Strategies to Address Low Drug Solubility in Discovery and Development

1. Complexities with Parenteral Administration of Poorly Water-Soluble Drugs

In comparison with the large number of drugs intended for oral administration, parenteral formulations constitute a more limited proportion of marketed products. However, the generation of useful parenteral formulations remains a key component of drug discovery and drug development for almost all drug molecules (regardless of the intended route of clinical administration) since data obtained after i.v. administration is necessary to generate fundamental pharmacokinetic parameters (volume of distribution, clearance, half-life, absolute bioavailability) to support compound optimization and progression, and such data are also useful to support early stage pharmacodynamic assessment of drugs intended for acute administration.

Low aqueous solubility significantly complicates the development of i.v. formulations since, in almost all cases, simple solution formulations are required. The following section describes several different approaches that can be used as i.v. solution formulations, depending on the physicochemical properties of the drug and the intended use (e.g., animal or humans). Depending on the formulation strategy adopted, increases in solubility of several orders of magnitude are possible. The risks associated with i.v. administration of poorly water-soluble drugs include, first, the potential for drug precipitation on administration and, second, the potential for formulation components to cause pain, phlebitis, inflammation, hemolysis, or unacceptable toxicology. Precipitation is more likely for formulations where pH adjustment or cosolvents form the basis for drug solubilization since both are likely to be altered significantly on dilution (see Sections IV and VI). In contrast, surfactant solubilization, complexation, and i.v. emulsions are far less sensitive to dilution (see Sections VII, VIII, and XI). Where dilution-mediated precipitation is possible, slow administration into large veins, where the blood flow is higher (and therefore the supply of plasma proteins and lipoproteins to provide in vivo binding is higher) is preferred, and this can also reduce the possibility of pain on injection, hemolysis, and phlebitis since the irritant component is maximally diluted. For studies in rats and dogs, this can be accomplished by implanting a dosing cannula into the jugular vein (or another large vein) with administration over a period of 5 to 10 min for rats and up to 30 min or longer in dogs. In contrast, administration into small veins in the extremities (where blood flow is much slower), such as the tail vein in rodent models, and rapid bolus injection, should be avoided with poorly water-soluble drugs, if possible, because of the potential for precipitation. Addition of small quantities of surfactants (<0.5%) or polymers may assist in preventing precipitation at the injection site.

Drug administration via other parenteral routes, including s.c., i.m., and i.p. administration, is on occasion warranted, particularly in proof-of-concept biology studies where there is evidence of poor oral exposure resulting from solubility limitations. With non-i.v. parenteral routes, the potential for the route and the formulation vehicle to have an effect (either desirable or undesirable) on the absorption profile is greater, and depot effects are often seen. For these administration routes, solution formulations are still preferred since the volume of fluid at the injection site is low, and, therefore, dissolution of suspension formulations may be limited. The lack of formulation dilution at the injection site dictates that the range of excipients and conditions to promote drug solubilization is even more limited than it is after i.v. administration. Ideally, non-i.v. parenteral formulations should be as close as possible to physiologic pH, and concentrations of cosolvents should be minimized.

2. Parenteral Formulation Approaches for Poorly Water-Soluble Drugs

The most common approaches for the development of parenteral formulations for poorly water-soluble drugs are summarized in the following discussion, and strategies for rapid identification of the most appropriate approach are discussed. The acceptable limits for each of these will depend on many factors including the species (animal or human), dose volume, rate of administration (e.g., infusion or bolus), parenteral route (i.v., i.m., s.c., i.p.), and duration of treatment (Gad et al., 2006; Li and Zhao, 2007). A summary of parenteral formulation approaches is given in Table 2, and these are briefly described as follows.

pH adjustment (see Section IV) is a powerful mechanism by which the solubility of weak acids and weak bases may be enhanced, but the approach is limited by the pK_a of the compound and the acceptable pH range of the formulation. In this regard, acidic solutions are generally more readily tolerated than are basic solutions, and working pH ranges of 2–9 (Li and Zhao, 2007) or 4–9 (Lee et al., 2003) have been suggested. The pH of parenteral solutions can be manipulated via the addition of small quantities of strong acids or bases, such as HCl or NaOH but is more usefully achieved via the use of a buffer, generally with as low a buffer concentration as possible while still maintaining the desired pH. Common buffer systems used in parenteral formulations include phosphate and bicarbonate at neutral or slightly basic pH and citrate, lactate, tartrate, acetate, and maleate for more acidic solutions (Flynn, 1980; Kipp, 2007).

Cosolvents (see Section VI) are also commonly used to promote solubility in parenteral formulations, and ethanol, propylene glycol (PG), low-molecular-weight polyethylene glycol (PEG 300 or 400), dimethylacetamide (DMA), and DMSO have been used. Combinations of cosolvents provide benefits since cosolvency is additive, whereas toxicity is often dependent on the particular solvent.

In many cases, judicious manipulation of solution pH in combination with cosolvents is sufficient to provide solubilization and is typically viewed as the first-line approach. For example, in a review of 317 discovery compounds, Lee et al., (2003) reported that >90% of compounds could be formulated using pH and cosolvents and that >60% of these could be formulated using cosolvent levels of less than 50% v/v.

However, where pH adjustment and cosolvents fail to provide the required solubility, alternative approaches must be used. Second-tier approaches include the use of cyclodextrins, surfactants, and lipid emulsions and, ultimately, more complex formulations such as liposomes, microemulsions, or nanosuspensions. For more complex systems, however, the potential effect, either positive or negative, on systemic clearance and distribution (and also absorption for non-i.v. parenteral formulations) should be carefully considered.

