Friday, January 21, 2011

Managing Bioanalytical Cross-Contamination


Min Shuan Chang, Ph.D., Elaine J. Kim, and Tawakol A. El-Shourbagy, Ph.D. Abbott Laboratories

Introduction

Analysis of drug concentration in biological samples is an integral part of the drugdevelopment process [1]. Concentration data from biological samples is required to study the absorption, distribution and elimination properties of a new chemical entity and to understand its dose-response relationship in both clinical and non-clinical studies. A clear understanding of pharmacokinetic variability becomes increasingly important in later stages of clinical studies, especially for pivotal bioequivalent studies such as those required for changes in formulation. Unlike a manufacturing process, contamination in an analytical laboratory does not directly impact the product. However, contamination leads to reporting biased results that
impacts the validity of the derived decisions and increases regulatory risk.
Laboratory contamination is defined as an unintended and undesirable transfer of analyte or interfering compounds into an analytical sample or sample extract before or during the analytical process. The following is a short list of the characteristics of laboratory contamination.
Detection
  • Usually observed as a positive bias.
  • Blank, placebo or pre-dose (first dose) has detectable analyte(s).
  • Results do not fit the trend of the concentration profile.
  • May or may not be detected by standards or quality control samples.
Source
  • Contaminated animals or subjects.
  • Contamination during sampling at the site.
  • Contamination during sampling at laboratory.
  • Contamination during analysis.
Contamination of test subjects and contamination during sampling at study sites have been discussed previously [2, 3, 4, 5] and are outside the scope of this article. Becausecontamination may not be predictable, detection is not assured by the current quality control program. For a batch containing a blank, a zero standard (blank with internal standard), a set of eight calibration standards, two sets of three quality control samples (QCs) and 80unknowns from the serialbleeding of five subjects or treatments, contamination (from a high concentration sample) may be detected in approximately 16 samples, i.e. blanks, pre dose samples, the sample at the end of elimination phase, two of the lowest standards and the low QC. Therefore, a single incidence of laboratory cross-contamination may be detected only 17 percent of the time. Therefore, occurrence of contamination may only be observed through assay statistics including a positive bias or residue for the low standards and QCs.
Laboratory contamination during the sample analysis process may be classified into two categories, carryover and cross-contamination. Carryover is the result of transferring an analyte from an earlier sample to later samples, usually due to the sharing of a common
container or transferring device. Carryover is serial in nature and occurs in both sample preparation and the analytical process. Carryover is generally distributed in a narrow zone (Figure 1). The effect of HPLC autosampler carryover may be estimated from the
results of a few samples and that the relative effect of carryover on concentration results may be calculated [6, 7, 8].
Figure 1
Cross-contamination is the transfer of an analyte from a sample to another that is processed with it. Cross-contamination is parallel in nature and may not involve a medium. For a bioanalytical assay, the major sources of cross-contamination are spills, aerosols and drips
during transfer. Because both the occurrence and magnitude are random, cross-contamination is difficult to estimate and manage. The term contamination will be used to describe both carryover and crosscontamination.

Recognize the contamination incident

Unlike accuracy and precision, contamination may not be detectable by the generally accepted QC program, but positive bias in low QC or low standards (more data points exhibit positive bias than negative bias or observation of large positive bias) are clues. A critical review of historical data would give good estimate for contamination potential of a method. If the results justify an investigation, performing experiments to understand the root cause will minimize
future risks.

No comments: