Analgesic combinations

Abstract

When the pathophysiology of a medical condition is multi-modal, i.e., related to multiple physiological causes or mediated by multiple pathways, the optimal strategy can be to use a drug or a combination of drugs that contribute multiple mechanisms to the therapeutic endpoint. In such situations, a rational multi-modal approach can also result in the fewest adverse effects. We discuss the quantitative analysis of multi-modal action using the treatment of pain as a practical example and give examples of its application to some widely used analgesic drugs.

Background

When the underlying pathophysiology of a medical condition is primarily mono-modal, i.e., related to a single mechanism, the most selective mono-modal (single mechanism) drug provides the greatest separation between the therapeutic (‘target’) and adverse (‘off target’) effects.

However, not all medical conditions are mono-modal. This is the case with pain. Nociceptive information is transmitted through numerous anatomical pathways and by multiple neurochemical substances. There is no single pain pathway and no single pain transmitter substance. It is unlikely, therefore, that all types of pain will ever be treated with an analgesic having only a single analgesic mechanism. For this reason a multi-modal approach has been advocated by us and by others.^40,42,46

Historically, pain has been characterized, and treated, principally on the basis of its degree –– e.g., ‘weak’, ‘moderate’, or ‘severe’. This was a pragmatic approach, and clearly justified given that the mechanism of action of the available analgesics was not yet known. Indeed, there was no basis for a pharmacologic understanding of analgesics until the discovery in the 1970’s of the mechanism of action of aspirin and other NSAIDs^a by Sir John Vane and colleagues^53,64 and of the opioids by several groups.^39,49,61 Clinical guidance regarding the treatment of cancer pain, such as the WHO (World Health Organization) ‘analgesic ladder’⁶⁶ generally involved matching the degree of pain with the analgesic category, namely, NSAIDs for Step 1 (‘weak’) pain and opioids for Step 3 (‘severe’) pain.

With the advances that came from more recent study of pain mechanisms there also came the recognition that pain is better categorized on the basis of its etiology rather than its degree.⁶⁵

Pain is Multi-Factorial

Most pains are multi-factorial in nature due to the anatomy and physiology of pain transmission (Fig 1).^18,24,65 As an example, when acute pain originates in the periphery nociceptive information is transmitted to the spinal cord by at least two types of primary afferent neurons (Aδ and C). Several pathways then transmit the signal to higher centers in the central nervous system. Chronic pain can involve peripheral or central sensitization phenomena or recruitment of new afferents that activate higher centers of pain processing. In addition, many neurochemicals are involved in the transduction (detection), conduction, modulation, or prolongation of the sensation or perception of pain. Thus not all pain is the same and understanding the mechanism of action of the pain generator provides a more target approach when choosing an analgesic.

Fig 1.

Anatomy of pain. Composite based on multiple sources. ¹⁸,²⁴,⁶⁵

Because of the multi-factorial nature of pain physiology, it is unlikely that a mono-modal analgesic mechanism will adequately treat all pains. It seems more likely that a variety of drug mechanisms, within the same agent or a combination of agents, would yield the most pain relief with the fewest adverse effects. Of the commonly available analgesic drugs, most have predominantly mono-modal mechanisms of analgesic action. ⁴¹

Mono-modal Analgesics

NSAIDs exert their analgesic action by inhibiting the conversion of arachidonic acid to prostaglandins catalyzed by COX (cyclooxygenase) isozymes. However, the inhibition of COX enzymes can lead to a range of undesirable and sometimes fatal short-and long-term organ toxicities, including gastrointestinal ulceration and bleeding. ^29,50 The discovery of the COX-2 isozyme and the development of effective anti-inflammatory drugs that were selective for COX-2, while sparing COX-1, appeared initially to offer a solution to this problem. However, some COX-2 inhibitors have been linked with an increased risk of cardio-renal effects. ^28,36,54 Most clinicians now try to minimize the long-term use of high doses of NSAIDs. Indeed, The American Geriatrics Society (AGS) released new guidelines addressing the pharmacological management of persistent pain in older persons in which a major change was the near elimination of utilizing NSAIDs because the risk of serious adverse events was deemed to outweigh the benefits.¹

