The mice with an intact immune system will produce robust immune responses to abrogate the implanted tumor tissue. Thus, immunodeficient mice are required to develop the PDX models. Nude CB17-SCID mice, NOD-SCID mice, NOG (NOD/SCID/IL-2γpartial deficiency) mice, and NSG (NOD/SCID/IL-2Rγcomplete deficiency) mice are the common mainstay mouse strains for PDX model establishment. Among them, nude mice are the most commonly used. These mice lack body hair and thymus. Given the absence of the thymus, these mice are T-cell deficient and, thus, cannot induce an adaptive immune response. The nude mouse model was first reported in 1966 by Flanagan (43), which allowed the growth of human cancer cells in mice. Subsequently, the use of CB17-SCID mice and NOD-SCID mice emerged successively. However, due to the high incidence of spontaneous lymphoma, CB17-SCID mice mostly die prematurely (44). In addition, the incidence of immunological leakage in CB17-SCID mice is extremely high.

Similarly, spontaneous lymphoma, in most cases, occurs later in NOD-SCID mice aged around 8.5 months (45). Consequently, neither mouse is suitable for long-term experiments. By the early 2000s, the Central Institute for Experimental Animals (CIEA) developed a severely immunocompromised NOG mouse, significantly improving the survival rate of human cell and tissue transplantation in immunocompromised mice. In 2005, the US Jackson Laboratory cultured NOD/SCID/IL2R gamma (null) mouse, also known as NSG mouse, with a higher transplantation rate and lower tumor graft rejection (46). The NSG mice lack mature T, B, and NK cells for genetic mutations. Thus, the NSG mice are currently the most ideal tumor graft receptors and are less prone to lymphoma development, and have a longer lifespan than other mice (Nude, SCID mice, NOD-SCID mice). The tabular comparison of several immunodeficient mice is shown in Table 1 . Therefore, the NSG recipient mice offer several advantages for transplantation with human hematopoietic stem cells (46, 47). The NSG mouse is also called “humanized mouse,” or the human immune system (HIS) model because it supports the growth of human hematopoietic stem cells. Thus, it developed an immune system and produced human T cells similar to those of humans (48). The processes of developing the HIS model are sketched in Figure 1 . The humanized PDX model with patient-matched immune components has several advantages over the alternative models for decoding tumor biology and anti-tumor drug development.

The implanted tissues or cells can be patient biopsies or tumor cells derived from ascites or pleural fluid ( Figure 2 ). The implantation tissues comprise small 1-3 mm³ clumps or single cell suspensions prepared by digesting small tissue fragments. Co-transplanting with the matrigel increases the engraftment success rates ( Figure 2 ). In 1990, Fridman et al. reported that the transplant success rate was significantly higher when lung cancer tissue and matrigel were intertwined and implanted into nude mice (49). To this end, tumor-associated fibroblasts and mesenchymal stem cells provide functional support for tumor progression and development. Furthermore, diversities that may also interfere with the ability of the graft to grow successfully were assessed and compared for ligand-receptor interactions between humans and mice. It was found that some mouse ligands do not activate the corresponding human receptors (50–52).

In brief, proper expression of some human ligands in immunocompromised mice may stimulate graft growth. Ectopic or orthotopic transplantation is traditionally performed via subcutaneous and intravenous injection and other forms. The subcutaneous transplantation model is the most extensively used owing to its simple operation and convenient measurement of tumor size. The implantation site is often the back or underarm of mice, which is an excellent model for drug response assessment. Nonetheless, given the limited tumor space, the transplant does not grow in the correct anatomical location. Therefore, this could not be a suitable approach for tumor metastasis research since the tumor does not grow in the correct anatomical location and microenvironment. A differential display analysis is required to compare graft biology and drug response on the in situ model and the ectopic model.

Priority should be given to tail vein injection to build an orthotopic transplantation model for tumor metastasis study. The advantage of tail vein injection is that the external environment of in situ tumor is closer to the human tumor, providing an experimental basis for subsequent tumor metastasis research ( Figure 2 ). Effective transplantation is the nascent basis of the experiment, and many factors can affect the implantation rate. The large vascular bed beneath the renal capsule facilitates the growth of the graft. Xin et al. implanted tumor tissue from NSCLC patients under the renal capsule to obtain an implantation rate of up to 90% (53) and successfully evaluated the sensitivity of patients’ chemotherapy regimen using PDX models. The subcutaneous transplantation success rate is only about 23% (54).

Furthermore, Chen et al. (55) reported that PDX was higher in squamous cell carcinoma, stage II, stage III, and poorly differentiated tumor specimens by following clinically confirmed NSCLC patients for up to 1.5 to 6 years. Additionally, some studies have shown that the implantation rate of PDX tumors after passage increases with the number of passages, and the long-time of tissue in vitro is associated with low implantation rates. Moreover, the implantation rate of brain NSCLC metastases is higher than the primary tumor (56), suggesting that the source of the samples potentially affects the implantation rate. Notably, the PDX was not established in the mouse generating human CD20⁺ immune cells. Therefore, adding rituximab during primary tumor implantation may decrease the frequency of lymphoma formation and promote PDX growth.

