Abstract
Spontaneous follicular and C-cell tumors of the thyroid gland in the Han Wistar rat were examined using two morphologic procedures. Firstly, in situ hybridization (ISH) was used to localize thyroglobulin (TG) and calcitonin (CT) mRNAs. Secondly, the proteins for these markers were detected using immunohistochemistry (IHC). The aim was to study the morphology of the tumors and to examine the usefulness of TG and CT markers in the differential diagnosis of these lesions. Follicular tumors with cystic, papillary and follicular patterns showed relatively consistent expression of TG mRNA by ISH, thereby confirming the diagnostic value of this technique. However, no staining for TG markers was observed in solid lesions. In general, C-cell tumors comprised well-differentiated cells that continued to express CT mRNA and peptides even after embolic spread and metastasis. Therefore, the performance of either ISH or IHC for CT markers can be used for diagnostic confirmation. Additional features noted in C-cell tumors included the appearance of tumor emboli or metastases in association with small primary lesions (less than 5 average follicular diameters in size) and the presence of eosinophilic (amyloid-like) material showing immunopositivity for CT peptides. Finally, evidence is provided for the sequestration of TG protein by proliferating C-cells.
Introduction
C-cell tumors of the thyroid gland are common spontaneous neoplasms in the laboratory rat (Hardisty and Boorman, 1990; DeLellis, 1994). Follicular tumors at this site are generally less frequent, although a moderately high incidence has been reported in the Han Wistar rat (Kaspareit-Rittinghausen et al., 1990). In carcinogenicity studies, differential diagnosis of thyroid lesions in the Wistar rat can be problematic as more than one type of tumor may be present and multiple tumors often co-exist. Small follicular lesions of a solid histologic pattern can be difficult to differentiate from C-cell carcinoma (Hardisty and Boorman, 1990). In addition, borderline proliferative C-cell lesions present diagnostic difficulties, particularly when attempting to distinguish between hyperplasia and adenoma (Botts et al., 1991). Although several publications describe the presence of calcitonin (CT) peptides in rodent C-cell tumors (DeLellis et al., 1979; Deftos et al., 1980; Martin-Lacave et al., 2002; Sawicki et al., 2006), no studies of thyroglobulin (TG) expression in follicular neoplasms appear to have been reported in the rat. Accordingly, it was decided to review the histopathology of thyroid tumors in the Han Wistar rat and to correlate the morphologic features with the expression of TG and CT mRNAs and proteins. To date, most work on rat thyroid tumors has involved the use of immunohistochemistry (IHC) alone but it was also considered important for the present study to utilize in situ hybridization (ISH) for the localization of mRNAs. The demonstration of both mRNA and protein for a particular marker, within a tumor cell, gives a more reliable indication of cell type, and this is of obvious importance when attempting to determine tumor histogenesis. Others have commented on the value of correlating the localization of mRNA by ISH and the localization of peptide by IHC in the study of pathologic changes in endocrine tissue (Neonakis et al., 1994).
The primary aim of the present study was to confirm the usefulness of TG and CT markers in the differential diagnosis of primary and metastatic thyroid neoplasms. A further intention was to determine whether “mixed” or “biphasic” tumors were present in the sample of thyroid glands investigated. Interestingly, such mixed (C-cell and follicular) thyroid carcinomas have been frequently reported in humans (Albores-Saavedra et al., 1985; Papotti et al., 2000), yet this lesion does not appear to have been recorded in other species, including the rat.
Materials and Methods
Tissues
All tissues came from a 30-month background carcinogenicity study, involving 500 untreated Han Wistar rats (Table 1). The animals were supplied by the Small Animal Breeding Unit, Glaxo Group Research Ltd, Bury Green, UK. Of these, 26% per cent of the rats survived until termination with the remainder being sacrificed for humane reasons at ages ranging between 18 and 29 months. The rats were fed a conventional rodent diet (Rat & Mouse Number 1, Expanded Maintenance Diet; Special Diets Services, Witham, UK) and were provided with water ad libitum throughout the study.