Acetaminophen (paracetamol) exerts its analgesic action by a mechanism yet to be elucidated. Multiple targets have been proposed, including COX-1, COX-2, a putative COX-3 isozyme, NO (nitric oxide) synthase, 5-HT (5-hydroxytryptamine, serotonin) receptors or neuronal reuptake sites, and a variety of others.⁵² None has been proven, but it is fairly certain that a central (brain and/or spinal cord) site of action is involved.⁴⁴ The risks related to the use of acetaminophen have prompted increased concern by physicians and regulatory agencies. Acetaminophen will now carry labeling that warns consumers about potential safety risks, including internal bleeding and liver damage, when containing it are taken to excess or taken along with certain other drugs, such as anticoagulants or steroids.¹⁴

Opioids exert their analgesic effects by actions that have been well characterized. All opioid analgesics are agonists at one or more of the three major 7-transmembrane G protein-coupled opioid receptors, named μ (for morphine), δ (for vas deferens), and κ (for ketocyclazocine). Opioid receptors are located along pain transmission pathways and throughout the CNS. Activation of opioid receptors decreases presynaptic release of neurotransmitters and hyperpolarizes the postsynaptic neuron; both of which inhibit transmission along the affected pathways. ⁴⁸ Opioids are commonly used for severe chronic pain because the dose can usually be adjusted to meet the required pain relief. In most cases, the limiting factor is ability to tolerrate side effects (mainly CNS and GI), but tolerance to the drug or lack of efficacy can sometimes be limiting factors.^25,26

Norepinephrine and 5-HT reuptake inhibitors decrease synaptic clearance and hence increase synaptic levels of the associated neurotransmitters. In both cases, descending pathways (midbrain to spinal cord) are enhanced and modulate the incoming pain signal. This mechanism is sometimes beneficial for particular types of pain and agents with dual effects on norepinephrine and 5-HT can provide more consistent benefits than selective reuptake inhibitors for the treatment of certain types of pain, such as fibromyalgia. However, difficulty in titration and adverse effects (e.g., nausea and vomiting) can limit the use of these agents. ²

Local anesthetics inhibit voltage-gated ion channels that are located on sensory neurons. Inhibition of Na⁺ influx inhibits depolarization of these (e.g., primary afferent) neurons and, thus, fewer action potentials are generated and fewer nociceptive impulses are transmitted to the CNS. ⁶² Systemic absorption from patch formulations of these drugs must be considered in patients receiving oral class 1 antiarrhythmic drugs.^12,19,34,47

Gabapentin and pregabalin appear to block specific subunits of voltage-gated Ca²⁺ channels, thereby inhibiting the release of excitatory neurotransmitters. ^16,17 It is not clear if this mechanism also accounts for the adverse effects of these agents, such as somnolence and dizziness, mild peripheral edema, gait and balance problems, and cognitive impairment in the elderly.^3,4,12

A variety of additional analgesics and adjuvant agents are also used, including α₂ agonists, anticonvulsants, antidepressants, corticosteroids, NMDA antagonists, topical agents, cannabinoids, triptans, and others. ⁴⁶

Selecting multi-modal mechanisms is a rational approach to achieving the goal of controlling multiple types of (multi-modal) pain. Combinations offer the possibility of several desirable outcomes.^22,27,38,40 However, the actual attributes of any particular combination must be rigorously tested.

Multi-modal Analgesics

Certain analgesics produce their effect by a combination of individual (mono-modal) and interactive mechanisms in a single drug. Examples of such ‘multi-modal’ analgesics include duloxetine, milnacipran, and venlafaxine (combined norepinephrine and 5-HT reuptake inhibition mechanisms),⁹ tramadol (combined opioidergic and monoaminergic mechanisms),⁴³ and the recently-approved tapentadol (combined opioidergic and noradrenergic mechanisms).⁶³ Also in this category are drugs having activity at multiple subtypes of a receptor, such as the mixed-action μ/κ opioids such as (to different extents) butorphanol and pentazocine.

Drug Combinations

The use of a combination of two drugs with overtly similar effects is common in medical practice. Prominent examples include analgesics, chemotherapy combinations, antihypertensives, antiemetics, etc. The usual aim in poly-drug use is to treat the situation with drugs that have different mechanisms of action. Because pain involves multiple mechanisms the use of combinations is especially rational. Also the use of two drugs almost always means a combination with lower doses of each, thereby minimizing the adverse events that might be associated with higher doses of a single drug. A complete review of the preclinical and clinical testing of specific analgesic combinations is beyond the scope of this review. However, Smith⁵¹ provides an extensive review of combination opioid analgesics, reviewing the many specific examples of the potential advantages (including both pharmacokinetic and pharmacodynamic) of these combinations and describing specific details of aspects of the preclinical and clinical testing of them.

