I

Preliminary Biomarker Discovery (Primary Cohorts-Internal)

Figure 3 Biomarker development and translation. Biomarker translation from discovery to clinical use involves five general steps or phases. Biomarker candidates are initially discovered in a primary (internal) cohort and confirmed in a secondary (external) cohort through clinical observations. Biomarker candidates that have the best fit-for-purpose and also satisfy a clinical need then enter prospective phase I and II clinical trials. Once a biomarker is used in these early clinical trials to demonstrate safety and efficacy, they may be considered "initially validated biomarkers." Following large-scale clinical trials (phase III), and once biomarkers have been used for decision making, they may be considered valid and may be adopted for clinical use. Biomarker assessment continues in postmarket phase IV trials. SOP-driven analytical quality assurance is performed throughout all processes of biomarker translation.

The validation process whereby biomarkers are translated from discovery to clinical use should be customized according to biomarker type, use, variability, and prevalence. However, the general process for validating any biomarker is the same. Before any biomarker can be applied clinically, it is subjected to analytical/method validation and also clinical validation. Biomarker validation can be performed in five general translational steps: preliminary biomarker discovery, biomarker discovery, safety and efficacy clinical trials, large -scale clinical trials, and continued surveillance (Figure 3).

Clinical biomarkers are validated in retrospective or prospective analyses and biomarker trials, or drug trials [12]. The validation process should reflect the clinical performance of the biomarker(s), based on existing clinical data, new clinical data, literature review findings, or current clinical knowledge [12]. Moreover, it should be an evidentiary and statistical process that aims to link a biomarker to specific biological, pathological, pharmacological (i.e., drug effect), or clinical endpoints [22].

Internal Validation

During the biomarker discovery phase, the main focus is to identify biomark-ers that distinguish between the treatment and control groups or correlate with the clinical observation of interest. Prior to this process, and depending on the sample size, plans can be made to allocate the subjects into two individual cohorts for the purpose of internal validation. It is important, however, to distinguish this allocation/splitting process from that used for the purpose of internal validation of classifiers:

• Internal validation of candidate biomarkers identified in the discovery cohort

• Use of different platforms

• Use of different statistical methods

• Internal validation of classifiers

• Some form of cross - validation method

There are several different approaches to separate the initial pool of patients or samples for internal validation and creating classifiers; the traditional and alternative approaches are outlined here. Traditionally, a discovery and a validation cohort are created (Figure 4). Genomic biomarker candidates may first be identified in the discovery cohort using microarray analysis, for example. One way of identifying a panel of biomarkers from the candidates is by use of classification methods. The samples from the discovery data set can be split into a training set, used to find a classifier, and a test set, used for internal validation of classifiers. The classifier is an equation combining the expression values of some of the candidate markers that best distinguish the treatment from the control group. Once the classifier has been developed, the panel of biomarkers may be validated again in the validation cohort before the external validation phase is carried out. In this sense, the data obtained in the validation cohort serves purely as an additional test set.

Although the traditional internal validation model is simple and logical, it may not be the most applicable strategy in the real world, given the complexity of most biomarker research today. There are two potential weaknesses to this approach. First, separating available samples into discovery and validation cohorts from the outset might unintentionally restrict the use of the samples or data to their arbitrary labels. In reality, during the developmental phase of biomarker research, different statistical analyses are often carried out to identify potentially useful biomarkers. Depending on the types of comparisons being made, a sample (or portion of data) could be used for discovery pur-

Project Management Made Easy

Project Management Made Easy

What you need to know about… Project Management Made Easy! Project management consists of more than just a large building project and can encompass small projects as well. No matter what the size of your project, you need to have some sort of project management. How you manage your project has everything to do with its outcome.

Get My Free Ebook


Post a comment