Familial follicular cell thyroid carcinomas in a large number of Dutch German longhaired pointers

Abstract Thyroid carcinomas (TCs) originating from follicular cells of the thyroid gland occur in both humans and dogs, and they have highly similar histomorphologic patterns. In dogs, TCs have not been extensively investigated, especially concerning the familial origin of TCs. Here, we report familial thyroid follicular cell carcinomas (FCCs) confirmed by histology in 54 Dutch origin German longhaired pointers. From the pedigree, 45 of 54 histopathologically confirmed cases are closely related to a pair of first‐half cousins in the past, indicating a familial disease. In addition, genetics contributed more to the thyroid FCC than other factors by an estimated heritability of 0.62 based on pedigree. The age of diagnosis ranged between 4.5 and 13.5 years, and 76% of cases were diagnosed before 10 years of age, implying an early onset of disease. We observed a significant higher pedigree‐based inbreeding coefficient in the affected dogs (mean F, 0.23) compared to unaffected dogs (mean F, 0.14), suggesting the contribution of inbreeding to tumour development. The unique occurrence of familial thyroid FCC in this dog population and the large number of affected dogs make this population an important model to identify the genetic basis of familial thyroid FCC in this breed and may contribute to the research into pathogenesis, prevention and treatment in humans.


| INTRODUCTION
Many dog breeds are predisposed to a variety of specific cancers due to consanguinity and inbreeding. 1 According to researches, cancer is one of the major cause of death in dogs, accounting for 8.7%-27% of all deaths. [2][3][4] Skin and soft tissues were the most common sites for tumour development, followed by alimentary, mammary, urogenital, lymphoid, endocrine and oropharyngeal. 2 Within the tumours in the endocrine organs, thyroid carcinoma (TC) is the most common type, which represents 1.2%-3.8% of all canine tumours and accounts for 90% of thyroid tumours. 5 undifferentiated carcinomas and thyroid carcinosarcomas (TCs) are also recognized. 8 In humans, TC is the ninth most common type of cancer and accounts for approximately 3.1% of all cancers. 9 The histologic growth patterns in humans are largely similar to those in dogs.
Additionally, TC shows no sex preference in dogs, although in humans, females have a three-fold higher risk than males. 7,10 The prevalence of TC in older dogs (between 10 and 15 years old) is significantly higher compared to earlier onset. 7 Thyroid tumours can be of familial or spontaneous origin. In humans, the majority of TCs are sporadic, and approximately, 5%-15% of them are considered to be of familial origin. 11,12 Due to the relatively low prevalence of familial TCs, the genetic causes are less investigated than sporadic types, thus are still poorly understood. 13 To the authors' knowledge, in dogs, there has only been one pedigree of apparent familial medullary TC reported. 14 Investigations and reports of familial thyroid tumours in dogs have been limited.
Over a period of more than 21 years, a relatively large number of TCs were diagnosed in the German longhaired pointers (GLPs) born in the Netherlands (Dutch GLPs). In this retrospective study, we review clinical and histopathological assessments of the GLPs with thyroid tumours and present genetic assessment including the inbreeding and heritability estimation based on pedigree.

| Study population
Medical records of the clinics belonging to Dutch and Belgian collaborating veterinary cancer centres and the database of two Dutch veterinary diagnostic pathology laboratories were searched for client-owned GLPs diagnosed with thyroid tumours between 1996 and 2017. Additionally, the owners of GLPs registered in the database of the Dutch GLP association were contacted to identify any dogs with a history of thyroid tumour. Once the dog was diagnosed with a thyroid tumour, the primary or referring veterinarian was contacted to obtain relevant information. If more than one dog was affected in the litter, the owners of the remaining littermates as well as dogs related to each of the parents were identified and contacted. Pedigree records were provided by GLP "Langhaar" association (www.germanlonghair.com) in order to perform a pedigree analysis.
Only GLPs with histopathologically confirmed follicular cell TC were included as cases in this study and used in genetic analysis. Surgical removal of the affected thyroid glands, when feasible, was centralized in one clinic. Tumour tissue obtained either at surgery or necropsy was collected from affected dogs, and part of the sample was formalin-fixed for histopathology, and the remaining sample was stored in RNAlater (RNA stabilization reagent: Qiagen, Hilden, Germany).
Cases were excluded if owners rejected to participate in the research. All the data and samples in this research were permitted to be used for scientific purpose and in publication.