Very often the drug combination is described as “additive”. Equally often the term “additive” is not properly defined. Also, exaggerated combination action is termed “synergistic” while reduced action is called “sub-additive”. A precise definition of these terms requires a quantitative component, that is, some metric that allows a differentiation between what is expected and what is unusual. Our aim here is to describe the several metrics that have been used to characterize drug interactions and these will apply whether the mechanism is known or unknown. In the context of pain reduction we often lack detailed information on mechanism; however, the finding of a non-additive (unpredicted) interaction can be a valuable first step in uncovering mechanism. In what follows we will consider the case of two agonist analgesics, which we designate as ‘drug A’ and ‘drug B’.

Dose-Effect Relations

The first step in the assessment of a drug combination^56,57 is obtaining information from each drug acting individually. This information takes the form of a dose-effect relation of each. [For convenience in expression we shall use the terms dose and concentration interchangeably here since dose-effect relation is more common than concentration-effect relation]. A common component of the dose-effect relation of a drug is the assumption that zero dose should give zero effect whereas a sufficiently large dose will yield the drug’s maximum effect. A typical over-the counter analgesic will produce some maximum effect, but it would not be the same as the maximum (pain relief) achieved by morphine. Very often a strong agonist such as morphine is used as a standard against which the test of a drug’s efficacy is measured and its maximum effect is taken to be 100%. With such a standard established the test drug A is commonly modeled according to the hyperbolic form

$$E=\frac{E_{A}a}{a+C_A}$$ (1)

where E is the effect, E _A is the maximum effect of drug A, a is the concentration (or dose) and C _A is a constant related to the drug’s potency. A similar form applies to drug B:

$$E=\frac{E_{B}b}{b+C_B}$$ (2)

In this common hyperbolic model, E _A = E _B implies that the relative potency is a constant. (More complicated forms, using exponents, may be required for a good curve fit; e.g., $E=\frac{E_B{b}^p}{b^q+{C_B}^p}$ and $E=\frac{E_A{a}^q}{a+{C_A}^q})$.^56,57 It is seen that the terms C _A and C _B represent concentrations that yield the half maximum effect value of drugs A and B, respectively. Equation [1] and Equation [2] yield sigmoidal curves when the effect is plotted against the logarithm of dose, and that kind of plot displays a somewhat linear curve that has historically been used because of the convenience and simplicity of straight line plots. The dose-effect equations for the two drugs contain the fundamental values needed to analyze the combination interaction. The basis for that analysis is the concept of dose equivalence.

Dose Equivalence and the Linear Isobole

For any level of effect that is achieved by both drugs one can easily find the concentrations (or doses) that are equally effective by equating the right hand sides of equation [1] and equation [2]; thus, the b-equivalent (b_eq) of concentration a of drug A is

$$b_{eq}=\frac{C_B}{\frac{E_B}{E_A}\left(1+\frac{C_A}{a}\right)-1}$$ (3)

In the special case in which E _A = E _B, the above becomes $b_{eq}=a\frac{C_B}{C_A}=\frac{a}{R}$ where R is the potency ratio C _A/C_B. This special case leads to the well-known linear isobole, a plot of all concentration pairs that give the same effect. In this plot a particular effect level E* is selected, often the half maximal effect, but any common effect is used. If drug B, acting alone, gives this effect in concentration B, we see that the pairs (a, b) are related to b plus the b-equivalent of a such that b + b _eq = B. This may be written b + a/R = B which can be re-arranged to the familiar intercept form shown as equation [4] and plotted in Fig 2.

$$\frac{b}{B}+\frac{a}{A}=1$$ (4)

The concentration pairs (a, b) are the points that constitute this straight line and these are the pairs in the combination that give the specified effect E*. Because the dose of drug B is added to its drug A-equivalent, the isobole is termed additive. It is notable that this derivation assumes that the equivalent dose acts in the combination in the same way that it acts when alone; in other words, there is no alteration or interaction so that the contribution of each constituent is related to its own potency. The derivation given here also makes clear that the straight-line isobole of additivity applies if and only if the relative potency is a constant across the entire range of effect (such as R). This graphical procedure, introduced and used by Loewe^31–33 has been widely employed to determine whether drug combinations are additive or non-additive. In these determinations the procedure is to test various combinations that give the specified effect. If the experimental combination (dose pair or concentration pair) plots as a point below the line, indicating that lesser doses are needed, then the combination is synergistic (Fig 2). It follows that the effect of a predicted dose pair (on the line) would lead to some effect greater than E*. An interaction has taken place so that the effect observed is enhanced. In contrast, testing may reveal that a dose pair plots as a point above the isobole in attaining the specified effect E*; this means a sub-additive interaction (Fig 2). Numerous preclinical studies of analgesic combinations have employed the isobolographic approach.^{6,7,11,13,15,23,30,35,44,45,60}

Fig 2.

