The present protocol establishes and characterizes a patient-derived xenograft (PDX) model of anaplastic thyroid carcinoma (ATC) and head and neck squamous cell carcinoma (HNSCC), as PDX models are rapidly becoming the standard in the field of translational oncology.
Patient-derived xenograft (PDX) models faithfully preserve the histological and genetic characteristics of the primary tumor and maintain its heterogeneity. Pharmacodynamic results based on PDX models are highly correlated with clinical practice. Anaplastic thyroid carcinoma (ATC) is the most malignant subtype of thyroid cancer, with strong invasiveness, poor prognosis, and limited treatment. Although the incidence rate of ATC accounts for only 2%-5% of thyroid cancer, its mortality rate is as high as 15%-50%. Head and neck squamous cell carcinoma (HNSCC) is one of the most common head and neck malignancies, with over 600,000 new cases worldwide each year. Herein, detailed protocols are presented to establish PDX models of ATC and HNSCC. In this work, the key factors influencing the success rate of model construction were analyzed, and the histopathological features were compared between the PDX model and the primary tumor. Furthermore, the clinical relevance of the model was validated by evaluating the in vivo therapeutic efficacy of representative clinically used drugs in the successfully constructed PDX models.
The PDX model is an animal model in which human tumor tissue is transplanted into immunodeficient mice and grows in the environment provided by the mice1. Traditional tumor cell line models suffer from several disadvantages, such as the lack of heterogeneity, the inability to retain the tumor microenvironment, the vulnerability to genetic variations during repeated in vitro passages, and the poor clinical application2,3. The main drawbacks of genetically engineered animal models are the potential loss of the genomic features of human tumors, the introduction of new unknown mutations, and the difficulty in identifying the degree of homology between mouse tumors and human tumors4. In addition, the preparation of genetically engineered animal models is expensive, time-consuming, and relatively inefficient4.
The PDX model has many advantages over other tumor models in terms of reflecting tumor heterogeneity. From the perspective of histopathology, although the mouse counterpart replaces the human stroma over time, the PDX model preserves the morphological structure of the primary tumor well. In addition, the PDX model conserves the metabolomic identity of the primary tumor for at least four generations and better reflects the complex inter-relationships between tumor cells and their microenvironment, making it unique in simulating the growth, metastasis, angiogenesis, and immunosuppression of human tumor tissue5,6,7. At the cellular and molecular levels, the PDX model accurately reflects the inter- and intra-tumor heterogeneity of human tumors, as well as the phenotypic and molecular characteristics of original cancer, including gene expression patterns, mutation status, copy number, and DNA methylation and proteomics8,9. PDX models with different passages have the same sensitivity to drug therapy, indicating that the gene expression of PDX models is highly stable10,11. Studies have shown an excellent correlation between the response of the PDX model to a drug and the clinical responses of patients to that drug12,13. Therefore, the PDX model has emerged as a powerful preclinical and translational research model, particularly for drug screening and clinical prognosis prediction.
Thyroid cancer is a common malignant tumor of the endocrine system and is a human malignancy that has shown a rapid increase in incidence in recent years14. Anaplastic thyroid carcinoma (ATC) is the most malignant thyroid cancer, with a median patient survival of only 4.8 months15. Although only a minority of thyroid cancer patients are diagnosed with ATC each year in China, the mortality rate is close to 100%16,17,18. ATC usually grows rapidly and invades the adjacent tissues of the neck as well as the cervical lymph nodes, and about half of the patients have distant metastases19,20. Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer in the world and one of the leading causes of cancer deaths, with an estimated 600,000 people suffering from HNSCC each year21,22,23. HNSCC includes a large number of tumors, including those in the nose, sinuses, mouth, tonsils, pharynx, and larynx24. ATC and HNSCC are two of the main head and neck malignancies. In order to facilitate the development of novel therapeutic agents and personalized treatments, it is necessary to develop robust and advanced preclinical animal models such as PDX models of ATC and HNSCC.