Rats were sacrificed by exsanguination from the abdominal aorta under isoflurane anesthesia. All major organs were removed and immersion fixed in 10% neutral buffered formalin. Thyroid glands (including masses) were fixed in situ on the trachea while the cervical lymph nodes were fixed whole. The samples were fixed for varying periods up to 1 month, before being dehydrated through graded ethanol, xylene, and embedded in paraffin wax. Following preliminary histologic examinations, thyroid glands containing proliferative lesions were selected for ISH and IHC from 25 male and 25 female animals.
In addition, sections of the lung and deep cervical lymph nodes were screened histologically from all animals and deep cervical nodes from 6 rats, containing metastases, were selected for ISH and IHC. None of the chosen tissues came from rats that were found dead, i.e., in all instances the postmortem interval was short as it was considered that autolysis may have adversely affected the ISH and IHC procedures. Serial sections (3 μm) from each sample were cut onto pre-coated silanized slides (Shandon, Runcorn, UK), and numbered for the following staining procedures: (1) hematoxylin and eosin (H&E); (2) CT ISH; (3) CT IHC; (4) TG ISH; (5) TG IHC. Attempts were made to ensure as far as possible that the ISH and IHC staining for each marker (i.e., CT or TG) were performed on serial sections.
All histopathologic diagnoses generally were made in accordance with the Society of Toxicologic Pathologists standard nomenclature guidelines (Botts et al., 1991). For each ISH or IHC procedure, an assessment was made of the number of positively stained cells on the following scale: (1), less than 20% positive cells; (2), 20–40% positive cells; (3), 40–60% positive cells; (4), 60–80% positive cells, and (5), 80–100% positive cells.
In Situ Hybridization
A cocktail of three 28 or 29 base pair cDNA oligonucleotide probes, complementary to exon 4 of the human CT mRNA sequence (Le Moullec et al., 1984) was synthesized. The sequences of the CT probes were as follows: Probe 1: 5′-GTGCCCAGCATGCAAGTACTCAGATTACC-3′; Probe 2: 5′-ACGTGTGAAACTTGTTGAAGTCCTGCGTG-3′; Probe 3: 5′-AGGTGCTCCAACCCCAATTGCAGTTTGG-3′. This probe cocktail has been shown to localize CT mRNA in normal C-cells in the rat without further modification (Thomas et al., 1993). A 27-base pair cDNA oligonucleotide probe, complementary to human TG mRNA sequence (Malthiery and Lissitzky, 1987), was also prepared. The sequence of the TG probe was as follows: 5′-TCCCTTCGG CGTCCACACACCAGCACT-3′.
This probe has been used by others to demonstrate TG mRNA in follicular cells in the rat thyroid gland (Thomas G.A., personal communication) and shows 92% homology with the rodent mRNA sequence (Marino et al., 1999). The TG probe was designed against a consensus sequence enabling two probes to bind to each TG mRNA molecule, thereby improving the sensitivity of the procedure. All cDNA probes were synthesized by Oswell DNA Service, (Southampton, UK) (Applied Biosystems, 1991). The probes were purified by high pressure liquid chromatography and labelled at both the 5′ and 3′ ends with a single molecule of digoxigenin (Boehringer, Bracknell, UK).
The protocol used for ISH has been described previously (Farquharson et al., 1990; Pilling et al., 1999: Pilling, 2003). Briefly, following dewaxing and rehydration, sections were pretreated in proteinase K (1 μg/ml for 40 minutes at 37°C) and placed in prehybridization buffer for 1 hour at 42°C. This buffer was made up as follows: commercial hybridization buffer (Sigma P1415; Poole, UK), 4X standard saline citrate (SSC), 0.01M tris-HCl, 4 mg/ml sodium pyrophosphate, 1 ml/L diethylpyrocarbonate water. The hybridization was carried out with the appropriate probe (TG concentration 0.3 ng/μl; CT 0.1 ng/μl) diluted in this buffer overnight in a moist chamber at 42°C. After washing in graded SSC, bound probe was localized with alkaline phosphatase-linked anti-digoxigenin antibody (Boehringer) at 1:500 dilution in 10% normal sheep serum (Sigma) for 1 hour.