| Clinical data
The following information was retrieved from the medical records, if available: signalment, physical examination findings including tumour size (longest diameter), location and mobility (determined by palpation), clinical signs, time to presentation and date of diagnosis.
Whenever performed, the results of additional diagnostic tests, including blood tests and imaging tests, were recorded. Blood tests included complete blood cell count, serum biochemistry profiles, basal circulating total thyroxine (TT4) and thyroid stimulating hormone (TSH) concentrations. Staging was performed using diagnostic imaging (thoracic radiographs, cervical ultrasonography and computed tomography [CT]). If available, the presence of ectopic thyroid tumour was recorded. For negative controls, the primary antibody was omitted.

| Histopathology analysis
Tissues were evaluated by two veterinary pathologists and classified according to the World Health Organization (WHO) classification of tumours of the endocrine system scheme. 8 If a tumour had multiple growth patterns, then classification was based on the most predominant pattern. If capsular penetration of the neoplasm was unclear, additional H&E sections were cut for additional evaluation.

| Genetic analysis
To assess the genetic relationship between dogs collected, the family tree of all unaffected and affected (suspected and histopathologically diagnosed) dogs were constructed using Kinship2 package in R. 15 Pedigree-based inbreeding coefficients (F) of all the dogs were estimated using the CFC (Coancestry, inbreeding (F) and Contribution) program 16 based on the whole pedigree of GLPs. To evaluate the contribution of inbreeding to the incidence of the thyroid tumours in the population, the rank sum test of F between affected dogs histopathologically confirmed and unaffected dogs born before 2007 was done in R using Wilcoxon test. We excluded 86 unaffected dogs born after 2007, because many of these dogs are closely related to the affected dogs, and they could be highly susceptible to FCC. Although they are unaffected at the time of analysis, they could become affected later in their lives, thus biasing the result.

| Heritability estimation
Heritability was estimated using ASReml 4.1 based on the pedigree relationship between the unaffected dogs and cases histopathologically confirmed. 17 Unaffected dogs born after 2007 were also excluded from the estimation. The model used is as follows.
where y is the phenotype, which is a binary trait, affected status coded as 1 and unaffected status coded as 0. α is the fixed effect of gender, female or male. δ is the random animal effect. e is the random residual.
Heritability calculation equation is: where v δ is the variance of the random animal effect, and v y is the variance of FCC phenotype.

| RESULTS
In total, 264 GLPs born between 1991 and 2017 were identified (

| Diagnostic findings
Thirty-three of 54 dogs (61%) underwent at least one diagnostic imaging, including CT of the cervical region and thorax (13 dogs), cervical ultrasonography (three dogs), thoracic radiographs (22 dogs) and abdominal ultrasonography (four dogs). Six of these 33 dogs had more than one test performed. Sixteen of 54 dogs (30%) had no imaging, while in five dogs (9%), required data were missing.
Based on diagnostic imaging, four dogs had involvement of the regional lymph nodes: two dogs ipsilateral retropharyngeal lymph node, one dog ipsilateral mandibular and retropharyngeal lymph node and one dog ipsilateral cervical superficial lymph node. Histopathology confirmed metastatic disease in three dogs. One dog underwent post-mortem examination, but the suspected lymph node was not evaluated.
Distant metastases were suspected in only one dog (pulmonary nodules); however, further diagnostics were not performed to confirm this.

