Challenges in the Analysis of Pesticide Exposure Samples
Increased interest in children’s exposure to pesticides has resulted in the generation of large numbers of samples for analysis. In this section we discuss several key issues and lessons learned regarding analysis.

Laboratory capacity.
As studies of the type described here grow larger and a series of longitudinal samples are collected from each participant, the sample size may become too large for the capacity of one or two laboratories. Multiple laboratories should be enlisted for large studies to avoid sample backlogs. As laboratory capacity is improved, it is imperative to produce comparable data across studies, as the U.S. EPA did in its interlaboratory comparison study among the North American laboratories performing DAP analyses (James et al. 2003).

Intra- and interpersonal variability in urine samples.
Several methods have been evaluated to “correct” for the variability in urine dilution across spot samples, the most popular being creatinine (Boeniger et al. 1993). Creatinine excretion varies because of many factors, including the size of the participant, so interindividual variation, especially among diverse populations, is large. Thus, creatinine-adjusted pesticide concentrations should never be compared among individuals of vastly different age groups (i.e., children vs. adults). Changes in creatinine excretion during pregnancy should be thoroughly evaluated before comparing with other women in similar age groups. The validity of creatinine adjustment may also be analyte dependent. Further studies to assess the variability of commonly measured analytes in urine should be conducted to identify the most effective sampling strategies for cohort studies. In all likelihood, sampling for nonpersistent chemicals will require multiple samples taken over the course of the study at regular intervals (e.g., weekly, monthly, semiannually).

Selectivity of analysis.
Selectivity can refer to either the ability of a measurement technique to differentiate a single analyte that is measured from other components of the matrix (i.e., reducing false positives) or the ability of the analyte measured to accurately, and unequivocally, identify exposure to the target chemical of interest. However, high selectivity techniques are costly and require specialized training for operation (Barr et al. 1999). Methods such as immunoassays and less specialized technologies may be employed, but harmonization should be performed to ensure that data generated using different methods are comparable.
The selectivity of the analyte measured to accurately reflect the exposure of interest may depend on the biomarker being measured rather than the measurement technique. Many OP pesticides, for example, can be metabolized to common DAP compounds, so it is not possible to derive chemical-specific exposure estimates from such data. Further complicating the issue, the DAPs, as well as compound-specific metabolites, may be present in environmental media as the environmental degradates of the pesticides (Curl et al. 2003b; Wilson et al. 2004). No studies to date have shown that these environmental degradates can be absorbed and excreted unchanged; but if this does occur, then DAPs and other pesticide metabolites detected in urine would represent exposure to both the pesticide and its degradate. Some metabolites are very selective for the chemical measured. For example, 2-isopropoxy-4-methyl-6-hydroxypyrimidine, a metabolite of diazinon, is selective for diazinon exposure, although potentially the environmental degradates could contribute to the urinary levels as well. In some cases, the parent pesticide can be excreted in urine, such as for the herbicide 2,4-D (2,4-dichlorophenoxy-acetic acid).
One way to unequivocally identify exposure to a particular pesticide is by measuring the intact pesticide, presumably in blood or similar samples, because the intact pesticide is not appreciable in urine. However, blood measurement levels are typically about 1,000 times lower than urinary metabolite measurements; this requires highly sensitive analytical techniques, driving up the cost of analysis. In addition, target chemicals in blood may exhibit some degree of instability. Finally, there are no laboratory methods available for many common use agricultural or home pesticides in blood. Saliva sampling is an attractive alternative to blood sampling, as discussed above.

Sensitivity of analysis.
The sensitivity of an analytical method—the ability of the method to measure the chemical at the desired level—should be considered before a study begins (Barr et al. 1999). The biologic half-lives of nonpersistent chemicals are relatively short, usually on the order of hours or days (Needham and Sexton 2000). Samples collected several days after an exposure event may require ultra-sensitive methods for analyte detection. These measurements must provide adequate sensitivity to allow detection of the chemicals of interest in a sufficient proportion of the population to provide a realistic representation of the populations’ exposure. The current method for analysis of OP pesticide metabolites developed by the Centers for Disease Control and Prevention was used for many but not all of the studies described in this article and has proven to be quite sensitive (Bravo et al. 2002).

Alternative matrices and/or biomarkers.
Pesticides have been measured successfully in saliva (Lu et al. 2003), meconium (Whyatt and Barr 2001), and amniotic fluid (Bradman et al. 2003). Matrices such as meconium may provide longer term dosimeters for exposure to nonpersistent chemicals; saliva may provide a measure of internal dose without the invasiveness of blood sampling. Preliminary studies evaluating the partitioning of chemicals in the various matrices should be conducted that will allow for comparison of data among matrices and validate the usefulness of alternative matrices for biologic monitoring. An alternative matrix that may prove useful is the gel matrix in disposable diapers. Extraction techniques for solid materials may prove practical for the gel matrix and might improve sample collection procedures for infants and children who are not toilet trained.

Quality assurance and control.
A vital component of all biomonitoring methodology is a sound quality assurance/quality control (QA/QC) program. QA/QC procedures supporting these studies have included proficiency testing, repeat measurements of known biologic materials, and round-robin studies to confirm reproducible measurement values among laboratories, as well as field spikes and field blanks to confirm sample integrity.

Sample storage issues.
The time frame for sample testing and long-term storage becomes an issue for large studies. The long-term stability of analytes has been demonstrated for some matrices but not for others, for example, blood. One final logistical complexity is physical freezer space for storage, and the substantial cost of maintaining that storage. Archiving samples in the smallest containers possible would enhance the ability to keep the samples long term under proper storage conditions.