This article introduces detailed methods for establishing the subcutaneous PDX model of ATC and HNSCC, analyzes the key factors affecting the tumor take rate in model construction, and compares the histopathological characteristics between the PDX model and the primary tumor. Meanwhile, in this work, in vivo pharmacodynamic tests were performed using the successfully constructed PDX models in order to validate their clinical relevance.
All the animal experiments were performed in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care guidelines and protocols approved by the Institutional Animal Care and Use Committee of West China Hospital, Sichuan University. NOD-SCID immunodeficient mice aged 4-6 weeks old (of both sexes) and female Balb/c nude mice aged 4-6 weeks old were used for the present study. The animals were obtained from a commercial source (see Table of Materials). The ethics committee of West China Hospital authorized the study with human subjects (protocol number 2020353). Each patient provided written informed consent.
1. Experimental preparation
2. Acquisition and transport of fresh tumor tissue
3. Tumor transplantation
4. Tumor tissue preservation, fixation, and protein freezing
NOTE: The remaining tumor tissues were used for seed preservation, fixation, and DNA/RNA/protein freezing, respectively.
5. Passaging, cryopreservation, and resuscitation of PDX model tumors
6. Determining the therapeutic efficacy of lenvatinib and cisplatin in the ATC PDX model
NOTE: The ATC PDX model was used to test the therapeutic effect of the tyrosine kinase inhibitor lenvatinib and the chemotherapeutic drug cisplatin25,26,27.
A total of 18 thyroid cancer specimens were transplanted, and five PDX models of thyroid cancer were successfully constructed (27.8% tumor take rate), including four cases of undifferentiated thyroid cancers and one case of anaplastic thyroid cancer. The correlation between the success rate of model construction and the age, gender, tumor diameter, tumor grade, and differentiation were analyzed. Although the model success rate of grade 4 tumor samples was higher than for samples with lower grades, and the success rate of undifferentiated tumor samples was also higher than that of highly differentiated samples, the correlation analysis results showed that these factors were not associated with the success rate of the PDX model (Table 1). Seventeen HNSCC samples were inoculated, and four PDX models of HNSCCC were successfully constructed. The correlation analysis between the tumor take rate in the model construction and the clinical parameters of the tumor samples demonstrated that the degree of differentiation was associated with model success rate, while age, gender, smoking history, tumor diameter, cancer grade, metastasis, and HPV infection did not affect the tumor take rate (Table 2).
The tumor growth curves for each PDX model were plotted to better understand the growth rates of the PDX models from different patients (Figure 1, Figure 2, and Table 3). The average tumorigenic cycles (time from inoculation to a tumor size of 1,000 mm3) of THY-004 from generations P0 to P5 were 68 days, 87 days, 29 days, 34 days, 28 days, and 26 days, respectively. The average tumorigenic cycles of THY-012 from generations P0 to P5 were 119 days, 61 days, 66 days, 55 days, 87 days, and 116 days, respectively. The average tumorigenic cycles of THY-017 from generations P0 to P5 were 27 days, 17 days, 30 days, 13 days, 22 days, and 15 days, respectively. The average tumorigenic cycles of THY-018 from generations P0 to P3 were 134 days, 70 days, 48 days, and 48 days, respectively. The average tumorigenic cycles of THY-021 from generations P0 to P3 were 53 days, 66 days, 35 days, and 49 days, respectively. The average tumorigenic cycles of OTO-017 from generations P0 to P4 were 118 days, 86 days, 67 days, 129 days, and 88 days, respectively. The average tumorigenic cycles of OTO-022 from generations P0 to P5 were 155 days, 55 days, 32 days, 37 days, 27 days, and 46 days, respectively. The average tumorigenic cycles of OTO-030 from generations P0 to P2 were 133 days, 93 days, and 104 days, respectively. The average tumorigenic cycles of OTO-031 from generations P0 to P5 were 144 days, 58 days, 33 days, 34 days, 52 days, and 50 days, respectively. The ATC samples were stably passed to the P3 generation and later, while two cases of HNSCC samples failed to form tumors after passing to the P1 generation. The growth rates of some samples were relatively slow in the P0 generation, but their growth rates were accelerated after passing to the P1 and later generations. The histopathological characteristics of the patient tumors with those of different generations of PDX models were compared. The results showed that PDX tumors and patient-derived primary tumors were morphologically almost similar (Figure 3), with slight differences that may be due to the heterogeneity in the sampling area between patients and different generations of PDX.
