Establishment of a patient‐derived drug‐resistant oral squamous cell carcinoma animal model

Abstract

Oral squamous cell carcinoma (OSCC) constitutes 90% of oral tumors. Advanced cases severely impair patients' life quality of life due to anatomical location and limited therapies. Conventional treatments often induce drug resistance or recurrence. Patient‐derived xenograft (PDX) models are widely used to simulate tumor progression and drug responses, serving as translational tools for precision medicine. This study aimed to establish drug‐resistant OSCC PDX models. Human OSCC tissues were transplanted into immunodeficient mice and passaged (P1–P2). At P2 (tumor volume: 40–80 mm³), mice received cisplatin (1 mg/kg, three times/week) with cetuximab (1 mg/kg, weekly), GSK690693 (10 mg/kg, five times/week), or rapamycin (4 mg/kg, five times/week). PDX tissues from groups with less‐therapeutic response (manifested as larger tumor volumes) were serially passaged to assess treatment efficacy. Tumor tissues with diminished drug sensitivity underwent histopathological analysis and identified stability of their tumor characteristics using hematoxylin–eosin (HE) and immunohistochemical staining after one additional passage and retreatment. Results demonstrated that successive passaging accelerates tumor growth. First‐generation treatments showed universal sensitivity. At P2, cisplatin–cetuximab and rapamycin groups remained sensitive, whereas GSK690693 efficacy declined. Continued passaging of GSK690693‐treated tumors confirmed resistance, as evidenced by exhibiting enhanced malignant characteristics at histological level. The GSK690693‐resistant model was established first, whereas resistant models of other treatment groups were established according to similar protocols. These findings suggest that sequential passaging and drug exposure in PDX models recapitulated clinical tumor evolution, enabling the development of drug‐resistant OSCC models. This study can offer methodological insights for precision therapy of OSCC.

Patient‐derived xenograft (PDX) tissues from less‐responsive groups (larger tumors) were serially passaged. Sequential passaging and drug exposure in PDX models recapitulated clinical tumor evolution, enabling the development of drug‐resistant oral squamous cell carcinoma (OSCC) models. This study can offer methodological insights for precision therapy of OSCC.

1. INTRODUCTION

Oral squamous cell carcinoma (OSCC), the most common malignant tumor in the head and neck region, accounts for over 90% of all oral malignancies. ¹ Commonly used therapeutic drugs for OSCC include chemotherapeutic agents such as cisplatin and 5‐fluorouracil (often used in combination regimens), targeted therapies such as cetuximab (an epidermal growth factor receptor [EGFR] inhibitor), and immunotherapies like pembrolizumab or nivolumab (PD‐1 inhibitors), especially for advanced or recurrent cases. ² In recent years, despite significant advancements have been made in surgery, radiotherapy, chemotherapy, targeted therapy, and even traditional medicine, prolonged therapeutic interventions inevitably induce drug resistance. ³ , ⁴ , ⁵ Approximately 60% of locally advanced patients experience relapse after treatment, with over 70% resistance to platinum‐based chemotherapeutic agents (primarily cisplatin), ⁶ , ⁷ ultimately resulting in therapeutic failure. Due to the unique anatomical location of oral tumors, surgical resection frequently causes facial deformity and functional impairment, adversely affecting aesthetics, quality of life, and psychosocial well‐being. GSK690693 is a pan‐AKT inhibitor designed to target all three AKT subtypes (AKT1/2/3) simultaneously by competitively binding to the ATP‐binding site of the AKT kinase domain, thereby blocking their phosphorylation and downstream signaling. ⁸ Its core mechanism of action involves inhibition of the PI3K/AKT/mTOR pathway, which disrupts cell proliferation, survival, and metabolism‐related signals while inducing tumor cell apoptosis. Preclinical studies ⁹ , ¹⁰ have shown that GSK690693 exhibits significant efficacy in breast cancer and castration‐resistant prostate cancer models, suppressing tumor growth, reducing metastatic lesion formation, and synergizing with chemotherapy or radiotherapy. Although GSK690693 has demonstrated antitumor activity in other cancers, its long‐term effects and resistance mechanisms in OSCC remain undefined. Although retrospective clinical studies have identified some drug resistance‐associated genes, ¹¹ they remain insufficient for exploring the dynamic evolution process of tumors, leaving the acquired resistance mechanisms poorly understood. Therefore, establishing clinically relevant tumor models to elucidate tumor progression and systematically investigate drug resistance mechanisms is a critical approach for developing precision therapies to overcome drug resistance.