| Histopathology
Thyroid FCCs were diagnosed in 54 dogs. Bilateral neoplasms were diagnosed in 29 dogs. The majority of the 83 carcinomas showed a follicular growth pattern (n = 37; Figure 1A), whereas compact (solid) (n = 15; Figure 1B), follicular-compact (n = 16) and papillary (n = 9; Figure 1C) growth patterns were seen in the other carcinomas. In three dogs, a carcinosarcoma, characterized by osteosarcoma and carcinoma ( Figure 1D), was diagnosed. In two dogs, a carcinoma not otherwise specified (NOS) was diagnosed. In one dog, diagnosed in 1996, which was the first case we found, the diagnosis was only thyroid tumour with signs of malignancy. In four carcinomas, welldifferentiated bone tissue was seen (metaplastic bone formation). An ectopic compact FCC was found at the heart-base during necropsy in one dog that also had follicular-compact type carcinoma in both thyroid glands.
Immunohistochemistry was performed on the neoplasms of 40 dogs. The neoplastic cells were vaguely to markedly positive for thyroglobulin in all tumours. The strongest immunoreactivity was typically noted in the colloid with lower staining intensity in the neoplastic cells. All neoplasms were negative for calcitonin.

| Relationship between affected dogs
According to the pedigree of all collected GLPs ( Figure S1) GLP905 has an unknown case status due to inaccessibility, but has one suspected affected full-sibling, GLP13, and one affected half-sibling, GLP47, with histological diagnosis.

| DISCUSSION
In humans, familial TC is diagnosed when two or more first-degree relatives are affected. 12 Here, we showed that the incidence of FCC is strikingly high in some families of Dutch GLPs, like in the pedigrees of GLP52 and GLP905. These two dogs have a most recent common ancestor, GLP306 ( Figure S2), born in 1989, with the F of 20.57%.
Furthermore, 78 probably affected GLPs (26 suspected and 52 histopathologically confirmed FCC cases) can be traced back to a common cross of six generations prior to GLP52, the cross between GLP319 and GLP296 ( Figure S3). With such close relationships between the majority of the affected dogs, the FCC in these dogs is considered to be a familial disease.
In this study, besides the 54 histopathologically confirmed cases, twenty-nine dogs were suspected to be affected by thyroid tumour based on clinical findings (e.g., presence of a mass lesion at the location of the thyroid gland), but because no histological assessment was performed, these suspicions could not be confirmed. Interestingly, these suspected cases are very closely related to the most affected GLPs with diagnosis ( Figure S5). Among them, twenty-two are closely related to the two prominent spreaders of the disease, GLP52 and GLP905 (Figure 3 Familial cancers usually occur at a relatively young age. TC normally occurs at the median age of 9-10 years in dogs, and its occurrence increases with age. 5 Figure 2B). Moreover, the two prominent spreaders, GLP52 and GLP905, are highly inbred, with inbreeding coefficients of 0.21 and 0.24, respectively. Both parents of GLP52 are from inbred crosses between half-siblings. We also see other extreme inbreeding examples, which produced affected dogs. For instance, GLP905 was crossed with its half-sibling GLP1119 and produced two affected dogs (one confirmed and one suspected).
Cancer incidence is complex and is determined by a combination of many factors, including genetic make-up, the environment and the lifestyle of the carrier, with genetics playing a large role. In humans, TC has the strongest genetic component among all the cancers, with genetic contribution exceeding other factors. 24 In these GLPs with TC, genetic factors may contribute more than environmental factors as well, with a heritability estimated to be 0.62.
The genetic basis of familial thyroid cancer is poorly defined in humans, as only 5% of familial FCC cases have well-defined germline mutations. 13,20 Research of TC in dogs can contribute to the knowledge of corresponding TC in humans. Dogs have been proposed as an ideal model for human cancer research, because many cancers have strong similarity in histological appearances, genetic causes, biological behaviours and response to conventional therapy. 25 Additionally, dogs share their environments with human pet owners, thus are partly exposed to similar risk factors, which can be exploited for epidemiological studies of cancers common in humans and dogs. 26 The affected GLPs we reported here can serve as a model to identify the genetic basis of FCC. We have a uniquely large number of affected dogs from one breed, and they are inbred (average F 0.23) and very likely share common genetic mutations that are associated with carcinogenesis.
The large sample size gives more possibility and power to further define the underlying mutation(s) of this disease by genetic and genomic techniques, like, e.g., whole genome association analyses.