Pharmaceutical Trace Analysis

of an analyte specific clean-up is useful in minimizing this problem. Identification of the species producing the effect and application of specific clean-up steps can eliminate problems. Low results can result from incomplete analyte extraction, loss or degradation of the analyte during sample clean-up, or matrix effects. These problems can be eliminated by a careful evaluation of each of the steps employed in the sample preparation. 1) Extraction efficiencies can be improved by dissolving the sample matrix whenever possible and determining the optimum solvent for analyte extraction. When dissolution of the matrix is not practical, an exhaustive extraction procedure with an optimum solvent should be used. 2) Loss or degradation of the analyte can be controlled by determining the recoveries of all analytes through each clean-up step. An understanding of the chemical and physical properties of the analyte will not only improve recoveries, but also can allow optimization of each clean-up step. 3) Matrix effects leading to low results generally are caused by exceeding the capacity of the clean-up procedure for either analytes, related compounds, or coextractive species. This "overloading" can change the chromatographic retention of the analyte. In general, these effects can be minimized by decreasing the sample size or using a highcapacity pretreatment step to remove the coextractives. Application of these precautions can result in a method which does not generate false positive or false negative results. A recent collaborative study to determine fortified levels of CDDs and CDFs in human adipose tissue at 5-50 pg/g has been completed by eight laboratories highly experienced in the determination of CDDs and CDFs [Albro et al., Anal. Chem. 57, 2717 (1985)]. By implementing the practices described above, laboratory 2 avoided generating either unaccountably high or low results (table 1). The apparent drawback of the laboratory 2 method is the relatively long analysis time per sample. However, when the standard deviation of the recoveries for each laboratory is calculated and used in the equation to determine the relative time to analyze N samples (measurements necessary to yield data of defined statistical reliability), method 2 can actually generate data with a specified precision in the shortest time. Table 1. Interpretation of recovery data from CDD/CDF collaborative study