Surfactants (see Section VII) can also be used to enhance drug solubility via micellar solubilization. Surfactants are generally limited to those that are nonionic because of their more favorable safety profile and may include polysorbates (e.g., Tween 80), Cremophors (including Cremophor EL and RH40), Solutol HS-15 (BASF Corporation, Washington, DC), vitamin E TPGS, and poloxamers (Pluronics, including Pluronic F68; BASF Corporation). Whereas surfactants provide for relatively robust solubilization, a complexity is the growing realization that many surfactants can affect transport processes, such as cellular efflux, and may also result in significant adverse effects, including hypersensitivity reactions (Gelderblom et al., 2001; ten Tije et al., 2003). Cremophor EL, in particular, seems to promote an immune reaction in some species (most notably dogs) when administered parenterally. Surfactants bring additional complexities to data interpretation and, where possible, might usefully be avoided in favor of pH changes, cosolvents, or the use of cyclodextrins.

Complexation agents (see Section VIII), such as cyclodextrins, provide a mechanism for enhancing solubility for many drugs and have been increasingly used as modified cyclodextrins, such as hydroxypropyl-β-cyclodextrin (HP-β-CD) and sulfobutylether-β-cyclodextrin (SBE-β-CD), have become more cost effective. HP-β-CD and SBE-β-CD have higher aqueous solubilities than do unmodified cyclodextrins (therefore allowing the support of higher concentrations of complexed drug) and, importantly, are becoming well characterized in terms of their safety profile. In rare situations where the binding constant between drug and cyclodextrin is particularly high (e.g., ∼10⁶ M⁻¹), there is the potential for alteration of the pharmacokinetic distribution and elimination processes (Charman et al., 2006), but most drugs exhibit significantly lower binding constants (<10⁴ M⁻¹) and are efficiently “released” from the complex via dilution and displacement by endogenous species (Stella and Rajewski, 1997; Stella et al., 1999; Stella and He, 2008). HP-β-CD and SBE-β-CD are commonly used in parenteral formulations in concentration ranges of up to 20 to 40% w/v in both humans and animals, and combining complexation with pH adjustment can, in some cases, provide additional solubility enhancement.

Intravenous emulsion formulations (see Section XI ) can be used for lipophilic, poorly water-soluble drugs that have high solubility in lipids. Typically, lipid emulsion formulations contain between 10 and 20% lipid, which limits drug load for compounds that do not have high lipid solubility. Intravenous emulsion formulations have been used for many years as nutritional supplements (e.g., Intralipid; Baxter Healthcare, Deerfield, IL) and are usually formulated using soybean oil as the dispersed phase and phospholipids as the emulsifier. Emulsions can be formulated de novo by dissolving drug into the oil phase of the emulsion before emulsification; however, this requires the use of specialized high-pressure homogenization equipment to provide emulsions with submicron particle sizes. Alternatively, drug may be incorporated into preformed commercial emulsions by dissolving the drug in a small volume of cosolvent with subsequent slow incorporation using sonication (El-Sayed and Repta, 1983).

More complex formulation approaches, such as the use of nanosuspensions (see Section IX), liposomes, or microemulsions, have all been used to facilitate parenteral administration but can substantially affect drug disposition. These approaches are better avoided when the intent of the formulation is to define fundamental pharmacokinetic parameters of a drug.

In many cases, combination approaches to solubilization in parenteral formulations are beneficial. The overarching premise for this approach is that the benefits of solubilization are often additive, whereas the effects on toxicity are typically a function of reaching a critical level for a single component. Maintaining concentration of one solubilizer below that threshold, but combining multiple approaches, therefore, may have benefit. In some cases, however, combination approaches add complexity without generating benefit or, indeed, are detrimental. Combination approaches are summarized in Table 3 and are further described in detail in each individual section of this review.

1. Formulations to Support Drug Discovery

In proof-of-concept pharmacology studies, the focus is generally placed on generating sufficient exposure to allow early indication of the efficacy of a particular pharmacophore or compound series. At this stage, only small quantities of compound are usually available, and control of crystallinity is limited, with early batches often including a significant proportion of amorphous material. Before a salt screening program, ionizable compounds may also be available only as the free acid or base or potentially as a non-optimal salt form. Under these circumstances, simple solution formulations are preferred since they circumvent the complexities of dissolution and, therefore, obviate any variability in exposure resulting from differences in solid-state characteristics. Solution formulations facilitate ease and reproducibility of administration via oral gavage, enable ready conduct of single and multiple dose studies, and support dosing strategies such as cassette dosing. The approaches taken to develop oral solution formations are similar to those used for parenteral formulations, however, the acceptable limits on formulation components are typically wider. A useful order of potential formulation approaches to achieve simple oral solution formulations is provided in Table 4. In early stage proof-of-concept studies, relatively aggressive absorption enabling formulations may be used to overcome solubility and dissolution rate limitations, depending on the frequency and length of dosing; however, caution should be used to ensure that the formulation components do not confound the efficacy readout, and vehicle controls are essential.

Oral delivery during lead optimization is typically focused on prioritizing lead compounds from within a series with more favorable absorption properties and on identifying the key structural and physicochemical factors that contribute to poor oral bioavailability, in an effort to guide subsequent structural modification. Understanding the basis for poor oral absorption is fundamental to designing a strategy to overcome this limitation. The approaches used for oral formulations of poorly water-soluble compounds in lead optimization are therefore usually less complex than those incorporated, for example, during discovery biology, and are designed to discriminate between compounds in the absence of enabling formulation effects. Consistency in formulation is important to distinguish between potential effects resulting from the formulation and intrinsic differences in structural and physicochemical properties of the lead.

Simple suspension formulations containing a suspending agent (e.g., cellulose derivatives) along with a low concentration of surfactant to reduce particle agglomeration are often used in early bioavailability studies (see Table 5). These formulations provide little in the way of “assisted” dissolution and therefore tend to discriminate between different compounds. A further advantage is that they can be readily extended into animal toxicology studies, provided sufficient compound exposure can be achieved. A complexity in the use of suspension formulations is the potential impact of particle size on dissolution rate and the possibility of batch-to-batch variability given that solid-state properties and particle size are not likely to be optimized at the time the initial pharmacokinetic and bioavailability studies are conducted. Assessment and monitoring of particle size are beneficial, and normalization of particle size across batches via laboratory scale milling may provide greater consistency.

