Skip to main content
Premium Trial:

Request an Annual Quote

MD Anderson's Mills Building Cancer Proteome Atlas Using RPPA Data from 70,000-plus Patient Samples


Led by Gordon Mills, chair of systems biology at the University of Texas MD Anderson Cancer Center, a group of researchers is developing a database containing proteomic characterizations of thousands of cancer patient samples and cell lines.

Named The Cancer Proteome Atlas, the database was launched in April and currently contains proteomic profiles of 4,495 tumor samples and more than 500 cell lines generated using reverse phase protein array technology. According to Mills, he and his colleagues have thus far collected RPPA-based proteomic data on more 70,000 patient samples, with primary areas of focus including leukemia as well as lung, head and neck, ovarian, endometrial, and breast cancer.

In addition, the researchers have performed a variety of large-scale screens to investigate the effect of drugs and other agents on the protein profiles of various cancer samples. Ultimately, Mills told ProteoMonitor, they plan to move the project into a "CLIA environment" where it can help "direct patients for targeted therapy."

The nearly 4,500 patient samples in the database at present have come primarily from the National Cancer Institute's Cancer Genome Atlas consortium, of which Mills is a participant. He and his team are currently preparing a paper integrating the proteomic data from these samples, he noted, calling it "a proteomic view of the Cancer Genome Atlas."

"What we are doing is integrating the proteomic analysis across 11 different tumor lineages and using that to cluster and characterize the different diseases and also [to identify] the proteomic clusters that transcend diseases," he said.

Additionally, Mills said, the researchers are performing analyses to determine protein signaling pathways that are differentially activated both within specific cancers and across multiple diseases, as well as to investigate levels of specific, targetable proteins.

This latter effort, he noted, could prove particularly useful for setting cut-offs for biomarker research – a key challenge for the field. "We can use the fact that we have run 4,500 patient samples to set a cutoff and say, [for instance] 'This really represents a high PI3 kinase,'" he said.

Of the 70,000-plus samples the group has run, roughly 6,000 of them have come from the TCGA consortium. This includes around 400 clear cell renal cell carcinoma samples analyzed as part of a TCGA study published last week in Nature. Measuring 180 proteins on an RPPA platform, Mills and his team identified several proteins linked to patient survival, finding specifically that reduced levels of AMP-activated kinase and increased levels of acetyl-CoAcarboxylase correlated with poorer outcomes.

Based significantly on this result, the TCGA researchers were able to link a metabolic shift toward increased fatty acid synthesis to poor patient outcomes, Mills said. He added that, of the four measures the researchers used to identify prognostic signatures for the disease – mRNA, miRNA, DNA methylation, and protein – the proteomic profile demonstrated the strongest correlation with survival.

Mills also led the proteomics portion of the TCGA's recent breast cancer study, which was published in Nature last September (PM 9/28/2012). In that work, the MD Anderson team ran 403 breast cancer samples on RPPAs, measuring levels of 171 cancer-related proteins and phosphoproteins, and identifying a distinct proteomic-based cancer subtype not detected by the study's genomic analyses.

That finding has since been shown to "be extremely robust across multiple datasets," Mills said, noting that in the study the researchers were "clearly seeing a new subtype of breast cancer that we hadn't characterized before."

The Cancer Proteome Atlas is intended to be an open-ended project, Mills said, adding that he and his colleagues will perform RPPA analysis on samples from "anyone in the world." Indeed, he said, roughly half of the 70,000 samples they have analyzed thus far have come from outside MD Anderson. The researchers don't, however, do fee-for-service work, meaning that analysis of outside samples "needs to be part of some sort of interaction on an academic level," said Mills.

Regarding his aim to add a CLIA component to the effort that would enable the researchers to use their RPPA analyses for guiding patient therapy, Mills cited work by George Mason University researchers Lance Liotta and Emanuel Petricoin as examples of the direction that research might take.

Liotta and Petricoin "are already running clinical trials based on the [RPPA] platform for patient selection, and they are doing a great job of it," he said.

In fact, the GMU researchers and their collaborators presented data at the American Society of Clinical Oncology's annual meeting last month in which they used a variety of molecular profiling techniques, including RPPA, to guide treatment in 25 metastatic breast cancer sufferers (PM 6/7/2013).

In 13 of these 25 subjects, the molecularly guided therapies extended progression-free survival by more than 30 percent compared to the patient's last treatment regimen, a result that Nicholas Robert, study co-author and an oncologist at Virginia Cancer Specialists, told ProteoMonitor well surpassed the researchers' criteria for success.

Also last month, Theranostic Health, a start-up launched by Petricoin and Liotta to commercialize their RPPA technology, introduced its first commercial assay – the TheraLink HER Family Assay, an RPPA-based test intended as a supplement to conventional HER2 testing for guiding therapy in breast cancer patients (PM 5/3/2013).

Beyond the RPPA research, Mills and his colleagues are working to add other forms of molecular data to their analyses.

"I run our mass spec core; I do an enormous amount of genomics; I do an enormous amount of transcriptomics," he said. "We try not to be a one-trick pony and work across the different platforms, because we think that is where you are going to get the most information. We look at [proteomics] as adding to analysis of DNA and RNA, not replacing it."

Integration of different kinds of omics data remains one of the more daunting challenges facing such work, Mills said, noting that his team is currently developing an initiative specifically aimed at tacking this problem.

"The challenges of moving across [omics] platforms are major, and [the field] is barely in its infancy," he said. "There are very few people who have been able to do this in an efficient manner."