Cocaine was the first local anesthetic to be introduced into modern clinical medicine. Cocaine is an amine extracted from the leaves of the plant Erythroxylon coca and other plants of the same genus, native to Peru and Bolivia. During the 1850s, the plant was discovered in South America by German scientists and its principal active ingredient, the (tertiary) amine cocaine (“coca-ine”), was subsequently extracted, purified, and introduced into clinical medicine in the latter part of the 19th century. Final elucidation of the chemical structure of cocaine did not occur until 1924 but, by then, many hundreds of synthetic substitutes had already been prepared and trialed based upon the anesthesiophore of the cocaine molecule. The introduction of procaine (Novocaine)^b in 1905 heralded a generation of a large family of ester caine local anesthetics having greater safety than cocaine and, most importantly, lacking its addictive/psychostimulant properties. Of these, chloroprocaine (Nesacaine) and tetracaine (amethocaine, Pontocaine) still remain popular with some anesthesiologists—the former for spinal anesthesia and the latter for topical anesthesia—along with some other ester caines, such as oxybuprocaine (Novesine) and proparacaine (Proxymetacaine), that are still used by some ophthalmologists.

As a class, ester caines in solution are subject to degradation by chemical hydrolysis, notably upon heat sterilization. This led to their replacement by amide caines that, as a class, resist chemical hydrolysis. The main pharmacologic benefit of ester caines, however, is that they are readily hydrolyzed metabolically by plasma and tissue esterases. Butamben, a benzocaine derivative of bygone years, received attention again in the late 1990s for its newly discovered ability to provide long-lasting and somewhat selective neural blockade after epidural administration of a suspension, but difficulties with its formulation appear to have precluded its continued development.

The early 1930s saw the introduction of dibucaine (cinchocaine, Nupercaine), a quinoline derivative with the ester linkage replaced by a carbamoyl group. This agent is rather toxic, and its use was confined mainly to spinal anesthesia. It still provides the basis of a laboratory test of serum cholinesterase activity. The next major development occurred in two rival Swedish companies, AB Astra and AB Bofors, where some significant variations resulted in a new generation of amide caines. In these, the labile ester linkage was replaced by the chemically sturdier amide group, and this is further protected from chemical, but not enzymic, hydrolysis by steric hindrance from the 2,6-dimethyl groups on the aromatic (xylidine) ring. Lidocaine (lignocaine, Xylocaine, 1940s) was first, followed by mepivacaine (Carbocaine, 1950s), bupivacaine (Marcaine, 1960s), prilocaine (Citanest, 1960s), etidocaine (Duranest, 1970s), articaine (Carticaine, 1970s), ropivacaine (Naropin, 1980s) and levobupivacaine (Chirocaine, 1990s). In etidocaine, the carbon of the amide linking chain is ethyl substituted. In prilocaine, the 2,6-xylidine ring has been replaced by an o-toluidine moiety, and the amide linking chain is methyl substituted, thereby protecting the amide link from chemical hydrolysis; prilocaine also has a secondary rather than tertiary amino group, as in its near-relatives. Articaine, which was originally introduced for dental anesthesia and relatively recently into mainstream anesthesiology, differs from the other commonly used agents in that the aromatic head is a thiophene ring; it has an amide linking chain to the amino group, but it also has an ester group.

By the late 1970s, routine use of the ester caines had waned, and lidocaine and bupivacaine had become the de facto standards of shorter- and longer-acting agents, mepivacaine and prilocaine continued to have a smaller following among some anesthesiologists and dentists for selected procedures, and etidocaine had developed an unfortunate reputation for producing motor block that outlasted sensory block. At about the same time, concern over the safety of etidocaine and bupivacaine was becoming prominent (7,8).

As safety issues in regional anesthesia have continued to come to the fore, the future place of bupivacaine has now become questionable (9,10), especially with the introduction of ropivacaine and levobupivacaine, which have similar anesthetic activities but greater margins of safety (10,11). Articaine is now finding a wider place among anesthesiologists due to recognition of its relatively greater margin of safety (12).

The next major development came in the 1980s, with the development of ropivacaine—the first synthetic chiral caine—followed, a decade later, by the development of levobupivacaine.

Many substances, including a wide variety of anesthetic drugs (e.g., halothane, enflurane, isoflurane, desflurane, ketamine, thiopental, mepivacaine, bupivacaine, prilocaine, etidocaine, ropivacaine, and articaine) have a (single) chiral center (from the Greek, cheir, meaning “hand”). This center, also referred to as an asymmetric carbon atom or center of asymmetry, has four different functional groups bonded to it. This structure imparts a particular type of stereoisomerism known as enantiomerism. Where it is required to distinguish between them, substances lacking a chiral center (e.g., lidocaine, procaine, tetracaine and sevoflurane) may be referred to as achiral.
Enantiomers (also referred to as enantiomorphs) are stereoisomers having nonsuperimposable mirror image relationships (i.e., one enantiomer cannot be superimposed upon the other without chemical bonds being broken and remade differently).

Enantiomers have identical physical and chemical properties, but differ in optical activity—the direction with which they rotate plane polarized light—a distinguishing property for which they became known as optical isomers, a now-obsolete term in the context of stereopharmacology. A single stereoisomer, free from its enantiomer, is referred to as enantiopure. A racemate (racemic mixture) consists of equal amounts of both enantiomers (and, therefore, has null rotation of plane polarized light); nonracemic mixtures consist of unequal mixtures of enantiomers. Racemization is the irreversible production of a racemic mixture from a nonracemic chiral starting material. Inversion is conversion of a chiral substance to its enantiomer. Enantioselectivity refers to a preference for one enantiomer over the other, whereas enantiospecificity refers to marked distinctiveness or exclusivity of that enantiomer. Diastereoisomers (or diastereomers) have multiple centers of chirality, and they can have multiple pairs of enantiomers (a pair for each chiral center), and normally differ in their chemical and pharmacologic properties (e.g., ephedrine and pseudoephedrine are diastereomers), and each consists of a pair of enantiomers.

