Research progress in the construction of animal models of autoimmune thyroiditis

Abstract

Autoimmune thyroiditis (AIT), also known as Hashimoto’s thyroiditis (HT), is an autoimmune disease that is characterised by elevated thyroid-specific antibody titres. The incidence of AIT is increasing year over year, making it urgent to establish a suitable animal model for this condition, in order to better explore its pathogenesis and potential pharmaceutical mechanisms for treatment. Owing to a lack of basic research on this disease, problems such as disparate modelling methods with unclear and varying success rates make it difficult for researchers to obtain effective information on AIT in the short term. This report summarises and analyzes the current literature on AIT and combines actual operability to explain the selection and specific implementation processes behind the uses of different modelling approaches, to provide a better overall understanding of autoimmune thyroid diseases.

Keywords:

• Hashimoto’s thyroiditis
• autoimmune thyroiditis
• animal model
• autoimmune disease
• thyroid

1. Introduction

Autoimmune thyroiditis (AIT) is a common chronic autoimmune disease that includes Hashimoto’s thyroiditis (HT). It is characterised by lymphocytic infiltration in the thyroid parenchyma and the presence of anti-thyroid antibodies [1]. A recent epidemiological survey of 31 Chinese provinces showed that the overall prevalence of AIT was 14.19%, with a significantly higher prevalence in women [2]. Thyroid peroxidase antibodies (TPOAb) exert cytotoxic effects on thyroid cells and can eventually lead to the development of hypothyroidism [3]. TPOAbs are also strongly associated with adverse pregnancy outcomes, such as miscarriage and preterm delivery, and can even lead to foetal neurodevelopmental delays [4]. However, there is still a lack of aetiological treatment options for AIT. In recent years, small-scale randomised controlled trials (RCTs) have demonstrated that selenium preparations [5], vitamin D3 [6], and certain herbal medicines and compounds [7] may reduce TPOAb levels. However, rigorous testing is still urgently needed to elucidate the specific mechanisms of action of these substances; hence, the construction of suitable animal models for AIT has become a primary focus for some researchers.

A robust animal model of AIT would exhibit similar serological indicators and pathological features of the disease to what is seen in human patients, and the development of the model should follow the course of the human disease. However, there are few basic studies on AIT in the literature, and the evaluation of existing models has not yet been standardised. This has led to problems, such as unclear reproducibility of various models and non-uniform model evaluation criteria, making it difficult for researchers to obtain effective information in the short term. Therefore, this paper aims to summarise and analyse the currently literature published, explain the different modelling methods and specific implementation procedures according to the selection of different experimental mouse strains, and summarise the existing model evaluation criteria to provide theoretical references for the experimental research of AIT.

2. Common experimental animal strains and modelling methods

2.1. Non-obese diabetic mice

Non-obese diabetic (NOD) mice are one of the few animal models of organ-specific autoimmune diseases. These mice can spontaneously develop autoimmune type 1 diabetes without the need for induction using antigens or immune adjuvants [8]. NOD.H-2^h4 mice represent a subtype of NOD mice that express the H-2KK and I-AK genes in an otherwise NOD-like genetic background. These mice do not develop diabetes [9], but do spontaneously develop AIT. According to the literature, there are two main approaches to modelling AIT using NOD mice.

