Transferring proteomic discoveries into clinical practice

Advancing beyond the genome, the field of proteomics now demonstrates great promise for providing improved means for early disease diagnostics, patient stratification, monitoring treatment and predicting relapses and more [1]. To this day, a variety of proteomic technologies has been developed and implemented in the quest of pinpointing candidate biomarker signatures, as reviewed elsewhere [2–5]. For several years, numerous papers have also been published highlighting the discovery of such tentative biomarker signatures in cells, tissues and/or body fluids (for reviews see [1,5–7]). Despite these exciting findings, the subsequent efforts of transferring these proteomic discoveries into clinical practice, providing validated signatures of clinical relevance have yet to deliver [8–12]. The reasons behind this impaired translation are manifold and complex, highlighting the challenges that clinical proteomics must face before its full potential can be explored and exploited.

The poor quality of many early studies has raised concerns as to whether relevant, multiplexed biomarker signatures could be discovered, and the viability of such proteomic efforts are questioned [8,9]. These shortcomings have highlighted the need for relevant study designs, precise sample handling, robust technology platforms for high specificity and sensitivity, as well as stringent statistics/bioinformatics, fueling the design and implementation of adequate standards and strategies [8,9,13,14]. Consequently, as the field has matured, the ‘old’ concept that a condensed panel of biomarkers will provide a much more selective and specific classifier suitable for, for example, disease diagnostics, compared with single analytes, has proven to be valid and will become a gold standard to aspire to [2,5,9,15].

Delineating panels of biomarkers, to be eventually validated for use in clinical assays, is truly a multidisciplinary effort, requiring the entire research team to be deeply involved from start to finish. It is naive to believe that basic scientists alone can define and exploit clinical problems. In a similar manner, it is unlikely that clinicians will grasp the complexity of technical issues and hurdles involved in state-of-the-art proteomic analyses. Hence, being an integrative part of a translational research center and/or effort, where the combined knowledge of leading experts, including both clinicians and scientists, will be instrumental to the outcome of the project.

One of the major hurdles in such translational research to date has been the lack of success in validating outlined biomarker signatures among in-house efforts, as well as laboratories and proteomic technology platforms worldwide [1,5,8,9,12,16]. This is commonly assigned to significant technical and biological variations. An array of actions should be taken to address this multifaceted and intricate issue. One approach would be to promote proteomic technologies capable of handling crude, nonfractionated clinical specimens to reduce sample complexity, instead of using samples that have been tampered with by various prefractionation methodologies, which are associated with inherent issues regarding reproducibility and yield [1,2,8]. However, this will in turn place high, but not unrealistic, demands on the sensitivity, specificity and selectivity of the proteomic assay [1,2]. Once a candidate biomarker signature has been indicated, the identity of the markers should be verified (e.g., by sequencing) to identify the proteins and/or peptides underpinning the signature to make the read-out independent of the measurement platform (i.e., readily allowing the signature to be measured by any suitable methodology). The observed signature(s) should always be benchmarked against the current gold-standard methodology, such as biopsies or a panel of clinical parameters, to demonstrate and assess usefulness and cost–effectiveness.

Furthermore, the protocols for collecting and handling the clinical samples are central to reducing the observed variability. While there appears to be a consensus among the research community that strict guidelines must be adopted, the implementation of such procedures in clinics has proven to be slow. This barrier is mainly due to practical, logistical and economical concerns that must be overcome in order to allow rapid advances in the field. Similarly, the properties of the sample format, such as stability, will also have a major impact on the end results. For example, focusing on blood-based proteomics, plasma samples collected in ethylenediaminetetraacetic acid tubes has in many cases proven to be the preferred sample format [16].

Aiming for an almost noninvasive test procedure, targeting plasma (or serum), which is probably the most commonly available clinical sample format, is an obvious choice. While we agree with the common view that targeting nonfractionated blood-derived clinical samples is challenging with currently available proteomic technologies [1,5,7], we argue against the notion that this sample format is not well suited for most disease-specific biomarker discovery studies [14,17]. In fact, affinity proteomics, illustrated by protein microarrays, has in recent years made significant progress, and is indeed capable of surveying blood-based proteomes even for low-abundance (pg/ml range) biomarkers in non-manipulated samples [2]. It could be argued that how well a plasma proteome reflects a given disease state depends on the specific condition at hand [12], but we and others believe plasma, directly or indirectly, acts as a systemic mirror of the numerous events taking place in vivo[1,5,7,9]. This makes it an excellent candidate for generating molecular patterns in both health and disease. In the process of identifying such disease-associated patterns, it is essential not to limit the proteomic comparison to only two groups, disease versus healthy cohorts, but also to include patients suffering from other conditions and related indications. This will give a more realistic scenario, resembling real-life situation in clinical practice, where the assay must be capable of distinguishing these patient cohorts with high specificity and sensitivity. A comprehensive map, describing the biological variation of plasma proteomes in healthy individuals in detail is still lacking, but will be required sooner rather than later in order to advance further towards transferring proteomic discoveries into clinical practice.

In most current biomarker discovery studies, clinical samples from patients with defined and variably advanced diseases are employed. Aiming for early disease diagnosis, the key question is, how well does this patient cohort and their biomarker signature reflect the realistic clinical situation? In other words, will this biomarker signature also work for early diagnosis? To address these issues, comprehensive prospective studies must be initiated, where numerous patients are continuously monitored over time.

Although academic-based researchers are getting better at protecting the intellectual property rights of key findings, universities should aim to further encourage and support such activities, as this is vital for the transfer of proteomic discoveries into clinical settings. The funding alternatives for exploring such findings in between each discovery phase, often funded by bodies such as governmental research councils, and the end product, funded by corporations, such as venture capital and others, need to be expanded.

Once the biomarker signature has been validated adequately, it is physically ready to be implemented into clinical settings for high-throughput routine analyses. Depending on the proteomic technology used for biomarker discovery, the set-up may be directly translated into clinical practice, while in other cases, an adaptation to another assay system will be required first. While traditional technologies, such as ELISA, are cost effective and easy to implement, the use of more complicated and expensive proteomic technologies should not automatically be abandoned [9,14]. Considering the benefits for patients, in terms of factors such as lowered risk, improved diagnostics and less invasive test procedures, more complicated set-ups are thus justified.

Overall, the opportunities for translational research, progressing from bench to bedside, is finally looking more promising, and adopting state-of-the-art proteomic technologies show improved potential for pinpointing clinically relevant candidate biomarker signatures that can then be transferred into clinical practice.

Financial & competing interests disclosure

This study was supported by grants from Gunnar Nilsson Cancer Foundation, Swedish National Research Council (VR-NT), VINNOVA and the Foundation of Strategic Research (Strategic Center for Translational Cancer Research – CREATE Health [www.createhealth.lth.se]. The authors are inventors and co-owners of patent applications that cover aspects of disease diagnostics for various cancer indications. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.