NEW YORK – A team from Baylor College of Medicine, the University of Texas MD Anderson Cancer Center, and the University of Alabama has uncovered protein-based subtypes for cancers originating at five sites in the body, including subtypes that turned up in prior expression-based analyses of the same cancer types.
"[W]e applied data from molecular profiling technologies to sort out protein alterations that could be linked to cancer," senior author Chad Creighton, co-director of cancer bioinformatics at the Baylor College of Medicine and researcher in bioinformatics and computational biology at MD Anderson Cancer Center, said in a statement.
"When looking at mRNA data, it's assumed that they represent what happens at the actual protein level, but this is not always the case," he explained. "Proteins can be modified after they are synthesized, for example, and this is something mRNA data would not show."
Using mass spectrometry-based proteomic profiles and RNA sequence data generated by the Clinical Proteomic Tumor Analysis Consortium (CPTAC) or other studies, Creighton and his colleagues analyzed 532 tumors from individuals with breast, colon, ovarian, renal, or uterine cancer, identifying protein-based tumor subtypes within and across the five cancer types. Their findings appeared online today in Nature Communications.
"We found that many of the subtypes that we had found based on mRNA observations that we made before held up when we looked at the protein level," Creighton explained, though he and his co-authors cautioned that there were "some notable differences between protein-based and mRNA-based results as well."
In an effort to complement tumor subtype clues gleaned from past studies, the researchers brought together proteomic data representing 100 uterine corpus endometrial carcinomas, 110 clear cell renal cell carcinomas, 125 breast cancer tumors, 100 ovarian cancer cases, and 97 colon cancers previously profiled for the National Cancer Institute's CPTAC project.
"Whereas the initial CPTAC-led marker analysis studies were each focused on a specific cancer type," the authors noted, "our present study aims to define pan-cancer proteome-based subtypes that would transcend tissue-based type across the CPTAC cohorts."
From these and other data, the team tracked down and validated 10 proteomic subtypes, comparing them to mRNA-based subtypes ascertained using data from large efforts such as the Cancer Genome Atlas project.
That protein-based tumor subtype set included three subtypes not found with gene expression data, for example, including tumors marked by proteins involved in Golgi apparatus function, endoplasmic reticulum activity, and a hemoglobin complex, respectively.
Creighton noted that four more of the proteomics-based subtypes did not appear to stem from cancer cells, but from the cells and tissues in the tumor microenvironment. Among them were two immune-related subtypes: a T cell-rich immune subtype and a subtype marked by proteins from the complement activation adaptive immune pathway, which has been linked to cancer-related inflammation.
"Activation of the complement system is an important component of tumor-promoting inflammation, which in turn has an important role in carcinogenesis and cancer progression," Creighton and his co-authors wrote, "and which involves macrophages and neutrophils."