The Importance of Sample Collection and Handling Methods in Biomarker Research

---

Despite its huge potential being touted enthusiastically by scientists, the media and regulatory bodies with equal degrees of optimism, the relatively new field of biomarker research still faces enormous and complex challenges. Because the progress that has been made in a relatively short period of time is remarkable, it is therefore all the more surprising that one of the field's most basic requirements — access to high quality clinical samples collected in a consistent manner — should remain a significant obstacle. Nevertheless, this is the case.

The collection of blood samples is a routine aspect of clinical practice and research, yet the specific methods used for this collection vary considerably from centre to centre. Some centres centrifuge immediately after blood is drawn, while others allow considerable time to pass before samples are spun down. Centrifuge speeds and temperatures can vary widely. Samples may be kept refrigerated (or on ice) throughout the handling process, or they may be left at room temperature. Samples may be frozen quickly or remain unfrozen for long periods. Protease inhibitors (and there are multiple options for these) may or may not be used, and there are several options for anticoagulants.

All of these sample handling options are assumed to have little effect on the results from routine clinical chemistries, which presumably explains why more rigorous standard and best practices for sample handling have not been introduced. Unfortunately, proteomics research does not appear to enjoy the same measure of flexibility: the methods used to collect, handle, ship and store samples can have a dramatic impact on the quantity, or even the presence, of a given protein within these samples.

Immediately after a sample is drawn, a cascade of biochemical processes take place that cause proteins to break down and be modified. That is, the proteins that were present in the blood at the moment of blood draw will not be precisely the same proteins that will be present when the sample is subsequently analyzed. Freezing the sample will block these processes but, generally speaking, the more time that passes between sample draw and storage, the greater the changes to the sample's protein profile.

The consequences of this and related phenomenon can be dire. Many studies compare the blood from healthy volunteers to blood drawn from patients. However, samples are usually drawn from healthy volunteers in a relatively controlled setting in comparison to the practical realities of drawing samples from patients. Clinic staff dealing with a gravely ill patient will likely have other things on their mind than transporting a patient’s sample to a centrifuge or freezer within a given time window, while a study established specifically to collect healthy volunteer blood will not have the same logistical issues. That is, if the two sets of samples were frozen at disparate time points, and time-to-freezing has an effect on the quantities of protein in the blood, then any observed differences in protein profiles between these samples could be entirely spurious.

Surprisingly, few fluid sample banks keep records of how each sample was handled. Although all banks have standard operating procedures, most allow some measure of flexibility in the timing of procedures, and many have no clear guidelines on issues such as time-to-freezing and time-to-spin. Furthermore, many centres keep no record of occasions when standard operating procedures (SOPs) were violated during the sample collection process. Yet without careful records and adherence to procedures, the usefulness of the samples may be severely compromised. Indeed, if there is one piece of practical advice that could be given to aspiring sample banks it would be to record what happened to each sample after it was drawn as carefully as possible. Obviously this is not a trivial recommendation for busy clinic staff that have no burning desire to complete more paperwork (much less keeping a quiver of stopwatches on hand for timing each sample’s progress), but this will at least allow future researchers to match sample handling characteristics between the test and control groups.

What are the effects of sample handling procedures on sample quality? Some proteins are known to be affected more than others by sample handling variables. Some cytokines, such as IL-1a and IL-1ß, appear to be relatively robust under a variety of handling conditions, while others, TNFa and IL-6 for example, show rapid degradation.1 Indeed, some have argued that sample handling effects on proteins such as VEGF are significant enough to make much of the literature difficult to interpret.2 Furthermore, some literature suggests that the optimal sample handling technique (e.g. the anticoagulant used) varies from protein to protein. Since it is clearly impractical to instruct sample collection centres to use multiple sample handling methods, the establishment of best practices will clearly involve some measure of compromise.

It remains unclear whether some sample handling variables are more important than others. It may be, for example, that controlling one or two of these variables will be adequate to ensure comparability between samples, or it may be that some of these variables produce multiplicative effects. Unfortunately, at present, the literature is not sufficiently mature to draw clear conclusions. Worse, for many variables there is almost no literature at all. For example, there is little data on centrifuge speed or duration effects on sample handling: since hospitals often use different types of centrifuges there is considerable variability in these parameters between sample collection centres. The effect of this variability is not currently known.

Another aspect of the sample handling literature that has been poorly addressed has to do with the logistics of sample collection within clinical settings. As alluded to above, the banking of samples will rarely be a top priority for clinic staff. A study might clearly demonstrate that a sample frozen in 10 minutes is better than a sample frozen in 30 minutes, but if clinic staff cannot freeze samples faster than one hour after blood draw for logistical reasons these data are of limited use. It may simply be impractical to set rigid collection or handling criteria in many cases (e.g. for samples collected in an emergency room or in association with surgery) unless dedicated staff are available to manage the logistics. At present, few sample banks enjoy this luxury, though there are some important exceptions. The Ontario Tumour Bank (Toronto, ON), for example, contracts specialized staff at each of the hospitals in its collection network to ensure samples are collected according to its rigorous SOPs. In any event, pragmatic recommendations for sample banking staff are urgently needed.

Although the sample handling literature is disturbingly limited, some general conclusions can be drawn. As might be predicted, the faster one freezes a sample, and the faster the sample is spun down, the less the degradation.3 There is some limited evidence that keeping samples cool during the sample handling process can reduce the effects of long latencies to freezing. It is also a truism that the fewer the number of freeze-thaw cycles, the better.1 Put another way, keep the samples cool, spin them quickly, and then freeze them quickly. And above all, keep track wherever possible of what was done to the sample and when.

Even with a superficial investigation into this issue, one cannot help but come to the conclusion that more research is needed. Although the present article has focused on blood samples, similar issues exist for tissues and other fluids. Sample handling standards and best practices need to be developed and widely implemented. Methodical, systematic testing of each of the sample handling variables, alone and in combination, would provide extraordinarily useful information to those interested in sample banking. Additionally, it will be important to work with clinic staff and hospital administration to ensure that the procedures are not only feasible but practical enough that they can be followed by busy research staff.

Fortunately, some sample handling initiatives are underway. HUPO has embarked on a research campaign on this and related topics4 and recommendations are expected shortly. The Ontario Cancer Biomarker Network (Toronto, ON) has also made this a major research focus in its central facility and will continue to work closely with the Ontario Tumour Bank and various research hospitals in the region to address the need. With so many sample handling variables, however, much remains to be done.

References
1. Flower, L, R.H. Ahuja, S.E. Humphries, V. Mohamed-Ali. "Effects of sample handling on the stability of interleukin 6, tumour necrosis factor and leptin." Cytokine 12 (2000): 1712-16.
2. Hormbrey, E, P. Gillespie, K. Turner, et al. "A critical review of vascular endothelial growth factor (VEGF) analysis in peripheral blood: Is the current literature meaningful?" Clinical and Experimental Metastasis 19 (2002): 651-63.
3. Marshall et al. "Processing of serum proteins underlies the mass spectral fingerprinting of myocardial infarction." J. Proteome Research 2 (2003): 361-72.
4. Rai, AJ, C.A. Gelfand, B.C. Haywood, et al. “HUPO Plasma Proteome Project specimen collection and handling: Towards the standardization of parameters for plasma proteome samples.” Proteomics 5 (2005): 3262-77.

Dr Kenneth R Evans is president and CEO of the Ontario Cancer Biomarker Network (Toronto, ON).