2.1.1. High iodine water induction

Studies have shown that, when administered drinking water supplemented with 0.05% sodium iodide (NAI) for 6–8 weeks, the incidence of AIT in NOD.H-2^h4 hermaphroditic mice approaches 100%. When inflammation reaches its most severe level, it becomes chronic for the next 3–4 months [10]. Based on this property, a subtype of NOD mice, NOD.H-2^h4 mice, have mostly been used to model AIT. Wang et al. [11] divided 108 male 4-week-old NOD.H-2^h4 mice into three groups, in which the model group was given 0.005% (50 mg/L) sodium iodide (equivalent to 100× the normal iodine intake) administered via drinking water, and the control group was given sterile water, for the duration of the study. At the end of the experiment, thyroid tissues from both the groups were excised for Haematoxylin and Eosin (HE) staining, and the incidence of thyroiditis was determined using light microscopy. The incidence in the control group was found to range from 16.7%–50%, while approximately 80% of the NOD.H-2^h4 mice developed lymphocytic infiltration in the thyroid gland and showed an increasing prevalence of AIT from 8 weeks onward, further increasing with more intake of the iodine-supplemented water. Han et al. [12] investigated the role of the Sphk1/S1P/S1PR1 signalling pathway in the pathogenesis of AIT in NOD.H-2^h4 mice. They supplemented the drinking water for the mice with 0.05% (500 mg/L) sodium iodide, modelled the disease progression for 8 weeks, and began an experimental intervention treatment on week 9. Compared to the control group, the model group had a 100% morbidity rate, a higher thyroiditis score, and a higher incidence of goitres (p < 0.01). Liu et al. [13] used the same modelling method to explore the efficacy of the Buzhong Yiqi Decoction for treating AIT; however, due to the small number of experimental mice in that study, the incidence of AIT was not reported. Similarly, Di Dalmazi et al. gave NOD H-2^h4 mice water containing 0.05% (500 mg/L) sodium iodide for 12 weeks, to treat HT [14].

Another study gave mice high-iodine water as well as anti-TGFβ monoclonal antibodies (mAb) at various times while the mice developed SAT, and found that TGFβ1 may promote the migration to, or retention of, inflammatory cells in the thyroid [15]. Other cytokines, such as IFN-γ, IFN-α, IL-1, and others, may also play certain roles in the pathogenesis of thyroid autoimmunity [16–18].

Immune dysregulation and autoimmune sequelae often occur as a result of immunotherapy for cancer. Among these, HT caused by immune checkpoint inhibitors (ICIs) is the most common. This sequela provides a new perspective for developing a model of HT [19]. Ippolito et al. [20] injected two monoclonal antibodies, anti-PD-1 and CTLA-4, and used NaI (0.05%) for the induction in the NOD H-2^h4 mice of AIT. The incidence and severity of thyroiditis were then assessed using pathological scores of thyroiditis, expression levels of inflammatory factors, and thyroid function tests. The results showed that thyroiditis was more severe after PD-1 was blocked, whereas its prevalence was higher after CTLA-4 was blocked. This may be because PD-1 mainly acts on peripheral tissues and produces a stronger response, while CTLA-4 acts on the early stages of the immune response and on secondary lymphocytes [21]. Another study evaluating the relationship between thyroiditis and papillary thyroid cancer (PTC) showed that pre-existing thyroiditis was protective against PTC, whereas concomitant thyroiditis was not. The thyroiditis model in this study was induced by giving mice high-iodine water (2 g/L) for 4 weeks [22].

The modelling method using high-iodine water in NOD.H-2^h4 mice is relatively simple and highly reproducible. According to the infiltration of thyroid lymphocytes observable by HE staining, the incidence of AIT gradually increases with prolonged exposure to the high-iodine water, and the model can be stabilised at approximately 8 weeks to then be used for the exploration of AIT pathogenesis and pharmacodynamic studies of Western or traditional Chinese treatments for the condition. However, because of the specificity of the NOD.H-2^h4 mouse strain, there is currently only one source for this strain, it is expensive to purchase, and it must be bred and reared on its own under SPF-level laboratory conditions, which results in a long modelling time. Experimenters conducting AIT studies for the first time may find this approach costly in terms of both time and money, particularly in the event of early animal mortality.

