia in both captive and wild fishes (e.g., Nigrelli, 1952; Lightner and Meineke, 1975; Mawdesley-Thomas, 1975; Blasiola et al., 1981; Bunton and Wolfe, 1996; Harada et al., 1996). Based on those descriptions and photomicrographs, it appears that some lesions were diagnosed as neoplasms because the investigator(s) interpreted the presence of ectopic follicles as evidence of metastasis. For example, Blasiola et al. (1981) reported numerous metastatic thyroid adenocarcinomas from a captive population of kelp bass. Diagnoses of carcinoma were based on local invasiveness to surrounding tissue as well as metastasis to distant organs. The authors stated that few morphologic criteria for malignancy existed in individual cells, but that the metastatic nature of the neoplasms was evidenced by the presence of follicles within various organs of the affected fish.

The diagnosis of thyroid carcinoma in bony fishes according to our system requires clear histologic evidence of anaplasia. Neoplastic cells should exhibit cellular pleomorphism, nuclear atypia, and/or an increase in mitotic activity. True metastasis may occur but is not absolutely necessary for a lesion to be classified as a carcinoma. Nevertheless, cells comprising metastatic foci must also exhibit anaplastic features. It is not sufficient to make a diagnosis of carcinoma based solely on the occurrence of normal appearing thyroid follicles in multiple tissues as this condition may simply be ectopic thyroid follicles or ectopic follicular cell hyperplasia. Well-differentiated and poorly differentiated carcinomas occur in fishes; but in both cases, at least some anaplastic features are apparent in the lesions.

The data gathered from the Registry of Tumors in Lower Animal (RTLA) database are especially important because they highlight the need for consistency in diagnostic terminology. Considerable overlap is evident among the 27 different diagnostic categories and terms that were found in this database (Table 1), and some terms are no longer commonly applied. Initiated in 1965, the RTLA collection represents the work of many contributors; therefore, this degree of variability is understandable. However, the primary reason for the confusion was the lack of any clear criteria in the literature. Following our review of these cases, the terminology and diagnoses have been revised based on the criteria presented in this paper (Table 2).

In their review of tumors and related lesions of thyroid tissue, Leatherland and Down (2001) retained the terms “simple hyperplasia,” “adenoma,” and “carcinoma” as the primary lesion categories and stated that there are potentially many morphological variants within these main categories. Our proposed classification scheme differs considerably from that proposed by those authors. For example, specific diagnostic criteria were not presented for the 3 major lesion types that they proposed; however, we provide descriptions of detailed cytologic features that distinguish the various hyperplastic lesions from adenomas and carcinomas.

Leatherland and Down (2001) did not subcategorize hyperplastic lesions; in our scheme, 3 types of hyperplastic lesions are recognized based on the extent and location of the hyperplastic thyroid tissue. Those authors provided a complicated scheme for the subcategorization of adenomas that included macrofollicular, trabeculate, microfollicular or fetal, afollicular or embryonal, papillary follicular, and colloid adenomas. Criteria appeared to be based on the size and shape of the follicles and the amount of colloid that was present within the follicles. In a previous publication, Leatherland (1994) used the same terminology to describe different types of goiters that, of course, are hyperplastic rather than neoplastic lesions. In our opinion, the adenomas they described appear to be nonneoplastic, and thus would be more accurately diagnosed as nodular hyperplasia. The development of such extensive subclassification systems results in complex terminology that appears to serve little practical purpose. Last, Leatherland and Down (2001) indicated that carcinomas exhibit pathological or clinical evidence of malignancy but did not describe or illustrate those features for any carcinomas. In contrast, we present several cases of follicular cell carcinoma in this paper and provide criteria that should allow a pathologist to make a diagnosis of carcinoma based on histomorphologic features.

Proliferation of thyroid follicular cells in laboratory rodents is generally considered to follow a developmental progression from hyperplasia to neoplasia (Hardisty and Boorman, 1990). Adenoma in the Fischer rat can occur as single or multiple masses in otherwise normal thyroid glands or in glands with focal or diffuse follicular cell hyperplasia. Similarly, we have observed adenomas in Japanese medaka that occurred as single or multiple masses in normal thyroid tissue or within areas of follicular cell hyperplasia, suggesting that a progression from hyperplasia to neoplasia may exist in fish.

Numerous reports support the fact that multiple factors, both extrinsic and intrinsic, may stimulate thyroid hyperplasia (see Hoover, 1984). Hoover (1984) provided an excellent example of how an environmental factor might influence proliferation of the fish thyroid. She reported an increased prevalence of pharyngeal thyroid growths in female guppies from a study designed to assess the effects of N-hydroxyfluorenylacetamide on fish that were maintained at different temperatures. In both control and treated fish, hyperplastic and ectopic thyroid lesions were significantly more prevalent in fish held at 27°C compared to fish held at 17°C. Despite evidence of marked proliferation in pharyngeal and nonpharyngeal sites, the lesions were appropriately termed “hyperplasia” rather than “neoplasia.” Such studies further emphasize the need to exercise caution in diagnosing thyroid growths as neoplastic.

We did not address proliferative thyroid lesions in elasmobranch fishes in this paper for several reasons, primarily because differentiating hyperplasia from neoplasia is less complicated in cartilaginous fishes than in teleosts. Because the elasmobranch thyroid is an encapsulated organ, local invasion of the surrounding tissues is a valid criterion for malignancy, whereas the concept of “invasiveness” is less meaningful in teleosts due to the diffuse distribution of their normal thyroid tissue. The majority of proliferative lesions reported from elasmobranchs are goiters, not neoplasms, and would be classified as nodular hyperplasia according to our system. However, we prefer to avoid using the clinical term goiter as a histopathological diagnosis. Crow et al. (2001) provided a thorough histological assessment of goiters in elasmobranchs based on the examination of 12 cases from the RTLA collection. They described the types of goiters that occur, indicated that goiters may result from both hypertrophy and hyperplasia, and discussed various types of conditions that can lead to thyroid enlargement.

It is difficult to evaluate and compare many of the reports of thyroid neoplasms in fishes. Many investigators do not provide thorough gross and microscopic descriptions of the lesions and adequate illustrations are often lacking. For neoplasms, it is essential to accurately describe any prominent histologic patterns and to furnish specific details concerning their cellular morphology. Realistically, thyroid lesions from different studies cannot be compared unless the texts are supplemented with high-quality figures that illustrate important diagnostic features. For true tumor comparisons, however, even high-quality images cannot supplant a light microscopic inspection of the histologic sections. Histologic slides of representative lesions should be deposited in a central repository and made available for other researchers to examine and appraise. One such repository is the National Cancer Institute’s Registry of Tumors in Lower Animals in Sterling, Virginia, USA. Similar to the way in which taxonomists routinely deposit a type specimen of a new species in an accredited museum, so new material will be available for future study, histologic slides that demonstrate key diagnostic features can be submitted to the RTLA to the benefit of the scientific community.