(A) The solid straight line is the additive isobole for two drugs with constant relative potency such that the specified effect requires doses a and b of drugs A and B, respectively, when each drug acts alone. Experimental dose combinations above the isobole (point P) indicate sub-additivity whereas a point below the line (point Q) indicates a super-additive (synergistic) interaction. (B) When one of the drugs lacks efficacy (e.g., drug A) the additive isobole is horizontal with intercept b representing the dose of drug B that gives the specified effect. In this case, as in the upper figure, experimental points P and Q denote sub-additivity and synergism, respectively. (C) An experimental dose combination (a′, b′) below the additive isobole calculates as a′/A + b′/B = γ, thereby lying on a parallel line (shown dotted) that has intercepts γA and γB, where γ is the interaction index. The radial distances to this experimental point is denoted by r while the distance to the additive isobole at (a, b) is R from which it follows (trigonometrically) that r/R = γ. It is equally evident (from trigonometry) that the ratio of the experimental total dose to the additive total, (a’+b’)/(a+b) = r/R = γ.

An especially interesting case is that in which one of the two drugs lacks efficacy. In this case the additive isobole is horizontal and, thus, has an intercept only on the vertical axis that represents the dose of the active drug for the specified effect. In this case, an experimental point (a, b) below the horizontal is synergistic while a point above is sub-additive. This situation is exemplified in the combination of ibuprofen and glucosamine. The latter has no effect in standard analgesic tests, yet the combination, in certain ratios, was shown to be synergistic in standard tests of antinociception. ⁶⁰

While the isobologram is familiar and convenient to construct and view, its mathematical basis is the concept of dose equivalence and, thus, there are non-graphical ways to test combinations that use the dose equivalence concept. In one application⁵⁸ we calculated the expected (additive) total dose of the combination tramadol and acetaminophen and showed that in several fixed ratio combinations the total found by experiment was less than the additive total, a clear demonstration of synergism. This drug combination was subsequently developed as the analgesic Ultracet^®.

Fixed Ratio Combinations: The Interaction Index

In many experiments with combinations the constituent doses are administered in a fixed ratio. This procedure allows a simple determination of the combination doses (or concentrations) that produce the specified level of effect. This is accomplished by fitting the dose-effect data in an appropriate regression procedure. When viewed on the illustrative isobologram of Fig 2, this fixed ratio is indicated by the broken radial line. The intersection of this line with the additive isobole gives the dose pair that is additive and it also allows a quantitative assessment of the departure from additivity. The ratio of the radial distances (experimental point/additive intersection) has been termed the interaction index, ⁵⁵ a value here denoted by γ such that a/A + b/B = γ (compare with equation [4]). With reference to Fig 2C it is seen that an experimental point (a′, b′) at radial distance r represents a total dose a′ + b′ = r cos θ + r sin θ. In contrast, the additive total dose on the isobole is at radial distance R so that this total = R cos θ + R sin θ. Thus, the ratio of experimental to additive totals is r/R = γ (see legend for further detail). Representative examples of combinations analyzed and quantitated with the interaction index are given in Tallarida et al⁵⁹ and in Codd et al¹¹. In the former, morphine and clonidine were given intrathecally to mice and tested in the 55 °C water tail immersion test. Three different fixed ratio combinations were used and yielded values of γ = 0.3, 0.05 and 0.1. In the tests by Codd et al¹¹ antiallodynia in mice was assessed with various combinations of tramadol and topiramate and gave values of γ = 0.2, 0.3 and 0.6. The examples further illustrate that synergism and the γ metric depend not only on the drugs but also on the ratio of the drug combination.