2.1.2. Allogenic thyroglobulin combined with freund’s adjuvant

Thyroglobulin (Tg), a known thyroid autoantigen, can induce an immune response in mice, producing thyroid autoantibodies and varying degrees of lymphocyte infiltration into the thyroid tissue, thus inducing the development of experimental autoimmune thyroiditis (EAT) [23]. Thyroglobulin is often fused with Freund’s adjuvant to form an emulsion for injection [24]. Freund’s adjuvant is a type of immune adjuvant that enhances the immunogenicity of an antigen or enhances the host’s immune response to it, and its high adjuvant activity is difficult to compare with that of other adjuvants, making it a popular choice for scientific research [25]. NOD mice are mostly modelled using porcine Tg (pTg) mice. Cai et al. [26] established a mouse model of HT to investigate whether the condition itself could trigger neuroinflammation accompanied by mood alterations. They randomly divided 8–9-week-old female NOD mice into control and HT groups after 7 days of adaptive feeding. The mice in the HT group were injected with 25 μg of Tg that had been fully emulsified in complete Freund’s adjuvant (CFA) at the base of the tail. Fourteen days later, an equal amount of pTg fully emulsified in incomplete Freund’s adjuvant (IFA) was injected at the same site, while the control group was injected with phosphate-buffered saline (PBS).

NOD mice created using this modelling method have mostly been used to explore HT-related alterations in neurological indicators. Although it has also been confirmed that NOD mice spontaneously develop autoimmune diabetes with age [27–29] a single injection of CFA can suppress the development of diabetes in NOD mice, such that the condition does not interfere with modelling HT. It should be noted that Freund’s adjuvant emulsified with Tg needs to be injected subcutaneously as soon as possible after preparation, because it is a viscous oil-based emulsion that can be difficult to inject.

2.2. Lewis rats

Lewis rats belong to an inbred strain derived from Wistar distant herd rats. They have high levels of thyroxine, insulin, and growth hormones in their serum, and are often used in immunological studies and diabetes research. An AIT model based on Lewis rats is often achieved through the administration of high-iodine water combined with allogeneic Tg injection. Pan et al. [30] explored the mechanism of action of Jia Yan Kang Tai granules for treating AIT. In their study, 5-week-old SPF-grade female Lewis rats were adaptively fed for 1 week, then divided into normal and modelling groups. The normal group was given distilled water to drink, while the drinking water for the modelling group contained 0.064% sodium iodide for the first week. At week 3, 0.1 mg of pTg was dissolved in 100 μL of PBS buffer and emulsified with 100 μL of CFA, to prepare an emulsion with a final concentration of 0.05%. Doses of 0.2 mL of this preparation were injected subcutaneously in the posterior necks and lower abdomens of the model rats. Following an interval of 2 days, the same amount of this agent was injected once again, in order to induce immunity. Over the next 4 weeks, 0.2 mL doses of this emulsified Ptg preparation were injected every week, to boost the immune responses of the rats, while they continued to consume sodium iodide water. In the control group, the rats were injected with PBS following the same schedule, and dosing was started at week 8 after their TPOAb and TGAb levels were determined through blood sampling from the abdominal aorta. Another study using Lewis rats for modelling did not use high-iodine water, opting for pTg injections, albeit at higher doses [31]. Because iodine can cause damage to the rat thyroid gland, prior research experience is needed to choose the appropriate concentrations in different experimental contexts, in order to achieve the expected disease severity.

2.3. Sprague-Dawley rats

Sprague-Dawley (SD) rats are a distant strain of white rats that were first bred in 1925 at the SD farm in Wisconsin, USA. They have faster growth rates and larger litter sizes than Wistar rats, and are commonly used for nutrition, endocrinology, and toxicology studies. Ma et al. [32] prepared an AIT model using iodine-rich water combined with heterologous Tg injection, dividing female (weight: 220–240 g) SD rats into control, model, Yang He Tang 5 g, Yang He Tang 15 g, and sodium selenite groups. They injected 5 mg of bovine Tg (bTg) emulsified with Freund’s adjuvant every 2 weeks, for 6 weeks, into multiple points (dorsal, abdominal, subcutaneous, and hindfoot) of each rat group except the control group (who were only given Lugol’s iodine water [5%]). Zhao et al. [33] used 100 μg of mTg plus Freund’s adjuvant for immunisation once per week, for 6 weeks. Cao et al. [34,35] dissolved 100 mg of pTg in 50 mL of PBS, mixed it 1:1 with CFA, and injected 100 μg of this emulsion subcutaneously into the backs of female SD rats, at multiple sites. On the 14th, 21st, 28th, and 35th days after the first injection, booster injections of pTg emulsified in IFA were administered at the same sites.