Another approach is to use simple solution formulations, typically prepared using cosolvents with or without pH manipulation. These formulations have the advantage of being independent of solid-state properties since compounds are in solution but have the disadvantage of likely precipitation after administration, particularly if the formulations prepared using cosolvents are near the saturation solubility limit. Generally, more specialized enabling formulations, such as surfactant-, cyclodextrin- and lipid-based formulations, are not ideal for lead optimization in that they may mask intrinsically poor physicochemical properties and take the focus away from structural optimization to improve solubility. An early reliance on the use of enabling formulations also complicates subsequent developmental activities.

As larger quantities of drug become available, more definitive information regarding solid-state properties is generated and salt screening programs are initiated for weak acids and bases (see Section IV). Isolation as a salt form can provide for significant enhancements in dissolution rate. The availability of larger quantities of material also allows more complex particle size reduction strategies to be explored including micronized formulations and nanosuspensions (see Section IX).

2. Use of Enabling Formulations to Promote Oral Absorption

In situations where attempts to increase polarity and water solubility lead to significant decreases in potency, and where solubility or dissolution has been identified as the rate-limiting process for absorption, a strategic decision may be made to progress a compound series via the use of enabling formulations to improve bioavailability and exposure. This approach may be taken in an effort to obtain rapid proof-of-concept data in humans against a new target or simply to enable early pharmacokinetic and safety data to be obtained for a novel compound series.

The preclinical assessment of more specialized oral formulations will depend heavily on the type of formulation being assessed. Generally, rodents require liquid formulations that can be easily administered via a gavage tube. Recent developments in the availability of minicapsules and tablets allows dosing of tablets or encapsulated materials to rodents; however, the requirement for specialized equipment to prepare these dosage forms and the low liquid volumes in the rodent GI tract (which may limit the rate of capsule/tablet disintegration) make this approach less common.

Alternate crystal forms such as cocrystals, polymorphs, or amorphous drug may explored, although the inherent instability of metastable crystal and amorphous forms dictates the need for close monitoring of crystallization properties, and formulation approaches to stabilize against solid-state transitions may be necessary, depending on the required stability period. Cocrystals provide some advantage in this respect as their crystal lattice energy is significantly higher than amorphous drug, allowing the generation of formulations with improved stability. Stabilization of polymorphs or amorphous drug may be achieved by the use of (usually polymeric) solid dispersion formulations, where reduced molecular mobility, often achieved via intermolecular interactions between carrier and drug, reduces the risk of changes to the amorphous-crystal structure during storage. Solid dispersions require relatively complex equipment to provide clinical formulations, such as spray drying or hot melt extrusion; however, trial batches can be formulated using simple benchtop equipment such as a rotary evaporator or freeze-driers.

As drug candidates progress through development, preclinical investigations in nonrodent species will be conducted, usually in dogs or nonhuman primates. The physical form of materials used in these studies will vary, and simple solutions or suspensions will often still be used. Larger species generally provide the first opportunity to assess the performance of prototype tablet and capsule formulations that might ultimately progress to the clinic. For ease of preparation on a relatively small scale, hand filling of dry powder blends or granules into hard gelatin capsules or liquid filling into preformed soft gelatin capsules can be used. Sealed hard-gelatin capsules provide an alternative for liquid fills and may be sealed by hand or, more effectively using specialized capsule-sealing equipment, realizing that such equipment may not be readily available at this stage. For poorly water-soluble drugs, the shift to encapsulated and solid dosage forms dictates renewed concentration on solid-state formulation strategies such as salts, cocrystals, and particle size reduction alongside more complex formulation approaches, including solid dispersions and lipid-based formulations.

3. Preclinical Toxicology Formulations

Preclinical toxicology evaluation is a particular challenge for poorly water-soluble drugs. The need to escalate doses to the point where in vivo exposure is sufficient to demonstrate limiting toxicity generally requires absorbed doses (in milligram per kilogram terms) 30 to 100 times greater than might be envisaged with a clinically relevant dose and formulation. Where water solubility is low, this is often impossible with solution or suspension formulations, since beyond a certain point, drug exposure remains constant regardless of increases in dose, dictating the need for more complex formulation approaches; however, this requirement is tempered by potential concerns regarding the toxicology of the large quantities of excipient often required to afford the higher exposure. Relatively little information is publically available that describes the quantities of different combinations of excipients that do not lead to toxic endpoints over a range of dosing periods. One notable exception is an excellent review by Gad et al. (2006), which provides a summary of the effects of a range of excipients in toxicity testing protocols of different lengths. Neervannan (2006) also provide some data regarding the maximum quantities of common excipients that might be safely used in acute and subchronic dosing protocols for up to 7 days. However, these communications provide data on a limited number of surfactants, lipids, and polymers and do not address the potential toxicity of the combinations of components often used.

Toxicity concerns for many of the excipients used in more complex formulations are attenuated by the realization that many polymers and surfactants have high molecular weights and are relatively polar, or amphiphilic. As such, the extent of absorption of these excipients after oral administration is low. Many surfactants are also digestible (Cuiné et al., 2008). Under these circumstances, the risks of systemic toxicity are low, and the oral LD₅₀s of most hydrophilic polymers and nonionic surfactants are high (Rowe et al., 2009). The major concerns therefore revolve around the possibility of GI side effects, including loose stools and emesis (Reppas et al., 1991; Andersen, 1994; Schulze et al., 2005; Ruble et al., 2006). By incorporation of vehicle controls, many of these effects can be accounted for during data analysis. Confounding data interpretations issues arise, however, where GI symptoms manifest as a consequence of the candidate drug’s desired biologic activity or toxicity profile.

