Biomarker Discovery in Emergency Care – A Suggested Development Pathway

Background: Biomarkers are widely used in medical diagnosis with rigorous pathways for the identication, development. However, there are no published pathways for emergency care. Aim: We aim to synthesise the existing literature on biomarker development in order to produce a Biomarker Development Pathway for a physiological biomarker relevant to emergency care. Methods: Medline, BIOSIS, the National Center for Biotechnology Information databases and the Web of Science for relevant articles. Material published on websites was searched using two internet search engines (Google Scholar and Ask). Websites of organisations involved in biomarker development were also ‘hand searched’ for relevant information. Results: We identied 536 papers, of which 14 were relevant to the methodology of biomarker development. One additional paper was found by hand searching from websites. Conclusions: There were no papers that explored the specic requirements for the development of a physiological biomarker in emergency care. The biomarker development pathway published here will guide emergency care researchers in a sequential approach to biomarker development, enable description of the stage of their technology development in grant applications, and give technology development companies a clear route ‘downstream’ to clinical use.

guide emergency care researchers in a sequential approach to biomarker development, enable description of the stage of their technology development in grant applications, and give technology development companies a clear route 'downstream' to clinical use.

Background
Biomarkers are widely used in medical diagnosis, with the term conventionally applied to a "biochemical measurement" (BNP etc). However a biomarker in fact has a much wider de nition as "a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention". 1 Within clinical chemistry, cancer and clinical genetics there are rigorous pathways for the identi cation, development and characterisation of a biomarker. These are seen as a fundamental underpinning for biomarker development and provide the standards by which a potential new biomarker can be judged. The concepts underpinning these pathways seem to be entirely appropriate for emergency care, although in the emergency situation the later evaluation stage probably needs a pragmatic rather than explanatory approach. 2 However there are no published pathways developed speci cally for emergency care and none of the current pathways consider the development of novel physiological markers.
In emergency care measurements of physiology are an important part of the process of diagnosis, whether individually (such as measurement of oxygen saturation) or in combination (such as an 'Early Warning Score'). According to the broad de nition these measurements of pathophysiology can also be thought of as biomarkers. A number of new technologies which provide novel physiological measurements in emergency care are being developed and marketed. Either as single measurements or as patterns of change they are sold as providing valuable extra information in emergency care, however there is no development pathway for a physiological biomarker in emergency care so it can be di cult for a clinician to understand whether or not a new measurement is of any bene t. Lack of a clear development pathway also makes it di cult to researchers to produce the data required for evaluation of clinical utility. This paper aims to synthesise the existing literature on biomarker development in order to produce a Biomarker Development Pathway for a physiological biomarker relevant to emergency care.

Methods
A literature search was undertaken using the search strategy in Table 1. Abstracts of papers identi ed were read for relevance and the references of relevant papers were used to identify further relevant publications. Searches were undertaken of Medline, BIOSIS, the National Center for Biotechnology Information databases and the Web of Science. Material published on websites was searched using two internet search engines (Google Scholar and Ask). Websites of organisations involved in biomarker development were also 'hand searched' for relevant information. Abstracts were scanned to identify those that were relevant.
There is no quality grading system for this type of publication, so relevant papers were simply categorised into (1) those that were the result of a formal consensus process (with a de ned consensus methodology used), (2) those that were the result of an informal consensus process (clinical conference or other meeting), and (3) publications of an individual/group.

Results
The search strategy identi ed 536 papers, of which 14 were relevant to the methodology of biomarker development ( Figure 1). There were no papers which explored the speci c requirements for the development of a physiological biomarker in emergency care. One additional paper was found by hand searching from websites.
Of the 15 papers relevant to the methodology of biomarker development there was one paper that was the product of a formal consensus process, three that were the result of an informal consensus, and 11 that were publications by an individual/group ( Table 2).
The pathways found were then amalgamated and the language changed to be appropriate for emergency care ( Figure 2).