The immune system is responsible for maintaining tumor regression (57). Programmed death ligand 1 (PD-L1) is present on the surface of tumor cells and inhibits the activation and proliferation of T cells, which results in immune escape of tumor cells by specifically binding the programmed death receptor-1 (PD-1) (58, 59). Moreover, T cell depletion is common in people with cancer or chronic infection (60, 61). The T cell depletion is characterized by loss of T cell effector function and continuously increased expression of immune-inhibitory receptors, one of the leading causes of immune disorders in cancer patients. Fortunately, numerous studies have demonstrated the vital role of PD-1 inhibitors in reversing T cell depletion (62–64). In other research, the use of cancer immunotherapy with antibodies, cancer vaccines, and immune checkpoint inhibitors (ICIs) as an emerging and efficient pathway has attracted widespread attention (65–67). Researchers have made unprecedented advances in immunotherapy by utilizing humanized mouse PDX models to accelerate acquired studies of clinically relevant tumors and evaluation of the clinical value of cancer immunotherapy. Humanized mice with human hematopoietic stem cells have been broadly reported as a potent tool in studying targeted PD-1 antibodies (68). Injecting PDX mice with fresh cord blood containing CD34⁺ hematopoietic stem cells to rebuild the immune system significantly shortens the time and regenerates functional, active immune cells. This overrides the delayed tumor transplantation caused by prolonged immune system establishment and the limitation of the incomplete immune system. Immunotherapy provides a new strategy for lung cancer treatment (69, 70). The optimized PDX models are ideal for testing the significant inhibitory effect of the antagonistic PD-1 checkpoint drug-pembrolizumab on lung cancer PDX tumors. Research shows that pembrolizumab-activated human immune cells could effectively limit tumor growth. Targeting cytotoxic T lymphocyte-associated protein 4, PD-1, and PD-L1 using antibodies to block immunomodulatory checkpoints on tumor cells, immune cells, and endothelial cells is an effective NSCLC therapy. Numerous studies have shown that immune checkpoint inhibitors such as PD-L1 and CTLA4 significantly increase the survival rate of advanced NSCLC patients.

Moreover, long patient follow-up studies reveal that the five-year survival rate of NSCLC patients receiving such treatment is as high as 16% (71). The anti-PD-1 drug pembrolizumab is currently approved as an immunotherapy regimen for patients with advanced NSCLC expressing PD-L1 (72–75). Moreover, Nivolumab and Atezolizumab can be used despite PD-L1 expression in NSCLC (76, 77). The studies cited above show great promise in oncology research via the use of the PDX model to study and evaluate the necessity of immunotherapy (78).

The US Food and Drug Administration has approved more than 20 drugs for the NSCLC treatment, including 14 targeted and immunotherapy drugs. However, the rapid emergence of drug resistance and the diverse resistance mechanisms continue to threaten the ability to treat NSCLC patients. While significant advancements have been made for NSCLC treatment, acquired resistance remains a significant barrier to therapeutic response (79, 80). The genetic alteration uncertainties in patients after treatment is an obstacle to conducting prospective studies. There is an urgent need to identify biomarkers for predicting acquired resistance or poor response of lung cancer to ICIs. The PDX model is ideal for drug resistance studies, including identifying mutations that promote treatment resistance (81). Therefore, the PDX model is critical in the next phase of drug development and reduces the lag in clinical trials. The SCLC patient’s response rate to first-line chemotherapy is about 70%. Nevertheless, most patients experience rapidly acquired resistance within months of treatment, rapid relapse, and poor efficacy of second-line treatment (82, 83). The development of PDX models has been invaluable for investigating acquired drug resistance of tumors.

Previous research reported that EZH2 promotes the development of acquired resistance of PDX by silencing SLFN11. However, after EZH2 knockdown, the PDX tumors’ resistance was suppressed, and the drug efficacy was optimized (84). The exact resistance mechanism of the original tumor was observed in the PDX model of LUAD (85). These studies demonstrate that PDX models accurately predict the patient’s drug response. The widespread uses of ICIs, including PD-1, PD-L1, and CTLA4 antagonistic antibodies, ushered in a new era of cancer immunotherapeutic. However, as PD-1 axis inhibitors become increasingly established in standard treatment options for lung cancer, more patients show resistance to these therapies. Research is needed to identify the mechanisms underlying the acquired resistance of ICIs in lung cancer to overcome the drug resistance of ICIs.

