J. Pharm. Pharmacol. 1993,45: 850-861 Received July 21, 1992 Review 0 1993 J. Pharm. Pharmacol. Effect of Detergents and Other Amphiphiles on the Stability of Pharmaceutical Drugs ANSELMO GOMES DE OLIVEIRA A N D HERNAN CHAIMOVICH* Departamento de Farmacos e Medicamentos, Faculdade de Cigncias Farmacduticas. Universidade Estadual Paulista. Araraquara. and *Departamento de Bioquimica. Instituto de Quimica, Universidade de Siio Paulo. Caixa Postal 20780, Siio Paulo, SP. CEP 01498-970, Brazil The diversity of applications of amphiphiles such as surfac- tants and phospholipid-like compounds in pharmaceutical technology renders the study of their interactions with pharmacologically active drugs of particular relevance (Gibaldi & Feldman 1970; Knight 1981; Voigth & Born- schein 1982; Attwood & Florence 1983). Amphiphiles aggre- gate in excess water to yield a variety of supramolecular structures (Fendler 1982; Israelachvili 1985). Basic know- ledge of drug-aggregate interactions in micelles, microemul- sions and liposomes must be taken into account in the development of novel, more specific and more stable phar- maceutical products (El-Nokaly & Friberg 1982; Attwood & Florence 1983; Jayakrishnan et al 1983; Martini et a1 1984; Gasco et a1 1988 a, b; Keipert et a1 1989). Reviews on micellar structure and micellar effects on chemical reactivity are available (Fendler 1982; Bunton & Savelli 1986). Here, we will focus on the effects of aqueous micelles on drug stability. The review is organized to emphasize the chemical reaction modified by micelles rather than drug type or family of therapeutic compounds. There have been a number of qualitative descriptions of micellar effects on drug stability. However, the applicability of these studies is severely limited, since at times it is difficult to interpret qualitative results with existing models or to generalize to conditions other than those reported. The main purpose of this review is to emphasife the relevance of quantitative studies of micellar effects on drug stability, especially where the study has allowed a thorough dissection of the aggregate effects in the rate or mechanism of the reaction. The existence of models for the quantitative analysis of micellar effects on reaction rates allows the design of adequate experimental conditions, analysis of experi- mental data and, most importantly, the prediction of the effects of micelles under a variety of conditions. Micelles Detergents, constituted by a polar head-group and a hydro- carbon chain equivalent to more than eight methylene groups, associate spontaneously in water to form dynamic aggregates denominated micelles (Tanford 1980; Fendler 1982; Israelachvili 1985). These aggregates, and many others Correspondence: H. Chaimovich, Departamento de Bioquimica, Instituto de Quimica, Universidade de SHo Paulo, Caixa Postal 20780, SHo Paulo, SP, CEP 01498-970, Brazil. formed by amphiphiles, confer unique properties on the solution. Micelles exhibit an interfacial region separating the polar bulk aqueous phase from the hydrocarbon-like interior (Tanford 1980; Israelachvili 1985). In oil-in-water micro- emulsions, the micelle interior may contain added oil. The interfacial region (also called the Stern layer) has a width equivalent to the detergent head-group and, in the case of ionic detergents, contains the ionic head-groups, a fraction of the counter-ions and water (Tanford 1980; Fendler 1982; Israelachvili 1985; Bunton & Savelli 1986). The Stern layer is extremely anisotropic causing properties of this region to change abruptly over a distance of a few angstroms. The anisotropy of the aggregate renders micellar solutions a special medium in which hydrophobic, amphiphilic or ionic compounds may be solubilized and reagents may be concen- trated or separated in aqueous solution (Fendler 1982; Bunton & Savelli 1986). A cartoon representing a micelle of a negatively charged detergent (Israelachvili 1985) is presented in Fig. 1 . In micellar solutions, as in other microheterogeneous systems, rate modifications may result from at least two unrelated factors: reagent compartmentalization and differ- ences in the free energy of activation and the mechanism of the reaction in the aggregate. Separation