Abstract
Modern balanced anesthesia consists of the use of a number of drugs to achieve the desired clinical endpoints, such as no awareness, no movement, muscle relaxation, control of blood pressure and control of heart rate in the face of the noxious surgical stimulus. Understanding the methodology used for the study of drug interactions and knowing the interactions that occur between commonly used anesthetic drugs is important for using them effectively and safely.
Keywords
drug interactions, synergy, additivity, response surface methodology, propofol, opioids, volatile anesthetics
Chapter Outline
Understanding Drug Interactions Among Commonly Used Anesthetic Drugs
Inhaled Anesthetics and Opioids
Inhaled Anesthetics and Other Agents
Propofol and Inhaled Anesthetics
Ketamine and Midazolam or Propofol
Drug Interaction Displays on Monitors
Pharmacodynamic-Based Target Control and Closed-loop Infusions
History
Pharmacology is characterized by detailed descriptions of the actions of individual drugs. In the case of anesthesia, no single drug has been found to be universally satisfactory and indeed general anesthesia is now regarded as a set of desirable clinical endpoints rather than a discrete phenomenon of its own. Useful endpoints include lack of awareness and a pleasant induction and recovery; lack of movement, adequate muscle relaxation, reasonable blood pressure control; and maintenance of homeostasis by suppressing autonomic reflexes while balancing the narrow therapeutic index of many of the drugs to ensure positive outcomes. Consequently, multiple drugs are used for all but the simplest of procedures.
This concept of multiple drug use was first described by Lundy in 1926, who used the term balanced anesthesia. His concept included the liberal use of local or regional anesthesia as well as hypnotics, volatile anesthetics, and opioids. At the time, use of volatile agents as a sole anesthetic was common, but unacceptable cardiovascular and respiratory depression frequently accompanied the high doses required to suppress movement to noxious stimuli—a finding still true of the volatile anesthetics in use today.
The theory behind balanced anesthesia is that using desirable drug combinations reduces dose requirements of individual drugs, minimizing the incidence of unwanted side effects in those with a narrow therapeutic index and so improving the quality of anesthesia. A review of the contribution of anesthesia to perioperative mortality found evidence to support this notion with the relative risk of dying within 7 days of surgery being 2.9 times greater if one or two anesthetic drugs were used, compared with the use of three or more. Today, balanced anesthesia is the standard approach to general anesthesia.
Many drugs used in anesthesia have overlapping actions and, when given in combination, can be used to produce effects distinct from those they create individually. To maximize the clinical utility of any drug, it is important to understand its effects when used alone and in combination. The interaction between propofol and fentanyl is a notable example. Propofol alone is effective at causing unconsciousness without intolerable adverse effects. The much larger doses required to prevent movement in response to surgical pain also suppress respiration and can unacceptably reduce blood pressure. On the other hand, fentanyl is incapable of reliably causing unconsciousness and even in very high doses does not reliably suppress movement to pain. When these two drugs are used in combination, there is only a modest reduction in the dose of propofol required to cause unconsciousness, but a dramatic reduction in the dose required to suppress movement to pain with less reduction in blood pressure. The relative dosage of the combination of propofol and opioids also allows a degree of independent control of two critical variables—unconsciousness and lack of movement in response to pain.
Early studies of drug interactions include that by Fraser in 1872, who described pharmacologic antagonism between physostigmine and atropine, and by Frei in 1913, who used isobolographic analysis to demonstrate increased effectiveness of combinations of disinfectants for killing bacteria compared with individual agents. The response surface method, which is now used to describe the entire dose response relationship of two drugs, was first proposed by Loewe in 1928, who described synergy as an “inflated sail” and additivity as a “tense sail.” The concept was introduced into modern anesthetic clinical pharmacology analysis by Minto et al. in 2000. The seemingly daunting task of describing an entire response surface for the interaction between two drugs was found to be tractable and as few as 20 intensively studied subjects are needed to adequately describe the interaction between two drugs. The tedious calculations are easily performed on a modern computer, something Loewe clearly lacked. Response surface methodology is now the basis of many studies of commonly used drug combinations. Although outwardly complex, the models can be used to increase the accuracy of anesthetic doses and improve prediction of expected effects in patients. They have been incorporated into some anesthetic monitors for real-time display and can be used to optimize drug titration, promoting outcomes such as reduced wake-up times or increased cardiovascular stability without increasing hypnotic depth.
This chapter introduces terminology used to describe drug interactions, the methodology used to study drug interactions, and the important interactions between commonly used anesthetic drugs. Some of these interactions are covered in other chapters of this text in more detail (see Chapters 2 , 10 , 11 , and 17 ). Interactions among combinations of analgesics or antiemetics as used postoperatively are not reviewed in this chapter; however, similar methodologies and rules can be applied when considering their use in combination.
