Grafting Tumours on Mice Reveals New Lung Cancer Classification and Testing Strategy

Among all the cancer types, lung cancer is one of the deadliest and most common, accounting for one-fifth of cancer deaths and 13% of all cancers. One hurdle for researchers and clinicians is that the proteome, or the complete set of proteins in a system, changes in cancer and remains poorly understood. This is significant given that it is proteins that make up and control the tumour. Strategies to gain insight into these cancers include using mass spectrometry to profiling the proteomes of patient tumours and patient-derived xenograft  (PDX) tumour models. Researchers develop these PDX models by taking a sample of a patient’s early-stage tumour during surgery and grafting it underneath the skin of an immunodeficient mouse. Given the lack of an immune response, the human tumour can grow on the mouse and can be transplanted or passaged again and grow stably on other mice, allowing researchers to study them without having to retrieve patient samples continuously.  

In a recent Nature Communications article, co-supervised by Drs. Michael Moran and Ming-Sound Tsao, first author and recent MoGen Ph.D. graduate Dr. Shideh Mirhadi and others generated 137 PDX models directly from NSCLC patient tumour grafts. They analyzed the proteomes of the PDX model tumours and demonstrated a new tumour grouping method based on proteotypes, which are computationally derived proteome subgroups defined by the patterns of expressed proteins. Through mass spectrometry analysis of the PDX tumours, the team defined five proteotypes associated with different patient outcomes and specific molecular and cellular features.  

These PDX models accurately replicated and resembled the original NSCLC tumours directly from the patients on a molecular and cellular level, they look the same under a microscope. Aspects such as the genetic mutations, the expressed RNA molecules and protein compositions were also the same between the PDX model and the patient, which built up confidence in this methodology’s reliability.  One notable advantage of these models is the easy differentiation between the tumour and stromal cells, with the former being of human origin and the latter originating from mice. Researchers define stroma as non-cancerous cells in the tumour that perform a structural and support role. Usually, it’s difficult to distinguish between the tumour cells and the stroma in tissue samples. However, with PDX models, it is easier to sort it out in a mass spectrometer. Notably, more aggressive tumours with worse patient survival outcomes were the ones that tended to graft and passage successfully in the mice. Essentially, if the lung tumour can grow in a mouse, the prognosis tends to be worse, and the tumour itself is more severe. “The PDX system allows us to focus on the most aggressive, hard-to-treat lung cancers,” notes Dr. Moran, “it can serve as a strong prognostic indicator, as first discovered by my colleague and co-lead on this study, Dr. Ming-Sound Tsao.” 

For the five proteotypes, each had its own distinct protein phosphotyrosine (pY) profiles and patient survival rates. Researchers define pY profiles by the portion of the proteome whose tyrosine amino acids have attached phosphate groups, which could turn on a signalling pathway potentially leading to cancer development if dysregulated. The group measured the pY modifications because defective signalling is a significant cancer component and an important drug target class. Analyzing these proteotypes and determining which proteins contain phosphorylated tyrosines gives insight into which drugs inhibit the phosphorylation to test. Dr. Moran explains that the most relevant aspect—and the heart of the paper—is the proteotypes serving as indicators of patient outcome and determining prognosis based on which proteome subgroup a patient’s tumour falls under. Further bioinformatic analysis of gene expression patterns revealed that each of the proteotypes had distinct sets of multiple active biological and signalling pathways, serving as targets for candidate drugs and therapeutics.   

These NSCLC PDX models provide critical insights and have proven valuable models in cancer research. This publication showcases the importance of analyzing cancers by directly inspecting the proteome since it’s the proteins that form the tumour, manifest cancer and tend to be the actionable drug targets. “This tells us there may be a better way to classify and treat lung cancers,” states Dr. Moran, “when we classify them by their proteomes, we can investigate those different groups and find new treatments specifically tailored for that subtype.” He notes that the PDX models can serve as a preclinical system to test new hypotheses and lung cancer treatments. In the future, this system of taking biopsies, grafting them on PDX models and analyzing them via mass spectrometry could co-exist alongside the microscope for classifying cancers. This research demonstrates how different methodologies provide new avenues for grouping and treating cancers. Dr. Moran noted: “It’s exciting that new technologies are being applied to re-examine cancers and recognize they may be classified and treated in ways we haven’t imagined before.” 

A big thank you to Dr. Moran for contributing to this piece!

Read the article here: Mirhadi, S., Tam, S., Li, Q. et al. Integrative analysis of non-small cell lung cancer patient-derived xenografts identifies distinct proteotypes associated with patient outcomes. Nat Commun 13, 1811 (2022). https://doi.org/10.1038/s41467-022-29444-9