The final detection step was carried out overnight at room temperature using nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl phosphate in a buffer solution containing 1.2% trizma, 0.5% sodium chloride and 0.2% levamisole (all detection reagents Sigma, Poole, UK). After dehydration and mounting, slides were examined and photographed. A positive signal was demonstrated by the presence of dark purple or black diformazan cytoplasmic reaction product.
The probe concentrations and proteinase K conditions were optimised using normal thyroid glands from rats on the same study and of a similar age. These tissue samples were subsequently used as positive controls for each staining run. Negative control sections, also used in each staining run, were either (a) pretreated with RNAase A (100 μg/ml) before hybridization with labelled probes; (b) hybridized with inappropriate probes of similar length and C:G ratio (E. coli, cytosine deaminase, three 30-mer oligo-probe cocktail; R&D Ltd, Abingdon, Oxford, UK), and (c) hybridized in the absence of labelled probes.
Immunohistochemistry
The sections were dewaxed and incubated with 0.5% hydrogen peroxide in methanol for 30 minutes at room temperature to inhibit endogenous peroxidases. Nonspecific protein binding was blocked using normal swine serum (Dako, Ely, UK) diluted 1:5 in tris-buffered saline. Rabbit polyclonal antibodies raised against human TG (1:10,000; Dako A0251) and human CT (1:200; Novocastra, Newcastle, UK) were diluted in normal swine serum and applied to the sections overnight at 4°C in a humidity chamber. In order to verify the results obtained with the above antibody, a second anti-TG antibody (1:1500; Biogenesis 8900-0424, Poole, UK) was obtained and used in the same procedure on a limited number of tissue samples.
After washing, biotinylated swine anti-rabbit immunoglobulins (Dako) diluted at 1:100 in 20% normal swine serum (Dako) were applied for 30 minutes. The sections were incubated with StreptABComplex/HRP (Dako) for 30 minutes and developed in freshly prepared diaminobenzidine tetrahydrochloride in a buffer solution containing 0.05M tris HCl and 0.02% hydrogen peroxide at pH7 (all development reagents Sigma, Poole UK). Finally, the sections were counterstained with Mayer’s hematoxylin.
The anti-CT and anti-TG antibody concentrations were optimised using normal thyroid glands from rats on the same study and of a similar age. These tissue samples were subsequently used as positive controls for each staining run. Negative controls included: omitting the primary antibody from the procedure and the replacement of the primary antibody with normal rabbit immunoglobulin fraction (Dako).
Results
The numbers and types of thyroid proliferative lesions evaluated are listed in Table 1.
H&E Sections
Follicular cell adenomas (FCA) exhibited four histologic patterns: cystic, papillary, follicular (macrofollicular), or solid (microfollicular). A small number of these tumors was regarded as “mixed” as they contained more than one histologic pattern. Morphologic patterns within follicular cell carcinomas (FCC) were similar to FCA, although the cystic pattern was not observed in malignant tumors. FCC was distinguished from FCA by the presence of cellular pleomorphism, a high mitotic index, widespread necrosis and penetration of the thyroid gland capsule. No metastases were noted in deep cervical lymph nodes or lung in association with FCCs.
Diffuse C-cell hyperplasia (CCH) was present in 49/50 thyroid glands investigated, the one exception being a male where the gland contained a large C-cell adenoma (CCA), but no other thyroid tissue was present for evaluation. Other proliferative lesions (C-cell and follicular) were invariably present in association with diffuse CCH. In a few cases of C-cell carcinoma (CCC) a separate CCA was also present in the same thyroid gland. Anaplasia, pleomorphism and penetration of the thyroid gland capsule were prominent features of CCC and allowed differentiation from CCA.
The morphology of all C-cell proliferative lesions conformed with previous descriptions, however an additional feature in the form of small acinar structures was present in several CCAs and CCCs. These acini usually comprised cells lining a central empty space although some acini contained hyaline, eosinophilic material. Another observation was that of tumor-cell emboli in vessels adjacent to the C-cell lesion. Such emboli were present in 4/9 overt CCCs and also in five smaller lesions that, in the absence of the emboli, would have been diagnosed as focal CCH or CCA.