Cosolvents, in particular ethanol, typically exhibit more significant toxicity profiles and, hence, cosolvent combinations are commonly used to maintain the concentration of one single component at a relatively low level. Finally, several surfactants have also been shown to inhibit efflux transporters (Batrakova et al., 1999; Yamagata et al., 2007a) and GI cytochrome P450 enzymes (Wandel et al., 2003; Bravo González et al., 2004; Christiansen et al., 2011) and may therefore affect drug bioavailability after oral administration. Evidence of systemic effects, however, is limited except after parenteral administration. After parenteral administration, surfactants such as Tween 80 and Cremophor EL may stimulate hypersensitivity reactions (Weiss et al., 1990; Gelderblom et al., 2001; ten Tije et al., 2003), especially in dogs (Lorenz et al., 1982).

4. Development of Clinical Formulations for Poorly Water-Soluble Drugs

The strategies used when developing clinical formulations typically fall into two categories. The first strategy is the development of simple formulations that maximize the speed to first-in-human studies but provide no functional link to the likely formulations used in later stage clinical trials and that might be intended for commercialization. Second is the development of mature formulations, with the expectation of progressing essentially similar formulations through clinical development to commercialization. Where possible, the former approach uses nonformulated drug to simplify dosage form development, and, where solubility permits, simple direct filling of drug in capsules or drug-in-bottle approaches can be used. The relatively recent application of direct capsule-filling instruments such as Xcelodose (Capsugel, Greenwood, SC) that facilitate accurate filling of doses ranging from 100 μg to 100 mg have greatly simplified this process.

For many poorly water-soluble compounds, however, low solubility limits the possibility of adopting an accelerated route to first-in-human studies, and in these instances, more sophisticated formulations are required to ensure exposure. Formulations with the potential to progress through clinical trials, thereby circumventing the need for repeated bioequivalence bridging studies, are typically evaluated. The escalation of formulation complexity in this case (Table 6) commonly follows a pathway similar to that previously described for preclinical formulations, although in this case, the potential to progress through to full scale tableting or encapsulation methods and the use of Good Manufacturing Practice processing equipment becomes more important. Where simple solid-state or particle size approaches to enhanced solubility such as salt formation, cocrystals, or nanocrystals fail to provide sufficient exposure, the major routes of progression are the use of lipid-based formulations or solid dispersions. The choice of approach is dictated largely by the physicochemical properties of the drug candidate, where lipid-based formulations lend themselves to poorly water-soluble, lipophilic drugs (i.e., those with sufficient solubility in lipids, surfactants, and cosolvents to allow the target dose to be dissolved in the capsule fill material), whereas solid dispersions are commonly explored for drug candidates with insufficient lipophilicity to allow the use of lipid-based formulations.

In conclusion, the general strategies that may be taken to identify and overcome the challenges of low water solubility in the drug discovery and development environment are summarized in the proceeding discussion. The tools that have been suggested as approaches to overcome the limitations of poor solubility are described in detail in following sections.

1. Ionic Equilibria

On addition to water, strong electrolytes such as hydrochloric acid and sodium hydroxide dissociate almost completely. In contrast, weak electrolytes, such as the weakly acidic and weakly basic molecules that constitute most ionizable drugs, dissociate only partially at most pH values. For a weak acid or weak base, the total concentration in solution at a given pH is therefore the sum of the concentrations of the ionized and un-ionized species. The degree of dissociation into the ionized and un-ionized form is described by the dissociation constant (K_a) for the equilibrium reactions shown in eq. 12 for a monoprotic weak acid (HA) (left equation) and weak base (BH⁺) (right equation) in water, with the associated expressions for K_a shown in eq. 13.(12)(13)

As the dissociation constant may vary by several orders of magnitude, the pK_a or negative logarithm of K_a (pK_a = −log K_a) is more typically used to compare acids and bases (Table 7). Decreasing pK_a denotes stronger acids (i.e., species that more readily donate a proton), whereas increasing pK_a reflects stronger bases (i.e., species that more readily accept a proton).

The Henderson-Hasselbalch equation (eq. 14) further describes the pH dependency of the ionization properties of weak acids (left equation) or bases (right equation) and relates the degree of ionization to the pH of a solution and the pK_a of the drug.(14)

From eq. 14, it is apparent that under conditions where the pH of a solution is the same as the pK_a of a dissolved weak acid or base, the concentrations of ionized and un-ionized species are equivalent. A change in pH by one log unit away from the pK_a subsequently results in a 10-fold change in the degree of ionization.

2. pH Solubility Relationships for Weak Electrolytes and Salts of Weak Electrolytes

The intrinsic solubility of a nonelectrolyte is governed by the relative strengths of the intermolecular forces in the solute (i.e., its solid-state properties) and the affinity of the solute for the solvent (i.e., the strength of solute-solvent interactions) (see Section I.B). This is also the case for a weak electrolyte but is complicated by differences in ionization of the solute with changes in pH that, in turn, alter polarity and the affinity of the solute for the solvent. Thus, a weak electrolyte will initially dissolve to an extent governed by the intrinsic solubility of the un-ionized form. Depending on the pH, a proportion of the dissolved species subsequently ionizes in accordance with the Henderson-Hasselbalch equation (eq. 14), leading to further dissolution of solid drug to maintain a constant concentration (i.e., the saturated solubility) of the un-ionized form. This will continue within an unbuffered solution until the saturated solubility of both un-ionized and ionized species is reached. As the ionized species is able to form favorable ion-dipole interactions with polar solvents, it is more soluble than the un-ionized species, and therefore, drugs are more soluble in polar solvents at pH values that favor the ionized form. The total solubility (S) of a weak acid (HA) or weak base (B) at a given pH is therefore the sum of the solubility of the ionized and un-ionized forms (Yalkowsky, 1999).(15)

