Midazolam concentrations in patients' plasma was determined after extraction with high performance liquid chromatography (HPLC), gas chromatography (GC) and gas chromatography/mass spectrometry (GC/MS). GC was selected for routine plasma assays in terms of selectivity, simplicity, precision, accuracy and sensitivity (0.02 microgram/mL); HPLC analysis was less sensitive (0.1 microgram/mL) than GC; GC/MS was used for analysis validation. Plasma protein binding of midazolam was determined by GC in patients' plasma after in vitro incubation with midazolam, ultrafiltration and extraction; 5% of the drug was unbound to plasma proteins. Midazolam distribution in lipoprotein fractions separated by ultracentrifugation of plasma obtained from patients on prolonged midazolam treatment was also assayed by GC.

Severely ill patients often require total parenteral nutrition including intravenous lipid emulsions concurrently administered with lipophilic drugs. Therefore we investigated whether therapeutic application of a mixed medium chain/long chain triglyceride infusion affects the disposition of midazolam necessary for sedation in patients on artificial respiration. The concentrations of midazolam were measured in unfractionated plasma, and in lipoprotein fractions isolated from ex vivo blood samples, including determination of triglycerides and cholesterol; the albumin level was also analysed. Midazolam in the VLDL fraction was only 0.246 microgram.ml-1, whereas the total plasma concentration averaged 1.101 micrograms.ml-1, and the midazolam content of the LDL plus HDL fractions amounted to 1.771 micrograms.ml-1. Albumin in these lipoprotein fractions was just as unequally distributed. A lipid infusion resulted in a significant elevation of total triglycerides from 157 to 221 mg.dl-1 and VLDL-triglycerides from 77 to 155 mg.dl-1. The triglyceride content of the LDL plus HDL fraction rose from 102 to 139 mg.dl-1. At the same time the midazolam concentration in unfractionated plasma and in the VLDL and the LDL + HDL fractions decreased to 0.899 microgram.ml-1, 0.130 micrograms.ml-1, and 1.265 micrograms.ml-1, respectively. Cholesterol and albumin concentrations were not affected. The data show for the first time that a significant increase in plasma triglycerides during an intravenous lipid infusion does not result in accumulation of midazolam in lipoproteins, probably because albumin binding of the drug is very strong. The lack of midazolam trapping is important with respect to the safety of concurrent use of lipophilic drugs and intravenous lipid infusions.