of these compo- nents requires quantitative analysis of the effects of the aggregate on the reaction. Several models allow the calcula- tion of rate constants in the micelle (Martinek et a1 1977; Romsted 1977; Quina & Chaimovich 1979; Fendler 1982; Bunton & Savelli 1986). Monomers, substrates and micelles redistribute in solution much faster than the rate of thermal reactions (Fendler 1982; Bunton & Savelli 1986). Monomers and substrates exchange in the p s time-scale and ion-ion exchange rates are rapid relative to the lifetime of the micelle (Almgren et a1 1977). Thus, most of the models for the quantitative analysis of micellar reactions either assume a micellar pseudophase or consider that the micellar ensemble can be treated with a cell model, where all cells contain micelles with identical composition. The association of neutral substrates to micelles can be described by association or partitioning constants (Ks). A review on the forms of analysing micelle-neutral substrate partitioning has been published (Sepulveda et a1 1986). The binding of ions or ionic substrates to ionic micelles cannot be described by a simple partitioning (Quina & Chaimovich 1979). Pseudophase models with explicit consideration of ion STABILITY OF PHARMACEUTICAL DRUGS 85 1 FIG. 1. A micelle of a negatively charged detergent (Israelachvili 1985). exchange (PPIE model) treat the binding of an ionic substrate X in an ionic micelle with similarly charged counter-ions (Y) using a selectivity coefficient for ion exchange (KXIY). A more general form of analysing the association of ions to ionic micelles, with no assumptions concerning the con- stancy of micellar surface charge, is based on the use of cell models and Poisson-Boltzmann ion distributions, with Volmer or Langmuir isotherms to take into account the ion specificity, or non-electrostatic component for ion binding (PBE model, Bunton & Savelli (1986)). Quantitative analyses and the accompanying theories for ion binding to zwitter- ionic micelles have recently been described (Baptista et a1 1992). Therefore, it is possible to analyse quantitatively the binding of uncharged or charged substrates to neutral, zwitterionic and ionic micelles using adequate models under a wide variety of conditions. Neutral and ionic species associate with micelles, particu- larly ionic micelles, through different mechanisms. Neutral substrates associate hydrophobically and the distribution constants may increase with added salt (Sepulveda et a1 1986; Oliveira et a1 1990, 1991). The increase in the activity coefficient of the substrate with salt may increase the difference in the standard chemical potential for water- micelle transfer (A&) (where the chemical potential does not contain any electrostatic component) (A&, = g - g) for water-micelle transfer (Lissi et a1 1985, 1986). For ionic substrates, the effects of salt will depend on a balance between the effect of ionic strength on A@,,, and the specific effect of salt on the electrical potential at the micellar surface. Using ion-exchange terminology the salt effects on substrate binding to ionic micelles may also be described as follows: if the non-electrostatic component of the ion exchange con- stant is large, added salts may increase the binding of hydrophobic ions (Quina Kc Chaimovich 1979). In the case of a hydrophilic ion, added salts will always displace the ion from the Stern layer and, consequently, inhibit the binding to the micelle. Difficulties and differences in interpretations in describing micellar effects also arise from the form and units that describe the association or binding of a neutral substrate to micelles (Sepulveda et a1 1986). Quantitative analysis of reactions in micellar solutions demands further assumptions concerning the separation of micellar and intermicellar reaction rates. For thermal reac- tions, which are slower than the monomer-micelle equilib- ration rates, most models assume two rate constants, namely one in the micelle (k,) and one in the intermicellar aqueous phase (kw). Estimates of absolute values of rate constants for bimolecular (or higher order) reactions in micelles are further complicated by assumptions concerning volume elements, necessary for the calculation of local