Chiral drugs found in nature are usually single enantiomers because they are synthesized enzymically, and such reactions are usually stereospecific. Most synthetic chiral drugs are racemates, unless special synthesis is used to preserve the stereochemistry of a chiral center, or a chiral asymmetric synthesis is used to introduce a chiral center. Racemates are normally produced because there is an equal chance that both stereoisomers around a chiral center will be formed unless there is an energetic advantage for either one. Enantiomers behave essentially identically in an achiral environment, but in the chiral (amino acids, sugars, etc.) environment of the body, they behave as separate entities with potentially different pharmacokinetic and pharmacodynamic properties due to differences in their binding to macromolecules such as receptors (thus effect), membranes (thus distribution), and enzymes (thus metabolism) (13).

Three notations are used to describe chirality and associated optical activity, and all three appear in various publications about local anesthetic agents. First, (+) and (-), or dextro and levo (or d and l, now obsolete terminology) are associated with the pairs of enantiomers that rotate plane polarized light, respectively, to the right and left; that is, they are based upon optical properties (first noted in the 1830s). However, the direction of rotation is a phenomenologic feature only. A molecule may rotate plane polarized light in one direction when dissolved in one solvent but in the opposite direction in another solvent. Second, a systematic method of associating the stereochemistry with the direction of optical rotation was developed in 1919 by Emile Fischer and indirectly based upon structure. The Fischer-Rosanoff convention is based upon a molecule’s configuration relative to that of (+)-glyceraldehyde (now known to be (R)-2,3-dihydropropanal, see below), which was arbitrarily assigned a D configuration (its enantiomer, (-)-glyceraldehyde being designated L-glyceraldehyde), or relative to (-)-serine (L configuration). The configuration of a molecule would be assigned the D configuration if it (or a chemical degradation product that retained the chiral center) had the same direction of optical rotation as the model substance (+)-glyceraldehyde, and the L configuration if the direction of rotation was the reverse. The direction of rotation, as in D-(+)-bupivacaine, also may be added. The D and L notations are still used in sugar and amino acid chemistry. Third, in 1955, the sequence rules of Cahn, Ingold, and Prelog introduced a method for the unequivocal designation of molecular configuration by giving a sequence of priority to the four atoms or groups attached to a tetrahedral chiral center. With the smallest atom or group extending away from the viewer, the arrangement of the largest to smallest groups proceeding clockwise is designated as R– (for rectus); its enantiomer is designated S– (for sinister) (Fig. 3-1). If more than one chiral center is present, they must be designated by their configuration at each position. As examples, levobupivacaine, which has one chiral center, is (S)-1-butyl-2-piperidylformo-2′,6′-xylidide; natural (-)-cocaine, which has two chiral centers, the position of each being specified, is 3(S)-benzoyloxy-2(R)-methoxycarbonyl tropane.^c

Contemporary literature still contains all three notations, depending upon the amount of information available and when the information was acquired, but only the Cahn-Ingold-Prelog notation denotes the absolute configuration. Racemates are designated (±)-, dl-, DL-, rac-, or RS-; nonracemic mixtures are designated by their enantiomeric excess (or ee), a measure for how much of one enantiomer is present compared to the other; for example, a sample with 40% ee in the R-enantiomer, would have the remaining 60% as racemic (with 30% of R-enantiomer and 30% of S-enantiomer), thus having a total of 70% R-enantiomer. The naming confusion is preserved by the use of international nonproprietary names such as levobupivacaine, which designates optical rotation rather than absolute configuration. The direction of optical rotation is only rarely of pharmacologic significance, so that it is often omitted from the name: (-)-(S)-bupivacaine is adequately designated S-bupivacaine (or by the nonproprietary name levobupivacaine). Application of the various forms of stereochemical notation is given in Table 3-2 using mepivacaine and bupivacaine as examples (522).

Despite our modern attention to chirality, much was known about the chiral pharmacology of local anesthetics from work performed during the 1960s and 1970s, mainly at the Swedish companies AB Astra and AB Bofors. Prilocaine, mepivacaine, bupivacaine, etidocaine, and articaine were introduced into clinical use as racemates before it was appreciated how important these agents might turn out to be. (Bupivacaine, and the other racemic local anesthetics discussed here, were named originally to designate the racemate, so that there is no need to specify that the particular agent is a racemate unless required by the context. Once in the body, enantioselectivity in pharmacokinetics makes a racemic agent into a nonracemic mixture, so that a measured plasma concentration of bupivacaine, for example, will normally comprise an unequal mixture of R– and S-bupivacaine.) After the concern was raised about the disproportionate cardiotoxicity of bupivacaine, ropivacaine (which is alone among the synthetic local anesthetic agents to be produced as a single enantiomer from its inception) was announced (14), and levobupivacaine (enantiopure S-bupivacaine) was later introduced based upon findings of its lesser toxicity than either bupivacaine or its enantiomer dextrobupivacaine (R-bupivacaine) (15,16). Nonracemic mixtures also have been trialed, for example, with bupivacaine (17,18) but have not been developed commercially.

Although pharmacologic enantioselectivity was first recognized by Louis Pasteur in 1858, it is only relatively recently that the pharmacologic consequences of enantiomerism in the context of anesthesiology has been rediscovered (19,20,21,22,23,24). A variety of studies, mainly using bupivacaine enantiomers in different models and bioassays has, not surprisingly, shown that enantioselectivity at voltage-gated ion channel and other receptors mediating different neurogenic effects ranges from negligible to considerable (25,26,27,28,29,30,31,32,33,34,35,36,37,38). Presently, drug regulatory authorities are attuned to new chemical entity racemic drugs; these must be demonstrated as being no less safe than their enantiopure counterparts (39).

Analytical separation of local anesthetic (and many other) pairs of enantiomers for pharmacokinetic and other purposes is based on their different affinities for chiral macromolecular stationary phases, typically α₁-acid glycoprotein (also known as orosomucoid), or cyclodextrin, packed into a chromatographic column (40,41,42). On a manufacturing scale, a racemate such as bupivacaine, an organic base, can be separated into its component R– and S-enantiomers by combination with a suitable enantiomer of an organic acid, such as tartaric acid, to form two diastereomer salts that have different solubilities and thus unique propensity for selective precipitation; this is a common laboratory process called resolution, and commercial methods have been patented for this process (43). Alternatively, chiral synthesis to preserve the stereochemistry at the chiral center of the reagents used can be used to produce the required enantiomer selectively (44). For any drug, the commercial method used ultimately depends on cost-effectiveness of the process and the chemistry of the particular drug.

