Challenges for omics technologies in the implementation of personalized medicine

We are currently witnessing the emergence of a new paradigm in health care where the current ‘one size fits all’ approach, in which diseases are treated based on an average drug response, which does not take into account disease heterogeneity in the larger population, is being replaced with personalized/precision/P4 (predictive, preventive, personalized, and participatory) medicine which utilizes a systems biology approach to accurately diagnose an individual patient’s disease at the molecular level, and then use that information to develop individualized targeted treatments specifically for that patient (the right drug at the right dose for the right patient). Unfortunately, there is currently a common public (and often governmental) misconception that data obtained from genomics alone provides all the necessary information to understand the biology of human health and disease and support precision/personalized/P4 medicine (PM). Clearly this is not the case. Genomics only informs on a persons genetic composition (the blueprint), but not which genes are actually expressed (the parts list) (transcriptome) or functional (the end product) (proteome). While human pathologies are encoded by both our genome and our environment, they are all produced by measurable changes in the human proteome. The genome, encoding approximately 23,000 genes, is essentially static and only changes when a mutation, gene methylation, or translocation occurs. By contrast, the proteome is far more dynamic, more complex, and far larger (estimates suggest up to 1 million distinct proteoforms). This is because proteins can undergo alternative splicing, harbor single amino acid polymorphisms arising from non-synonymous single-nucleotide polymorphisms, or undergo posttranslational modifications. Following the initial release of the first drafts of the human genome in both Nature and Science in 2001, it was immediately realized that an in-depth understanding of the human proteome would be required to fully understand the complex biology relating to health and disease, and the Financial Times immediately ran a feature article entitled ‘The Next Holy Grail, Deciphering the Whole Protein Set,’ highlighting the importance of proteomics [1]. Importantly, it must be remembered that almost all current drugs target proteins. Recent proteomic advances mean that the technologies required to achieve virtually full proteome coverage have now matured. At the mass spectrometry level, comprehensive, reproducible datasets can now be generated rapidly using sensitive, quantitative, massively parallel, targeted proteomic approaches. In particular this has enabled comparative studies allowing the identification of potential biomarkers or interactome studies revealing key disease-related signaling pathways leading to potential new drug targets. Thus, a recent publication showed single-shot proteomics could provide evidence for more than 90% of the proteome of a human cancer cell line with >6200 proteins present in 10/10 replicates. In mouse brain tissue, >10,000 proteins were detected in only 100 min, with sensitivity extending into the low-attomolar range [2]. This method (BoxCar) also showed a greater dynamic range in human plasma samples, a key hurdle when using this biological matrix. In another example, the top-down proteomic technologies, in which intact proteins are analyzed, were used to detect and quantify mutation-specific consequences of KRAS biochemistry including post translational modifications in human colorectal cancer patients with the same genotype [3]. At the clinical level, proteomics-based techniques are becoming established. For example, the FDA has approved MALDI-TOF-based proteomics instrumentation and assays for microbiological testing, with more than 2000 instruments now in clinics worldwide, as well as the OVA 1 ovarian cancer assay that was based on a SELDI-derived proteomic biomarker panel. Advances in micropurification techniques, development of fully automated high-throughput array technologies, the availability of highly specific validated monoclonal and polyclonal antibodies [4], the development of technologies capable of preparing highly homogeneous subcellular or tissue preparations (or even single cell analysis (e.g. CyTOF [5]), and advanced bioinformatics further complement the proteomics toolbox. However, even here further improvements can be made as, for example, in the optimization and validation of improved protein extraction protocols allowing even deeper mining of the proteomes [6]. These technologies, in turn, will form part of the omics pipeline comprising techniques such as genomics, epigenomics, transcriptomics, proteomics, peptidomics, interactomics, metabolomics, and microbiomics that together will support PM. There is no doubt such an approach to PM is both viable and effective as evidenced by several recent studies. A comprehensive personal omics characterization (Integrative Personal Omics Profile), which combined genomic, transcriptomic, proteomic, metabolomic, and