Error Estimates

The linear isobole of additivity, though applicable only in cases of a constant relative potency, is nevertheless convenient for its simplicity and for estimating the variance (square of the standard error) of the additive total dose. Hence, it has been widely used and yields generally acceptable results even in cases in which the relative potency departs somewhat from a constant value. All points (a, b) on the additive isobole can be expressed as fractions (f and (1 – f )) of the respective potencies A and B, i.e., a = fA and b = (1 – f )B. Therefore, a combination with constituent amounts chosen such that doseB/doseA = (1 – f)B/fA has a total quantity T = fA + (1 – f)B. The variance of the additive total T is given by a standard formula V(T) = f ² V(A) + ( 1 – f)² V(B). Having this total variance allows estimation of the component variances from the proportions p _A and p _B in the combination, e.g., V(a) = p _A ² V(T) and V(b) = p _B ² V(T). [Strictly speaking, one does not know f (and 1 – f) precisely because A and B are estimated quantities. In practice, however, these fractions are reasonably estimated from the mean A and B.] We illustrate the error calculation with a numerical example:

Suppose that A = 50 with standard error 8 and B = 20 with standard error 5, each determined from regression analysis. Let the chosen fraction f = ½ so that the additive drug A-component is 25 and that for drug B is 10, giving a total dose T = 35. The proportions in the combination are therefore p_A = 25/35 and p_B = 10/35. Then V(T) = (1/2)²(64) + (1/2)²(25) = 22.25 from which V(a) = (25/35)²(22.25) and V(b) = (10/35)²(22.25). The square root of each yields the standard errors, S.E.(a) = 3.37 and S.E.(b) = 1.35, which are often graphically indicated on the additive point of the isobologram. The Student t-test for the difference of two means (experimental – additive) is the appropriate statistical test. The additive mean and standard errors are calculated as described above, and the experimental values of a and b are determined from the experiment. This information has been added to the revision.

Response Surface

For two drugs with a constant relative potency the drug-B equivalent of dose a of drug A is seen to be a/R as shown above. Thus, the additive effect (a, b) combination is (b _eq, b) = (a/R, b). When inserted into the concentration-effect relation of drug B we get

$$E=E_B\frac{(a/R+b)}{(a/R+b)+C_B}$$ (5)

This relation expresses the effect as a function of a and b and its graph yields a surface whose height above the planar (a, b) point is the additive (expected) effect for zero interaction (see Fig 3). In contrast to the isobole method which determines and compares doses that give a specified effect, the response surface provides a view of the effect height of the experimental combination in relation to the surface height (the additive effect). Because this compares the two effect heights it is visually more obvious than the isobologram, although both methods provide the same information regarding departures from simple additivity. It is therefore an alternative method that shows additivity (and departures from additivity) and does so over the entire effect range. A response surface analysis was applied to show analgesic synergism for the combination of morphine and clonidine. ⁵⁹ In this kind of analysis the additive surface is constructed from equation [5]. The test for an interaction examines the actual effect attained with combination (a, b) and its height will place it either on the surface (additivity), below the surface (sub-additivity) or above the surface (synergism). The test of these heights (effects) in relation to the surface height is a more challenging statistical endeavor since exact variance formulas for the additive surface height do not exist. An approximate method known as the ‘delta’ method may be employed to estimate this variance.¹⁰

Fig 3.

The response surface for doses of drugs A and B is a plot of surface height (effect) against concentrations a and b of drugs A and B, respectively. Combinations (a,b) represented by a point in the plane give an effect whose magnitude is the height of the surface above the (a, b) point when the interaction is simply additive. The surface therefore allows a comparison to the actual effect of the dose combination to the expected additive effect. An observed effect significantly above the surface indicates synergism while an effect level below the surface indicates sub-additivity.

Variable Potency Ratio

Loewe³³ recognized that a variable potency ratio would yield an additive isobole that was curvilinear but his publications provided no mathematical derivation that defines this case. The most clear cut case of a varying potency ratio is seen with two agonists that produce significantly different maximum effects. This situation was addressed by Grabovsky and Tallarida²⁰ and is summarized here for the case of agonists A and B that yield maximum effects E _A and E _B respectively as defined in equation [1] and equation [2]. The dose equivalent of dose b is given by equation [3]. Thus, for a selected effect that requires dose B of drug B when it acts alone, it follows that b + b _eq = B as previously noted but, in this case, b _eq is not a/R; instead it is given by equation (3). This yields the additive isobole given by

$$b+\frac{C_B}{\frac{E_B}{E_A}\left(1+\frac{C_A}{a}\right)-1}=B$$ (6)

This graph, clearly nonlinear, is shown in Fig 4. Further details are given in reviews.^56,57

Fig 4.

The additive isobole for drugs with a variable relative potency (due to different maximum effects) and which, acting alone, require doses A and B. The additive isobole is curved in this case of a variable relative potency.