An agent’s potency derives from its structural, stereochemical, and physicochemical properties: These properties govern the fraction of a dose reaching and residing at the relevant receptor, the agent’s affinity for the receptor, and its (biochemical and biophysical) efficacy once it has bound to the receptor.

The potency of local anesthetic agents is determined early in drug development by blockade of isolated nerve electrophysiologic studies and/or neural blockade procedures in intact laboratory rodents. Their relative potency is often specified in terms of their equi-effective anesthetic concentrations, and some well-known relationships between agents are given in Table 3-1. Equi-effectiveness is sometimes judged by clinical bioassay, for example, by subjects having similar characteristics of neural blockade (e.g., dermatomal–time diagrams) (45), and/or in amount of supplemental analgesic used by patients after surgery when the local anesthetic agent is used for some neural blockade procedure (46). Equal clinical effectiveness is, of course, a desirable standard of judgment, but it is a fairly imprecise standard in that, with reasonably similar agents being tested in typically small cohorts, most studies are more likely to fail to detect a difference between agents rather than show no difference between agents (type II error). More stringently, potency can be specified in terms of the (molar) dose and/or concentration of local anesthetic agent found in preclinical studies that is required to produce a defined extent and/or degree of neural blockade in a particular model or preparation, or the frequency of its occurrence in a certain population.

By-and-large, the same chemical and physicochemical changes that influence anesthetic potency among the various agents also influence potency for toxicity, whether the agents be ester- or amide-type, with some adjustments for differences introduced by also considering chirality. Among these agents, the more potent substances are more lipophilic and less hydrophilic, and this could engender situations in which the physical presentation of the local anesthetic agent is limited by the soluble amount of dose. Greater potency is also usually associated with a longer duration of neural blockade;
for example, intrathecal doses of bupivacaine 4- and 8-mg have approximately the same time courses of neural blockade as ropivacaine 8- and 12-mg (47).

Various nerve blocks are now being used to provide a means for assessing the relative potency of local anesthetic agents for producing relevant effects (e.g., sensory or motor block) in a particular population (e.g., primiparous patients in first-stage labor). From studies using the “up-down” sequential single-dose technique, the median effective dose (ED₅₀)^d has been determined as a primary measure for comparing potency between agents or for assessing the usefulness of drug combinations (48,49,50). It has been found, for example, that the mean (and 95% confidence intervals) of the relative analgesic potency ratios in intrathecal labor analgesia were 0.65 (0.56–0.76) for ropivacaine-to-bupivacaine, 0.80 (0.70–0.92) for ropivacaine-to-levobupivacaine, and 0.81 (0.69–0.94) for levobupivacaine-to-bupivacaine, suggesting a potency hierarchy of spinal bupivacaine >levobupivacaine >ropivacaine (51). Similarly, in separate studies, the relative ropivacaine-to-bupivacaine analgesic potency ratio for epidural labor analgesia was found to be 0.6 (0.49–0.74), and that for ropivacaine-to-levobupivacaine potency, the ratio was 0.98 (0.80–1.20) (52,53). The different ratios for ropivacaine-to-levobupivacaine between intrathecal and epidural dosing presumably reflect the more complex milieu of epidural analgesia, as described later.

Some have criticized the use of ED₅₀ as the appropriate measure of relative potency, because it represents only the 50th centile; that is, the mid-point of the dose-response curve, and that the ED₉₅ would be a more useful measure for extrapolation to clinical practice (48,54,55). Some of the criticism is based on claims that the observed relative potencies of local anesthetics in clinical use do not fit with the observed ED₅₀‘s. The ED₉₅ is beyond the linearized (ED₂₀ to ED₈₀) portion of the slope of the (presumed sigmoidal) dose-response curve, so that the relationship between the ED₅₀ and the ED₉₅ need not bear the same dose proportionality relationship, which arises when comparing drugs with different slopes. Indeed, the slopes of the dose-response curves for different end points (e.g., sensory and motor blockade) may, and probably do, differ both between closely related agents, as well as within agents (56,57).

The up-down sequential dose method has also been used for assessing the effectiveness of treatments under particular conditions. For example, it has been shown that the ED₅₀ for bupivacaine in late labor was greater by a factor of 2.9 (95% CI 2.7–3.2) compared with the MLAC in early labor (58). The up-down sequential dose method is also used assessing combination treatments with additives such as epinephrine or opioids (59,60). Others have also used the more formal isobolographic analysis with several combinations of doses for assessing the ED₅₀ for combinations of drugs to determine whether a preferred dosage ratio exists (61).

Molecular weights of the commonly used local anesthetic agent bases vary from 220 to 288 Da (Table 3-1). These agents have essentially the same skeleton with the increases in molecular weight being due to alkyl substitution; hence, lipophilicity and pK_a also change with molecular weight. Because the diffusion coefficient is inversely proportional to the square root of the molecular weight, and differences in molecular weight among related local anesthetic agents are relatively small, the diffusion coefficient does not contribute significantly to differences in their rates of diffusion. Changes in molecular weight produced by alkyl substitution, however, for example in pK_a, are accompanied by changes in other physicochemical properties, and these potentially lead to other related effects (Fig. 3-2).

Ex vivo studies of the dural permeability of drugs have been performed with cadaveric tissues, and these demonstrate a relatively minor influence of molecular weight. Although a relatively simple model, cadaveric membrane-diffusion cell preparations are themselves subject to experimental vagaries of tissue choice (species, site and thickness, etc.) and preparation (notably fresh versus frozen) that probably explain some of the inconsistencies seen in data from different laboratories (67,68). Despite the simplicity, it is still difficult to identify the impact of the various factors and covariates (molecular weight and lipophilicity, for example, because these run together when a homologous series of drugs is used), and corrections need to be made for nonionized (c.f., total) concentration of drug (69). In vivo studies are even more difficult to interpret because of the competing influences of vascular activity, systemic absorption, and uptake into fixed spinal tissues, as well as cardiovascular sequelae of concurrent general anesthesia in the preparation of living subjects. These factors have not yet been investigated systematically, and it is only recently that microdialysis has provided a method for beginning to study in vivo dural permeability and to estimate the bioavailability to cerebrospinal fluid (CSF) of intrathecal and epidural local anesthetics (67,70,71,72).