Combinations of eq. 14 and eq. 15 generate the following solubility equations for weak acids (left, eq. 16) and weak bases (right, eq. 16) where total drug solubility at a particular pH is a function of the solubility of the un-ionized form (S₀) and the pK_a of the weak acid or base.(16)

From eq. 16, it is evident that for a monoprotic weak acid or base, changes in pH of 1 unit away from the pK_a lead to 10-fold (1 log unit) differences in solubility. For diprotic species, similar relationships may be developed (eq. 17) where total solubility is dependent on the pK_a of two ionization centers (Yalkowsky, 1999). Accordingly, diprotic species are expected to show a 100-fold (2-log unit) change in solubility with a change in pH of one unit (Garren and Pyter, 1990; Venkatesh et al., 1996). The top equation is for a weak diprotic acid, and the bottom equation is for a weak diprotic base:(17)

The solubility equations for weak electrolytes shown in the preceding should be viewed with two significant caveats. First, eq. 16 and eq. 17 suggest that drug solubility changes indefinitely with changing pH. Ultimately, however, the achievable solubility is regulated in most cases by the presence of counterions that may be intentionally present (in the case of a salt) or unintentionally present (in the case of ions present in blood, GI fluids, buffers, and so forth). In this case, the total solubility is regulated by the limit of solubility of the ionized weak acid or base in the presence of the corresponding basic or acidic counterion, ultimately resulting in precipitation of the solid salt of the acid-base pairs. The solubility of the salt is therefore the limiting condition to the aqueous solubility of drug in the ionized form when a counterion is present. Where multiple counterions are present, the limiting salt solubility typically reflects the salt with the lowest solubility, or the salt formed with the counterion in highest concentration if common ion effects are significant. Salt solubility, including aspects of common ion effects, is described in more detail in Section IV.4. Second, classic ionization texts usually ignore the possibility of ion-pair formation or self-association. Where ion-pair formation or self-association leads to the formation of supramolecular complexes in solution (including multimers, micelles, and liquid crystalline species), apparent solubility values may be much greater than anticipated (Serajuddin et al., 1987; King et al., 1989; Kriwet and Muller-Goyman, 1993; Fini et al., 1995).

Exemplar pH solubility profiles for a weak acid and a weak base, and the salts formed by association with a corresponding basic or acidic counterion, are presented in Fig. 7. The total drug solubility (S in eq. 16) of a weak base is low at high pH where the un-ionized form (B) predominates (region 1 of the pH solubility profile for a weak base). Where the solubility of the ionized drug species (BH⁺) is significantly in excess of that of the un-ionized form, B (which is typical), decreasing pH results in an increase in the proportion of BH⁺ and an increase in total solubility. The increase in solubility becomes particularly significant from pH values of approximately 1 pH unit above the pK_a (i.e., where the degree of dissociation increases beyond ∼10%) to pH values below the pK_a where, as described previously, solubility increases 10-fold for every unit pH decrease (for a weak base). At a critical point (and critical pH; denoted pH_max), the equilibrium solubility of the ionized species (as the salt) is reached, and no further increases in solubility are seen with decreasing pH. This limiting solubility is set by the solubility product of the salt (K_sp):(18)

When the concentration of solid phase (i.e., salt) is in excess and, therefore, approximately constant:(19)

In region 2 of Fig. 7 (i.e., at the solubility limit), excess solid phase is present as the salt, and the saturated species in solution is largely the ionized form. At pH values <pH_max, the total solubility of a weak base is given by [BH⁺] plus a small (and decreasing) quantity of B (free base) (Kramer and Flynn, 1972; Serajuddin, 2007). The total solubility in region 2 of Fig. 7 is given by eq. 20. for weak acids (top equation) and weak bases (bottom equation).(20)

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

The pH solubility profiles for (A) a weakly basic and (B) a weakly acidic compound. In region 1, the un-ionized free acid or free base is present as the solid phase and as the saturation or equilibrium species in solution. Total solubility is a function of both un-ionized and ionized species (eq. 16), and at pH values greater than (for a weak acid) and less than (for a weak base) pK_a, solubility increases because of increased quantities of the ionized species in solution. Solubility is ultimately limited in the “salt plateau” region (region 2) by K_sp (eq. 19)_. In region 2, the salt is present as the solid phase, and the ionized form is the saturated or equilibrium species in solution. At pH values greater than (for a weak acid) and lower than (for a weak base) pH_max, the total solubility decreases slightly because of a reduction in the quantity of free acid or free base in solution (eq. 20). Where pH change continues as a result of the further addition of acidic or basic counterion, salt solubility in region 2 may decrease much further (dotted line) as a result of the common ion effect.

Scheme 1 summarizes the solid-solution equilibria for a weak base (B), where the counterion is provided by a strong acid (HX) to form a salt (BH⁺X^-)_solid. This scheme brings together the relationships between dissociation (described by K_a and the dominant factor in region 1 of Fig. 7) and the equilibrium solubility of the salt (described by K_sp and the dominant factor in region 2 of Fig. 7). In some cases, additional equilibria, including the potential for hydrate formation or self-association, may be present but have been omitted here for the sake of clarity.

• Download figure
• Open in new tab
• Download powerpoint

Scheme 1.

Phase solubility scheme for salt formation between a weak base (B) with a strong acid (X^-). Adapted from Pudipeddi et al. (2002).

In a solution of a monovalent salt, dissociation leads to equal concentrations of [BH⁺] and [X⁻]. From eq. 19, the maximum solubility of [BH⁺] (i.e., the limiting value for salt solubility) is equal to the square root of the solubility product (√K_sp). K_sp is a constant for a particular salt and defines the maximum concentration of drug ions and corresponding counterions that a solvent can maintain in solution (Bhattachar et al., 2006). Essentially, the higher the K_sp value, the greater the solubility of the salt. K_sp is therefore a critical parameter in understanding the solubility of salts or the solubility of ionized drugs where salt formation in situ is likely. Unfortunately, techniques that accurately predict K_sp do not exist, largely as a result of the unpredictability of the nature of the crystal lattice. Screening programs that are able to rationally and rapidly assess the limit of solubility of different salt forms are therefore critical to support effective development programs, and are described in more detail in Section IV.E.2.