Tumor heterogeneity, gene mutations, abnormal signaling pathways activation, and tumor microenvironment alterations play important roles in tumor progression and drug resistance. ¹² Due to the continuous nature of spatiotemporal evolution and dynamic progression in tumors, cells and genetic variations exhibit high heterogeneity and rapid evolution both within and between tumor foci, posing challenges for achieving precision diagnosis. ¹² , ¹³ , ¹⁴ A study analyzing phenotypic transformations in 61 non–small cell lung cancer (NSCLC) patients after developing tyrosine kinase inhibitor (TKI) resistance reported cases of small cell lung cancer, squamous cell carcinoma, large cell neuroendocrine carcinoma, and sarcoma. ¹⁵ However, there is a lack of animal models capable of thoroughly exploring tumor heterogeneity and its dynamic evolutionary process, which hinders simulation studies aimed at understanding these patterns. Therefore, selecting clinically relevant tumor evolution models to elucidate tumor progression and systematically investigate resistance mechanisms is critical for developing precision therapies to overcome drug resistance. Currently, animal models for OSCC drug resistance research include genetically engineered mouse models (GEMMs), syngeneic tumor models, cell‐derived xenograft (CDX) models, and patient‐derived xenograft (PDX) models. ¹⁶ PDX model, constructed by directly transplanting surgical or biopsy tumor specimens from patients into immunodeficient mice, effectively retains the histological architecture, molecular features, and genetic characteristics of the original tumors, with genomic stability exceeding 80%. ¹⁷ As dynamic biological systems, PDX models enable monitoring of cancer evolution through successive passaging in mice while also capturing interpatient heterogeneity. This allows the identification of molecularly defined tumor subtypes, associated biomarkers, and therapeutic targets, making PDX a pivotal tool in translational research. As summarized in Figure 2A, residual tumor cells in PDX models under long‐term drug exposure undergo adaptive phenotypic switching to evade therapeutic pressure, with drug‐tolerant persister cells demonstrating selective survival advantages. ¹⁸ , ¹⁹ , ²⁰ Building on this evidence, our study successfully established a GSK690693‐resistant OSCC PDX model by serially passaging tumors from mice treated with GSK690693, a targeted small‐molecule inhibitor. Meanwhile, the first‐line clinical treatment drugs for OSCC were also used to observe the dynamic evolution process after treatment and construct the drug resistance models. The establishment of the drug resistance model provides a vital methodological reference for the research on the drug resistance mechanism of OSCC and precise treatment.

FIGURE 2.

Serial propagation in drug‐treated patient‐derived xenograft (PDX) mice. (A) Methodological diagram. (B) Experimental operation protocols. (C) Tumor growth kinetic in GSK690693‐treated PDX mice (n = 6, 95% confidence interval [CI], $\overline{\mathrm{\chi }}$ ± $\mathrm{s}$), ****p < 0.001. (D) Tumor growth kinetic in rapamycin‐treated PDX mice (n = 6, 95% CI, $\overline{\mathrm{\chi }}$ ± $\mathrm{s}$), ****p < 0.001, ***p < 0.005. (E) Tumor growth kinetic in CDDP + CTX‐treated PDX mice (n = 6, 95% CI, $\overline{\mathrm{\chi }}$ ± $\mathrm{s}$), ****p < 0.001.

2. METHODS

2.1. Clinical tumor tissue samples

Primary OSCC tissues were obtained from Peking University School of Stomatology (Beijing, China). Using clinical samples was approved by the Biomedical Ethics Committee of the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences (approval number: QC‐23002). All procedures involving human participants complied with the Declaration of Helsinki (revised 2013). Informed consent was obtained from recruited patients prior to sample collection.