When tested with an ex vivo preparation of monkey or rabbit spinal meninges, the permeability coefficients of a
heterogeneous series of drugs including lidocaine and bupivacaine were found to be independent of molecular weight (and molecular axis and molecular volume); however, a bell-shaped relationship was found with lipophilicity (as measured by the octanol-to-pH 7.4 buffer distribution coefficient) (73). When tested ex vivo with a series of homologous mepivacaine-like local anesthetics (i.e., progressively increasing molecular weight and lipophilicity), a parabolic relationship between permeability in lumbar and cervical meninges of the rabbit and lipophilicity was found. The permeability of human dura mater ex vivo to a variety of opioids and local anesthetic agents was found to depend on simple diffusion, independent of molecular weight, driven by the concentration gradient, and was not enantioselective (69,74,75). There is, however, some evidence to suggest that molecular weight might be relevant to the diffusion of local anesthetics in the sodium (Na⁺) channel of the nerve membrane (76).

Recent correspondence in the literature has also drawn attention to the relationship between molar and mass label quantities of some local anesthetic agents that could have implications for differences in dose requirements and the relative potency of similar agents (77,78,79). It is worth reiterating here some elements of that discussion because it also involves international differences that can occur in labeling requirements.

Local anesthetic agents are normally prepared as %weight/volume solutions of the hydrochloride “salts” of the relevant base, such that the molecular weight is equal to the base + HCl (= 36.5 Da) if anhydrous, additionally + H₂O (= 18 Da) if monohydrate, etc. An aqueous solution for injection will be made of the salt, but the concentration of active ingredient may be specified either as that of the salt or the base; the water of crystallization is normally allowed for and adjustments are made. The relative concentrations of bupivacaine and levobupivacaine have the same molecular weight (= 288 Da), but whereas the concentration of bupivacaine is specified as the HCl salt (= 324.5 Da anhydrous), that of levobupivacaine is specified as the base (= 288 Da), at least for the European regulatory authorities. Thereby, the mass concentration of levobupivacaine HCl to make the solution has to be greater by 12.6% to achieve the same molar concentration of base as in a solution of bupivacaine HCl. Levobupivacaine could, therefore, appear to be more potent than bupivacaine if this difference in concentration were to be ignored. The message is plain: check the locally issued product label/information as to how the concentration of local anesthetic agent is specified!

The inclusion of an amino group in the structure of most local anesthetics confers upon them the “split personality” of a weak base, meaning that they exist in solution partly as the nonionized free base and partly as the ionized cation (conjugate acid) (Equation 3):

Accepts proton (association):

The position of equilibrium depends on the dissociation constant (K_a) of the conjugate acid and on the local hydrogen ion concentration (Equation 4). Thus,

where the square brackets indicate concentration or, more properly, activity. By rearranging Equation 4 and taking logarithms, the Henderson-Hasselbalch equation (Equation 5) is obtained:

where pK_a is defined, by analogy to pH, as the negative logarithm of the dissociation constant of the conjugate acid under particular conditions of solvent and temperature (82,83,84).

Equation 5 shows that the pK_a is equal to the pH at which the local anesthetic is 50% ionized, that is, when

The greater the pK_a of a base, the stronger is that base—that is, it has a greater ability to attract a proton, and the smaller is the proportion of nonionized form at any pH. The ester-type agents have higher pK_a values (8.5–9.1) than the amide types (7.6–8.2) (Table 3-1), which accounts, in part, for relatively poor penetrance and the need to inject these agents close to neural tissue. The effect on pK_a values of differences in structure within the two main types of agents is complex, involving steric factors as well as the inductive effects of alkyl substituents on the amine nitrogen. For example, the greater pK_a of bupivacaine compared with mepivacaine is explained by the effect of greater alkyl substitution, making the nitrogen atom relatively more negative (and able to attract a proton). In contrast, the lower pK_a of etidocaine and of prilocaine reflects the effect of alkyl substituents on the bridging carbon atom in sterically hindering the approach of hydrogen ions to form the conjugate acid and in decreasing stabilization of the resultant cation by hydration.

Ionization is relevant to the solubility and activity of local anesthetics (Table 3-1) and their equilibrium distribution in various body compartments. Because the ionized forms are much more water-soluble than the free bases, the drugs are dispensed as their HCl salts, producing acidic solutions. This also helps to stabilize the esters since they are more readily hydrolyzed in alkaline conditions. The drugs are much more lipid-soluble in their free-base forms than in their conjugate acid forms. Thus, the nonionized fraction becomes essential for passage through lipoprotein diffusion barriers to the site of action on the nerve membrane. The balance between ionized and nonionized forms is shifted in favor of nonionized by the addition of bicarbonate (or other base), at the expense of decreased water solubility and the risk of reducing the effective drug concentration by drug precipitation (85).

Grouls et al. determined the effect of pH on the net flux across human meninges in vitro of lidocaine and bupivacaine. They found that the values at pH 4.0 were less than 20% that of the values measured at pH 7.4 (86). Decreasing ionization by alkalinization of the solution for injection will effectively raise the initial concentration gradient and flux (mass transferred per unit time) of diffusible drug (87,88) thereby increasing the rate of drug transfer. This maneuver has been successful in decreasing the latent period of a peripheral nerve block (89,90,91) but has not been universally successful for central neural blockade (92,93,94,95). Once at the nerve membrane, ionization is
again necessary for complete anesthetic activity (96). Increasing the pH of aqueous local anesthetic solution also increases the surface tension (97), and this has been shown to decrease the drop rate when lidocaine solution was being delivered under gravity feed.

The aqueous phases on either side of many body membranes differ in their pH values. Consequently, although nonionized drug concentrations in these phases will be the same at equilibrium, different concentrations of ionized and, therefore, of total drug will exist on either side of the membrane if a pH gradient exists. For a weak base, the equilibrium ratio (R) of total drug concentration across a membrane is given by Equation 6. Thus,

where the subscripts 1 and 2 refer to the two aqueous compartments. Total drug concentration will be greater in the compartment having the lower pH because a greater proportion will be in the ionized form. Thus, for example, lowered pH owing to infection in tissues surrounding a nerve results in less nonionized drug, which is the form that can cross the axonal membrane. Similarly, local anesthetics will diffuse from blood to acidic gastric contents, thereby maintaining a concentration gradient due to “ion trapping.”