3. Factors Affecting Salt Formation at pH_max

The pH_max defines the limiting pH above which (for a weak acid) and below which (for a weak base) the solid phase constitutes the salt rather than the free acid or free base. To facilitate salt formation, the bases or acids that are used to provide the counterion must be sufficiently strong to shift solution pH past pH_max (into region 2 of Fig. 7). In general, salts of weak bases will form more readily with more strongly acidic counterions such as hydrochloride compared with more weakly acidic counterions such as citrate. As a general rule, salt formation will occur only between a weak acid or weak base and counterion if the difference in pK_a for the two is greater than 2 (Pudipeddi et al., 2002; Stahl and Wermuth, 2002; Black et al., 2007). It is also evident that conditions that cause a shift in pH_max will change the ease of salt formulation. Thus, conditions that result in an increase in pH_max for a weak base and a decrease in pH_max for a weak acid will assist in salt formation by reducing the pH shift that must be generated by the counterion to move into region 2 of Fig. 7. For example, decreasing the pH_max for a weak acid will allow salt formation with a more weakly basic counterion, thereby increasing the possible choices of counterion.

At the pH_max value, the salt and the free acid or free base are present in the solid state, and this situation is unique to this transition point in the pH-solubility profile. Under these circumstances, eq. 16 and eq. 20 are both applicable. Setting eq. 16 equal to eq. 20 and solving for pH (i.e., pH_max) results in eq. 21 and a relationship that provides an indication of the dependence of pH_max on pK_a, the intrinsic solubility of the un-ionized free acid or free base (S₀) and K_sp (Bogardus and Blackwood, 1979; Serajuddin, 2007). Note that the equation on the left applies for weak acids while the equation on the right applies for weak bases:(21)

From eq. 21, it is clear that the value for pH_max is dependent on the properties of the drug namely, intrinsic solubility (S₀), relative acidity/basicity (pK_a) and the solubility of the salt (K_sp). The effect of changing these three terms on pH_max is illustrated for a typical weak base in Fig. 8. Increasing the pK_a (i.e., increasing basicity) (Fig. 8A), increasing S₀ (i.e., increasing intrinsic solubility) (Fig. 8B), and decreasing K_sp (i.e., forming salts with lower solubilities), all assist salt formation by increasing pH_max (realizing that lower solubility salts may not be preferred in many cases).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

The effect of (A) pK_a, (B) intrinsic drug solubility (S₀) and (c) salt solubility (K_sp) on the position of the pH_max of a weak base. Salt formation of a weak base becomes more attainable (an increase in pH_max) with increasing pK_a of the base, an increase in intrinsic solubility, or a decrease in salt solubility. Adapted from Pudipeddi et al. (2002).

4. Determinants of Salt Solubility

The maximum solubility of a salt is the boundary condition for in vivo solubility and dissolution and, therefore, the parameter against which alternate salts are screened (at least for their impact on solubility). The accurate determination of salt solubility, however, is not a trivial task due to the potential for interconversion between the salt form and the free base or free acid (see Section IV. D.1.c), and the impact of a number of system variables. These are described below.

b. Choice of Counterion to Maximize Salt Solubility

Table 8 provides a list of common counterions that have been used to isolate salts of weakly acidic and weakly basic drugs for the development of oral and parenteral drug products. An indication of the frequency of use is also provided (Paulekuhn et al., 2007). As K_sp is counterion specific, the solubilities of salts formed with a variety of different counterions can vary significantly. For example, solubility differences of up to 100-fold have been described for a series of amine salts of diclofenac (O’Connor and Corrigan, 2001), and differences in solubility of up to four orders of magnitude have been described for salts of ibuprofen (David et al., 2012). Similarly, for the weak base enalapril, salts formed with a range of aliphatic or aromatic carboxylic acids exhibit solubilities ranging from 22 mg/ml up to 350 mg/ml (Kumar et al., 2008a).

View this table:

• View inline
• View popup

TABLE 8

List of acidic and basic counterions commonly used in pharmaceutical formulations

A better understanding of the factors that affect salt solubility (and therefore counterion selection) may be gained by examination of the determinants of the molar free energy of solution (ΔG_soln) (Chowhan, 1978):(22)(23)

Solubility is a function of the free energy gain achieved by solvation of the anion and cation (eq. 23) offset by the free energy of the crystal lattice (eq. 22). Salt formation has the capacity to alter both the free energy of the crystal lattice (ΔG_lattice) (by changing the nature of packing and the strength of intermolecular interactions in the solid state) and the degree of solvation (ΔG_solvation). Solvation is typically enhanced by salt formation since dissociation into ions facilities the formation of ion-dipole interactions with water molecules that are more energetically favorable than, for example, hydrogen bond interactions between water and un-ionized drug. In contrast, the crystal lattice energy is typically increased by salt formation via the generation of stronger intramolecular ionic interactions, and this is usually evident in an increase in melting point (Corrigan, 2007; David et al., 2012). From eq. 22, the increase in (ΔG_lattice) is expected to be detrimental to solubility. However, the higher solvation energies resulting from ionization typically offset the energy cost of breaking ionic interactions in the crystal lattice, leading to a net increase in solubility for salts compared with the free acid or base.

