Case Study: Regulatory Considerations in the Analysis of Human Patient Samples in an Academic Core Lab

摘要

ANALYTICAL CHALLENGE: Absolute quantitation of propofol in 1,600 human plasma samples. REGULATORY CONSIDERATIONS: Before our core lab could accept this project, regulatory and safety aspects had to be taken into account – For clinical samples, would CLIA or GLP certification be required? Would HIPAA apply? Is propofol (infamous as the drug that killed Michael Jackson) a Controlled Substance? What about biosafety regulations? These issues will be defined and explored. In brief for this particular case study, CLIA was not required because results would not be reported back to patients or used for diagnosis. GLP was not required because data would not be used in FDA or other regulatory submissions. HIPAA did not apply because samples and data were not personally identifiable. Propofol is not classified as a controlled substance; if it were, many controlled substances can be obtained as analytical standards (in solution at low concentration) without triggering federal DEA regulations. As for biosafety, the patients were not known to have any infectious diseases, and samples were handled under “universal precautions”. SCIENTIFIC METHOD: Human plasma samples were spiked with D17-propofol (internal standard) and cleaned up via liquid-liquid extraction. The heptane extracts were analyzed without derivatization on a Scion TQ GC-triple quadrupole mass spectrometer (Bruker Daltonics) using electron ionization and SRM mode. An isocratic oven program (195C) minimized cycle time: total injection-to-injection time was 2.5 minutes. Four transitions were monitored, 2 each for propofol and D17-propofol. RESULTS: The method proved to be reliable for the analysis of 1,600 plasma samples spread over 2 months. Each 100-vial autosampler tray of samples took just over 4 hours to analyze. Calibration curves and QCs were run with each batch of samples and demonstrated consistent method performance over time. LLOQ was 4nM (400 amol on column), LLOD 2nM; response between 1nM and 4mM was linear.