Discussion
When considering the evaluation of a biomarker the ideal pathway depends on how the marker might be used in clinical practice. There are three basic ways in which a biomarker could be useful in emergency care with each of these different types of biomarker requiring a slightly different development pathway: a) As a Diagnostic Biomarker -to de ne whether or not the patient has a speci c condition. b) As a Co-diagnostic Biomarker -to risk stratify patients with a particular condition, or to de ne a subgroup that might bene t from a particular intervention. c) As a Treatment Target Biomarker -to de ne a goal for therapy in a speci c condition.
The conventional language used to describe a biomarker development pathway is heavily in uenced by the biochemical methods that are usually used. The stages are: 1) Prototype Design and Discovery -the technological development required to measure a new parameter.
2) Biomarker Identi cation -comparison of a broad range of measurements using the new technology between a small group of 'disease' (usually advanced stage disease) and 'normal' subjects in order to identify parameters (potential biomarkers) that might differentiate.
3) Biomarker Quali cation -the quanti cation of the relationship between biomarker and disease (to see if the potential biomarker will 'qualify' as clinically useful).
a. Retrospective studies (maybe from a disease registry) looking at the biomarker in patients with known outcomes (disease or normal).
b. Prospective studies in a representative group (all disease severities) using biomarker to predict outcome (disease or normal).
4) E cacy Trials -to quantify the effect of biomarker use on outcomes.
Prototype Design and Discover is often undertaken by technologists without medical input. There are some notable exception in emergency care where emergency physicians are involved at the forefront of technological development 3 , but it is unfortunately more common for an emergency physician to be Biomarker Identi cation is the process of taking a rst look at the data produced and identifying any potentially useful variables. This stage in the Biomarker Development Pathway has its origins in biochemical and genetic technologies, where advances in technology in genomics, proteomics and metabolomics now allow the measurement of many thousands of variables. It is impossible to evaluate all of these variables, so the process of Biomarker Identi cation (with samples from patients with advanced disease) gives an initial ltering, usually based on quantitative methods but without the ability to look at aspects such as the clinical signi cance of the associations found. At this stage there is also an evaluation of both biological plausibility (does the putative biomarker t with what we think we know of the disease process) and clinical plausibility (is there the potential for the putative biomarker to be useful in clinical care). With so many variables being measured it is certain that false positive association will be found at this stage.
Biomarker Quali cation is the process by which the relationship between the proposed biomarker and the speci c disease is quanti ed. The potentially confusing term 'Quali cation' is derived from the biochemical literature, being used as the biomarker was said to 'qualify' for clinical use. This is observational study with well-de ned inclusion criteria and de nition of the endpoints of 'disease' or 'normal'. These studies are usually in two stages, with a retrospective case-control analysis of whether 'real world' patients (not just those with advanced disease) know to have the disease have different levels of the biomarker from normals, followed by a prospective analysis (cohort study) of whether the biomarker can identify the disease. E cacy trials (randomised Controlled Trials) are the nal part of the Biomarker Development Pathway, and are unfortunately often missed out. However tempting it is to market a product that can "make a diagnosis" after the biomarker quali cation procedures it is not su cient to simply show that there is a good relationship between the Biomarker and the disease. In order to in uence clinical practice it must be shown that the use of the Biomarker improves patient outcome (in the same way as a new drug would be expected to show a bene t before being marketed).
The terminology used in these four key stages of biomarker development can be altered to t with emergency care research as shown in Figure 2. This Biomarker Development Pathway integrates the technology and clinical elements, with a combination of medical device developments and clinical studies at each stage. This emphasises the need for close collaboration between emergency physicians and those developing the methods and technologies, this collaboration being more likely to be effective if it is initiated as early as possible within the pathway. For example if a company rst meets a clinician with a fully formed product it is unlikely that it will have been optimised for the clinical situation.
The different stages of biomarker development have different potential sources of funding (Table 3). There is a complex funding landscape as the development of a biomarker crosses the intersections between Industry / Public funding and basic science / clinical research. Early contact with the relevant research program manager is the best way to navigate between the various organisations -there may initially seem to be both overlapping functions and gaps, however the different organisations are increasingly well co-ordinated and will either collaborate or signpost the researcher to the most appropriate place.

Conclusion
There were no papers that explored the speci c requirements for the development of a physiological biomarker in emergency care. The biomarker development pathway published here will guide emergency care researchers in a sequential approach to biomarker development, enable description of the stage of their technology development in grant applications, and give technology development companies a clear route 'downstream' to clinical use.   Tables   Table 1 Search Table 3: a) Choice here depends on the level of engineering risk and the breadth of applicability.
b) If the technology development is more risky/time consuming/costly than the initial clinical studies funding is more likely to be DPFS. Development of late stage technology is the focus of DPFS.
c) A 'proof of concept' study would probably be DCS, a multi-center follow up study would be EME and a trial with health economic analysis would be HTA. Figure 1 Screening process Figure 2 Biomarker Development Pathway for Emergency Care (technology development steps in grey)

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. PRISMA2009checklist.doc