New Horizons in Sedative Hypnotic Drug Development

In this issue of Anesthesia & Analgesia, 2 research groups report their progress in the development of 2 novel anesthetic drugs: remimazolam1,2 and methoxycarbonyl (MOC) analogs of etomidate.3,4 As their names suggest, remimazolam is a new rapidly metabolized benzodiazepine and MOC-carboetomidate and MOC-etomidate are rapidly metabolized etomidate analogs. Through the drug-development process, both well-known and innovative methods were used to build these drugs and explore their behavior. The purpose of this editorial is to highlight some of these methods and their clinical implications. DESIGNED FOR SPEED Recent advances in clinical pharmacology have led to the concept of “soft.” Soft refers to drugs that are designed to be safer with an increased therapeutic index and a rapid predictable metabolism to inactive metabolites.5 This should not be confused with psychoactive agents without addictive properties (e.g., cannabis, mescaline, and lysergic acid diethylamide [LSD]) that carry the same label (i.e., soft versus hard). Aspects of soft that are of particular interest to anesthesiologists include rapid onset of effect and quick recovery. In the realm of IV agents, leveraging from experience with remifentanil, esmolol, mivacurium, and rapacuronium, drug developers are quick to incorporate a molecular structure, in this case an ester moiety, that could yield rapid predictable drug metabolism. Both remimazolam and MOC analogs of etomidate were designed with an ester moiety, and preliminary analyses reveal a rapid kinetic profile for both. Although these results are compelling, not all esters are the same. Depending on neighboring molecular structures, metabolism rates may vary. Some may metabolize too quickly, requiring fast infusions to achieve a desired effect at the same time creating excessive metabolites. Others may metabolize too slowly and accumulate with continuous infusions. Clinical pharmacologists often evaluate numerous analogs to find a structure with a favorable ester metabolism that also maintains a desirable drug effect. CLEANING UP ETOMIDATE Beyond speed, to build a better etomidate, Pejo and colleagues4 explored the synthesis of an etomidate analog that does not suppress adrenocortical function. They identified a portion of etomidate, specifically its pyrrole ring, which inhibits 11β-hydroxylase, an enzyme known to have a key role in adrenal steroid synthesis. Armed with this knowledge, they created a new analog of etomidate, carboetomidate, by modifying the pyrrole ring. Early work with this analog was promising; inhibition of cortisol synthesis in vitro and suppression of steroid synthesis in a rodent model were markedly decreased. The clinical implications of this new form of etomidate are important. After only a routine induction dose, etomidate in its current form can suppress adrenal function up to and beyond 24 hours.6 This can be dangerous in septic and critically ill patients who require steroids to maintain their immune function and metabolic homeostasis. Some authors have even called for abandoning etomidate as an induction drug for septic patients or at a minimum, providing supplemental corticosteroid therapy after etomidate administration.7 An etomidate analog that is less likely to cause significant adrenal suppression may offer clinicians another option when considering the pros and cons of propofol or etomidate for the induction of anesthesia, as a sedative component to a total IV anesthetic, as a rapid-acting sedative for procedural sedation, or for prolonged sedation in an intensive care unit setting in septic or critically ill patients. VISUALIZING DRUG BEHAVIOR THROUGH SIMULATION The group developing remimazolam utilized model building and simulation to characterize remimazolam and to formulate how to go about conducting the next phases of its evaluation. Some of the simulation tools they used to accomplish these goals include the context-sensitive half-time (CSHT) and Monte Carlo Simulations. Although at first glance, these tools may be overly complex to be of interest to an anesthesiologist, they do offer some unique insights into anesthetic drug behavior that impact clinical decision-making. The CSHT was initially introduced as a tool to compare the kinetic profile of IV anesthetics when administered as prolonged infusions8 and estimate the time required for plasma concentrations to decrease by 50% (half-time) based on infusion duration (context). Plots of the CSHT illustrate clinically relevant differences in how drugs are cleared from the blood. Those known to accumulate in peripheral tissues have a different CSHT profile than those that are more rapidly metabolized. IV agents that contain an ester moiety (i.e., remimazolam and the MOC etomidate analogs) have a rapid CSHT that becomes independent of infusion duration for longer infusions when compared with similar drugs without this biochemical structure (i.e., remifentanil versus fentanyl, alfentanil, or sufentanil). Based on their arterial kinetic model of remimazolam, Wiltshire et al.2 present simulations of the CSHT for infusions that last up to 8 hours. As expected, remimazolam has a rapid CSHT (6 minutes) independent of infusion duration after 2.5 hours. To put remimazolam into perspective, consider the CSHTs for common sedative hypnotics and opioids presented in Table 1. The CSHT profile for remimazolam is similar to remifentanil even for infusions up to 8 hours. For short-duration infusions (<2 hours), remimazolam is similar to propofol and etomidate, but for longer infusions, remimazolam has a much faster decrease in drug concentration. Kinetic models for the MOC etomidate analogs are unavailable in humans, but when they do become available, it will be interesting to see how simulations of their CSHTs compare with the other drugs, especially their parent compound, etomidate, listed in Table 1.Table 1: Context-Sensitive Half-Times (in Minutes) for Selected Sedative Hypnotics and Opioids for Continuous Infusions from 2 to 8 Hours9Of note, CSHT is a tool to describe kinetic but not dynamic drug behavior. It does not determine how long a drug effect will persist once an infusion is terminated. Effect is a function of dose (infusion rates). For example, assuming linear kinetics (i.e., the kinetics don't change with increasing dose), the CSHTs after 2 hours of propofol