NEW YORK – Investigators at the Chinese Academy of Sciences (CAS) and other centers in China and the US have spelled out the genetic alterations and regulatory changes that mark almost 150 primary or metastatic colorectal cancer (CRC) cases in China, identifying protein and protein phosphorylation patterns that appear to track with treatment response or metastasis.
For a paper published in Cancer Cell on Thursday, the team outlined results from genomic, proteomic, or phosphoproteomic analyses of almost 500 clinical tissue samples from 76 individuals with non-metastatic CRC tumors or from 70 metastatic CRC cases, including tumor, matched normal tissue adjacent to the tumor, and corresponding peripheral blood samples.
Senior and co-corresponding author Rong Zeng, a systems biology, life science, and technology researcher affiliated with CAS and Shanghai Tech University, and her colleagues reasoned that multi-omic analyses that encompassed genomic and proteomic clues may contribute to "precise and individualized care," by uncovering features not detected using genomic data alone, particularly when it comes to more advanced stage IV cases.
"Under current recommendations, stage I-III CRC can be treated with a variety of therapies that result in good [five]-year survival rates. However, the progression and survival prediction for stage IV patients represents a highly challenging obstacle for successful treatment selection," they explained. "The relationship between proteome pattern and prognosis can potentially facilitate the precise treatment and evaluation for stage IV patients."
Based on results from these exome sequence, array-based methylation, and liquid chromatography- and tandem mass spectrometry-based protein analyses, together with network analyses, the researchers identified recurrent alterations in the Chinese CRC cases that were more or less common than those previously reported in CRC data generated for the Cancer Genome Atlas or at Memorial Sloan Kettering.
Their data also pointed to three proteomics-based CRC clusters, representing tumor subtypes with varied molecular signatures and prognostic patterns, along with protein phosphorylation features that appear to correspond to metastasis risk and response to drug treatment.
To get a better look at drug responses in vivo, the team developed a machine learning method based on almost two-dozen mini patient-derived tumor xenograft models, including samples from nine CRC cases with matched tumor and liver metastatic tissues available.
"For patients with druggable mutations, we proposed a strategy that exploits the protein-phosphorylation relationship to effectively select the most suitable targeted therapy," the authors reported, adding that "the accumulation of multi-omics data in conjunction with efficient drug testing can establish an accurate index for determining the most suitable drugs for a given tumor type."