2.2. Experimental animals

Severely immunodeficient NSG mice (NOD/SCID/IL2Rγnull, aged 5–8 weeks, purchased from SPF Biotechnology Company, Beijing, China; SCXK [Jing] 2024–0001) were used for PDX model establishment and housed in the specific pathogen‐free (SPF) animal facility of the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences. All mice were provided with adequate food and water, maintained at a constant temperature of 22°C, and subjected to a strict 12‐h light/dark cycle (light phase: 7:00 a.m–7:00 p.m., 100 lux). Animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences (approval number: QC‐24002). Throughout the experiments, animal usage and any potential pain experienced by the mice were minimized in accordance with ethical guidelines.

2.3. Construction of OSCC PDX models and drug treatment

Fresh tumor tissues resected from patients were sectioned into 2 × 2 × 2 mm fragments under sterile, low‐temperature (2–8°C) conditions and subcutaneously implanted into mouse flanks within 24 h. After reaching mung bean‐sized volumes, tumors were passaged (PDX P1–P2).

For P2 PDX mice, when tumor volumes reached 40–80 mm³, they were randomly divided into four groups (n = 6 per group): negative control: PBS (300 μL/mouse, intraperitoneal injection), first‐line clinical therapy: cisplatin (1 mg/kg, three times/week, intraperitoneal injection) combined with cetuximab (1 mg/kg, weekly, intraperitoneal injection), AKT inhibitor: GSK690693 (10 mg/kg, five times/week, intraperitoneal injection), mTOR inhibitor: rapamycin (4 mg/kg, five times/week, intraperitoneal injection). The drug dosages were determined based on pre‐experiments (Figure S1) and relevant literature. ²¹ , ²² , ²³ Three dose levels of the AKT pathway inhibitor GSK690693 were as follows: low (3 mg/kg, five times/week, intraperitoneal injection), medium (10 mg/kg, five times/week, intraperitoneal injection), and high (30 mg/kg, five times/week, intraperitoneal injection), with six mice per group treated in pre‐experiments. Treatment was administered for 21 days. Tumor growth and animal behavior (activity, feeding/drinking, weight changes, and physical abnormalities) were monitored regularly. Tumor volume (V) was measured every 2 days using a caliper to determine the longest diameter (a) and shortest diameter (b) of the subcutaneous tumor. The tumor volume(TV) was calculated as TV (mm³) = (a × b²) × 0.5. Tumor growth curves were plotted based on these measurements. Tumor growth inhibition(TGI)% was calculated as TGI = (TVcontrol‐TVtreatment)/TVcontrol. Relative tumor volume growth rate (T/C) % was calculated as TGI = RTVtreatment/RTVcontrol. Relative tumor volume (RTV) = Vafter‐treatment/Vpre‐treatment. According to China Food and Drug Administration (CFDA) Guideline for Pharmacodynamics Studies of Antineoplastic Drugs, we use T/C and TGI to evaluate drug response, when T/C >40% or TGI <40% is regarded as the drug being ineffective.

When tumor volumes reached ≥1000 mm³ or the treatment cycle was completed, each harvested tumor sample was divided into three parts: one was snap‐frozen for cryopreservation; one was fixed in 4% paraformaldehyde (PFA), embedded in paraffin, and sectioned; one was used for serial transplantation and passaging.

2.4. Construction of drug‐resistant OSCC PDX models

After the drug treatment cycle was completed, tumor tissues were dissected, and the tumor with the poorest drug sensitivity (largest volume) was selected and retransplanted into new NSG mice for continuing treatment with the same drug. If the tumor growth and pathological status remain stable after drug administration and the resistance is determined, it is considered that the drug resistance model has been successfully established (Figure 2B).