at a rate of 300 or 50 μg/kg/min are the same. The duration of effect, however, would of course be longer for the higher infusion rate. For instance, the times required for plasma propofol concentration to reach 1 μg/mL after termination of a 2-hour 300 or 50 μg/kg/min infusion are 10 and 3 minutes, respectively. If not interpreted correctly, the CSHT can be misleading when discussing duration of effect. This nuance is well appreciated by Wiltshire et al. In their simulations, they also present a form of CSHT that goes beyond just describing kinetic behavior; the CSHTs for 2 measures of sedation are changes in the Bispectral Index (BIS) scale and the Modified Observer's Assessment of Alertness/Sedation (MOAA/S) scale. The MOAA/S scale ranges from 0 for unresponsive in the presence of verbal and tactile stimuli to 5 for awake and responsive. Using fixed infusion rates of remimazolam and midazolam, they describe the time required for a clinically significant change to occur in terms of infusion duration. In place of a 50% decrease in plasma concentration, Wiltshire et al.2 used the time required for the BIS value to return by 50% from the difference between its lowest (maximal effect) and baseline levels and the time required for the MOAA/S score to change from 4 to 5 (from a lethargic to a normal response to name spoken in normal tone). It is important to point out that the “context” of this pharmacodynamic CSHT considers both infusion duration and the infusion rate, unlike its kinetic cousin whereby the context refers to only the infusion duration. In other words, the pharmacodynamic CSHTs they report only apply to single infusion rates used to generate them. As with the propofol example above, the pharmacodynamic CSHTs would be different for other infusion rates (i.e., longer for higher infusion rates). This pharmacodynamic version of the CSHT is perhaps of more interest to anesthesiologists than the kinetic one; it provides better insight into the duration of effect once an infusion is terminated. It also clarifies how some drugs should not be used as continuous infusions where a rapid decline in drug effect is desired. As the authors illustrated, after an 8-hour midazolam infusion, the pharmacodynamic CSHT for the BIS was 6 hours! Another aspect of the pharmacodynamic CSHTs that merits consideration is the nature of the pharmacodynamic model used to estimate the half-time; specifically, the conditions under which the pharmacodynamic model was created. Pharmacodynamic models of sedation are frequently created in the absence of a constant stimulus. This is not exactly ideal and primarily because a good surrogate of pain encountered in the perioperative arena is difficult to consistently mimic in a reproducible manner and in such a way that doesn't harm study participants. That being said, clinicians rarely use sedatives in a setting where there is no stimulus. At best, pharmacodynamic models created in the absence of a stimulus provide an estimate of how patients may behave at the end of a procedure not associated with significant postprocedure pain. To take pharmacodynamic CSHTs of sedation to the next step, pharmacodynamic models need to be created in the presence of stimuli encountered in the perioperative arena. In reporting their phase I clinical trial results for remimazolam, Antonik et al.1 and Wiltshire et al.2 utilized a sophisticated approach to characterize the variability in their kinetic models that is worth highlighting. Through the use of simulation, kinetic models were used to explore various dosing regimens and their associated drug-concentration profiles over time and thereby minimize the number of experiments in human subjects. Simulations, however, only present a population estimate and do not easily convey intersubject variability. In their work with remimazolam, Wiltshire et al. use measures of kinetic model parameter variability (in this case, the percentage of relative standard errors) to estimate how remimazolam may behave across a population of healthy subjects. Based on a recirculation model structure, they built a 5-compartment model with a total of 10 kinetic parameters, each with an estimate of variability. A Monte Carlo approach was used to create 1000 new models by randomly selecting new kinetic parameters from the distribution of each parameter's variability. They simulated a dosing regimen to predict plasma remimazolam concentrations for each of these 1000 models. To put these simulations into clinical context, they used their pharmacodynamic model for MOAA/S to predict sedation levels over a 40-minute period based on the plasma concentration estimates. They found a dosing regimen that minimized excessive and inadequate sedation. Some of the 1000 models had brief episodes of an MOAA/S score <2, but the majority had an MOAA/S score between 2 and 4, and none remained at 5. This exemplifies the utility of simulation to explore drug dosing and associated effects that can be used to better guide phase II clinical trial development and perhaps decrease cost and the number of subjects needed to identify effective doses. As with the pharmacodynamic CSHT, the Monte Carlo simulations make predictions of sedation using a pharmacodynamic model created in the absence of a constant stimulus. It is likely that in the presence of a painful stimulus, more subjects may have higher MOAA/S scores. Another limitation to models and simulations is that they often struggle to account for special populations whose response to anesthetic drug therapy may vary significantly from what is observed in healthy volunteers. As any clinical pharmacologist will tell you, there is more work to be done to better describe how body habitus, cardiovascular depression, age, and chronic exposure to opioids and benzodiazepines influence model predictions of drug concentrations and effect. In conclusion, the authors have used innovation to create softer (i.e., more predictable, faster), cleaner (i.e., less adrenal suppression) anesthetics and to expedite their development and better characterize behavior through model building and simulation. Through their efforts, remimazolam and the MOC etomidate analogs will likely bring interesting advances to our specialty in the coming years! DISCLOSURES Name: Ken B. Johnson, MD. Contribution: This author wrote the manuscript. This manuscript was handled by: Tony Gin, FANZCA, FRCA, MD.