The importance of solvation of both anion and cation to the overall solvation energy dictates that hydrophilic counterions, by virtue of higher solvation energies, might be expected to form more soluble salts than hydrophobic counterions (Menegon et al., 2011; David et al., 2012), and indeed this is often the case. However, the choice of counterion affects both solvation and lattice energy, and as such, definitive structure-solubility relationships for differing counterions do not exist (Stahl, 2003; Corrigan, 2007; Serajuddin, 2007). Thus, monovalent and divalent alkali metals (K⁺, Na⁺, Mg²⁺, Ca²⁺) are widely used counterions for salt formation for weakly acidic drugs (Table 8), and increasing valency and decreasing counterion ionic radius might be expected to increase hydrophilicity and facilitate solvation. However, salts of monovalent and larger cations (of equivalent valency) are typically the most soluble (Chowhan, 1978; Anderson and Conradi, 1985; Forbes et al., 1995). For example, Fig. 9 depicts the solubility of metal salts of 7-methylsulfinyl-2-xanthonecarboxylic as a function of pH (Chowhan, 1978) and illustrates that maximum salt solubility is evident for larger monovalent counterions [i.e., K⁺ (most soluble) > Na⁺, > Ca²⁺ > Mg²⁺]. These trends suggest that counterion effects on lattice energy (where larger, less highly charged species reduce intermolecular attractive forces and typically result in lower crystal lattice energies) are more critical determinants of the solubility of a salt than corresponding effects on solvation (where higher-charge densities are expected to increase solvation). Hydrophilic counterions may therefore show higher solvation energies but result in lower solubility (compared with a less hydrophilic counterion) by virtue of higher lattice energies. For example, the lactate and mesylate salts of the experimental antimalarial drug [α-(2-piperidyl)-3,6-bis(trifluoromethyl)-9-phenanthrenemethanol] are more soluble than salts formed with the freely soluble hydrochloride counterion (Agharkar et al., 1976), and saccharin salts of lamotrigine are more soluble than those formed with the more hydrophilic succinic acid and fumaric acid counterions (Galcera and Molins, 2009). Kumar et al. (2008a) have also shown that malate salts of enalapril are more than 10 times more soluble than maleate, besylate, and tosylate salts, despite each of the counterions showing very high aqueous solubilities. Finally, the solubilities of 22 salts formed between various p-substituted benzoic acid derivatives and benzylamine have been shown to vary significantly, despite broadly similar energies of solvation (Parshad et al., 2002). Across different salt series, lattice energy effects on solubility therefore appear to dominate over solvation effects in many cases.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9.

The pH-solubility profiles for 7-methylsulfinyl-2-xanthonecarboxylic acid and its salts at 25°C. Symbols represent experimental data. The point at which horizontal lines cross the upward solubility curve represent the pH_max values for respective salts. Solid symbols indicate that the starting material was the acid, and open symbols indicate that the starting material was the salt. Adapted from Chowhan (1978).

Consistent with the potential importance of differences in lattice energy on salt solubility, several authors have described correlations between the melting point of pharmaceutical salts (as a measure of lattice energy) and saturated aqueous solubility (Anderson and Conradi, 1985; Gould, 1986; Thomas and Rubino, 1996; Ledwidge and Corrigan, 1998; O’Connor and Corrigan, 2001; Guerrieri et al., 2010). Figure 10, for example, shows the relationship between melting point and solubility for a range of flurbiprofen salts formed with different organic and inorganic amine cations and provides good evidence of a relationship between decreasing melting point and increasing solubility (Anderson and Conradi, 1985). The properties of the counterion that affect packing within the salt lattice, and therefore the lattice energy, include symmetry and size, the ability to form hydrogen-bonds with the drug (in addition to ionic interactions), and the capacity to delocalize charge (Chowhan, 1978; Gould, 1986; Parshad et al., 2004). For example, salts of the weakly basic drug candidate (UK-47880) formed with planar and aromatic acids were found to be highly crystalline and to have higher melting temperatures compared with an oily, low melting salt formed with two aliphatic and more flexible carboxylic acids (citric acid and dodecyl benzene sulfonic acid) (Gould, 1986). Malate and maleate salts of ephedrine also show higher melting temperatures compared with fumarate and tartrate salts (Black et al., 2007). This has been suggested to reflect the small size of the malate counterion and the symmetrical nature of the maleate counterion, both factors that favor hydrogen bonding within the salt lattice such that the lattice energy is increased (Gould, 1986). Similarly, diclofenac salts formed with the freely soluble tris(hydroxymethyl)aminomethane (TRIS) counterion were found to be less soluble than expected, a situation later ascribed to the ability of TRIS to hydrogen bond within the salt lattice (Fini et al., 1996; O’Connor and Corrigan, 2001). In this example, the TRIS salt had the highest melting temperature and lowest solubility of a range of diclofenac salts formed with organic amine counterions and therefore provides another example of the potential for more polar counterions to result in lower salt solubilities via stronger packing within the crystal lattice (Corrigan, 2007). In contrast, increasing the alkyl chain length of amine counterions led to a decrease in the solubility of salts of several weak acids (ibuprofen, gemfibrozil, etodolac), although increasing the alkyl chain length on the counterion lowered the salt melting temperature (David et al., 2012). The overall effect of counterion polarity on solubility therefore reflects the net effect of changes to both lattice energy (where more polar counterions typically increase lattice energy, decreasing solubility) and hydration (where increasing counterion polarity typically increases hydration and increases solubility).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 10.

The relationship between log K_sp and the melting temperature of 1:1 amine salts of flurbiprofen. Basic counterions were 1) = 2-amino-2-methyl-1,3-propanediol, 2) = ammonium, 3) = tromethamine, 4) 2-amino-2-methylpropanol, 5) = tert-butylamine, and 6) = adamantanamine. Adapted from Anderson and Conradi (1985).

