Maximizing prediction probability PK as an alternative semiparametric approach to estimate the plasma effect-site...

The k(e)(0) value is the first order rate constant determining the equilibration of drugs between plasma or end-tidal concentration and effect-site (e.g., brain) concentration. Parametric and semiparametric approaches have been used for estimating individual k(e)(0) values and describing the drug-response curve. In this study, we introduce a new semiparametric approach calculating k(e)(0) values for isoflurane, sevoflurane, and desflurane by maximizing the prediction probability P(K).

Data from 45 patients scheduled for a radical prostatectomy were analyzed. After lumbar epidural catheterization, patients received remifentanil and propofol solely for induction of anesthesia. Thereafter, epidural analgesia was initiated, and isoflurane, sevoflurane, or desflurane (15 patients each) was added to maintain unconsciousness. At least 45 min later, end-tidal concentrations were varied between 0.5 and 2 minimum alveolar anesthetic concentration. We estimated an individual k(e)(0) value for each patient by optimizing the prediction probability P(K) (P(K)-based k(e)(0)) or by minimizing the area within the hysteresis loop (area-based k(e)(0)). Data are mean +/- sd.

Both semiparametric approaches led to comparable k(e)(0) values with 0.18 +/- 0.06 min(-1) (P(K) based) and 0.15 +/- 0.04 min(-1) (area based) for isoflurane and 0.17 +/- 0.08 min(-1) (P(K) based) and 0.16 +/- 0.11 min(-1) (area based) for sevoflurane. k(e)(0) values for desflurane (P(K) based: 0.30 +/- 0.17min(-1); area based: 0.32 +/- 0.25 min(-1)) were significantly higher than for isoflurane and sevoflurane.

Maximizing the prediction probability P(K) for estimating k(e)(0) seems to be a promising method that researchers could use on an exploratory basis.