Abstract
Physicians, especially nephrologists, love to measure things. We revel in serum, plasma or urine concentrations of so many different analytes, obsessing at times over small changes. On occasions, the analytes (or to give them their trendier modern title, biomarkers) we choose to focus on are truly revealing and diagnostically helpful. An example is the measurement of N-terminal pro-B-type natriuretic peptide (NT-proBNP) in the diagnosis and management of heart failure, though the implications of adding this diagnostic and therapeutic refinement, carefully rehearsed in the article by McIntyre, 1 show well that caution continues to be needed regarding over-enthusiastic adoption of new technologies lest these become performance standards before the evidence is mature. Sometimes, tests which are diagnostically useful in the non-renal setting, such as NT-proBNP or cardiac troponin, are profoundly influenced by the presence of kidney disease, the loss of renal excretory and metabolic functions seemingly degrading their diagnostic performance in this setting. This disappointment is all too common for nephrologists. However, in all the above cases, we have a touching faith and belief in our ability to accurately and precisely measure the concentration of the analyte/biomarker in question. Furthermore, by doing so, a patient might benefit from it.
While we non-laboratory doctors are often aware in general terms of the different analytical techniques that our local and national laboratories use, we have much less knowledge of the implications of these for clinical practice. Nephrologists, for example, know something about the interpretation of plasma creatinine concentrations, if often rather less about how measurements are made in the laboratory, and which methodologies are more or less reliable. But sometimes the Byzantine world of the ‘formula wars’ (different approaches to the manipulation of plasma creatinine concentrations to derive formulae which can better predict glomerular filtration rate 2 ) leaves many kidney specialists paralysed and rigid with tedium. It should not be so, of course, as to misunderstand the derivation of the key values we use in clinical practice is to do our patient populations a grave disservice.
So when we come to an ‘old friend’ like parathyroid hormone (PTH), with which many of us have grown up (and old), we think we are on safer ground. It has a long pedigree – measurements of PTH have been part of clinical and research practice for decades now. It is something that guideline bodies such as Kidney Disease Outcomes Quality Initiative (K-DOQI), Kidney Disease Improving Global Outcomes (KDIGO) and the National Institute of Health and Clinical Excellence (NICE), for example, have used to instruct us in how best to manage our patients and their chronic kidney disease – mineral bone disorder (CKD-MBD
3
). Many trials in CKD-MBD management use PTH concentration thresholds for trial inclusion and for trial/protocol decisions.
4
NICE released its technology assessment of the use of Cinacalcet therapy based on plasma PTH concentration thresholds. K-DOQI told us in 2003 that PTH concentrations should be between 150 and 300 ng/L (17–33 pmol/L).
4
KDIGO, in 2009, in contrast, mandated us nephrologists to engineer our patients’ plasma PTH concentrations into a range which is two to nine times the upper limit of the normal PTH range for the local laboratory (typically approximately 150–600 ng/L [17–66 pmol/L]).
5
The United Kingdom Renal Registry (
All this would be fair, and reasonable, practice if there were any evidence whatsoever that the achievement of this, or that, PTH value had any worthwhile meaning for renal patients – especially when those values fall within the ‘desired range’ – and the people who look after them. Unfortunately, the evidence we seek in this regard is slender and insubstantial, with little real support for the interventions we routinely undertake as clinicians. PTH concentrations at the extremes of the biological range may be of some interest clinically, but most results which are between say 200 and 700 ng/L (22–77 pmol/L) do not reliably guide clinicians as to the state of skeletal health. While controversial, the fact is that nephrology remains a specialty long on opinion and dogma, and short on quality evidence.
Into this debate we must now add another, potentially lethal, dose of reality regarding PTH measurements. And that is the fact that PTH is genuinely tricky to measure accurately: there are many different assays currently available, and no universal standard. 7 Different hospitals can use one of several different commercial assays. The historical story of the different assay approaches to PTH measurements over the last three decades is an interesting and illuminating one. 8 We have known for some time that what we think we are measuring as PTH is not one molecular species and that different assays perform to a different specification and standard from each other; furthermore, sometimes results obtained can depend on the manner in which the blood is drawn and then handled.
The paper in this issue by Almond and colleagues 9 now examines this problem using a real-life setting, applying the results obtained from split blood samples using six commonly available current PTH immunoassays in UK clinical practice. Almond and colleagues found that there is a marked inter-assay variability – not by a trivial amount, but 4.2 times taking the lowest to the highest values. The reasons for this marked inter-assay variation are complex 3,7 but as the authors clearly show, if one relates a precise numerical range or threshold (such as UK Renal Association guidelines did pre-KDIGO or NICE has done for the use of Cinacalcet therapy), the majority of patients might be at risk of undergoing significantly different clinical interventions as a result. Cinacalcet therapy would be permitted for patients whose PTH concentration say appeared to be 800 ng/L (88 pmol/L) or higher. However, the same patients might, in another hospital using a different assay, have their PTH concentration measured as only 500 ng/L (55 pmol/L) and therefore simply continue with vitamin D therapy as they would not be eligible for Cinacalcet. This simple, but compelling, point is superbly illustrated in Figure 2 in the Almond and colleagues’ paper.
When you also consider a recent piece of work by Lamb's group which shows that it would take 26 repeated measures of PTH concentrations to be reasonably certain of the ‘true’ value, and that natural cyclical and other phasic causes of PTH concentration variation mean that it takes a 72% change in a PTH concentration to be certain it is a real change, 10 we can now see all too clearly how badly suited plasma PTH measurement is to the real-world setting in which we use it. It is remarkable as to how these important factors have been ignored or dismissed or buried away by various guidelines groups, to their lasting shame. These inter-assay differences are not only of academic interest, but also of real clinical interest and importance and must be taken into account to a much greater extent than is currently the case. In another recent article, Sprague and colleagues 11 clearly sound a clarion call of warning about the clinical governance issues raised by not having some common ground across PTH concentration guidance. Almond and colleagues, 9 as also recently described by Cavalier's group, 12 show that by referring to assay-specific multiples of the normal range, it is possible significantly to reduce the mis-classification of patients by reliance on absolute PTH concentrations. Reduce yes, but abolish no, and PTH measurement remains a very unsatisfactory practice even in 2012. Both for patient-level outcome predictive value, and also for a better guide to skeletal health and integrity, bone-specific alkaline phosphatase may have more to offer us. 13
We need to better understand what it is that PTH concentrations are telling us about our kidney patients. Our ability to diagnose bone lesions, and thus guide skeletal therapy decisions, using PTH as primary guidance is poor. Plasma PTH has a half-life of 2–4 min and is exquisitely and acutely dependent on plasma calcium (or other allosteric modulators of the calcium-sensing receptor) concentrations. The ability of plasma PTH concentrations to predict skeletal or patient outcomes remains lamentable, which is why our continued reliance on this single measurement is potentially so troublesome. 3,8,11 Almond and colleagues 9 remind us clearly that in requesting investigations, we must understand the inherent problems in both the measurement and the interpretation of biological analytes. We must also understand, whether or not an accurate measurement has been made, and whether measurement of any marker is likely to increase the possibility of patient benefit at the level of hard patient outcomes. Measurement for measurement's sake will not do in 2012. For if there is no clinical benefit, then there is no utility, just futility.
DECLARATIONS