The aqueous solubility of a local anesthetic is related directly to its extent of ionization and inversely to its lipid solubility (Table 3-1; Fig. 3-3). Despite quite large differences in the lipid solubility of the amide local anesthetic bases, the differences in aqueous solubility of the conjugate acids are smaller. Benzocaine and butamben, which lack an (ionizable) amino group, are almost insoluble in water. For this reason, their use is essentially confined to topical anesthesia, or injection for prolonged neural blockade after solubilization with agents such as dextran or suspension in polyethylene glycols. Mixtures of local anesthetic agents with excess bicarbonate will lead to precipitation of the local anesthetic base after the solubilizing acid (normally hydrochloric acid) has been neutralized by the inorganic base.

A low aqueous solubility was thought to be a limiting factor when selecting an agent for subarachnoid block, principally due to the risk of the agent precipitating in CSF and causing neural irritation. It is known that the aqueous solubility is less with more lipophilic agents, and solubility decreases with increasing pH. Transient neurologic disturbances after spinal local anesthetic administration has been reported sporadically for many decades. However, the problem has not vanished by substituting more dilute lidocaine or mepivacaine solutions, suggesting that water solubility of the local anesthetic is not the most critical factor (98,99). Determination of the roles of a particular local anesthetic agent, the particular method of pharmaceutical preparation, and/or any additives have been the subject of much research, but relatively little sound science has emerged (100). How much is due to the local anesthetic, its concentration, and its properties such as surface activity, or the procedure itself, remains controversial (101,102).

Various experiments have been performed with local anesthetics to improve their water solubility, some with pharmaceutical manipulation to affect their local disposition, and some with alterations to the chemical structure. Whether these become useful clinically remains for future studies.

The distribution coefficients of drugs measured in organic solvent/aqueous buffer systems in vitro are commonly used to reflect their relative in vivo distribution characteristics or lipophilicity (83,525). The values obtained normally differ quantitatively between systems due to both the polarity of the solvent and the extent of drug ionization at the chosen pH, such that their best use is in making comparisons between drugs within the same system (522). Net lipid solubility is independent of ester or amide grouping. Tetracaine (amino ester) is regarded as being highly lipid-soluble, as are ropivacaine, bupivacaine, levobupivacaine, and etidocaine (amino amides). Rosenberg et al. (103) have shown good rank order correlation between the n-heptane/buffer distribution coefficients of bupivacaine, etidocaine, ropivacaine, and lidocaine and their in vitro distribution into rat sciatic nerve and human epidural and subcutaneous fat. The intrinsic potencies (reciprocals of ED₅₀) of amino ester, amino amide, and piperidine amide local anesthetics have been found to be proportional to their calculated solubility in nerve fibers as determined from their lipid solubility (104). However, the order breaks down when a substance has too great a lipid solubility, and it is suggested that a limiting octanol distribution coefficient of about 5 pertains to useful membrane penetration. This is shown in Figure 3-2, where the n-hexyl derivative of mepivacaine fails to significantly penetrate the meninges after epidural administration in rabbits. After lipid solubility reaches this critical level, the apparent anesthetic potency, along with the intrinsic systemic toxicity, decreases (producing bell-shaped curves). This is caused by the drug so favorably distributing into lipoidal barrier membranes that insufficient partitions move into axoplasm (to produce neural blockade) as well as vital organs (producing lethal effects). Nevertheless, local tissue toxicity (cell leakiness) increases continuously with increasing lipid solubility throughout the test series of drugs (79,105).

The relationships for the three classes of agent are essentially parallel but, for the same degree of calculated solubility of the local anesthetic base, the intrinsic potencies were found to be in the order amino ester >amino amide >piperidine amide (106). However, the relationship between potency and physicochemical properties is complex. The distribution coefficients are the result of the composite chemical group effects. Different relationships might apply in vivo, especially for
chiral agents. The relative proportions of base and conjugate acid of the individual agents depends on their pK_a. Most relationships are developed in terms of base concentrations only when it is known that both forms are required for membrane stabilizing effect. Overall, a high lipid solubility would be expected to promote drug entry into membranes by increasing the diffusion rate, but this has to be balanced with a high fraction of drug in the nonionized state (106,107,108). The net effect on onset of maximum anesthetic action is difficult to predict, however, because a faster rate of diffusion is offset by a greater capacity for uptake into a membrane.

It has been found that high lipophilicity alone also does not predict the relative rate of transfer of local anesthetics and chemically similar drugs through meninges in ex vivo models (73). The ex vivo apparent permeability coefficients of homologous N-alkyl piperidine amides and a series of heterologous amide and ester caines follow a bell-shaped curve with increasing lipophilicity; their in vivo rate of absorption from site of administration in the epidural space to CSF constantly decreases with increasing lipophilicity, but their fractional availability to CSF increases (64,67). This apparent paradox reflects the relative balance in competing routes of local distribution into local fixed structures and clearance by local vasculature. Further complexities pertain even within homologous series. Because pK_a values also increase with increasing alkyl substitution, the nonionized fraction increases significantly. Moreover, even when lipid solubility and nonionized fraction are taken into account, the intrinsic potency of the agents is subject to differences in preferred molecular geometry for fit to the local anesthetic receptor site (108). However, it is generally found that by promoting interaction with the hydrophobic components of receptors, a high lipid solubility will increase potency and duration of effect.

Most drugs reversibly bind to/associate with proteins (and various other macromolecules) to some extent. This association, when it occurs within the body, can provide a transport medium and/or a temporary repository for the drug. Drug binding to proteins is a secondary property originating from physicochemical and stereochemical properties that has received much attention with local anesthetic agents.

Thermodynamic characterization of drug binding treats it as an adsorption process and involves the molar concentration of binding protein(s), the molar concentration of binding sites (or number per molecule of protein), and the association/dissociation constants of the interaction. Pharmacologic characterization, however, is more concerned with determining which protein(s) are involved, and the extent of binding at various drug and protein concentrations. If more than one drug and/or protein is involved, as typically occurs with drug binding in plasma, the relative extent(s) of binding will be governed by the (competitive) relative affinity constants of the separate drug–protein interactions.