The anti-tumor efficacy of lenvatinib (a multi-target tyrosine kinase inhibitor approved for the treatment of advanced thyroid cancer28) was evaluated in the PDX model of ATC. As shown in Figure 4A, lenvatinib treatment significantly inhibited the tumor growth in the ATC PDX model compared to the normal saline control group (P < 0.05). At the end of the experiment, the tumor tissue was excised and weighed to determine the tumor weight. Compared with the control group, the tumor weight of the lenvatinib treatment group was lower, although a statistical difference was not achieved (Figure 4B). In addition, no obvious changes in general status and body weight were observed in the mice treated with lenvatinib (Figure 4C). Due to the excessive frequency of cisplatin administration during the experiments, the mice showed significant toxicity, manifested by weight loss and even death. The anti-tumor efficacy of cisplatin is shown in Supplementary Figure 1.
Figure 1: Tumor growth curve of ATC PDX models from different patients. Each color represents the specified generation, and each curve represents a single tumor. One to three mice were inoculated at the passage 0 (P0) generation, and five mice were inoculated in the subsequent passages (P1-P5). Please click here to view a larger version of this figure.
Figure 2: Tumor growth curve of the HNSCC PDX models from different patients. Each color represents the specified generation, and each curve represents a single tumor. One to three mice were inoculated at the P0 generation, and five mice were inoculated at P1 and later generations. Please click here to view a larger version of this figure.
Figure 3: Histopathological study. Histopathological comparison between the patient primary tumors and corresponding PDXs (passage 1 and passage 3) of ATC (THY-012, THY-017) and HNSCC (OTO-017) (hematoxylin-eosin staining, 100x). The pathological subtypes of THY-012 and THY-017 were anaplastic thyroid carcinoma, and the pathological subtype of OTO-017 was squamous cell carcinoma. Scale bars = 100 µm. Please click here to view a larger version of this figure.
Figure 4: The therapeutic efficacy of lenvatinib in the ATC PDX model. Changes in the (A) tumor volume, (B) tumor weight, and (C) body weight of ATC PDX-bearing mice after treatment with lenvatinib (10 mg/kg). Statistical analyses were performed using T-tests to compare levatinib with control. *P < 0.05 versus control was considered to be statistically significant. Please click here to view a larger version of this figure.
Parameters | Class | Tumor take rate (%) | P |
Age (years) | <60 | 16.67 (1/6) | 0.615 |
≥60 | 33.33 (4/12) | ||
Gender | Male | 16.67 (1/6) | 0.615 |
Female | 33.33 (4/12) | ||
Tumor diameter | <6cm | 37.50 (3/8) | 0.608 |
≥6cm | 20.00 (2/10) | ||
Pathologic TNM stage | I | 0.00 (0/1) | 1 |
Ш | 0.00 (0/1) | ||
31.25 (5/16) | |||
Differentiation | High | 0.00 (0/7) | 0.059 |
Poor | 25.00 (1/4) | ||
Undifferentiated | 57.14 (4/7) |
Table 1: Correlation between the ATC tumor take rate and the clinical characteristics of the patients.
Parameters | Class | Tumor take rate (%) | P |
HPV | Negative | 33.33 (2/6) | 1 |
Unknown or positive | 36.36 (4/11) | ||
Age (years) | <60 | 33.33 (3/9) | 1 |
≥60 | 37.50 (3/8) | ||
Gender | Male | 50.00 (5/10) | 0.304 |
Female | 14.29 (1/7) | ||
Smoking status | Ever | 44.44 (4/9) | 0.62 |
Never | 25.00 (2/8) | ||
Tumor diameter | <3cm | 40.00 (4/10) | 1 |
≥3cm | 28.57 (2/7) | ||
Pathologic TNM stage | I | 75.00 (3/4) | 0.423 |
25.00 (2/8) | |||
Ш | 0.00 (0/1) | ||
33.33 (1/3) | |||
Distant metastasis | Y | 28.57 (2/7) | 0.633 |
N | 44.44 (4/9) | ||
Differentiation | High | 12.50 (1/8) | 0.036* |
Moderate to high | 100.00 (2/2) | ||
Moderate | 0.00 (0/2) | ||
Moderate to poor | 66.67 (2/3) | ||
* P < 0.05 |
Table 2: Correlation between the HNSCC tumor take rate and the clinical characteristics of the patients. *P < 0.05.