In all these cases, the vessels involved appeared to be small venules or lymphatics located at a discrete distance from the edge of the lesion and frequently just outside the thyroid capsule. Metastases in the deep cervical lymph nodes were seen in association with 6/9 CCCs. The morphologic appearance of each metastasis was representative of that seen in the primary lesion. Interestingly, metastases were seen in 2 cases, of the 6 identified, where the size of the primary C-cell lesion was smaller than 5 average follicular diameters.
Thyroglobulin Expression
Normal follicular tissue was identified in the majority of thyroid glands although frequently this tissue was adjacent to a proliferative follicular or C-cell lesion. Considerable variation in the shape and size of the follicles was observed. Consistent with the differences in follicular size, marked interfollicular variation occurred with the strength of staining for TG mRNA and protein. Strong staining for TG mRNA was observed in small follicles with cuboidal epithelium, whereas in the large follicles with attenuated epithelium, only occasional positive cells were seen (Figure 1). Staining for TG protein was not present in all thyroid follicular cells in the normal gland. TG signals were occasionally observed in cuboidal follicular epithelium, typically as small apical granules, but no positive staining was noted in attenuated epithelium. In the large majority of follicles, the colloid stained strongly for TG protein.
Staining for TG expression in three cases of follicular cell hyperplasia (FCH) gave variable results (Table 2). Positive staining for TG mRNA was present in the majority of follicular cells in two of these lesions. TG protein was observed in a minority of cells in one lesion, with these cells also staining positively for TG mRNA. One FCH lesion was completely negative for TG markers.
TG expression (mRNA or protein) was observed in all histologic patterns of FCA, with the exception of the solid pattern, present in two tumors, which exhibited no staining (Table 2). TG mRNA or protein were present within a proportion of cells in both follicular and papillary lesions (Figure 2A, 2B) although in general significantly more cells stained positively for TG mRNA than protein. In the cystic tumors, TG mRNA was observed in a minority of cells in all lesions but signals for TG protein were absent. TG mRNA or protein were present in FCCs in all follicular and papillary areas but not solid regions (Figure 3A, 3B). In the positive areas the numbers of cells giving signals for TG mRNA or protein varied considerably, and there was no consistent mRNA:protein ratio in the neoplastic cells, but in general it appeared that the same cells stained positively for both TG mRNA and protein. The colloid, where present, stained strongly for TG protein in the majority of cases of FCA or FCC.
Surprisingly, IHC staining with anti-TG antibody was observed in the large majority of C-cell lesions of all types including CCH and CCA (Table 2). Identical results were obtained with both of the anti-TG antibodies used. The pattern of staining was similar in most cases, with only a small proportion of cells giving positive signals. The staining was present in scattered groups of cells that were usually located in close proximity to a trapped follicle (Figure 4A). A strong, membranous staining pattern was frequently observed in these groups of cells, with diffuse cytoplasmic staining being noted in some instances. In CCCs, only well differentiated, morphologically normal cells gave positive signals for TG protein. Groups of cells staining positively with the anti-TG antibody, and present in serial sections, were also positive for CT mRNA and peptides (Figure 4B), however no signals for TG mRNA (Figure 4C) were obtained from these cells. In CCCs, no staining for TG protein was seen in C-cell emboli or metastases in cervical lymph nodes.
Calcitonin Expression
In normal thyroid glands, CT mRNA and peptides were present in cells whose morphology and parafollicular distribution identified them as C-cells. These cells exhibited strong ISH and IHC cytoplasmic signals for both CT mRNA and peptides with little intercellular heterogeneity in the staining intensity.
CT mRNA and peptides were observed in all C-cell lesions (Table 2). Strong signals for CT mRNA were present in all CCH and CCAs with the large majority of cells exhibiting strong cytoplasmic staining. IHC staining for CT peptides was also strong with the exception of a few lesions that contained a minority of cells showing positive staining. In all primary CCCs the majority of cells stained positively for CT mRNA and peptides, although unlike normal C-cells, there was considerable intercellular heterogeneity in the intensity of ISH and IHC staining. In general this staining appeared to correlate with the degree of structural differentiation with the morphologically normal cells giving the strongest signals (Figure 5). In all C-cell lesions, cells staining positively for CT mRNA, and present in serial sections, were also positive for CT peptides.