The emergence of the PDX model in drug research development and addressing cancer treatment challenges provides tools for better research. Using the PDX model carrying tumor as the embodiment to evaluate drug efficacy before treatment, it can not only screen for the best anti-cancer treatment but also avoid some adverse drug reactions, which will have a significant impact on patients’ treatment decisions ( Figure 3 ). As indicated earlier, the PDX model identifies fewer common microwaves, nanoparticles, gene therapies, and the potential of new anti-cancer drugs for future intervention in tumor development, thus providing a new therapeutic indicator (86). For instance, a PDX model has been used to evaluate three clinical standard protocols for NSCLC patients. The results showed that patients had different sensitivity to the three regimens. Moreover, the patients were not equally sensitive to the same regimen. These findings suggest that the PDX model determines the treatment response and predicts individualized treatment regimens for patients with NSCLC (61). Furthermore, the PDX model predictably addresses the effects of small molecule compounds, antibodies, and microorganisms on lung cancer (87). There are three molecular LUAD subtypes (88–90), with the highly proliferative PP being the most malignant. No validated targeted therapy and immunotherapy have been developed for clinical use against the PP subtype. Therefore, developing specific therapeutic agents will restore hope to patients with the PP subtype. Using nucleic aptamer, Sun et al. found that the leucine-rich PPR-motif containing (LRPPRC) is a specific nucleic aptamer binding protein significantly overexpressed in PP subtype and is strongly linked with the disease stage and patient prognosis. Gossypol acetic acid (GAA) was successfully screened using aptamer-assisted high-throughput methods to specifically bind LRPPRC and cause its degradation in a ubiquitin-proteasome-independent manner. After treatment with GAA, a robust anti-tumor function was shown in the LRPPRC-positive LUAD-PDX models but not in the LRPPRC-negative PDX models. These findings suggest that GAA specifically targeted LRPPRC knockout is a promising therapeutic strategy targeting the PP subtype (91). The above research has entered phase II clinical practice, extending the treatment strategy for lung cancer and highlighting the role of PDX models in developing efficient treatment regimens. Among the functions mentioned earlier, the increase in studies targeting PDXs suggests that PDX models can accurately predict drug response in clinical patients. Therefore, PDX tumor cells can be used for high-throughput screening of anti-cancer therapeutics (92). A study evaluating the use of zebrafish in PDX models reported that transparent zebrafish larvae could visualize individual tumor cells and their response to treatment, acting as a rapid drug screening platform (29). In other words, researchers can fetch the information from the models directly. An intensive drug screening was performed in more than 1,000 PDX models (93). Based on the obtained data, the PDX model has become an essential part of the preclinical screening for new anti-cancer drugs. However, there exist theoretical differences in drug absorption, distribution, and pharmacokinetics between mice and humans, and there may be no natural drug reaction in humans (94, 95). This may result in undesirable consequences after drugs enter clinical trials to determine the pharmacological discrepancies between the two for better future development and testing of anti-tumor drugs.

With the development of individualized, targeted therapies for genotypes, the preclinical studies and clinical trials are conducted simultaneously to elucidate disease mechanisms. In this context, a PDX model can be used as a personalized “Avatar model” for precision medicine in cancer treatment (96). The pre-clinical and clinical trials are conducted simultaneously in tumor patients with specific genetic structures and the corresponding PDX models in co-clinical trials. The trials compare the response between patients and the models in determining the mechanism of action of drugs and finding new targets and biomarkers (97) ( Figure 3 ). A group of PDX models directly derived from T2 or T3 stage NSCLC patients was established by Iduna et al. (54). The downregulation of EGFR was observed in the sensitive PDX models after the treatment with cetuximab but not in the resistance models. Iduna confirmed the association of KRAS mutations with the EGFR resistance phenotype, demonstrating that KRAS wild-type status is a clinical biomarker for targeted therapy of NSCLC (54). With such fidelity of the PDX models, it is envisioned that if the pre-clinical studies are performed simultaneously or earlier, the discovery and validation of KRAS mutations as a biomarker for drug resistance will be facilitated. Consequently, such validation linking impactful biomarkers to treatment efficacy will lead to direct clinical effects. The fibroblast growth factor receptor (FGFR) gene is one of the molecular targets of LUSC. As is known to all, dovitinib is a tyrosine kinase inhibitor targeting FGFR. Previous studies have demonstrated that dovitinib has potential anti-tumor capability against tumor cell lines with FGFR1 amplification. Related reports indicated that FGFR1 amplification is associated with poor prognosis in LUSC patients (98). A co-clinical trial concurrent with phase II trials in LUSC patients demonstrated that PDX models accurately replicated patient response to dovitinib. These findings revealed that FGF3 and FGF19 genes were significantly enriched in dovitinib -sensitive tumor cells. In addition, the findings suggest that activation of genes involved in FGFR signaling may be a critical factor in sensitivity to dovitinib (99). The above evidence indicates that the PDX model has strong potential in identifying predictive biomarkers and advancing precision medicine and drug development (100).