Manifestations of Genomic-Level Intratumoral Tumor Heterogeneity in Multi-Region Patient-Derived Xenograft Models of Non-Small Cell Lung Cancer

Academic Background and Research Motivation

Patient-Derived Xenograft (PDX) models are widely used in cancer research. PDX models are established by transplanting human tumors into immunodeficient mice. Therefore, PDX models are considered to simulate tumor biological characteristics more accurately than traditional cell lines, retaining in vivo cell-cell and/or cell-matrix interactions, three-dimensional structures, and their most recent derivative characteristics. Many studies have shown that the drug response of PDX models is consistent with that of individual patients or patient groups, which makes PDX models used in personalized medicine and preclinical drug trials, becoming an important model in cancer biology.

However, the genomic fidelity of PDX models is crucial for their application in preclinical oncology. A key issue is whether these models can accurately reflect the heterogeneity of primary tumors at the genetic level. Against this background, researchers, including Robert E. Hyde et al., attempted to systematically evaluate the genomic characterization of PDX models in Non-Small Cell Lung Cancer (NSCLC).

Research Source and Authors

This study was co-authored by Robert E. Hyde, Ariana Huebner, David R. Pearce, and other researchers from various institutions such as University College London, Francis Crick Institute, etc. This paper was published in “Nature Communications,” accepted on March 28, 2024.

Research Methods and Processes

The study was based on the investigation of 48 PDX models from 22 patients derived from the TRACERx research project. The specific steps include:

1. Multi-region Tumor Sampling and PDX Model Establishment: Using a defined sampling protocol, the researchers sampled from multiple regions of the primary NSCLC and created PDX models through subcutaneous injection into NOD SCID Gamma (NSG) mice. The team created PDX models from patient region-specific tumor samples, observing the implantation success rate and the time characteristics of passaging extensions of these models.

2. Whole-Exome Sequencing (WES): WES data of initial generation (P0) PDX models and third-generation passaged PDX models (P3) were compared with their matched primary tumor samples, allowing researchers to analyze genomic heterogeneity between tumor samples and PDX models. WES data processing involved not only sequence alignment and variant detection but also an adaptive mouse reference genome to separate reads originating from the mouse.

3. Genomic Bottleneck and Heterogeneity Analysis: Researchers indicated that the establishment of PDX models often leads to a genomic bottleneck effect, where PDX models typically represent a single tumor subclone, though independently derived models can represent different tumor subclones. This bottleneck limits the ability of PDX models to fully reflect intratumoral heterogeneity.

4. Signal Filtering and Data Analysis: A specifically developed NSG-adapted reference genome was used to effectively filter out background signals caused by the presence of mouse DNA in the samples. Further, various tools and algorithms such as BamCpm, GATK, Somatic Variants were used to ensure high-quality and accurate experimental data.

Main Research Results

1. Establishment and Expansion of PDX Models: The success rate was approximately 33.1%, and 50.0% at the patient level. Multi-region sampling significantly increased the success rate of model establishment for all pathological subtypes. For the initial P0 PDX models, the average growth time was 85 days. With an increase in the number of passages, the growth rate of PDX models accelerated, with a median growth time of 51 days for P1-P3.

2. Histological Consistency: Most PDX models maintained high histological consistency with their primary tumors. However, in a few cases, histological differences appeared in the initial P0 or passaged P3 models, such as increased clear cell components and changes in epithelial cell differentiation.

3. Genomic Analysis: PDX modeling was performed in the form of single or multiple clones. It was found that PDX models showed significant genetic differences compared to their original tumor regions, partly due to the bottleneck effect during PDX modeling, leading to a single subclone dominating the PDX model. Additionally, PDX models still exhibited certain genetic variations and evolution during passages.

4. Genomic Similarity: Mutation distance and copy number distance were calculated, showing that P0 PDX models had the highest genomic similarity to their original tumor regions, whereas P3 PDX models exhibited variations primarily due to genetic evolution during passages.

Research Conclusions and Value

The main conclusions of the study are as follows: 1. Limitations of PDX Models: A single PDX model often cannot fully reproduce the genomic heterogeneity of the primary tumor, especially in the presence of multiple tumor subclones. This bottleneck phenomenon is particularly evident in the early stages of model establishment.

1. Importance of Multi-region Sampling: Multi-region sampling significantly enhances the ability of PDX models to represent intratumoral heterogeneity. Establishing multiple PDX models reflecting different primary tumor regions can help comprehensively capture the intrinsic heterogeneity of tumors.

2. Optimization of Modeling and Analysis Methods: To ensure the accuracy of PDX data, an adaptive mouse reference genome was developed as an innovative tool, improving the accuracy and feasibility of PDX data.

Research Highlights

The innovations in methodology, such as the development of an NSG-adapted reference genome, and the systematic evaluation of genomic fidelity in PDX models, provide significant academic contributions. The conclusions provide new perspectives and evidence for the application of PDX models in preclinical research and personalized medicine.