Interestingly, mesylate (methanesulfonate) salts appear to be finding increasing favor for the delivery of weak bases both orally and parenterally (Paulekuhn et al., 2007; Elder et al., 2010; Branham et al., 2012). The wider use of the mesylate counterion in salt selection strategies for solubility enhancement may be in part attributed to its high-charge density but also to its retention of a degree of hydrophobicity, which appears to affect molecular packing in the crystal lattice. As a result, salts formed with the mesylate counterion are likely to show high solvation energies compared with salts formed with other organic counterions (because of the high-charge density) but lower lattice energies compared with salts formed with inorganic counterions (which also show a high charge density). For example, mesylate salts of α-(2-piperidyl)-3,6-bis(trifluoromethyl-)9-phenanthrenemethanol(Agharkar et al., 1976), RPR200765 (Bastin et al., 2000), and haloperidol (Li et al., 2005b) all showed lower melting temperatures compared with equivalent hydrochloride salt forms. The mesylate ion also has a very low pK_a (the lowest of all common organic acidic counterions) (Table 8), in turn facilitating salt formation with weak and very weak bases (Stahl, 2003). Mesylate salts are also less susceptible to the common ion effect in vivo, and this is described in more detail in Section IV.D.1.d.

Finally, the choice of counterion requires appreciation of the impact of the counterion on a range of pharmaceutical factors, including solubility, but also on other properties, such as flow, compression, and stability. Currently, the ability to predict differences in the solid-state properties of salts isolated with differing counterions is limited by a lack of understanding of the factors that control packing arrangements during crystallization and the potential for ionized drug molecules to self-associate in solution. This in turn limits accurate de novo prediction of counterion effects on solubility. Consequently, it is more typical to select the most appropriate salt via a salt-screening process. Salt screens are described in more detail in Section IV.E.2.

1. Buffered Systems Used in Parenteral Formulations

The most commonly used pharmaceutical buffers comprise weak acids (HA) and the conjugate base (A⁻). Since physiological pH covers mainly neutral and acidic pH, buffers of weak bases (B) and their conjugate acids (BH⁺) are less frequently used. To exemplify, the acid-conjugate base pair of acetate buffer is illustrated in Scheme 2.

• Download figure
• Open in new tab
• Download powerpoint

Scheme 2.

Dissociation of a commonly used (acetate) buffer.

In this system, an increase in hydrogen ion concentration is buffered by the conjugate base (CH₃COO⁻), and the addition of hydroxide ion is similarly neutralized by the acid (CH₃COOH). Buffer capacity is widely used to measure the ability of a buffer to resist a change in pH and is a measure of the quantity of acid/base required to change the pH of a buffered solution by 1 pH unit. Increasing buffer concentrations leads to increases in buffer capacity, and at a constant buffer concentration, maximum buffering capacity is achieved when the solution pH is close to the pK_a of the buffer.

Details of a wide selection of buffers have been previously published (Flynn, 1980; Merck-Index, 2006). Buffer systems most commonly applied to parenteral pharmaceutical formulations and their effective buffer pH range are listed in Table 9 (Lee et al., 2003; Li and Zhao, 2007; Nema and Brendel, 2011). The most effective buffer range is usually limited to one pH unit either side of the pK_a, although buffer combinations may be used to extend the buffer range.

View this table:

• View inline
• View popup

TABLE 9

Buffers widely used in parenteral formulations, including details of their buffering range and concentration used in some example marketed formulations

During parenteral administration, pain on administration is likely when using solutions buffered to high or low pH. Formulations are therefore typically limited to a working pH range of 4–9, although a wider pH range (2–9) may be used acutely in a preclinical setting (Li and Zhao, 2007). Extremes of pH may also lead to chemical instability, and in these cases, drug stability should be monitored in the formulation for the intended period of use (Pasini and Indelicato, 1992; Islam and Narurkar, 1993; Carstensen, 1995; Zhu et al., 2002; Chen et al., 2006). Buffer concentrations normally exceed 0.05 M and are generally less than 0.5 M (Wang and Kowal, 1980; Sinko, 2006). Increasing buffer capacity does not increase drug solubility but can prove more effective in maintaining drug in solution on dilution or with the addition of a formulation additive. High buffer concentrations, however, increase the risk of pain on injection and in high volume may alter physiological levels of biologically important buffer components such as lactate (Nema and Brendel, 2011). The buffer will also contribute to the overall ionic strength of the formulation, and high buffer concentrations can promote salting out of drug from solution (McDevit and Long, 1952; Raghavan et al., 1996; Yalkowsky, 1999).

A pH adjustment is a simple and powerful mechanism for solubility adjustment and, as such, is typically the first approach for solution formulations of ionizable drugs. For example, naproxen is a weak acid with a pK_a of 4.57 and intrinsic solubility (of the free acid) of 16 μg/ml (Fini et al., 1995). As a comparatively “strong” weak acid, the low solubility of naproxen can be increased dramatically in a pH range acceptable for parenteral administration. Indeed, solution concentrations of ∼45 mg/ml might be expected in pH 8 phosphate buffer and ∼450 mg/ml in pH 9 sodium bicarbonate buffer. Under these circumstances, pH adjustment could be used to deliver a wide range of doses (Lee et al., 2003). In some cases, and in particular for weak(er) acids of higher pK_a and weak(er) bases with lower pK_a values, pH adjustment alone may be insufficient to allow the administration of an appropriate dose. For example, attempts to formulate a drug candidate that was a weaker acid than naproxen (e.g., a compound with pK_a = 7), but of equivalent intrinsic solubility, would provide a maximum concentration of only ∼1.6 mg/ml in bicarbonate buffer (pH 9). In this case, pH adjustment alone is unlikely to offer a viable solubilization strategy, and a combination of strategies would need to be explored, such as the addition of a cosolvent, surfactant, or cyclodextrin. The advantages (and complications) associated with these combination approaches are discussed in the following section, and the benefits of cosolvents, surfactants; cyclodextrins as individual solubility enhancers are described in more detail in Sections VI, VII, and VIII, respectively.

2. Impact of Cosolubilizers and Electrolytes on pH-Mediated Solubilization