Acinar structures in C-cell lesions were consistently positive for CT mRNA and peptides. The eosinophilic material present in the acinar lumina stained positively for CT peptides (Figure 6A, 6B) but negatively for CT mRNA. Also, this eosinophilic material did not stain positively for TG markers. Strong signals for CT mRNA and peptides were observed in the cells in the vascular cancer emboli associated with these C-cell carcinomas. Metastatic cells in the cervical lymph nodes showed positive staining for CT mRNA and peptides in the metastatic cells (Table 2; Figure 7), with the majority of cells giving positive signals for both markers. In all cases, no signals for CT mRNA or peptides were observed in the follicular cells or colloid of the normal gland, nor in any of the proliferative lesions obviously having follicular origins.
Discussion
This paper describes the first use of ISH and IHC to investigate TG expression in the normal and neoplastic thyroid gland of the rat. In normal thyroid tissue, the strength of staining for TG mRNA and protein correlated with the height, and therefore with the degree of activity, of the follicular epithelium. IHC staining for TG protein was usually in the form of small apical granules, which presumably represent resorbed colloid droplets on the apical border of the follicular cell (Thomas and Williams, 1994).
These results confirm that TG expression can be reliably used to confirm the histogenesis of follicular neoplasms, of all histologic patterns, with the exception of solid lesions. The latter were consistently negative for TG mRNA and protein, as presumably they are the most undifferentiated forms in the progression of the neoplasm, and may be unable to synthesize and store TG protein. No metastatic cells of FCC in lung or lymph nodes were identified in this study, which is unsurprising as metastasis of malignant follicular tumors appears to be an uncommon event in rats (Hardisty and Boorman, 1990). The results of the TG IHC staining of rat follicular tumors presented here are broadly in agreement with those published for human tumors (de Micco et al., 1993).
Calcitonin expression was chosen as a marker of C-cell lineage in this study. Although other peptides are known to be produced by C-cells (Zabel et al., 1987; DeLellis, 1994), CT is generally regarded as the most important specific marker of these cells and is considered a prerequisite for the diagnosis of CCC (medullary thyroid carcinoma) in humans (DeLellis et al., 1978). Hyperplastic and neoplastic C-cell lesions did not exhibit the same range of morphologic patterns as follicular lesions, in the present study, and on the whole the cellular composition was relatively well differentiated.
Strong staining for CT markers was present in all C-cell lesions with the numbers of positive cells decreasing only slightly with progression from CCH through CCA to CCC. This reduction in staining corresponded with an increase in anaplastic and pleomorphic cells in CCC that were often negative for CT expression. CT mRNA and peptides were also present to some degree in all C-cell emboli and cervical lymph node metastases. These results are consistent with those of others (Deftos et al., 1980; Martin-Lacave et al., 2002), who reported the presence of CT peptides in a large number of proliferative C-cell lesions in the rat although ISH staining for CT mRNA was not performed by these authors.
A consistent observation in the present study was that in general more cells stained positively for TG or CT mRNA, than they did for the respective peptides, in both follicular (except solid patterns) and C-cell tumors respectively. Neonakis et al. (1994) also found more consistency with ISH rather than IHC staining for CT in human tumors and suggested that the discrepancy was due to tumor cells losing their storage but not their synthetic capacity.
Tumor emboli and metastases, all staining positive for CT markers, were observed in association with several small C-cell lesions within the thyroid gland itself. These lesions were less than 1 mm in diameter, and certainly not exceeding “5 average follicular diameters,” which is the appropriate size to be graded as focal hyperplasia in the STP classification scheme (Botts et al., 1991). Each of the cases was examined carefully, using all available serial sections, to try and ensure as far as possible that the periphery of a larger tumor was not being sampled. Although this could not be definitively ruled out in every case, it seems unlikely that it could have accounted for all of these observations.
The fact that such small lesions, comprising well-differentiated cells with minimal atypia, are capable of vascular invasion and metastasis does not seem to be widely acknowledged in the literature. One can speculate that neoplastic C-cells may develop metastatic capabilities more quickly than other cell types. The reasons for this are unclear, but may relate to the re-establishment of latent migratory abilities that C-cells exhibit in the embryo during the period of so called “neural crest dispersion” (Pearse and Takor Takor, 1976).