concentrations. In spite of the assumptions, quantitative analysis of micellar reac- tions has been successful in providing insights into the effects of these aggregates on the kinetics and chemistry of reactions occurring in supramolecular aggregates. The PPIE model is of great conceptual and operational simplicity and has been used with success to analyse quantitatively a wide variety of reactions in micelles and other amphiphile aggregates such as microemulsions and vesicles (Fendler 1982; Kawamuro et a1 1991). The PPIE model, however, is only applicable when the concentration of added ions is relatively small with respect to the total concentration of added ionic surfactant. The PPIE model also fails with high concentrations of highly hydrated ions, such as OH- or F- (Bunton & Savelli 1986). PBE models have provided the needed complement to PPIE to analyse quantitatively reactions under conditions which can be extended to the analysis of the effects of micelles and other aggregates on the stabilization or destabilization of pharma- ceutically important drugs. Esters The mechanism of ester hydrolysis has been elucidated in detail (Bruice & Benkovic 1966; Jencks 1987). A description of the mechanism of water attack on esters is shown in Scheme 1 P R C - Y e, + H2X x = o Y = OR’ OH (7 I R - C - Y R - C - Y I I +X X H - “ ‘A 0 II x It R - C - X R - C - Y + H Y SCHEME 1 + HZO The stability of the tetrahedral intermediate (Scheme 1) depends on several factors, including the presence of acid- base catalysts and the ester structure. The effect of micelles on ester hydrolysis has been extensively investigated (Fendler 1982; Bunton & Savelli 1986; Correia et a1 1991). The rate-limiting step for alkaline hydrolysis of esters in ionic micelles is, as in water, OH- attack. Micelles, however, increase the sensitivity of the reaction to polar effects (Correia et a1 1991). As a rule, in OH--mediated ester 8 52 ANSELMO GOMES DE OLIVEIRA AND HERNAN CHAIMOVICH hydrolysis, postively-charged micelles increase and nega- tively-charged micelles reduce the reaction rate. Uncharged micelles decrease the reaction rate and zwitterionic micelles can either increase or decrease the rate of ester hydrolysis, depending on the hydrophobicity of the ester and on the reaction conditions. For bimolecular reactions, such as the OH - attack on esters, the second order rate constants for the reaction in water (k2,4 and micelles (kz,,,) are within one or two orders of magnitude (Romsted 1984). In general, k2,,, for OH- ion attack tends to be lower than k2w, the inhibition increasing with the hydrophobicity of the ester (Vera & Rodenas 1986a, b). The reactivity ofwater attack on esters in micelles is similar to that in bulk water (Kurz 1962; Menger et a1 1978). The effect of micelles on the hydrolysis of the antipyretic, analgesic and anti-inflammatory acetylsalicylic acid (ASA) and on several esters of the local-anaesthetic type has been investigated. The degradation of ASA exhibits a pH-dependent and a pH-independent region. In the pH-independent region the ionized carboxylate enhances the reaction rate by intramole- cular base catalysis (Scheme 2) (Fersht & Kirby 1967; Connors et a1 1986). + CH3COOH SCYEME 2. Non-ionic detergents decrease up to twofold the rate of hydrolysis of the undissociated form of ASA (Cid & Moran 1976; Ismail & Simonelli 1986). Positively-charged micelles of hexadecyltrimethyl-ammonium bromide (HTAB) inhibit (twofold) the spontaneous decomposition of the ionized form ofASA (Vera et a1 1983). This small inhibition has been associated with a decreased availability of water at the reaction site (Broxton 1982). However,&here is experimental and theoretical evidence strongly suggesting that at the micellar surface, where ASA is located, water is as reactive as in bulk solution (Kurz 1962; Menger et a1 1978; Tanford 1980; Fendler 1982; Israelachvili 1985; Bunton & Savelli 1986). The shape of the function relating the observed rate constant for OH- attack on ASA with the concentration of positively-charged HTAB is biphasic, the rate increasing with detergent up to a maximum where the rate is higher than in water decreasing thereafter, at higher detergent concentra- tion, to a value lower than that in water (Vera & Rodenas 1984,1988a, b). These effects were analysed quantitatively by the PPIE model and the results show that the value of k2, is one-hundredth that of k2w. The rate effects of cationic micelles are strictly dependent on medium composition and salt (Vera & Rodenas 1984; Rodenas & Vera 1985). The reaction of OH- ion with long-chain derivatives of aspirin was used to demonstrate different orientations of substrate with respect to the micellar interface. In micelles, 5- alkyl derivatives expose the ester group at the interface and render these derivatives more exposed to OH- ion attack ” y 3 c=o SCHEME 3. (Broxton et a1 1987). NMR analysis of 5-octyl and 5-heptyl derivatives of ASA in micelles shows that the carboxylate group of ASA is perpendicular to the micellar surface and, for the more hydrophobic esters, a change in the relative orientation of the carboxylate and the ester group occurs (Scheme 3) (Broxton et a1 1987). The structural data are consistent with the effect of micelles on the kinetics of the various derivatives. Hydroxyl- functionalized, positively-charged detergents exhibit effects similar to those described for HTAB (Broxton et a1 1989). The effects of micelles on other pharmacologically active esters such as ethyl p-aminobenzoate (benzocaine), 2-diethyl- aminoethyl p-aminobenzoate (procaine), and endo-a- hydroxy-benzeneacetic 8-methyl-8-azabicyclo-[3,2,I]-oct-3- ester (homatropine) have been reviewed up to 1980 (Linda et a1 1981). Most detergents stabilize local anaesthetics to different extents. Inhibition of the rate of hydrolysis, under different sets of conditions, depends on the detergent charge and drug hydrophobicity (Linda et a1 1981). The alkaline degradation of procaine is inhibited by HTAB up to twentyfold at pH 9.1, increasing the shelf-life at 25°C (Razvi & Beg 1981, 1982; Razvi et a1 1984). Penetration of the drug into the hydro- phobic core of the positively charged micelle was suggested as a rationalization of the results. This interpretation is intriguing in view of the amount of data indicating that penetration of substrates, especially substituted phenyl deri- vatives into the micellar core is by no means a general phenomenon (Broxton et a1 1987; Zanette & Chaimovich 1992). HTAB stabilization of benzocaine in alkaline solu- tions (pH 10.5-12.5) has also been observed and attributed to poor OH- binding in HTAB (Zarina et a1 1986). Binding of OH- ions to HTAB micelles and local OH- concentration at the micellar surface, has been clearly demonstrated (Chaimovich et a1 1979; Quina et a1 1980). Procaine and benzocaine are also stabilized by micelles of sodium dodecyl- sulphate (SDS) and Tween 80 (Razvi & Beg 1981; Razvi et a1 1984). Therefore, while it is clear that micelles increase the stability of local anaesthetics, it is not evident why, especially in the case of positively charged detergents, these water- soluble drugs seem to be compartmentalized differently from OH- ion. The pH dependent incorporation of local anaesthetics such as benzocaine, procaine and tetracaine in phospholipid liposomes indicates that the non-dissociated form of the drug STABILITY OF PHARMACEUTICAL DRUGS 853 & C H 2 - r H - C H p - N H - CH S - C H - C H 2 I - N- CH ' CH3 OH &'CH3 I + R-COOH R SCHEME 4. partitions preferentially in the bilayer (Schreier et a1 1984; Habib & Rogers 1987; Bianconi et a1 1988). The rate of alkaline hydrolysis of procaine and tetracaine is progressi- vely decreased by liposomes of dimyristoylphosphatidyl- choline and dipalmitoylphosphatidylcholine. A maximum protection (by a factor of 2) was obtained with liposomes composed of the negatively charged phosphatidylserine (Bianconi & Schreier 1991). A two-state model was used to analyse the kinetic results and to propose that the rate of alkaline hydrolysis of the local anaesthetics is negligible when incorporated in the bilayer (Bianconi et a1 1988; Bianconi & Schreier 1991). Quantitative analysis of the effects of bilayers on the pK, of local anaesthetics has been described (Schreier et a1 1984). The undissociated form of the local anaesthetic is preferentially solubilized in the bilayer and the solubilization site changes upon protonation (Schreier et a1 1984; Habib & Rogers 1987). The