Study of Drug Interactions
Terminology
Drug interactions are usually considered in the dose or concentration domain. A simple experiment that illustrates interactions between two drugs takes half the dose of each that alone causes a certain level of effect. Assume that level of effect is equal to 1. If the drugs are additive, one would expect these two half doses given together to produce the same effect as giving the whole dose of either drug alone (i.e., 0.5 + 0.5 = 1). This expected effect for a dose combination becomes the null hypothesis against which one can assess the presence (or absence) of a positive or negative interaction. If the observed effect is greater than expected (<0.5 + <0.5 = 1), synergy exists. A commonly used drug combination in which this occurs is propofol and midazolam. Together, the individual doses are reduced to about one third so that 0.33 + 0.33 ≡ 1. If the observed effect was less than expected (>0.5 + >0.5 = 1), infra-additivity exists ( Fig. 6.1 ). If instead one combines half the effect of two drugs, the answer would be very different. This is because the sigmoid log-dose effect relationship is highly nonlinear, meaning drug effects cannot be simply added.
The term infra-additive is used when less effect than that expected from simple additivity is observed. The combination of midazolam and ketamine exemplifies the effect of infra-additivity. Midazolam has only a moderate effect on the ketamine dose required to suppress response to verbal command and no effect on the ketamine dose required to suppress movement to a noxious stimulus.
The term antagonism is reserved for interactions in which there is an absolute reduction in the effect of one drug in the presence of the other. For instance, the analgesic effect of fentanyl is reduced in the presence of naloxone. In this example, naloxone is incapable of causing analgesia when given alone. An underlying assumption of this definition of additivity is that a drug cannot interact with itself, so two half doses of the same drug must be additive.
Shift in Dose-Response Curve
The dose-response relationship for most anesthetic drugs can be described using the standard sigmoid E max model. Its characteristics include (1) a threshold drug concentration that must be surpassed before any effect is seen, (2) an increase in effect proportional to logarithmic increases in drug concentration, and (3) saturation after which additional increases in drug concentration no longer produce an increase in effect ( Fig. 6.2 ).
When a second drug is introduced, the simplest model of an interaction describes its influence, at a single fixed dose, on the dose-response relationship of the first drug. Using the previous example, intravenous fentanyl 1 µg/kg, administered immediately before intravenous propofol, shifts the propofol dose-response curve for unconsciousness to the left. This equates to a 20% reduction in dose for that endpoint. Likewise, the curve for suppression of movement to a noxious stimulus shifts to the left by 50%. This model demonstrates that fentanyl affects the noxious endpoint more than the hypnotic endpoint, which is expected from our knowledge of the individual drugs.
This approach to studying interactions accounts for the dose requirements for the two drugs when given together, but describes the interaction at only one dose of fentanyl. It does not describe whether the interaction with the second drug is additive or synergistic or whether there is a therapeutic advantage to using the combination. To quantitate the interaction, and to decide whether synergy exists, a description is needed of the drug’s effects when given individually, as well as when given together, and for a range of doses. Use of isobolograms is an alternative approach to interaction analysis that partly addresses these limitations.
Isobolograms
The traditional approach to quantitating a drug interaction for two drugs is to construct graphs of the dose pairs that together produce a single level of effect. These isoeffect lines are called isoboles . Comparison of an isobole with a line of additivity signals the interaction type: a straight line indicates additivity; inward bowing indicates synergy; outward bowing indicates infra-additivity or antagonism (see Fig. 6.2 ).
Isobolograms are easy to construct and analyze when both drugs are individually capable of producing the endpoint in question. Numerous studies in the literature use this methodology and their conclusions are comparable to those of studies using more complex methodologies. Typically a departure from additivity of less than 10% has been regarded as additive on the basis that even if statistically significant, it is clinically irrelevant. When one drug cannot achieve the endpoint in question, it is still possible to determine whether synergy exists, but the exact degree of departure from additivity cannot be accurately determined. A limitation of using isobolograms is that the result is only applicable to the effect level investigated. To gain a full understanding of the interaction between two drugs for all effect levels, it is possible to overlay a series of isoboles ranging from minimal effects to the maximal effect attainable (E max ).
Analysis of isobolograms involves determining the dose-response relationship for each drug individually and then exploring the dose-response relationship of the drugs in combination. Simple experiments investigate dose pairs that maintain a fixed dose ratio of the individual drugs. It is also possible to combine any number of dose combinations and responses and to calculate an isoeffect line using nonlinear regression—this is now the preferred method ( Fig. 6.3 ). The standard isobologram assumes that the potency ratio of the two drugs remains constant for all dose pairs. In some instances this assumption does not hold true, such as when individual dose-response curves reach different levels of maximum effect. These cases require special consideration in that the line of additivity may in fact be curved. Consequently this can lead to false identification of synergistic relationships. Although isobolographic methodology is appropriate for studying interactions of these types, they are better described by more sophisticated analyses.
