Bone turnover markers for assessment of anti-resorptive effect in clinical practice: A good idea meets the problem of measurement uncertainty.

OBJECTIVES: Bone turnover markers (BTM) are measures for understanding the effect of anti-resorptives upon osteoclast activity. Post-hoc trial data suggests reduction in BTM of 40% may represent a target for defining appropriate response to therapy. We modeled clinical application of this target threshold in an individual patient setting where assay measurement uncertainty and biological variation are included. DESIGN: Using serum C-telo-peptide (ß-CTX), we constructed hypothetical scenarios of ß-CTX measurement pre and post bisphosphonate therapy. Using typical ß-CTX assay characteristics (analytical coefficient of variation, CV 5.0%) and published intra-individual ß-CTX data for post-menopausal women (CV 18.0%), we calculated the post-therapy ß-CTX that must be seen on single repeat measure for 95% confidence that the observed result was ≥40% below baseline. Sensitivity analyses considered greater and lesser variations in the combined sources of variation. RESULTS: The one-tailed 95% reference change value for any detectable therapeutic decrease in ß-CTX was 22%. However, to have 95% confidence of having achieved a reduction ≥40%, an observed ß-CTX decrease of ≥56% is required. Larger decreases are needed for scenarios of greater analytical or intra-individual variation. CONCLUSIONS: Although population data suggest a ß-CTX decrease of 40% is commensurate with adequate therapeutic response to anti-resorptives, application to an individual patient where measurement and natural variation are present is problematic. ß-CTX decreases much >40% are required to be confident of having achieved the optimal treatment response. It is uncertain whether this is a legitimate change to be expected in all individual patients and therefore clinical application of this threshold is uncertain.

The concept of precision extended to the atherogenic index of plasma.

BACKGROUND: The atherogenic index of plasma (AIP), defined as the base 10 logarithm of the ratio of plasma triglyceride (TG) to high density lipoprotein cholesterol (HDL-C), has been employed as a predictor of cardiovascular risk. We seek to quantify the analytical precision of the AIP using the coefficients of variation (CVs) of the TG and HDL-C assays. METHODS: Error propagation methods are employed to develop a simple formula for the standard deviation of the random analytical error in the AIP assuming that the errors in the TG and HDL-C assays are normally distributed. An alternative derivation assuming log-normal distribution of errors gives nearly identical results while avoiding subtle technical problems. RESULTS: The SD of the AIP is given by sigma(AIP) approximately = 1/ln(10) square root(CV(TG)(2) + CV(HDL-C)(2)) and this formula will provide SD results that are accurate within 0.4% for CVs of TG and HDL-C less than 5%, as compared with results of Monte Carlo simulation. We also explain that the concept of CV cannot be applied to the AIP since it is a logarithm. CONCLUSIONS: The formula provides a simple means to quantify the precision of the AIP from precision data available for the TG or HDL-C assays.