2.5. Histopathological analysis via hematoxylin–eosin staining

HE staining was performed on formalin‐fixed paraffin‐embedded patient tumor tissues, PDX tumor tissues, and tumor tissues from all treatment groups. Staining protocol: tissue samples were dewaxed and rehydrated through a graded alcohol series, and then sections were stained with hematoxylin for 2 min, rinsed, differentiated in 1% acid alcohol for a few seconds. Further, counterstaining was carried out with 0.5% eosin aqueous solution for 1–2 min. Finally, sections were dehydrated, cleared, and mounted.

2.6. Immunohistochemistry for tumor marker detection

For immunohistochemistry (IHC) staining, paraffin‐embedded PDX and treated tumor tissue sections were dewaxed and rehydrated through a graded alcohol series. Endogenous peroxidase activity was blocked with 3% H₂O₂. Tissues antigen retrieval was performed using citrate buffer, followed by blocking with goat serum for 30 min at room temperature. The sections were then incubated overnight at 4°C with the following primary antibodies: Ki67 (ab1667, Abcam, 1:200), CK5/6 (ZM‐0313, OriGene, ready‐to‐use), E‐cad (ab40772, Abcam, 1:500), EGFR (ab52894, Abcam, 1:500), P53 (ab32049, Abcam, 1:200). The next day, secondary antibody (HRP‐conjugated goat anti‐rabbit, 37°C, 30 min) was applied. Diaminobezidine was used for chromogenic detection, and counterstaining was performed with hematoxylin, followed by dehydration, clearing, and mounting. Quantitative analysis of immunohistochemical results was conducted using the Q‐score calculation.

2.7. Statistical analysis

Data were analyzed using GraphPad Prism 9.0, expressed as mean ± standard deviation (SD) ($\overline{\mathrm{\chi }}$ ± $\mathrm{s}$). Multigroup comparisons employed two‐way analysis of variance (ANOVA). p < 0.05 indicated statistical significance.

3. RESULTS

3.1. Establishment of OSCC PDX models

Fresh tissues were implanted in NSG mice using the method shown in Figure 1A to construct the OSCC PDX model and retransplanted for passage. HE staining was performed on patient‐derived primary tumor tissues and each generation of PDX tumor tissues. Histopathological analysis revealed that both primary tumors and P1–P2 generation PDX tumors exhibited nest‐like growth patterns with significant cellular atypia and active proliferation (Figure 1B–D). No notable morphological differences were observed between patient tumors and PDX P1–P2 xenografts. These pathological results confirm successful construction of OSCC PDX models.

FIGURE 1.

Establishment and characterization of oral squamous cell carcinoma (OSCC) patient‐derived xenograft (PDX) models. (A) Methodological diagram of establishing OSCC PDX models. (B–D) Concordant histomorphological features between patient‐derived tumor tissues and PDX models. Scale bar: 250 μm.

3.2. Reduced drug sensitivity in GSK690693‐treated PDX mice during successive passaging of all treated OSCC PDX mice

To simulate the tumorigenesis and development process of OSCC PDX mice after drug treatment, we designed the clinical treatment group, the rapamycin group, and the GSK690693 group. All of them were serially passaged while maintaining their original treatment protocols, as shown in the experimental flowchart in Figure 2B. With successive passaging, we found that the GSK690693 accelerated tumor growth in the P2 generation, with no significant statistical difference compared to controls (Figure 2C). In contrast, after two passages in the rapamycin treatment group and the cisplatin combined with cetuximab treatment group, the tumor growth in the P2 generation still exhibited therapeutic efficacy despite faster tumor growth compared to P1, showing significant statistical differences versus the control group (p < 0.005 and p < 0.001, respectively; Figure 2D,E; Table S1). As shown in Table 1 and Table S1, the TGI were 75.89% (GSK690693‐P1), 27.97% (GSK690693‐P2), 86.69% (rapamycin‐P1), 63.45% (rapamycin‐P2), 77.96% (CDDP + CTX‐P1), 68.96% (CDDP + CTX‐P2). Meanwhile, the T/C were 22.83% (GSK690693‐P1), 56.97% (GSK690693‐P2), 17.14% (rapamycin‐P1), 26.27% (rapamycin‐P2), 26.23% (CDDP + CTX‐P1), 16.93% (CDDP + CTX‐P2).