Besides being more lipophilic, the longer-acting local anesthetics also exhibit higher degrees of binding to plasma and tissue proteins (Table 3-1), with greater plasma binding of the agent in its base form at higher pH values (109,110,111,112,113,114). This suggests that the binding forces are predominantly hydrophobic. Steric factors also apply, so that the enantiomers of chiral local anesthetics usually exhibit differences in degree of binding to proteins, although the differences are usually smaller than those between agents where differences in lipophilicity predominate. Differences between the enantiomers of some drugs may be large enough to account for differences in some receptor-related effects (e.g., for toxicity) or in pharmacokinetics (e.g., for distribution and/or metabolism).

Adsorption of local anesthetics to binding sites, or solubility within membranes or tissues, although producing relatively high apparent partition coefficients, may result in slower net penetration rates. This may be considered either as a lowering of diffusion coefficient or as a decrease in ΔC, the effective concentration gradient of diffusible drug (Equation 2). Binding of the drugs to proteins associated with the aqueous phases on either side of a membrane will affect the transfer and equilibrium distribution of total drug, analogously to ionization. Only the unbound drug will diffuse readily and, again, this will modify net drug transfer rate by an influence on ΔC.

In contrast to many drugs, the primary effects of local anesthetics at both the pharmacologic level (neural blockade) and clinical level (analgesia and anesthesia) can be measured fairly objectively. The anesthesiologist is concerned particularly with the onset, spread, quality, and duration of block of one or many nerves with differing morphology. Ultimately, however, these variables depend on the drug’s concurrent distribution at, and dissipation from, the site of injection. Therefore, it would be valuable to have chemical measurements of the agents at these sites as a function of time in order to establish pharmacodynamic relationships. Regrettably, such data remain sparse, even from animals. Therefore, knowledge of the local disposition, or neurokinetics, of local anesthetics remains largely theoretical and is deduced from observation of the spatiotemporal changes in anesthetic effect, rather than from direct measurements of intraneural and perineural drug concentrations. Most data have been obtained from studying a particular agent under particular conditions. Despite these limitations, it is possible to draw some general conclusions.

Competing factors that affect local disposition of local anesthetics include dispersion by bulk flow of the injected solution, diffusion, and binding and vascular uptake of the agent (or relevant solute). Local blood supply is critical, as this is responsible for local clearance/systemic absorption of drug; metabolic breakdown seems less important, and is probably negligible with amide-type local anesthetic agents. It has been suggested that the local blood supply is responsible for delivery of drug to deeper structures of the spinal cord via the radicular arteries (115,116) but this is now considered unlikely (117). In addition to these factors, the pharmaceutic presentation of the agent can
be manipulated so that the rate of release of agent controls the balance between local and systemic uptake of the agent.

Although lipophilicity is an intrinsic physicochemical property of a drug resulting from its molecular structure, the effective lipophilicity can be altered by pharmaceutic presentation. A variety of drugs have been formulated with innovative methods to increase their local concentrations and/or residence times. Some examples include cyclodextrins to increase the local concentration of local anesthetics by making them more hydrophilic, and liposomes to make them more lipophilic. Both formulations have been found to be longer-acting after subarachnoid administration in laboratory animals than the simple aqueous solutions of the same drugs. The mechanisms of prolongation of action are presumed to differ but have in common the alteration of the rate of drug diffusion. Drugs combined with cyclodextrins are believed to have a slower rate of vascular absorption than aqueous solutions. Liposome or polylactide microsphere encapsulation provides a depot from which drug is released slowly. Liposomal preparations of local anesthetic agents have both a longer duration of action and a lower propensity for systemic toxicity than equivalent doses in aqueous media injected perivascularly, epidurally, or intrathecally in experimental animals (118,119,120,121) and in epidural, intercostal, or subcutaneous anesthesia in humans (122,123,124). Moreover, liposomal preparations of local anesthetic agents have a higher rate of penetration of the dermis than equivalent aqueous solutions (125).

Commonly accepted thresholds for “CNS toxic” plasma concentrations from local anesthetic absorbed after uncomplicated
perineural administration range from 5 to 10 μg/mL for lidocaine, mepivacaine, and prilocaine, and from 2 to 4 μg/mL for bupivacaine, ropivacaine, and levobupivacaine. Higher values have a higher probability of associated toxic effects, and vice versa.

It is intuitive that a greater circulating blood concentration of any local anesthetic agent is more likely to be associated with a greater risk of systemic toxicity than a lesser one. However, a deterministic relationship between blood drug concentration and toxicity is tenuous for several reasons. First, the rate of change of plasma drug concentrations is an important, but often neglected, element in the relationship because it relates to the degree of equilibration between tissue and blood drug concentrations—that is, how well the blood drug concentrations provide a reliable surrogate for the relevant tissue drug concentrations. Second, toxicity normally occurs concomitantly in different regions, and there can be both direct and indirect effects, for example, with cardiac effects deriving from direct myocardial effects and from CNS cardiac control mechanisms. Third, the concurrent physiologic and pharmacologic condition of the subject (e.g., blood gas/electrolyte/acid–base status, state of consciousness, etc.) can have marked influence on the outcome (148,151,152,166,167).

Several scenarios need to be distinguished pharmacokinetically: unusually rapid or normal persistent drug absorption from successful perineural administration, or unintended intravascular (usually IV, but occasionally intra-arterial) administration. Systemic toxicity from rapid absorption is usually acute, but relatively short-lived, whereas that from persistent absorption develops more slowly and recedes more slowly because the greater mass of drug deposited in the vital tissues takes longer to dissipate. Intravenous administration of sufficient dose usually produces sudden-onset, serious, acute generalized (whole body) effects that are usually rapidly receding, with a duration and severity related to the administered dose. Although reported relatively rarely, severely toxic effects can result from unintended intra-arterial injection of very small doses of local anesthetic agent into the afferent vasculature of vital organs (153,154), and this has been exploited with intended and appropriate doses as a useful research technique for studying regionally selective toxicity (155,156). Thus, it is important to remember that direct toxic effects in one region (notably, the brain) can have profound effects in another region (notably, the heart) via neural and cardiovascular system (CVS) mechanisms. Crucially, both the toxic effects and the associated blood drug concentrations depend on the state of consciousness, so that extrapolation of data between anesthetized and conscious subjects is especially hazardous (473,473a).

