Clinical pharmacology quality assurance program: models for longitudinal analysis of antiretroviral proficiency testing for international laboratories

Abstract

Among National Institutes of Health HIV Research Networks conducting multicenter trials, samples from protocols that span several years are analyzed at multiple clinical pharmacology laboratories (CPLs) for multiple antiretrovirals. Drug assay data are, in turn, entered into study-specific data sets that are used for pharmacokinetic analyses, merged to conduct cross-protocol pharmacokinetic analysis, and integrated with pharmacogenomics research to investigate pharmacokinetic-pharmacogenetic associations. The CPLs participate in a semiannual proficiency testing (PT) program implemented by the Clinical Pharmacology Quality Assurance program. Using results from multiple PT rounds, longitudinal analyses of recovery are reflective of accuracy and precision within/across laboratories. The objectives of this longitudinal analysis of PT across multiple CPLs were to develop and test statistical models that longitudinally: (1) assess the precision and accuracy of concentrations reported by individual CPLs and (2) determine factors associated with round-specific and long-term assay accuracy, precision, and bias using a new regression model. A measure of absolute recovery is explored as a simultaneous measure of accuracy and precision. Overall, the analysis outcomes assured 97% accuracy (±20% of the final target concentration of all (21) drug concentration results reported for clinical trial samples by multiple CPLs). Using the Clinical Laboratory Improvement Act acceptance of meeting criteria for ≥2/3 consecutive rounds, all 10 laboratories that participated in 3 or more rounds per analyte maintained Clinical Laboratory Improvement Act proficiency. Significant associations were present between magnitude of error and CPL (Kruskal-Wallis P < 0.001) and antiretroviral (Kruskal-Wallis P < 0.001).

Figure 1. Box plots of log recoveries and absolute recoveries by CPL, for two different types of nominal concentrations

Upper and lower left (1a and 1b): Box plots of log recovery, defined as natural log of reported/nominal concentration (RC/NC), where NC is defined as the weighed-in value (WIV) and the final target (FT) in the upper and lower plots, respectively. Dotted gray line at 100% indicates perfect agreement between RC and NC. Dashed black lines indicate recoveries of 80% and 120%. Upper and lower right (1c and 1d): Box plots of absolute recovery, defined as 100* [NC + abs(NC-RC)]/NC . Dashed gray lines indicate absolute recoveries of 100%, 105% and 110%; dashed black lines indicate absolute recoveries of 120% and 140%. The box's bold center line is the median. The upper and lower edges of box represent the 25th and 75th percentiles. The whiskers extend to either: (1) 1.5 times the interquartile range (the distance between 25th and 75th percentiles) or the (2) most extreme point, whichever is closer to the box. Points outside the whiskers are displayed individually. These points are outliers relative to the box plot, but without positing a particular distribution for the concentrations, no probabilistic statement is inferred.