Demand is growing for biomarkers that can quickly assess whether a drug is modulating its intended target. However, collecting and storing blood or other biological samples needed to find those markers has its tricky side, with nucleic acids and proteins each posing unique challenges. Amgen's Scott Patterson, Senior Director for Medical Sciences, described some of these in his presentation at the G.O.T. Summit. 

One of the earliest types of clinical biomarkers reflects pharmacodynamics (PD), or how a drug acts in the body. Such markers can be protein or DNA levels in blood or other fluids. As with any other type of biomarkers, the biggest challenge has been getting really informative markers that are also reliable. Today, drug developers also want to be able to use markers in both preclinical and clinical studies. The catch: Methods of sample collection and preparation have to be consistently maintained between the research lab and the clinic. Of course, that's much harder than just controlling them in the research setting.  

There are some ways to circumvent environmental variables. For example, a clinic might transport a cold, frozen, or fixed sample to the analytical lab to be assayed. Or, perhaps at least a portion of the assay could be conducted at the clinical site, eliminating the pitfalls of transportation. It is critical to examine whether the act of collection itself induces changes in the analytes to be measured. Variables such as over/underfilling a collection tube or leaving a tourniquet on too long can affect the composition of blood samples. If a collection tube is not prechilled, or if a very small sample thaws out, then RNA integrity could be compromised. A delay in sample assaying following collection could also have a deleterious effect on results. If these types of effects can be understood, then assays could be modified accordingly. Said Patterson, "Experimental noise can be related to how a sample was prepared—artifacts introduced during the collection process can affect results." 

For example, incubation can introduce significant changes to DNA. Incubation-induced changes to a sample from timepoint A to timepoint B can, in fact, exceed the genetic differences between samples from two different donors. As Patterson pointed out, a nucleic-acid sample can change over the course of 24 hours to the extent that it is more similar to a sample obtained from a different donor than to its original state. 

Low molecular weight (LMW) analysis of serum and plasma is a prime example of how preanalytical variables can come into play. Time-course studies in serum have shown that ex-vivo proteolysis plays a big role in sample variation. Amgen's experience, which is supported by the literature, suggests that serum samples cluster to collection parameters more closely than to the disease state. "Molecular weight analysis is an extreme example of how results can become skewed due to factors introduced by sample preparation," said Patterson. LC-MS/MS analyses have identified these LMW "features" as proteolytic fragments of common blood proteins—basically, fragments of host response. Such fragments may not be useful biomarkers, since they are not disease specific.  

LMW serum profiling thus becomes an attempt to reproducibly monitor an uncontrollable process. Following such efforts, Amgen concluded that it is probably too susceptible to non-biological biases to produce reliable results in clinical studies. Something that is as seemingly insignificant as which position the patient was in during phlebotomy (e.g. supine, sitting, or standing) can skew results. While disease-specific fragments are probably present in the sample, host-response profiling can be performed at higher sensitivity and dynamic range by other methods. 

Whole blood is the best source for target modulation biomarkers, assuming the pathway of interest exists in blood cells. Detection of secreted proteins following a biochemical challenge is one of the most commonly used assays for measurement of therapeutics that block inflammatory processes. While secreted proteins have proven useful, Amgen has tried to develop methods that would allow the identification of biomarkers that are more proximal to the receptor.  

One such method is multi-parameter flow cytometry, which can replace and/or augment traditional whole blood assays by providing more proximal readouts of cytokine response. Whole-blood stimulation can be conducted right at the clinical site. Amgen stimulated whole blood with TNFalpha or LPS and incubated it at 37 degrees Celsius or room temperature overnight. It turns out some cell types respond differently following overnight storage. In this case the monocytes underwent a drastic change following overnight storage as opposed to the T cells and NK cells. Patterson emphasized, "You must understand your sample and therefore which components of your assay it will be necessary to pay close attention to."  

How samples are collected, stored, and transported prior to assay is crucial to obtaining sound results. As demonstrated, the very act of collecting the sample can alter its analytes. The beauty of bringing validated assays from research to the clinic is that it will allow for sample analysis (or at least a portion thereof) to be conducted on site, eliminating the need to store and transport the sample with the attendant undesired effects or to be aware of the changes that could occur and plan accordingly.