Most data documenting local anesthetic toxicity in humans are gathered, essentially opportunistically and retrospectively, from acute incidents in patients undergoing neural blockade, or from prospective studies in a small cohort of healthy volunteers undergoing administration of the drugs for experimental purposes. For ethical reasons, human subjects can be given only mildly toxic doses when local anesthetics are administered intravenously for research, typically to the threshold of CNS subjective symptoms (152,157,158,159,160,161,162). Information about more serious toxicity must, therefore, be derived either from clinical circumstances in which the objectives are preservation of life and well-being, rather than acquisition of scientific data, or from laboratory animal models. Laboratory rodent and isolated-tissue models are widely used and are especially useful for elucidating mechanisms and for comparisons between drugs (163,164,165); however, these are generally limited by the restricted data that can be obtained, or because of their destructive nature, or because of the isolation of tissues from their normal milieu. On the other hand, large experimental animals (dogs, pigs, sheep) can be prepared to allow pharmacokinetics and pharmacodynamics to be observed concurrently in a context rather similar to patient treatment. The ultimate goals of such experiments are normally to set “fail-safe” guidelines for human drug use on the grounds that responses among mammalian species are fundamentally the same, and to elucidate principles that can be expected to apply to humans using methods, techniques, and drug doses that are not normally offered in humans.

Although reported values such as those in Table 3-4 provide some useful guidelines, they refer to the mythic “average subject” and have to be interpreted in the light of many factors. These include whether measurements are made of plasma, serum, or blood drug concentrations; total or unbound (free) drug concentrations; relative enantiomer concentrations (if a racemic local anesthetic); active drug metabolite concentrations, as well as how the drug got into the plasma (intravenously or by vascular absorption), the rate of drug administration, the site of blood sampling (arterial or venous), how soon after drug administration the measured samples were drawn, and, most importantly, the physiologic status of the patient and, in particular, whether the patient was conscious and/or premedicated at the time (473a). In a previous era, the specificity and sensitivity of the drug assay procedure also could have been a factor, but this is rarely a concern with contemporary techniques. Thus, the values in Table 3-4 are more useful when circulating concentrations are not changing rapidly and there is time for equilibration between drug concentrations in plasma and vital tissues. Some individual subjects demonstrate toxic symptoms at lower drug concentrations than the commonly accepted values, whereas others do not demonstrate toxic symptoms despite having values within, or even greater than, those values (473b). In some cases, this is because of the development of locally high drug blood concentration. For example, local anesthetic drugs injected into patients with intracardiac right-left shunts (169) or injected inadvertently into the carotid or vertebral artery during attempted stellate ganglion block, bypass the lung, resulting in a high probability of CNS toxicity (153,154).

In the absence of local metabolism, all of a dose deposited perineurally will eventually become absorbed into the systemic circulation. The concentration gradient is the main driving force for dissipation of drug both from its (heterogeneous) perineural site of administration into local tissues and uptake into the blood. The net resulting rate of systemic drug absorption will, therefore, be approximated by the sum of exponentials^f representing different rates of absorption from different local tissues, with the overall rate being governed by the agent’s distribution coefficient into, and the local perfusion of, the dominant tissue(s). A similar pattern pertains to drugs in the systemic circulation. Concentration gradients drive the exchange of drug between blood (strictly, plasma water) and tissues (strictly, extracellular fluids), including regions that excrete and metabolize drug. This pattern can also usually be represented by the sum of exponentials, in which the drug concentration approaches zero as time increases because of its excretion and/or metabolism lumped together as elimination (see systemic disposition, which is discussed later in the text).

If blood drug concentration–time profiles are available after perineural and IV administration, then it becomes possible to calculate the rate(s) of drug absorption. The underlying principle of this is that an IV injection reflects the whole-body (systemic) disposition of the drug; that is, distribution and elimination after it has arrived in the systemic circulation. Conversely, a perineural administration reflects the concurrent whole-body disposition of drug while arriving and once in the systemic circulation. The mathematical procedure of numerical deconvolution is used to determine the rate of leaving the (perineural) site of administration; this is depicted in Figure 3-8, where “pools” are shown to represent functional regions of drug deposition (rather than precise anatomic spaces). These calculations were originally performed for local anesthetics after IV regional anesthesia (170); subsequently, the method was extended to epidural anesthesia, with determination of the IV pharmacokinetics on a separate occasion (171). More recent technical developments have allowed application to a variety of subjects undergoing epidural and subarachnoid blocks, in which a concurrent small dose of deuterium-labeled agent is administered intravenously (172,173,174,175,176,177). The two forms of the
agent (native and labeled) are then separated analytically by mass spectrometry, and the rate of absorption of the native form used for the block is again determined by numerical deconvolution.

Such pharmacokinetic calculations provide a mathematical-statistical description of a functional process whereby the drug acts a tracer; nevertheless, the description has physical elements. Because local anesthetics are relatively lipophilic compounds they will partition into tissues around their site of administration. Thus, their vascular absorption rates will be directly related to blood flow and inversely to local tissue binding (distribution coefficient). Important local determinants of systemic absorption include those affecting the tissue distribution, including the site of injection, the lipophilicity and vasoactivity of the agent, the presence of additives such as vasoconstrictors, other formulation factors intended to modify local drug residence and release, the influence of nerve block, and (patho)physiologic features of the patient.

It has been found that systemic absorption after epidural administration can be described, in its simplest form, by a bi-exponential function; that is, the drug behaves as if it has been deposited into two portions with different absorption rate constants, roughly reflecting regions with identifiably different low-affinity/high-perfusion and high-affinity/low-perfusion characteristics. Although a mathematical construct, this can be likened to the drug physically partitioning between aqueous (faster absorption) and fatty (slower absorption) pools; the fraction of the dose going to the slower absorption pool increases with the more lipophilic agents (Table 3-5; Fig. 3-9) and/or when a vasoconstrictor is used (171,172,178). T_max for a particular type of nerve block tends to occur at about the same time because the early absorption is dominated by the faster absorption rate constant, which is similar for all agents. However, the half-life of the drug in plasma is dominated by the slower absorption rate constant, and this is normally slower than the elimination rate constant. The real anatomic situation is, however, much more complex than a simple two-pool model.