Sample name | Generation P to P0 | Generation P0 to P1 | Generation P1 to P2 | Generation P2 to P3 | Generation P3 to P4 | Generation P4 to P5 |
THY-004 | 68 | 87 | 29 | 34 | 28 | 26 |
THY-012 | 119 | 61 | 66 | 55 | 87 | 116 |
THY-017 | 27 | 17 | 30 | 13 | 22 | 15 |
THY-018 | 134 | 70 | 48 | 48 | – | – |
THY-021 | 53 | 66 | 35 | 49 | – | – |
OTO-017 | 118 | 86 | 67 | 129 | – | – |
OTO-022 | 155 | 55 | 32 | 37 | 27 | 46 |
OTO-030 | 133 | 93 | 104 | – | – | – |
OTO-031 | 144 | 58 | 33 | 34 | 52 | 50 |
Table 3: The average tumorigenic cycle (time from inoculation to a tumor size of 1,000 mm3) of the ATC and HNSCC models.
Supplementary Figure 1: The therapeutic efficacy of cisplatin in the ATC PDX model. Changes in the (A) tumor volume, (B) tumor weight, and (C) body weight of ATC PDX-bearing mice after treatment with cisplatin (3 mg/kg). Statistical analyses were performed using T-tests to compare cisplatin with the control. *P < 0.05 versus control was considered to be statistically significant. Please click here to download this File.
This study has successfully established the subcutaneous PDX models of ATC and HNSCC. There are many aspects to pay attention to during the process of PDX model construction. When the tumor tissue is separated from the patient, it should be put into the ice box and sent to the laboratory for inoculation as soon as possible. After the tumor arrives at the laboratory, the operator must pay attention to maintaining a sterile field and practice aseptic procedures. For needle biopsy samples, because the tumor tissue is particularly small, inoculation after mixing the sample with the matrix gel would be more conducive to establishing the model. The primary tumor tissue should also be preserved, fixed, and frozen as much as possible for future research. During inoculation, the air in the trocar needs to be expelled as much as possible after the tumor pieces have been put into the trocar before use. After tumor inoculation, the tumor growth should be observed in mice for 1-4 months, and mice without tumor growth for more than 6 months can be euthanized29.
Immunodeficient mice are generally chosen as the host for PDX model construction29,30. From the P0 generation to the P2 generation, non-obese diabetic-severely compromised immune deficient (NOD-SCID) mice or NOD Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice are generally used. In the P3 generation and beyond, the samples are considered to be stably passed, so nude mice can usually also serve as the host, and tumors can also grow normally. Additionally, the total operation time, tumor isolation time, disease-free survival, and overall survival rate of the patients, the tumor malignant degree, and the histologic subtype were all associated with PDX model tumorigenicity31,32,33,34. The transplantation site also has an impact on the success rate of PDX modeling, and studies have shown that renal capsule and orthotropic transplantation have a high tumorigenic rate33,35. In addition, the use of Matrigel may also improve the tumorigenic rate36,37. It has been reported that human papillomavirus (HPV) infection affects the success rate of transplantation in HNSCC tumors; HPV-negative tumors have a superior take rate compared to HPV-positive tumors38,39. This study did not reach the same conclusion, probably due to the small sample numbers and incomplete information on HPV infection.
