Abstract
When introducing new measurements into the laboratory either as research tools or diagnostic tests, laboratories often rely on information supplied by the manufacturer to decide whether a commercial reagent kit is likely to be useful for a particular purpose. With well established ‘routine’ tests in clinical biochemistry, verification of all of the manufacturer's claims in this regard is not feasible or necessary when, for example, commissioning new multi-channel analysers. 1 Usually when the intended use for the test is clear, simple verification procedures can be followed and further validation studies are not needed. The correspondence between Fleming 2 and Koerbin et al. 3 in this issue highlights the need to target precision verification procedures at clinically relevant decision points within the assay measurement range. This can be particularly important where assays such as troponin T or troponin I have been developed to extend their measurement range into lower concentrations than has been attempted with previous generations of the assay. At this concentration of analyte where ‘real life’ performance may differ from the more ideal conditions in which the validation study was carried out, it is important to establish that your laboratory is able to achieve the minimum performance required for clinical purposes.
However, when novel biomarkers are being evaluated for their clinical utility, the situation is different. The assays are often at an earlier stage of application and consequently supporting data indicating, for example, the effect of preanalytical variables on assay performance may be sparse. Simple verification procedures are often inadequate and research workers need to be aware that they should undertake further appropriate validation studies, particularly where manufacturers do not provide the data on which their assertions are based. Wind et al. 4 in this edition report that measurement of carbonic anhydrase IX (CA IX) in ethylenediaminetetraacetic acid (EDTA) plasma is unreliable when using a commercially available immunoassay which employs the M75 detection antibody. This is despite the same measurement in serum being valid and a statement provided by the manufacturer supporting its use with blood tubes containing EDTA as an anticoagulant. They cite six previously published clinical evaluation studies that used EDTA plasma with assays using the M75 antibody. It has been known for some time that the catalytic activity of CA IX is profoundly affected by a conformational change consequent on metal ion binding in the catalytic domain of the molecule; indeed this property of carbonic anhydrase has been exploited by using it as a component of a reagent system to measure serum zinc. 5 It is an intriguing feature of this study that there is now indirect evidence for metal ion binding in the N-terminal domain that might affect both catalytic activity and the affinity of antibody binding via a conformational change.
So what are the broader implications of discovering an important confounding factor in previous studies designed to evaluate the clinical utility of measuring CA IX in blood? Preanalytical variables often come to the attention of clinical biochemists after a new test has become established and is in widespread clinical use. This is a fundamental problem of translational research that needs to be addressed. Detailed studies of preanalytical variables need to be undertaken earlier in the development of clinical assays. The clinical evidence that supports the use of a test in a defined patient group must be based on well validated assays taking into account the demands of biological variation, phlebotomy, sample processing, sample storage and analysis. Otherwise we may lose a potentially valuable marker altogether or fail to translate an effective research tool into clinical practice.
Manufacturers often have only limited access to samples from those patient groups in which the biomarker is likely to undergo clinical evaluation. Furthermore, the clinical utility of a biomarker may be assessed in a much wider range of clinical situations than could have been anticipated by the manufacturer. Consequently, analytical validation studies performed by manufacturers are often, understandably, based on small numbers of subjects and may not include any patients suffering from the disease to which the assay is subsequently applied. In these circumstances, access to the manufacturer's raw data and to supplementary information about how validation studies have been conducted is the first requirement to ensure that the limitations of these studies are understood.
Secondly, a collaborative effort between researchers and manufacturers should be made in the preliminary stages of any research project involving novel biomarkers. Well conducted validation studies relevant to the circumstances of testing and the patient groups being studied are useful to both parties. New candidate biomarkers are being identified at a rate that outstrips our capacity to evaluate their clinical utility. Where research effort is being put into clinical evaluation, the effect of preanalytical variables must be determined. 6,7 Sharing and publishing information on preanalytical aspect of validation is essential. Detailed validation information should be a prerequisite for publishing research studies on biomarkers.
DECLARATIONS