Bioequivalence and Bioavailability Forum

❝ On what basis appropriate matrix (whole blood, plasma, serum, urine, CSF etc.) is selected for method development and sample analysis in BA/BE studies

• Whole blood: Very rarely used. In some cases the drug bind strongly to erythrocytes and free fraction in plasma is bot low and highly variable. Furthermore, only the ery-fraction is clinically relevant. Example: chlorthalidone.
  
• Serum: Rarely used. Only if you face analytical problems with all (!) potential anticoagulants, the drug is stable till complete clotting, and you can exclude back-conversion of metabolite(s). Even if you master all these obstacles likely you will see turbid matrix after thawing, which gets even worse in validating freeze-thaw-cycles. I haven’t seen a method in serum for years.*
  
• Plasma: That’s the way to go. Any anticoagulant which passes validation is fine.
  
• Urine: History. Haven’t seen a method (in BE of course; PK is another cup of tea) for ages. BE based on urine data is only reasonable if the drug is excreted unchanged (low first pass and/or presystemic metabolization). Was used for bisphosphonates in the past (BA <2%). Since it is very difficult to assess the rate of absorption from urine, EMA accepted extent of absorption from urine (cumulative amount) but still wanted to see C_max from plasma.
  
• CSF: Never. Unethical.
  
• Other stuff: There are some studies published with exotic matrices: Saliva, tears, … These studies essentially tried to demonstrate correlation with plasma, allowing noninvasive therapeutic drug monitoring. In BE: Never.

• Nitsche V, Schütz H, Eichinger A. Rapid high-performance liquid chromatographic determination of nifedipine in plasma with on–line precolumn solid-phase extraction. J Chromatogr, Biomed Appl. 1987;420(1):207–11. doi:10.1016/0378-4347(87)80175-5.
  We used the method (validated for serum) in ten thousands of samples, but occasionally had issues with precipitated protein even in a cooled autosampler. So we switched to plasma. In the cross-validation we discovered that the between-method bias was close to nil, but the variability in plasma was more than twice as high compared to serum. Tried many anticoagulants, always the same story – never found out the reason. At the end we accepted the higher variability and rarely had blocked pre-columns again.