TABLE 1.

Tumor growth inhibition (TGI)% and relative tumor volume growth rate (T/C)% of all groups.

Group	TGI%	T/C%
GSK690693‐P1	75.89	22.83
GSK690693‐P2	27.97	56.97
Rapamycin‐P1	86.69	17.04
Rapamycin‐P2	63.45	26.27
CDDP + CTX‐P2	77.96	26.23
CDDP + CTX‐P2	68.96	16.93

Note: TGI% and T/C% of P2‐GSK690693 are 27.97% and 56.97%, respectively. The TGI% is <40%, whereas T/C% is >40%.

3.3. Histopathological analysis of tumor tissues in the GSK690693 insensitivity group

To verify whether drug resistance occurred in the GSK690693 treatment group, we performed HE staining. At low magnification, tumors in model groups displayed nest‐like growth; sensitive groups (cisplatin–cetuximab and rapamycin treatment groups) exhibited extensive necrosis, whereas resistant tumors (GSK690693 group) showed solid, sheet‐like growth (Figure 3A). At higher magnification, in the model group, the tumor cells appeared well‐differentiated, and intact keratinized beads could be seen in some areas (Figure 3B,C). Sensitive groups featured residual tumor cells adjacent to large areas of necrotic regions. The tumor cells became smaller, with shrunken, fragmented nuclei and defective keratin pearls (Figure 3B,C). However, resistant tumor tissues demonstrated poorly differentiated. In large areas of solid tumor tissue, reduced stroma, enlarged nuclei, and coarse chromatin were observed (Figure 3B,C). The above results indicate that the GSK690693‐treatment group underwent morphological changes after continuous passaging.

FIGURE 3.

Hematoxylin–eosin (HE) and immunohistochemistry (IHC) of control, sensitive, resistant group. (A–C) In the sensitive group, tumor tissue necrosis and cell degeneration were observed. Tumor tissue in model group grew in the form of nests. The tumor tissue in resistant group showed solid and flaky growth. (D) Fragmented CK5/6 expression was observed in necrotic tumor tissues of the sensitive group. The model group demonstrated intact CK5/6 expression, whereas the resistant group displayed patchy CK5/6 expression with rare keratinization. Scale bar: 500, 250, and 100 μm.

To further illustrate the changes in its histological morphology and prognosis judgment, we used IHC to detect the expression of keratin CK5/6 for morphological analysis. As shown in Figure 3D, CK5/6 was manifested intact in control group, whereas the sensitive groups exhibited fragmented CK5/6 staining within necrotic tumor tissue. In contrast, in non‐sensitive groups, it was diffusely positive with sheet‐like and reduced keratinization (Figure 3D).

3.4. Immunohistochemical analysis of related resistant markers in the GSK690693 insensitivity group

To further verify whether drug resistance occurred and prognosis judgment at potential molecular level, we performed IHC staining on Ki67, EGFR, E‐cad, and P53. The results show that Ki67 proliferation indices were high (>80%) in resistant and control groups but low (<15%) in sensitive groups (Figure 4A,B). E‐cad expressed weakly or deficiently in resistant groups but high in sensitive groups (Figure 4C,D). Meanwhile, EGFR was highly expressed in the resistant group but weakly expressed in the sensitive group (Figure 4E,F). For P53, both groups appear wild‐type or show normal expression (Figure 4G,H).

FIGURE 4.

Immunohistochemistry (IHC) analysis of resistance‐related markers in all groups. (A, B) The sensitive group exhibited low Ki‐67 expression, whereas the resistant group and model group showed high Ki‐67 expression. Quantitative analysis via Q‐score (n = 6, 95% confidence interval [CI], $\overline{\mathrm{\chi }}$ ± $\mathrm{s}$), **p < 0.01. (C, D) E‐cad expression is middle positive in the sensitive group and weak in the resistant group. Quantitative analysis via Q‐score (n = 6, 95% CI, $\overline{\mathrm{\chi }}$ ± $\mathrm{s}$), **p < 0.01. (E, F) Epidermal growth factor receptor (EGFR) is weak in the sensitive group and strong in the resistant group. Quantitative analysis via Q‐score (n = 6, 95% CI, $\overline{\mathrm{\chi }}$ ± $\mathrm{s}$), *p < 0.05. (G, H) P53 expression looks normal or wild in both the groups. Scale bar: 250 μm.

