Precision medicine has been described as "the right drug at the right time for the right patient," which recognizes the obvious fact that some individuals respond to treatment while others do not. When precision medicine became a thing around 20 years ago, its mention was often accompanied by somber predictions of "the end of the blockbuster model", which wasn't even remotely possible given today's health economics. Now precision medicine concerns itself more with selecting from among available medicines already approved for large populations rather than drawing from a long menu of orphan drugs approved for smaller and smaller patient populations.
Essential diagnostics
Toward that end, precision medicine is primarily about classifying patients. Today that means diagnostics, principally molecular diagnostics of the 'omics type. Stratification involves correlating a patient's relevant biomarkers (molecular plus imaging, etc.) with historical (or in some cases, non-clinical or experimental) responses to treatments.
Genomics represents the potential for a particular phenotype, while proteomics expresses that potential in action. Proteogenomics—the analysis of genes and proteins simultaneously, and preferably from the same sample—provides both pieces of information while shedding light on the interactions between genes and proteins.
The overriding issue, or as chemists would call it the rate-limiting step, is the proteomic part: While it's possible to make as many copies of genes as we like through PCR-type amplification, protein amplification is not yet possible, so whatever the sample contains is all you get.
Commercial precision medicine today relies almost entirely on genomics, which guides not just choice of drug but also helps caregivers monitor side effects, disease progression, and response to treatment. Genomics increasingly, perhaps exclusively, guides new drug development. The leukemia drug imatinib (Gleevec), and the lung cancer medicines Iressa and Tarceva, were developed as more-or-less "personalized" treatments because they are only prescribed to patients carrying a specific genetic signature.
Given the lack of a protein multiplier or amplifier, proteomics will experience somewhat greater difficulty entering wide-scale precision medicine than did genomics, and proteogenomics will have an even harder time.
The good news for proteomics is that MALDI-TOF instruments, which provide gentle ionization for delicate proteins, are now available on the benchtop and more than 5,000 have been installed in clinics worldwide. The bad news is that outside of academic proteomics and neonatal screening almost nobody has heard of MALDI-TOF.
Proteomics is also limited by low sequence detection coverage and the inefficient detection of somatic mutations. Proteins exist in several isoforms, each of which (at least during discovery) must be assumed to be relevant. Some, like phosphorylation, are reversible. Knowing which isoforms do what, a research project in and of itself, is required before claiming anything about proteomics, much less proteogenomics.
Greater than the sum
Yet research continues on the likely assumption that proteogenomics, compared with either proteomics or genomics, is greater than the sum of its parts. Robert Georgantas, Ph.D., Senior VP of Research and Translational Science at Biodesix, is certainly of that mind.
"We tend to think of genomics, transcriptomics, and proteomics in the context of the simple DNA-to-RNA-to-protein paradigm that we learned in basic biology," Georgantas explains. "But in reality, the transition from DNA to RNA to protein is highly complex with many steps, each potentially changing what we see in the end-protein phenotype. Proteogenomics has the potential to explain and unify this complex pathway."
Using information from the level of DNA and RNA ensures that all expressed proteins important to a disease condition are detected. "Likewise, using proteomics information informs as to which DNA mutations and RNA splice variants are truly involved in a disease."
Biodesix is at the forefront of 'omics as it relates to medicine. The company offers genomic, transcriptomic, and proteomic commercial tests, with additional assays in the pipeline for assessing immune response, risk of recurrence, and treatment guidance for lung cancer. The company also offers, through its Biopharma Services business, genomic and transcriptomic next-generation-sequencing assays spanning targeted panels to whole genome/transcriptome assays. For these, Biodesix takes advantage of advanced mass spectrometric methods such as deepMALDI, PASEF-timsToF LC-MS, and an unbiased proteomic method from proteomics specialty firm Seer.
The bottleneck
Next-generation sequencing has enabled exhaustive assays of the genome (whole genome and whole exome) and transcriptome (RNA-seq), and these assays are getting faster and cheaper by the day. Only recently have MS-based proteomic workflows come close to matching this level of throughput.
"Previously proteogenomic studies were constrained by the speed and reproducibility of peptide-based LC-MS," Georgantas explains. This throughput constraint prevented assaying the large samples sets required to statistically power proteogenomic informatics. "But recent advances in instrumentation, for example Bruker's Bruker timsToF™, and protein fractionation/isolation methods have enabled high-throughput, reproducible proteomics. In short, we are now able to generate the 'Big Data' sets needed to cogently explore proteogenomics."
The other inevitable hurdle with multiplexing high-content assays is processing and interpreting large data sets. With proteomics, even getting these data sets was a challenge.
"We need bioinformatics and computational biology methods to link genomic, transcriptomic, and proteomic data into proteogenomic structure. In other words, we still need good tools to make sense of all of this the data," Georgantas adds.
"The main challenge for proteogenomic development of diagnostic tests is the availability of sample sets large enough to power proteogenomic studies, with the correct sample collection types to allow genomic, transcriptomic, and proteomic assays from the same subject and, preferably, the same sample," Geogantas tells Biocompare. "We are fortunate at Biodesix to maintain a lung cancer biobank in excess of 140,000 samples that can be leveraged for future proteogenomic studies."
Proteogenomics is new, exciting, and useful but it will take quite a bit of development and industrialization before your primary care doctor uses it or orders a proteogenomic test similarly to ordering one for cholesterol or thyroid hormone.
"Adoption of proteogenomics will require demonstration of clinical utility to detect disease, diagnose disease, provide treatment guidance, and perhaps do some other things. A promising technology is just a promise until it shows concrete utility to assist a clinician," Georgantas says.