Small acinar structures, occasionally with eosinophilic lumenal contents, were noted in several C-cell tumors. In all cases these acinar cells stained positively for CT mRNA and peptides, but negative for TG mRNA and protein, a pattern consistent with C-cell differentiation. Glandular differentiation in rodent C-cell neoplasms does not appear to have been previously reported, but the presence of acinar and tubular structures is a well-recognized feature in the glandular variant of MTC in humans (Lloyd, 2000). The reasons for the formation of glandular structures may relate to the de-differentiation of neoplastic C-cells, as glandular structures have been observed in the ultimobranchial body (Toran-Allerand, 1986) from which adult C-cells are derived.
The extracellular eosinophilic material seen both within acini and elsewhere in C-cell tumors stained uniformly positive for CT peptides but negatively for TG protein. This material may represent a form of amyloid, as it is known that amyloid contains at least part of the amino acid sequence of CT (Williams et al., 1987). The presence of amyloid in C-cell lesions in the rat has been reported (Burek, 1978; Hardisty and Boorman, 1990; DeLellis, 1994) but as yet the CT immunopositivity of amyloid in this species has not been recognised.
An unexpected finding in the present study was the IHC staining of C-cells with the anti-TG antibody. This was consistently observed in each type of C-cell lesion, in both hyperplasias and neoplasms. The cells also stained positively for CT mRNA and protein, as would be expected in C-cells, but were negative for TG mRNA. It is believed that these observations are consistent with the findings by others (Holm et al., 1987; de Micco et al., 1993) of TG in human C-cells (MTC). These authors suggested that the staining was due to reabsorption, by C-cells, of TG that had diffused into the interstitial tissue.
Credence for this hypothesis comes from the early work of Daniel et al. (1967) who showed, in rats, that the movement of TG from follicles into the lymphatics, via the interstitium, is a normal physiologic process. It is possible to speculate that proliferating C-cells are in some way interfering with the transport of TG from the follicular cells into the blood vessels or lymphatics, and/or indulging in nonspecific endocytosis of the protein. Further evidence that the positive staining of C-cells is nonartefactual and specifically due to the presence of TG, is the consistent close proximity of positive cells to trapped follicles in primary lesions, and also by the complete absence of TG signals in neoplastic C-cells within the emboli and distant metastases. These results clearly indicate that, in the diagnosis of thyroid lesions, any results from the use of anti-TG antibodies should be interpreted with care.
Numerous reports describe the occurrence of mixed (C-cell and follicular) thyroid carcinomas in humans (Albores-Saavedra et al., 1985; Livolsi, 1987; Papotti et al., 2000), although this entity does not to date appear to have been described in the rat, or any other species. In these lesions, areas of classical C-cell appearance are intermingled with follicles or papillary structures, with the diagnosis being dependent on the unequivocal demonstration of both C-cell and follicular markers in tumor cells, by ISH or IHC. The histogenesis of the lesion is controversial. However, no such tumors, with mixed histologic patterns, were identified in the present study.
A number of thyroid glands were identified in which both a follicular, and a C-cell proliferative lesion, were present in close proximity, but these were considered to be merging (collision tumors). This is not surprising given the high incidence of these lesions in the sample of thyroid glands investigated. The absence of any tumor showing both C-cell and follicular differentiation would appear to support the concept of the origin of C-cells and follicular cells from separate germ layers, namely the neural crest and endoderm, respectively (Pearse and Carvalheira, 1967).
In summary, the large majority of thyroid tumors, in the present study, expressed the mature differentiated phenotype of either C-cell or follicular neoplasms. In general, C-cell tumors comprised well-differentiated cells that continued to express CT mRNA and peptides even after embolic spread and metastasis. Therefore, the performance of either ISH or IHC for these markers can be used for diagnostic confirmation. Follicular tumors presented a more diverse range of morphologic patterns but, with the exception of solid (microfollicular) patterns, the production of TG mRNA was a consistent feature. In the latter case, ISH for TG mRNA would be the preferred diagnostic tool.