In conclusion, based on animal experiments and HE and IHC analyses, the tumor growth in GSK690693 treatment group accelerated after continuous passage, suggesting that the tumor tissue developed resistance to this inhibitor at the histological level.

3.5. Establishment and stable passage of GSK690693‐resistant OSCC PDX models

To establish a drug‐resistant OSCC PDX model, the tumor tissue with the largest volume was selected and continued to be reimplanted into four NSG mice treated with GSK690693, with the model group serving as the control and the treatment‐insensitive group designing as the P1 generation of the drug‐resistant model. It was observed that the tumor growth rate in the P2 generation of the drug‐resistant model was significantly faster than that in the control group (Figure 5A; Table S2). After 21 days of treatment, the average wet weights of the tumors dissected from the drug‐resistant and the control group were 0.23 g and 0.12 g, respectively (Figure 5B; Table S2). Macroscopic observation revealed that the tumor volume in the drug‐resistant group was larger than that in the control group (Figure 5C). Histological examination revealed that the drug‐resistant OSCC PDX model P2 generation exhibited poorly differentiated tumor tissue, with solid growth patterns and focal nest‐like structures, reduced stromal components, enlarged tumor cells, and coarsely clumped nuclear chromatin, which effectively retained the histopathological characteristics of the drug‐resistant P1 generation tumor tissue (Figure 5D). IHC analysis demonstrated a high Ki67 proliferation index (80%) and positive CK5/6 expression (+). The above results indicate the successful establishment of a GSK690693‐resistant OSCC PDX model.

FIGURE 5.

Establishment and characterization of the GSK690693 resistance model. (A) Tumor growth curve after P2 readministration. (B) Tumor weight chart. (C) Gross image. (D) Biological characteristics of P1–P2 remain consistent and exhibit stable passage. Scale bar: 250 μm.

4. DISCUSSION

Patients with recurrent OSCC face significant impacts on their esthetic appearance and physical and mental health due to the unique anatomical location of the disease. However, there are currently no effective treatment strategies for patients experiencing recurrence/metastasis caused by OSCC drug resistance. The primary approach remains chemotherapy combined with targeted therapies (e.g., cisplatin–cetuximab), but drug resistance frequently develops. ⁶ , ⁷ , ²⁴ Methods to test different precision oncology strategies based on complex and dynamically evolving molecular information still pose major challenges in clinical practice. ²⁵ , ²⁶ Therefore, constructing drug‐resistant models, elucidating resistance mechanisms, and implementing individualized targeted therapies are crucial for improving prognosis and extending patient survival.

Studies have shown that PDX models can monitor clonal selection and dynamic functional responses. For example, in a glioblastoma xenograft study, ²⁷ researchers subcutaneously transplanted surgically resected glioblastoma samples into immunodeficient mice. They observed that PDX tumor growth rates significantly increased with successive in vivo passages, accompanied by secondary gliosarcoma to sarcoma transformation. HE and IHC analysis between primary glioblastoma, secondary gliosarcoma, and gliosarcoma PDX tissues indicated that although PDX models largely retain the histopathological features of primary tumors, their malignancy increases with successive in vivo passages, and there is a significant correlation between tumor growth rates and histopathological characteristics. ²⁷ , ²⁸ Similar phenomena have been validated in PDX models of head and neck cancers and other tumor types. ²⁹ , ³⁰ For instance, Yao et al. ³¹ conducted a “phase II clinical trial‐mimicking” study using 49 head and neck squamous cell carcinoma (HNSCC) PDX models, successfully identifying multiple key biomarkers of intrinsic cetuximab resistance (e.g., ANKH amplification, PARP3 upregulation). These biomarkers were further validated in another HNSCC PDX cohort. And the team further developed acquired cetuximab‐resistant PDX models to explore resistance mechanisms. ³¹ Through in vivo and in vitro experiments, they demonstrated that combination‐targeted therapies could effectively overcome cetuximab resistance while also revealing dynamic clonal evolution during resistance development. These findings provide critical insights for clinical treatment strategies in cetuximab‐resistant HNSCC patients on PDX resistance models. ³¹

