Repeats

Every bioanalytical laboratory should develop an SOP for repeat assays. The policy defined in the SOP must be implemented before starting a study. Most examples requiring reassay result from some kind of analytical or technical difficulty:

a. Bad chromatography includes interfering peaks making the integration impossible, no peaks, radical differences in chromatographic pattern, chromatographic column failure etc.

b. Lost sample broken sample tubes, leaking pipet tips, leaking screw caps, etc.

c. Bad processing human or robotic errors like reagent or internal standard omission, adding excess of reagent or internal standard, etc.

These errors should result in a documented audit trail of deficient chromatograms, notes to the study file listing lost samples or errors in the processing, or computer print-outs in case of robotic systems. No numerical results are associated with failed analysis and samples should be repeated as single samples.

Occasionally, clinical samples exhibit concentrations above the validated range (AQL = above quantitation limit). Such samples should be diluted with the same matrix and repeated singly. If a concentration obtained is only slightly above the highest standard, the Conference Report recommends either redefining the calibration curve range by adding a higher standard or analyzing diluted sample. The DIA Consensus permits an extrapolation above the top of the calibration curve limited to one standard deviation or 15%.

Study samples sometimes provide results which formally and chromatographically look correct, yet seriously contradict previous results. The goal of a bioanalyst or pharmacokinetist is to provide results for which there is a scientific basis. At the same time, it is appropriate to challenge suspected results. One may suspect a pharmacokinetic outlier if a pre-dose sample from naive subjects contains a measurable drug concentration, if a profile exhibits a halved or doubled maximum contrary to known pharmacokinetics, or if concentrations are very different (500-1000%) than expected. Such samples, which could be called "suspected outliers (SO)" provide numerical values and repeats should be done in duplicate. A bioanalytical laboratory should also develop a decision tree dictating a verdict in every foreseeable case to eliminate arbitrary decisions. A very good decision tree has been developed by Lang and Bolton [3], Briefly, a 15% agreement between data is considered a confirmation if the repeats are done in duplicate, or 30% if only one repeat was possible. If results are too far apart, no result is reported (NR).

Suspected pharmacokinetic outliers should be evaluated and excluded using a well known outlier test. In bioanalysis, the volume of the study samples are generally limited and samples can rarely be assayed with sufficient number of replicates for any meaningful outlier test to be applied. It should be also noted that occasional outliers do not influence the outcome of a study if correct numbers of subjects are selected to ensure appropriate statistical power .

Repeat assays are required as a matter of policy by Canadian Health Protection Branch, or by some clients in the contract industry to verify correctness of the data. These repeats should flagged as "confirmation points" or "client-requested."

Finally, if samples are repeated in error, these results cannot be ignored and must be evaluated according to the appropriate decision tree.

Was this article helpful?

0 0

Post a comment