Research on SD rat models has undergone significant changes over recent years. Currently, there are no reported modelling rates in the existing literature, and there is a lack of consistent methods. Further research is warranted to determine whether thyroid globulins from different animals have different immune effects in SD rats.

2.4. CBA/J mice

CBA/J mice are inbred mice that are susceptible to AIT, prone to induced immune responses, and are commonly used for oncological, immunological, and physiological studies. Mouse Tg (mTg) combined with Freund’s adjuvant is mostly used to immunise CBA/J mice to develop AIT. Qi et al. [36] investigated the role of the NF-κB pathway in the pathogenesis of AIT by dividing female CBA/J mice into control and model groups at 8 weeks of age. The control group was given water + CFA, while the mice in the model group were given 100 μg injections of mTg + CFA subcutaneously for initial immunisation, then 100 μg of mTg + IFA again at 10 weeks of age as a booster. Samples were collected at 14 weeks of age, and the models were evaluated using HE staining. Another study [37] used pTg for immunisation. However, according to a study by Guo et al. [38], significant follicular epithelial hyperplasia developed within the thyroid lobules of CBA/J mice who were immunised this way, which resulted in a condition that differed significantly from the human histopathological features of AIT. Therefore, they did not recommend this approach.

This modelling method is relatively simple, requires only two injections, and is less harmful to the mice. It should be noted that mTg should be purified and identified at the time of use, and that different levels of mTg purity may affect the resultant condition [38].

2.5. Wistar rats

Wistar rats, another distant strain of white rats, were first bred by the Wistar Institute in the United States, and are commonly used in many laboratories worldwide. They have a stable sexual cycle, strong fertility, fast growth and development, a docile temperament, strong resistance to infectious diseases, and a low incidence of developing spontaneous tumours. Che et al. [39] selected 6-week-old female Wistar rats, dissolved 50 mg of pTg in 0.5 mL of deionised water emulsified with an equal volume of CFA, and injected 1 mL of this mixture subcutaneously into each rat for initial immunisation, to construct another AIT model. Booster immunisations were administered on days 7, 21, and 28. The control group was given normal water and feed, while the modelling group was given high-iodine water (500 mg/L NaI). In the end, the success rate of this modelling approach was as high as 90%. To explore the regulatory effect of ginseng on Th1/Th2 imbalances, as well as the expression of related immune and inflammatory factors in rats with AIT, Chen et al. [40] conducted experiments using 6–8-week-old female Wistar rats. To induce EAT, 100 mg of pTg was emulsified in 100 mL of CFA and administered subcutaneously as a primary immunisation. The second subcutaneous immunisation was administered on day 14, using the same amount of pTg emulsified in IFA, for five consecutive weeks. Compared to the control group, the model rats had significantly higher levels of anti-TG and anti-TPO (p < 0.05), indicating successful induction of EAT.

Wistar rat models vary depending on whether iodine-rich water is administered, as well as the frequency of IFA injections [41]. The experimental process for this model is slightly more complex, and the concentrations of pTg used vary among different reports in the literature. This requires researchers to explore appropriate dose concentrations; however, the cost of this approach is relatively low. Wistar rats are easy to obtain, and have no special requirements for feeding.

2.6. Other animal models

Female C57BL/6 mice (6–8 weeks old) were have also been used to model AIT [42]. The model group was given water supplemented with 0.05% NaI, and pTg (200 μg/mouse) in CFA was injected subcutaneously at multiple points on the back, abdomen, and neck during the first week, followed by pTg (200 μg/mouse) in IFA 2 weeks later, and the mice were harvested 4 weeks after the second immunisation. The same modelling approach was also used in female BALB/c [43] as well as C57BL/6 mice that were interbred with BALB/c mice, which are highly pure, genetically stable, and can be used to replicate the results of existing studies. Female Kunming mice have also been used for the construction of AIT models, but the methods used for these mice vary significantly [44,45], and the evidence reported in the literature for this approach is limited; therefore, we will not discuss them in detail in this review.

In addition to the common rat and mouse models, zebrafish models can also be used to study thyroid-related diseases. Zebrafish embryos develop rapidly, and researchers can use this model to study thyroid development, the genetic basis of the disease, or to conduct large-scale drug screening experiments to discover potential treatments [46]. However, because zebrafish do not have true thyroid tissues and their thyroid biology is different from that of mammals, this approach suffers from significant limitations. Therefore, research goals must be carefully considered and selected based on the characteristics of the zebrafish. Because their thyroid structures and functions are similar to those of humans, Cynomolgus monkeys can also be used to study the pathological process of thyroid diseases [47] or to evaluate the effectiveness of new drugs and toxicological studies [48]. However, these monkeys are costly to raise, many laboratories do not have the appropriate facilities, and the methods for this approach are still in early stages; therefore, it is not common in HT research. The spontaneous development of HT has only been reported in a 4-year-old female cynomolgus monkey in one toxicological study [49]. Schumm-Draeger et al. constructed a feline model by supplementing cats with a genetic predisposition to AIT with high-iodine water for 80 days, then observed their thyroid pathologies and immune indicators. They believed that this experimental model seemed suitable for further experiments, but have not published any follow-up reports [50]. Since then, other scholars have observed that chickens of the obese strain (OS) are hereditarily affected by spontaneous AIT that resembles human HT in terms of clinical, histopathological, serological, and endocrinological aspects. These scholars believe that the OS of chickens has the potential to be used as an animal model for autoimmune diseases in the future [51].

Compared to other species, mice have multiple advantages as animal models for AIT: 1) AIT is an autoimmune disease, and the physiological structure of the murine thyroid gland has many similarities to the human one, and can produce similar immune responses, making the experimental mouse model more representative. 2) Experimental mice are the most commonly used animal model in laboratory research, and their genomic backgrounds are clear. This allows researchers to easily control the genetic backgrounds of their animals and ensure genetic consistency between experimental groups, thereby improving the reproducibility and comparability of experimental results. 3) The technology is relatively mature, easy to use, and highly controllable. 4) Experimental mice are inexpensive and easily obtained, making large-scale research possible.

3. Evaluation indicators of successful modelling

In order to be used to conduct robust pathogenic or pharmacodynamic studies, a successful animal model should resemble or be similar to the human disease, or even completely mimic its pathological changes. Therefore, it is particularly important to establish reasonable and standardised indicators of modelling success before starting a study. The success indicators for AIT are essentially based on the diagnosis of the human disease, but the evaluation criteria chosen by various research teams are not consistent, owing to the different modelling approaches used. This has resulted in much heterogeneity in experimental results. Even for the same modelling method, differences exist among different research teams, and it is difficult for researchers to choose a uniform standard based on a large body of published research, that can be used as a quick reference. In response to this challenge, we have summarised and extracted some indicators of successful modelling that could be used as indicators, based on what has been published so far in this field.

3.1. Pathomorphological evaluation

Most studies have opted for pathomorphological evaluation, wherein the thyroid glands of randomly selected mice or rats were harvested for HE staining, to observe thyroid tissue lesions by light microscopy. Gao et al. mostly referred to the international modelling standard, which is the ratio of the area of lymphocytic infiltration into the thyroid gland to the whole gland area, at a > 2% cut-off [52]. The severity of thyroiditis can also be determined by grading the thyroid tissue according to the area of lymphocytic infiltration [12]. Grade 0 is normal, grade 1+ indicates at least 125 lymphocytes infiltrating one or more glands, grade 2+ is present when there is up to 25% infiltration; grade 3+ is assigned for 25–50% infiltration, and grade 4+ indicates > 50% infiltration. This grading method can be applied to several rat and/or mouse strains.