Based on previous research findings, this study utilized the targeted small‐molecule inhibitor GSK690693, the FDA‐approved drug rapamycin, and the first‐line clinical treatment regimen (cisplatin + cetuximab) to treat OSCC PDX mice. Continuous passaging was performed while maintaining the original treatment protocols. When tumor growth volume increased, indicating drug insensitivity, the tumor tissues from these groups underwent pathological analysis, and the largest tumor tissues were selected for further passaging. After serial passaging to P2 generation, the drugs were readministered. Stable tumor growth parameters and histomorphological characteristics were considered, indicating the successful establishment of a drug‐resistant model. Experimental results demonstrated that the GSK690693 treatment group exhibited significantly accelerated tumor growth phenotypes after the second passage (P2). The potential toxicity and maximum tolerated doses were considered, as shown in pre‐experiments with three dose levels of the AKT pathway inhibitor GSK690693: low (3 mg/kg, five times/week, intraperitonially), medium (10 mg/kg, five times/week, intraperitonially), and high (30 mg/kg, five times/week, intraperitonially), with six mice per group treated over 21 days. Results showed significant tumor growth inhibition in all OSCC PDX mouse groups compared to controls (Figure S1). Although the medium‐ and high‐dose groups exhibited smaller tumor volumes and weights than the low‐dose group, the high dose (30 mg/kg) led to weight loss and mortality (two/six mice), indicating potential toxicity. Based on these findings, we selected the medium dose (10 mg/kg, five times/week, intraperitoneally) to balance therapeutic efficacy and safety (Figure S1). Histopathological evaluation demonstrated hallmark features of diminished tumor differentiation in this group, including extensive solid tumor architecture with the following microscopic characteristics: reduced stromal compartment, cellular enlargement accompanied by elevated nuclear‐to‐cytoplasmic ratios, and prominent hyperchromatic nuclei displaying coarse chromatin clumping. Additionally, IHC results indicated high expression of the proliferation marker Ki67 (>80%) and strong patchy positivity for CK5/6, with significant loss of keratinization components. Moreover, EGFR, E‐cad, and P53 are markers closely related to drug resistance. IHC analysis showed that EGFR and E‐cad are strong and weak in resistant group. Compared to the model and sensitive groups, these tumors displayed morphological alterations and exhibited poorer differentiation than pretreatment tumor tissues under our experimental conditions. These pathological changes align with previously reported malignant progression phenomena in PDX models during passaging. ²⁸ , ³⁰ , ³² Through validation via continuous P1–P2 passaging and retreatment, tumors maintained stable rapid growth characteristics without significant shifts in histopathological indicators, confirming the successful establishment of drug‐resistant OSCC PDX models. For the two other drug‐sensitive groups, the largest tumors were continuously passaged and treated with their original drugs to simulate tumor evolution. If reduced sensitivity emerged, the same methodology was applied for resistance evaluation, model establishment, and timely therapeutic regimen adjustments. The standardized modeling approach developed in this study provides a robust technical platform for in‐depth exploration of OSCC resistance mechanisms and the development of precision treatment strategies.