The derivation leading to the nonlinear isobole provides a new methodology that can be extended to situations in which a single drug interacts with two binding sites (receptors) that each contribute to a common effect. Concentration-effect relations of the individually occupied sites will almost certainly exhibit equi-effective concentrations that vary with the effect level. Thus, the derivation leading to the nonlinear isobole is applicable and readily extended to the dual-site case. This extension is based on the fact that dose-effect data can be transformed to occupation-effect data. Therefore, when only a single receptor is occupying the agonist (e.g., when the other is blocked or the preparation has knocked out one receptor) the agonist concentration is used to get the receptor occupation from the law of mass action. The resulting occupation-effect relations of each of the two receptors will likely reveal different maximum effects and these individual relations are used to get an additive isobole (analogous to equation [6]). Thus, equation [6] now has occupation values instead of the concentrations a and b as well as constant C _A and C _B derived from the occupation-effect graphs. In contrast to the usual two-drug approach in which doses (or concentrations) are under the control of the experimentalist, here we have a single drug in varying concentrations that leads to a radial path that is defined from its affinity for each receptor. The intersection of the path with the nonlinear isobole of occupation provides the additive point, just as it does with the common linear isobole. The experimentally derived occupation pair (necessarily on this path) may be either ‘on’ or ‘off’ the isobole, thereby indicating synergism or sub-additivity just as it does in the standard two-drug case.

This extension may be useful in ligand-based drug designs that attempt to find molecules that bind to two sites that contribute to the analgesic effect. The newly-approved drug tapentadol is an example of two modes of action combined in a single molecule (μ-opioid receptor agonism plus norepinephrine reuptake inhibition). ⁶³

Combinations in Clinical Practice and Therapeutic Index

Fixed-ratio combinations

In clinical practice it is becoming more common to use analgesic combinations, both in fixed-dose ratios and ad hoc dose-ratios, to treat chronic moderate to severe pain and this is especially relevant when managing ‘mixed’ pain disorders. Under very close supervision and diligent patient compliance, ad hoc dose-ratios can provide customized therapy. However, the inherent problem associated with ad hoc ratio dosing –– difficulty maintaining the dose-ratio within the optimal range –– limits the more widespread utility of this strategy because of poor interaction indexes, adverse-event inflation, and discouragement of self-titration by the patient. Fixed-ratio dose combinations produce a more standardized reproducible clinical effect. For example, a fixed–ratio combination of acetaminophen 325 mg plus tramadol 37.5 mg provided standardized reproducible improvement in pain scores and side effect profiles in multicenter randomised controlled trials for the management of chronic back pain and fibromyalgia.^5,8,37

It should be pointed out that when a drug combination is administered multiple times, differences in time courses of their ADME (absorption, distribution, metabolism, or elimination) can result in greater accumulation of the longer acting drug and a functional shift in drug concentration ratios.

Therapeutic index

A key rationale for the use of drug combinations is the reduction of adverse effects. This suggests that tests to assess synergy/additivity/sub-additivity using only a single therapeutically-relevant tests are not sufficient and may not even be relevant to clinical potential. Rather, drug interaction research benefits from using several potency/efficacy tests to permit assessment of changes in therapeutic index. For example, a synergistic analgesic interaction may not be useful if synergism is also observed for limiting adverse effects. Alternatively, a merely additive analgesic interaction may be tolerable if sub-additive or antagonistic interactions are observed on endpoints related to adverse effects.

Even though certain analgesic combinations seem intuitively to make sense and have positive preclinical support, the clinical experience is not always as positive. For example, the combination of anticonvulsants with opioids for treating pain with a neuropathic component is common, and clearly documented in pre-clinical studies,¹⁰ but the outcome may not be as favorable as hoped. In a recent study in patients experiencing pain with a neuropathic component, subjects were treated with either a combination of gabapentin plus oxycodone or gabapentin plus plabcebo. The combination therapy resulted in improved pain relief, but it also increased the number and complexity of side effects.²¹

Summary and perspective

The optimal strategy for treating a multi-faceted medical condition is one that uses a drug or combination of drugs that contribute through multiple mechanisms to the therapeutic endpoint. A rational multi-modal approach can also result in reduced adverse effects. The determination of whether a particular combination is mechanistically undesirable (sub-additive), desirable (additive), or highly desirable (synergistic) requires rigorous quantitative analysis followed by adequate clinical assessment.

Perspective.

This article reviews the medical relevance of the quantitative evaluation of drug combinations, using pain and combinations of analgesics as specific examples. Such measure can help clinicians who seek to maximize therapeutic effect while simultaneously minimize adverse effects.