3.2. Serological evaluation

Serological indicators are mostly measured by enzyme-linked immunosorbent assay (ELISA) for serum levels of both TPOAb and TgAb, and the success of the modelling is evaluated using levels of both antibodies, or of TPOAb alone. For example, Dong et al. [53] considered Lewis rats with serum TPOAb titres of ≥ 60 IU/mL as successful models, and Zhao et al. [54] defined serum TPOAb and TgAb levels > 10× those of normal CBA/J mice as successful. However, these methods have only been reported by individual research teams, so more experiments are needed to support their reliabilities.

3.3. Serological indicators combined with pathomorphological evaluation

The use of serological indicators combined with pathological morphology to assess modelling success is similar with what is used to diagnose the human disease. Che et al. [39] concluded that their modelling was successful when their model group of Wistar rats showed increased FT3, FT4, TPOAb, TgAb, and decreased TSH levels at 10 weeks compared to the control group (p < 0.05), while also exhibiting moderate to severe damage to thyroid follicular structures and lymphocytic infiltration under light microscopy. This method is applicable to the initial stages of the disease in animal models, but is limited. If the disease continues to progress and FT3 and FT4 levels decrease, a clinical diagnosis of AIT can still be made, and the lack of a threshold value representing the elevation of serum indicators in animals makes the evaluation more ambiguous.

4. Discussion

The first animal model of AIT induced by neonatal mouse thymectomy was developed in 1976, but it had a robustness and poor stability [55]. Since then, researchers have discovered spontaneous AIT models, such as those observed in cats [50], chickens of the OS [56], beagle dogs [57], and NOD mice. However, limited by requirements of feeding conditions and costs, as well as a lack of relevant detection kits, these models are rarely used today. In recent years, the most commonly used spontaneous models have been NOD.H-2^h4 mice. Different concentrations of high-iodine water can be selected for tailoring this model, according to different experimental design requirements. The principles of iodine overdose-induced AIT include the following: (1) Triggering an autoimmune reaction, inducing abnormal expression of the MHC-II antigen, which enhances the aggressiveness of immune cells that then begin to attack the thyroid tissue [58]. (2) Iodine can combine with TG to generate iodinated TG, which increases the immunogenicity of TG and facilitates the presentation of antigen peptides, thus increasing the number of pathogenic T cells [59]. (3) Iodine is a key component in the synthesis of thyroid hormones; thus, excess iodine can lead to thyroid dysfunction that then stimulates the immune system to produce autoantibodies. (4) Excess iodine has a direct damaging effect on thyroid cells [60]. The other type of AIT model comprises experimentally-induced models. Researchers are continuously exploring various experimental methods for this purpose. For example, In 2010, Fang et al. induced susceptible donor mice with mTg combined with Freund’s adjuvant or lipopolysaccharides, then harvested their activated splenocytes. After culturing these in a complete medium containing mTg, they were then transferred into syngeneic recipient mice to induce EAT, which proved to be valuable for studying granulomatous thyroiditis [61]. Costagliola et al. [62] used a cDNA injection method and a combined electroporation method to inoculate specific muscle groups of C3H/Hen mice to produce long-term humoral and cellular immunity, thereby maintaining and promoting the immune response of thyroid autoantigens. However, this method is still in preliminary stages, is complex, and its success rate is low. Currently, a more mature in vitro induction method is based on exogenous Tg combined with freund’s adjuvant, which is suitable for a variety of experimental mouse strains. This is also an ideal modelling approach for EAT. It is widely used, and the experimental methods have been refined and validated.