With the continuous optimization of PDX technology and advances in translational research, this platform is accelerating the translation of personalized medicine from conceptual ideals to clinical implementation. It is critical to emphasize the following limitations of this study: (1) incomplete molecular characterization: current validation remains confined to histopathological verification of tumor morphological fidelity. Effective guidance of precision‐targeted therapies requires PDX models to maintain primary tumor characteristics at both biological and molecular levels. ²⁵ , ³³ , ³⁴ (2) Limited clinical correlation validation: due to heterogeneity in clinical treatment regimens, this study employed model control groups rather than replicating patient‐matched therapeutic protocols or conducting synchronized clinical trials. However, emerging clinical evidence confirms that PDX modeling—whether used independently or integrated with complementary approaches (e.g., genomic profiling)—significantly enhances patient survival rates. ²⁰ , ²⁵ , ³⁵ , ³⁶ , ³⁷ , ³⁸ (3) Limited subsequent application of the resistant model investigation. (4) Only select the “largest tumor” for passaging to represent resistance without clonal tracking or single‐cell analysis: our approach provides a foundational framework but may not fully capture intratumoral diversity. Looking forward, our research priorities will focus on the following: (1) multiomics integration: systematic dissection of molecular fidelity in PDX models through the convergence of sequencing (genomic/transcriptomic), proteomic profiling, and functional molecular biology to identify and validate resistance biomarkers. (2) Real‐time clinical translation: implementation of parallel clinical trials to establish dynamic synchronization between PDX model development and patient therapeutic interventions. (3) Preclinical drug screening and biomarker discovery design: we will elaborate on how the model can be used to identify novel therapeutic targets or evaluate drug combinations against resistance mechanisms. And we will highlight the model's potential to uncover dynamic molecular signatures associated with resistance (e.g., EMT markers, DNA repair pathways), which could serve as predictive biomarkers for patient stratification. ¹⁹ , ²⁵ (4) Clonal tracking or single‐cell analysis: we recognize their value and plan to integrate these techniques in future iterations to better characterize clonal dynamics and heterogeneity. ¹⁹ , ²⁵

5. CONCLUSION

This study adopted the serial passage method to simulate the occurrence and development of tumors in clinical patients after treatment. Specifically, experimental observations were conducted on the OSCC PDX models after cisplatin + cetuximab, GSK690693, and rapamycin treatments. It was found that GSK690693 showed drug resistance after the second passage. Based on this, the drug resistance animal model of GSK690693 was successfully established. This model not only provides an important tool for the study of resistance mechanisms in OSCC but also lays a methodological foundation for the subsequent research on precise treatment strategies.

AUTHOR CONTRIBUTIONS

Chuanni Feng: Data curation; formal analysis; investigation; methodology; project administration; validation; visualization; writing – original draft. Hao Liu: Funding acquisition; resources. Yalan Lu: Funding acquisition; resources. Yanfeng Xu: Methodology. Xinghan Wu: Methodology. Jinlong Wang: Methodology. Chuan Qin: Resources; supervision; validation; writing – review and editing. Binbin Li: Resources; supervision; writing – review and editing. Yanhong Li: Project administration; supervision; validation; writing – review and editing.

FUNDING INFORMATION

This research was supported by the National Natural Science Foundation of China (82173399), the Young Elite Scientists Sponsorship Program by CAST (grant #: 2022QNRC001), the Beijing Natural Science Foundation (7252096), the Beijing Natural Science Foundation‐Haidian Original Innovation Joint Fund Project (No: L222145).

CONFLICT OF INTEREST STATEMENT

Chuan Qin is an editorial board member of Animal Models and Experimental Medicine (AMEM) and an author of this article. To minimize bias, she was excluded from all editorial decision‐making related to the acceptance of this article for publication.

ETHICS STATEMENT

The research protocol involved in this article has been approved by the Biomedical Ethics Committee of the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences. Primary OSCC tissues were obtained from Peking University School of Stomatology (Beijing, China). Informed consent and the use of clinical samples were approved by the Biomedical Ethics Committee of the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences (approval number: QC‐23002). All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the Institute of Laboratory Animal Science, CAMS & Comparative Medicine Center, PUMC (IACUC approval number: QC24002).

Supporting information

Figure S1

Table S1

Table S2

Feng C, Liu H, Lu Y, et al. Establishment of a patient‐derived drug‐resistant oral squamous cell carcinoma animal model. Anim Models Exp Med. 2025;8:1513‐1523. doi: 10.1002/ame2.70066