alkaline hydrolysis of tetracaine-like local anaesthetics is totally suppressed upon incorporation in zwitterionic bilayers at low ionic strength. This phenomenon has recently been analysed quantitatively, with a two-state model, using electron spin resonance spectroscopy to determine partition- ing, orientation and rates of alkaline hydrolysis of a drug analogue (Bianconi & Schreier 1991). The effects of neutral (Tween 80, Tergitol), anionic (SDS and sodium taurocholate) and cationic (cetylpyridinium chloride) detergents on the enzyme-catalysed hydrolysis of chloramphenicol stearate have been determined (Bernabei et a1 1981). In all cases, rate increases with detergent concentra- tion to a maximum and then decreases at higher concentra- tions. These results are likely to be due to complex detergent effects on both enzyme activity and antibiotic stability. The rate of alkaline hydrolysis of acetylcholine is unaf- fected by neutral micelles (Nakagaki & Yoroyama 1986a, b). Positively- and negatively-charged micelles, as well as mix- tures of ionic with neutral detergents, inhibit the hydrolysis of acetylcholine. Inhibition by SDS is expected, since acetylcholine may bind to negatively charged micelles while OH- ions are largely excluded from the interface, especially at low ionic strength (Quina et a1 1982). Positively charged detergents, however, also decrease the rate of acetylcholine hydrolysis probably due to substrate orientiation in the interface. The positively charged group of acetylcholine supposedly lies at the interface while the reaction centre is shielded from OH- attack by insertion in the hydrophobic interior (Nakagaki & Yoroyama 1986a, b). Thus, for some local anaesthetics and acetylcholine the general rule of rate enhancement of OH- ion attack by positively-charged micelles does not seem to apply. This phenomenon is particularly interesting since in both cases interactions of substituted N and ester groups with the alkylammonium surface are involved. Understanding the molecular mechan- ism of this interaction is of interest from both the fundamen- tal and applied points of view. The decomposition of propranolol esters involves two parallel reactions, namely hydrolysis (kl) and 0 to N rearrangement (kz) (Scheme 4) (Buur et a1 1988; Irwin & Belaid 1988). The overall degradation rate is increased up to 30-fold by a cationic detergent (dodecyltrimethylammonium bromide) and decreased (fivefold) by SDS (Irwin & Belaid 1988). The rate increase produced by dodecyltrimethylammonium bro- mide is solely due to catalysis of the intramolecular 0 to N rearrangement. The catalytic effect was rationalized in terms of increased polarization of the carbonyl group in the micelle-bound substrate. The catalytic factor varies from 130 to 530 and depends on substrate hydrophobicity. Anionic detergents, on the other hand, inhibit both the intra- and intermolecular reactions. Although ionic micelles affect the degree of ionization of propranolol esters, this effect was only taken into account to explain the inhibitory effect of SDS (Irwin & Belaid 1988). Several intramolecular acyl transfers are catalysed by micelles (Cuccovia et a1 1977; Oliveira et a1 1990). In these, as in other systems, the effect of the interface on the acid dissociation constant of the nucleophile has to be corrected before proposing that the micellar effect is due to an effect on the activation energy of the intramolecular nucleophilic attack (Cuccovia et a1 1977; Oliveira et a1 1990). Amides The mechanism of amide hydrolysis, involving several potential rate-determining proton transfers, has been the subject of intense research in the last decades. A simplified representation of the mechanism of amide hydrolysis (Jencks 1987) is shown in Scheme 5. Kinetic effects of non-ionic detergents (ethoxylated lano- lin and polysorbate), HTAB and SDS on the alkaline hydrolysis of the anti-inflammatory drug indomethacin (1- (4-chlorobenzoyl)-5-methoxy-2-methylindoyl-3-acetic acid) have been described (Dawson et a1 1977; Krasowska 1980; Suleiman & Nagib 1990). Neutral detergents decrease, while 854 ANSELMO GOMES DE OLIVEIRA AND HERNAN CHAIMOVICH Hot slow R C - N H z R l + H z O = t fast _t 0 / /+ fast R C - O H z + R l N H z - + 0 RC