Overall, for a given blood sampling site, C_max and T_max are inversely related in that the greater the T_max then the lower the value of C_max. Extensive data on C_max and T_max of the amide local anesthetics after various routes of injection have been tabulated elsewhere (145,178). So too have data on (poly)exponential equation exponents and coefficients describing the rate(s) of systemic absorption after epidural administration (171,172,173,174,175,176,177,178). Such data are mainly useful for comparison between drugs, or recipients of drugs.

Differences between agents need to be assessed under comparable conditions regarding drug administration, subject choice, and subject preparation. Inspection of C_max and T_max after epidural injection of plain solutions of the amide local anesthetics, and allowing for differences in sampling regimens, shows that the increment in peak whole blood drug concentration per 100 mg of dose is about 0.9 to 1.0 μg/mL for lidocaine and mepivacaine, slightly less for prilocaine, and approximately 50% to 60% as much for bupivacaine, etidocaine, ropivacaine, and levobupivacaine (167,172,178,179,180). Although differences in disposition kinetics contribute to this order, it appears that, despite similar times of peak concentrations, net absorption of the longer-acting, more lipophilic agents is slower. This is consistent with experimental data on residual concentrations of
the agents in epidural fat after injection into sheep (178) and is confirmed by deconvolution calculations of the time course of drug absorption in humans (173,174,175,178,180).

Differences in the net absorption rates of the various agents have implications for their accumulation during repeated and continuous administration. Whereas systemic accumulation is more marked with the short-acting amides, extensive local accumulation is predicted for the longer-acting compounds, despite their longer dosage intervals (171,181,182). Observations of relatively low blood concentrations of prilocaine with respect to the toxic threshold, particularly after brachial plexus block (Fig. 3-10) and IV regional anesthesia, support the claim that this compound should be the agent of choice for such single-dose procedures (183). For prilocaine and for articaine, a high systemic clearance, rather than slow absorption, is responsible mainly for the relatively low blood drug concentrations (12,183).

Although the rate of systemic absorption of local anesthetics is controlled largely by the extent of local binding, their relative intrinsic vasoactive properties could also modulate local perfusion and hence uptake. However, the effects are a complex function of drug, dose, enantiomer, and type and tone of blood vessel, and their relevance to the overall relative absorption of drugs after peripheral and central nerve blocks is difficult to evaluate (172). Nevertheless, it has been suggested, for example, that an increase in epidural blood flow (whether mediated locally or by change in cardiac output or both), with assumed increase in the systemic absorption rate of bupivacaine, is an important factor leading to regression of analgesia during the continuous epidural infusion of bupivacaine (184). Furthermore, the vasodilatory effect of bupivacaine on cerebral pial, cutaneous, and epidural blood vessels contrasts strikingly with the relative net vasoconstrictor effect of ropivacaine and levobupivacaine, a difference that may contribute to their relative anesthetic profiles (185–187a). Others, however, have found the effects on cutaneous blood vessels to be biphasic: vasodilatory at high concentrations and vasoconstrictive at low concentrations—clearly, the choice of exploratory model and conditions is critical in evaluating the net effects (188,189). Although R-bupivacaine was found to be two to three times more potent than S-bupivacaine in blocking nerve fibers in vitro (190), this intrinsic difference in action may be modulated in vivo by the dominant stereoselective effects on blood vessels. Thus, the S-enantiomer amide local anesthetics are longer acting than the R-enantiomers after separate subcutaneous injection, apparently reflecting a greater vasoconstrictor activity and, therefore, a presumed slower systemic absorption (14,191). Similar logic about the greater vasoconstrictor activity of ropivacaine than bupivacaine has been used to rationalize the slower apparent rate of ropivacaine absorption. For example, after caudal administration of 2 mg/kg doses of 0.2% solutions in children undergoing hypospadias repair, there was no significant difference in mean C_max of the two agents, but the mean T_max of ropivacaine occurred significantly later (65 minutes, c.f. 20 minutes). As the plasma concentrations were measured in peripheral venous blood, it is not possible to confirm this explanation because the C_max T_max values depended upon the rate of equilibration of drug between blood and
forearm tissues, and was complicated further by the patients being under isoflurane anesthesia (192).

Anatomic features, such as vascularity and the presence of tissue and fat that can bind local anesthetics, are primary influences on their rate of distribution within, and removal from, specific sites of injection. As revealed by imaging or dyes at various sites of local anesthetic injection, other features, such as the presence of septa or other fibrous tissues (and any resultant pressure gradients), also contribute to distribution of solution and thus spread of anesthesia (131,193,194,195,196). This is pertinent to all procedures of neural blockade, but particularly so for the central neuraxis, where the detailed anatomy is complex and obscured from the anesthesiologist, variable between individuals, and still largely unresolved with respect to the resultant block (197,198).

On the whole, and under comparable circumstances, as judged by C_max values with regard for the effective dose, the net absorption rate decreases independently of the local anesthetic agent used, in the order intercostal block >caudal block >epidural block >brachial plexus block >sciatic and femoral nerve block (Fig. 3-6) (148,149,199,200). As indicated earlier, arterial plasma drug concentrations precede and exceed those in peripheral venous plasma, regardless of the agent. A study with epidural ropivacaine found that the forearm arteriovenous gradient diminished exponentially, with the difference essentially extinguished about 1 hour after administration (201), as would be expected from the time needed for drug concentration in the tissues to equilibrate with that in the venous blood. However, different conclusions might have been reached had the same study been performed by determining the drug arteriovenous concentration gradient in a lower limb because the block itself would have influenced the venous tone of the upper and lower limbs differently (171). Concurrent administration of general anesthetics and some medications can alter the time courses of arterial and venous plasma drug concentrations through local and systemic cardiovascular effects, and this may be observed when data from surgical patients and nonsurgical patients or healthy volunteers are compared.

Studies of epidural administration, as discussed earlier, found that the systemic absorption of several local anesthetics could be described as a bi-exponential process, differing only in the net rates; a similar description was proposed for paravertebral ropivacaine (192), and a mono-exponential process was described for subarachnoid lidocaine (172). These descriptions are consistent with anatomic findings, as demonstrated in sheep, of prolonged sequestration of more lipid-soluble local
anesthetic drugs in epidural fat. Thus, it is clear that the net rates of absorption are mainly sensitive to the lipophilicity of the agent and the fattiness of the region where the agents are deposited.