5. Conclusion

After years of research and exploration, the animal model for AIT is gradually maturing. High-iodine water induces immune-susceptible mouse strains to develop the condition. Iodine, an environmental factor, is closely associated with the occurrence and development of AIT. The iodine induction method is easy to implement, the model is stable, and can be used to study both the pathogenesis and pharmacodynamics of AIT. Exogenous Tg combined with Freund’s adjuvant is a relatively mature method for modelling EAT. SD rats, Wistar rats, Lewis rats, or CBA/J mice can be used for this approach, which are all easy to obtain and have low experimental costs. Moreover, relevant experimental techniques have now been proven that can be used to explore the impact of possible pathogenic factors, such as the environment and immunity, on AIT-related genotypes. This review compiled various modelling methods for AIT, from worldwide sources. The extracted literature is summarised in Table 1, with the aim of providing comprehensive guidelines for researchers in this field.

Table 1. Classification of modelling methods for AIT animal models.

Experimental animals	Modelling method	Modelling time	Model Evaluation Criteria	Modelling success rate
NOD mice	0.005% (50 mg/L) sodium iodide water induction [12]	8 weeks	The ratio of the area of lymphocytic infiltration in the thyroid to the area of the entire gland is >2%	80%
	0.05% (500 mg/L) sodium iodide water induction [13]			100%
	25μg of porcine thyroglobulin combined with Freund’s adjuvant [26]	7 weeks	Morphological examination (significantly enlarged thyroid, disordered and damaged thyroid follicles, mononuclear cell infiltration) + thyroiditis score + serological indicators (increased TPO-Ab, Tg-Ab)	Not mentioned
Lewis rat	0.064% sodium iodide water feeding combined with allogeneic thyroglobulin induction [30]	7-8 weeks	TPOAb serum titer > 60IU/mL	90%
SD rat	Bovine thyroglobulin combined with Freund’s adjuvant induction [32]	6 weeks	Thyroid morphological examination (see lymphocyte infiltration) + serological indicators (significantly increased TPO-Ab and Tg-Ab)	Not mentioned
	Mouse thyroglobulin combined with Freund’s adjuvant induction [33]
	Porcine thyroglobulin combined with Freund’s adjuvant induction [34,35]	5 weeks
CBA/J mice	Mouse thyroglobulin combined with Freund’s adjuvant induction [36]	6 weeks	Serum levels of TPOAb and TgAb were more than 10 times higher than normal CBA/J mice	80%
	Porcine thyroglobulin combined with Freund’s adjuvant induction [37]	Not mentioned		Not mentioned
Wistar rat	High-iodine water (500 mg/L NaI) feeding combined with Porcine thyroglobulin injection [39]	4-6 weeks	Serological indicators (FT3, FT4, TPOAb, TgAb, TSH) + pathomorphological evaluation	90%
	Porcine thyroglobulin combined with Freund’s adjuvant induction [40]	7 weeks		Not mentioned

This review does, however, have some key limitations worth noting: 1) Although most of the studies discussed chose female rats or mice, a small number used male animals or equal male/female ratios. Some studies suggest that sex hormones affect immune function [63], but we did not generalise and compare models involving different sexes in this review. 2) Owing to variations in experimental methodologies between different laboratories, identical modelling protocols may yield different success rates. There is no complete standardisation for this type of modelling as of yet, and no systematic experimental research is currently available that summarises the success rates of different modelling approaches. To provide a reference for researchers, we selected only the reported success rates from various individual reports in the literature. Future comprehensive and dynamic comparisons of the construction processes for various AIT models are needed to provide more representative experimental data.

Although experimental mouse models have many advantages for studying AIT, they are still just models and cannot replicate the complex physiological and immune processes that occur in humans with AIT. Therefore, laboratory research results should be closely combined with clinical research to evaluate the applicability of AIT models, thereby forming a model construction method that is universal and highly recognised by industry experts to provide a convent basis for researchers in this field.

Disclosure statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability statement

Data sharing is not applicable to this article as no new data were created or analysed in this study.

Additional information

Funding

This work is supported by Scientific and technological innovation project of China Academy of Chinese Medical Sciences [CI2021A01611] and the Clinical Research Centre Construction Project of Guang’anmen Hospital, CACMS [2022LYJSZX23].