Mild hypothermia alters midazolam pharmacokinetics in normal healthy volunteers

Abstract

The clinical use of therapeutic hypothermia has been rapidly expanding due to evidence of neuroprotection. However, the effect of hypothermia on specific pathways of drug elimination in humans is relatively unknown. To gain insight into the potential effects of hypothermia on drug metabolism and disposition, we evaluated the pharmacokinetics of midazolam as a probe for CYP3A4/5 activity during mild hypothermia in human volunteers. A second objective of this work was to determine whether benzodiazepines and magnesium administered intravenously would facilitate the induction of hypothermia. Subjects were enrolled in a randomized crossover study, which included two mild hypothermia groups (4 degrees C saline infusions and 4 degrees C saline + magnesium) and two normothermia groups (37 degrees C saline infusions and 37 degrees C saline + magnesium). The lowest temperatures achieved in the 4 degrees C saline + magnesium and 4 degrees C saline infusions were 35.4 +/- 0.4 and 35.8 +/- 0.3 degrees C, respectively. A significant decrease in the formation clearance of the major metabolite 1'-hydroxymidazolam was observed during the 4 degrees C saline + magnesium compared with that in the 37 degrees C saline group (p < 0.05). Population pharmacokinetic modeling identified a significant relationship between temperature and clearance and intercompartmental clearance for midazolam. This model predicted that midazolam clearance decreases 11.1% for each degree Celsius reduction in core temperature from 36.5 degrees C. Midazolam with magnesium facilitated the induction of hypothermia, but shivering was minimally suppressed. These data provided proof of concept that even mild and short-duration changes in body temperature significantly affect midazolam metabolism. Future studies in patients who receive lower levels and a longer duration of hypothermia are warranted.

Fig. 1.

Core temperature (mean ± S.D.) over time curves of four treatment groups. The temperatures were recorded every 2 min during the first 30 min and every 10 min thereafter. The cold (4°C) burden area was calculated from each individual temperature-time curve. There was a significant difference between warm and cold + magnesium groups (p = 0.01) of the cold burden area. ■, warm; ▴, warm + magnesium; ○, cold; ♢, cold + magnesium.

Fig. 2.

Individual subject midazolam time-plasma concentration profile from the four treatments (A–F) and the average time-concentration curves for six subjects (mean ± S.D.) (G). Cold (4°C) infusions are denoted by dashed lines, and warm infusions are denoted by solid lines. ■, warm; ▴, warm + magnesium; ○, cold; ♢, cold + magnesium.

Fig. 3.

1′-Hydroxymidazolam formation clearance and midazolam systemic clearance from noncompartmental analysis estimation with mean for four treatment groups. A difference was observed in 1′-hydroxymidazolam formation clearance between warm and cold + magnesium (p = 0.0168). A trend toward significance between warm and cold + Mg for systemic clearance (p = 0.0568). ■, warm; ▴, warm + magnesium; ○, cold; ♢, cold + magnesium.

Fig. 4.

Model diagnostic plots. Goodness-of-fit plots for the population pharmacokinetic model from NONMEM analysis. Individual predicted concentrations versus observed concentrations and the predicted concentrations versus observed concentrations are shown. A and B, the straight lines are the lines of unity. C and D, population-predicted concentrations versus weighted residual and time versus weighted residual.

Fig. 5.

The relationship between core body temperature (from the highest body temperature 37.8°C to the lowest body temperature 34.8°C observed in this study) and individual midazolam systemic clearance estimated from the final population pharmacokinetic model.

Fig. 6.

The simulated time-concentration profiles of midazolam in core temperature 32, 34, and 36.5°C. The curves were generated from WinNonlin software based on the population pharmacokinetic parameters estimated from NONMEM. The simulation curve for a core temperature 32°C (*) has the highest AUC and C_max followed by 34°C (♦) and 36.5°C (■). The simulations reflect the model-predicted reductions in midazolam clearance at 32°C (a 42.8% reduction compared with that at 36.5°C) and 34°C (a 26.0% reduction compared with that at 36.5°C).