Response Surface Models
Response surface methodology allows one to create a complete description of the dose-response relationship between two drugs for all levels of effect for a given endpoint. That response surfaces characterize the entire spectrum of drug effect is an important advantage of this approach in anesthesia clinical pharmacology. Anesthesia practice is unique in that anesthetists must target profound drug effects during an operation and then reverse the effect rapidly when the operation is finished. Thus to study anesthetic clinical pharmacology fully, it is necessary to understand the entire concentration-effect relationship and how to transition from full effect to recovery efficiently.
These models all assume that the interaction is constant over time (i.e., stationary). In creating response surfaces, it is also assumed that the surface is smooth and that individuals in the study population show a similar degree of drug interaction. The latter is important because it is not practical to study more than a small part of the entire dose-response surface in each patient. Response surface models are typically built in carefully controlled volunteer studies in which multiple measurements can be made; the resulting models are then validated in patients. The concept embraces previous approaches to drug interaction analysis, including the shift in the dose-response curve and the isobologram. It also provides a strong visual representation of the interaction that is easy to remember given the complexity of the mathematics involved. Response surfaces give a mathematical description of the magnitude of an interaction for all dose pairs of two drugs, which enables these models to be used in real time to predict the likely response of patients to the drugs used. They also allow identification of optimal drug dose combinations. For example, remifentanil 0.8 ng/mL:propofol 2 to 3 µg/mL was the optimal concentration pairing to prevent movement and blood pressure changes in response to esophageal instrumentation without causing intolerable ventilatory depression. The study is significant because it identifies the optimal dose ratio for a useful clinical endpoint while minimizing the risk of an unwanted side effect.
Response surface methodology is suitable for studying all the common types of drug interactions among agonists, partial agonists, antagonists, and inverse agonists ( Fig. 6.4 ).
Trial Design
The basic steps to deriving a response surface are described in Fig. 6.5 . Three different mathematical approaches have been chosen—Greco, Minto, and Bouillon et al.—but the basic principles underlying them are similar. There are also other valid approaches. Although each model draws a slightly different shape for the surface, the differences are not large. The three approaches are empirical in that they make no assumption about the nature of the underlying interaction.
The Greco model is simple to apply and uses a single interaction parameter. In doing so, it assumes the interaction is constant for all drug pairs and gives the surface a symmetric shape. This is an advantage when coverage of the surface is uneven but may lead to forcing the data to fit a shape that the data do not actually possess. (This model is used in Fig. 6.11 and 6.12 .)
The Minto model is capable of capturing virtually any shape across the response surface. It preserves the sigmoid shape of the dose-response curve for all drug combinations and is suitable for describing all commonly encountered drug interactions in anesthesia, including those between a pharmacologic agonist-antagonist combination (see Fig. 6.4 ). Interactions can be applied to each of the variables in the sigmoid E max equation individually. This facilitates describing interactions between those drugs whose individual response curves are diverse in terms of curve shape, slope, and maxima, and allows the investigator to determine the required level of model complexity, although the model has been criticized for having too many parameters. (This model is used in Fig. 6.7 .)
The model of Bouillon et al., also known as the hierarchical model, sequentially processes drug effects for one drug, then the other drug. It is simple to implement but has the quirk that parameter estimates will vary depending on their order of introduction into the model. Although the order by which drugs should be introduced into the model might appear logical, there are no formal criteria established and the order of analysis that results in the best fit to the data can vary for different endpoints within the same dataset. The model makes the assumption that the relationship follows a sigmoid dose-response curve but does not make a mathematical assumption about the shape of the combined surface. There is no interaction parameter, but instead goodness-of-fit can be assessed against other models describing the data. (This model is used in Fig. 6.9 .) This model has also been extended to describe the effects of three or more drugs.
Generic trial designs for the three methods for studying drug interactions are summarized in Fig. 6.6 . Suitable methodology and power analysis for response surface trial design have been determined by simulation analysis. It is essential that trials cover most of the surface for robust models to be derived. Endpoints can be either continuous (e.g., blood pressure or bispectral index) or quantal (e.g., response to verbal command or surgical incision). The type of interaction is identified by comparing an additive model, where the interaction parameter is excluded or is made to equal zero, with one that has an active interaction parameter. The choice of the best model is made by visual inspection of the surface, comparison of residual plots, and statistical testing of the reduction in log likelihood between related models. The response surface method can be extended to three or more drugs, although clear visualization becomes impractical beyond three drug interactions ( Fig. 6.7 ).
Related posts:
Full access? Get Clinical Tree