Different from the orthotopic and renal capsule transplantation models, the subcutaneous model is more convenient for observing the growth of tumors and is also conducive to operation40,41,42. Based on the tumor growth data of the ATC and HNSCC PDX model, we found that the growth rates of tumors from different patients were inconsistent, reflecting inter-tumor heterogeneity. The tumor growth rate of the P0 generation from most PDX models was relatively slower than for the latter passages, which was likely due to the adaption of the mouse microenvironment. Notably, the growth rate from some patient-derived tumors increased in different passages after the P1 generation, consistent with the shortened passage interval reported by Pearson et al.43. Histopathological examination demonstrated that the PDX tumors retained the morphological characteristics of the primary tumors. The correlation between the PDX model and the clinical ATC patients was also reflected in the results of the in vivo pharmacodynamic tests, which demonstrated that Lenvartinib exhibited a good anti-tumor effect, consistent with clinical reports25,26,27.
However, the PDX model also has certain disadvantages. For example, the tumor formation time is relatively long, which is unsuitable for patients with advanced or aggressive tumors. In addition, the time and monetary costs of high-throughput drug screening are too high44. Indeed, combining the PDX model with tumor organoids and establishing a patient-derived organoid (PDO) model corresponding to the PDX model would compensate for this deficiency44,45,46. Orthotopic transplantation models can be used to study the pathogenesis and metastatic mechanisms of tumors40,41,47. The lack of a functional immune system is another disadvantage of the PDX model, so increasing numbers of experiments are using humanized mice to construct the PDX model for tumor immunology research48,49,50.
The authors have nothing to disclose.
This work was supported by the Sichuan Province Science and Technology Support Program (Grant Nos. 2019JDRC0019 and 2021ZYD0097), the 1.3.5 project for disciplines of excellence, West China Hospital, Sichuan University (Grant No. ZYJC18026), the 1.3.5 project for disciplines of excellence-Clinical Research Incubation Project, West China Hospital, Sichuan University (Grant No. 2020HXFH023), the Fundamental Research Funds for the Central Universities (SCU2022D025), the International Cooperation Project of Chengdu Science and Technology Bureau (Grant No. 2022-GH02-00023-HZ), the Innovation Spark Project of Sichuan University (Grant No. 2019SCUH0015), and the Talent Training Fund for Medical-engineering Integration of West China Hospital – University of Electronic Science and Technology (Grant No. HXDZ22012).
2.4 mm x 2.0 mm trocar | Shenzhen Huayang Biotechnology Co., Ltd | 18-9065 | |
Balb/c nude mice | Beijing Vital River Laboratory Animal Technology Co., Ltd. | 401 | |
Biosafety cabinet | Suzhou Antai | BSC-1300IIA2 | |
Blade | Shenzhen Huayang Biotechnology Co., Ltd | 18-0823 | |
Centrifuge tube | Corning | 430791/430829 | |
Cryopreservation tube | Chengdu Dianrui Experimental Instrument Co., Ltd | / | |
Custodiol HTK-Solution | Custodiol | 2103417 | |
Dimethyl sulfoxide(DMSO) | SIGMA-ALORICH | D5879-500mL | |
Electronic balance | METTLER | ME104 | |
Electronic digital caliper | Chengdu Chengliang Tool Group Co., Ltd | 0-220 | |
fetal bovine serum(FBS) | VivaCell | C04001-500 | |
IBM SPSS Statistics 26 | IBM | ||
Ketamine | Jiangsu Zhongmu Beikang Pharmaceutical Co., Ltd | 100761663 | |
Lenvatinib | ApexBio | A2174 | |
NOD SCID immunodeficient mice | Beijing Vital River Laboratory Animal Technology Co., Ltd. | 406 | |
Pen-Strep Solution | Biological Industries | 03-03101BCS | |
Petri dish | WHB | WHB-60/WHB-100 | |
Saline | Sichuan Kelun | W220051705 | |
Scissor | Shenzhen Huayang Biotechnology Co., Ltd | 18-0110 | |
Tweezer | Shenzhen Huayang Biotechnology Co., Ltd | 18-1241 | |
Vet ointment | Pfizer Inc. | P10015353 | |
Xylazine | Dunhua Shengda Animal Medicine Co., Ltd | 070031777 |