Our first installment of this Proteogenomics Blog Series defined proteogenomics as combining nucleic acid sequencing data with information on the proteins that are produced as a result of those sequences. This approach provides a deeper understanding than what is possible looking at DNA, RNA, or protein alone.
The National Cancer Institute (NCI) defines proteogenomics as “the study of how information about the DNA in a cell or organism relates to the proteins made by that cell or organism. This includes understanding how genes control when proteins get made and what changes occur to proteins after they are made that may switch them on and off." NCI also addresses the promising role multiomic studies will have in future disease research: "Proteogenomics may help researchers learn more about which proteins are involved in certain diseases, such as cancer, and may also be used to help develop new drugs that block these proteins."
In this chapter, our focus will be on the impact of proteogenomics in oncology research. Proteogenomics is at the bleeding edge of precision medicine and in cancer research, playing a new role in both detection and treatment.
Molecular analysis of cancer started in the 1990s, with research indicating that overexpression of HER2 in women’s breast cancers made them responsive to Herceptin. The NCI joined the NHGRI to create The Cancer Genome Atlas (TCGA) database that mapped mutations in patients’ tumors, looking for commonly occurring changes.
TCGA has been highly successful, leading to the molecular classifications of tumors into subtypes that respond similarly to the same treatment approaches. That said, genomics doesn’t account for 100% of cancers, and there are non-responders who harbor the same mutations as responders. Additional information is needed — and it's now coming from the workhorse of the cell, proteins.
Screening protein levels in cancers is essential because the level of the RNA transcripts does not always predict the level of protein that is being made in the cell. In one type of kidney cancer, both phosphorylated MAPK1 and CDK1 kinases are upregulated at the protein level, but not at the RNA level. Insights like these are important to understanding the true mechanisms of cancer.
Until recently, protein analysis was either very low plex, very expensive, or both. With new developments in mass spectrometry, the ability to analyze nearly the whole proteome has improved the plex. However, the price remains fairly high, limiting the number of samples one can analyze. Additionally, if the sample is a biopsy, cells are not always easy to obtain or abundant. This further limits the scope of many studies.
High plex plasma-based protein biomarker assays like Olink’s Proximity Extension Assay (PEA) and SomaLogic’s SomaScan have truly revolutionized proteogenomics research. One blood draw can provide sufficient material for DNA, RNA, and protein analyses. These new tools have opened the door for plasma-based protein signatures that can deliver diagnosis, prognosis, and new treatment options.
The Cancer Moonshot program narrows the focus of proteogenomics to attaining actionable information that leads to new drug targets to help treat and/or diagnose cancer. Of FDA’s approved cancer drugs, less than 200 are proteins, but that is about to change.
The Clinical Proteomic Tumor Analysis Consortium (CPTAC) integrates proteogenomic data from more than one thousand cancer patients across ten cancer types in effort to identify druggable protein targets alone or in conjunction with the loss of tumor suppressor genes. As shown below in Figure 1, this approach has shown great success in identifying novel targets, specifically in the case of combined protein overexpression/hyperactivity with genomic alterations.
Fig. 1: Potentially druggable, non-pan-essential targets shared by at least five cancer types.
Inflammation has long been suspected to have a role in increased cancer risk, but a large-scale analysis to evaluate a cause-and-effect relationship has been lacking — until now. In a meta-analysis of nearly 60 thousand people, researchers used proteogenomics to assess the role of 66 inflammatory markers in adult cancer risk. This research identified:
Four causal relationships, with only increased levels of pro-adrenomedullin being strongly associated with an increased risk of breast cancer.
Increased levels of circulating IL23R linked to an increase in pancreatic cancer risk.
Decreased levels of prothrombin and IL1R-like were associated with decreased risks of triple-negative breast cancer and basal cell carcinoma, respectively.
None of the other 62 circulating inflammatory markers demonstrated a significant relationship with altered risk of any of the cancers studied. With this information, researchers can focus their attention on other potential causative factors rather than investigating individual inflammatory markers.
Using meta-analysis of proteogenomic data from the FinnGen cohort, the UK Biobank, and seven genome-wide association studies (GWAS), scientists have identified 13 proteins associated with colon cancer (CRC) risk. Upregulation of only two proteins, GREM1 and CHRDL2, is associated with increased CRC risk, whereas the other eleven proteins display decreased expression correlated to greater CRC risk. The strongest evidence was found for GREM1, CLSTN3, CSF2RA, and CD86, which are expressed in stem cells, epithelium, and monocytes. Four proteins, POLR2F, CSF2RA, CD86, and MMP2, are existing drug targets for other cancers and autoimmune conditions. These are potential candidates as novel therapeutics in colon cancer.
Proteogenomics was used to identify 20 novel factors associated with aggressive and early onset prostate cancer. Researchers identified PPA2, PYY, and PRSS3 as biomarkers of aggressive disease. Early onset associated proteins found are EHPB1, POGLUT3, and TPM3. Interestingly, MSMB was discovered as a marker that correlates inversely with both early onset and aggression and can serve as a marker for benign regions of the tissue. Another exciting finding from this study is that, of the 20 proteins found to be associated with increased prostate cancer risk, ten have available drugs on the market.
In this oncology-focused article, we have covered cases of proteogenomics research leading to the discovery of diagnostics, biomarkers for risk stratification, existing therapeutics, and novel drug targets. The UK Biobank data generated using Olink’s PEA technology featured prominently in many of the aforementioned studies.
At the date of this blog’s publication, the UK Biobank has just signed a deal to conduct the world’s largest human proteome study (600,000 samples) using Thermo Fisher Scientific’s Olink platform. With the development of high-plex protein biomarker assays combined with decades of available GWAS data, new discoveries will continue to emerge from these large datasets.
Eager to learn more? Stay tuned for the next article in our Proteogenomics Blog Series, where we'll dive into the role of proteogenomics technologies in neurological research.