Cyclodextrins

Abstract

β-cyclodextrin (β-CD) and other cyclodextrins (CDs) have utility for solubilizing and stabilizing drugs; however, some are nephrotoxic when administered parenterally. A number of workers have attempted to identify, prepare, and evaluate various CD derivatives with superior inclusion complexation and maximal in vivo safety for various biomedical uses. A systematic study led to SBE-β-CD (Captisol), a polyanionic variably substituted sulfobutyl ether of β-CD, as a non-nephrotoxic derivative and HP-β-CD, a modified CD developed by Janssen. SBE-β-CD and HP-β-CD have undergone extensive safety studies and are currently used in six products approved by the Food and Drug Administration (four for Captisol and two for HP-β-CD). They are also in use in numerous clinical and preclinical studies. This article will focus on the issues that led to the development of these two CDs, their safety, characterization, and applications, and especially their ability to improve drug delivery. SBE-β-CD interacts very well with neutral drugs to facilitate solubility and chemical stability, and because of its polyanionic nature, it interacts particularly well with cationic drugs. Complexes between SBE-β-CD and HP-β-CD and various drugs have been shown to rapidly dissociate after parenteral drug administration, to have no tissue-irritating effects after intramuscular dosing, and to result in superior oral bioavailability of poorly water-soluble drugs. The pharmacokinet-ics, tissue distribution, and cellular effects of some representative CDs, including SBE-β-CD and HP-β-CD, are reviewed. The safety profiles of CDs are discussed, with emphasis on the biological effects of some CDs on the gastrointestinal tract, kidney, and reproduction and development and the carcinogenic potential of CDs. In addition, human experience with CD derivatives, specifically SBE-β-CD and HP-β-CD, indicates that these two CDs are well tolerated in humans and have no adverse effects on the kidneys or other organs following either oral or intravenous administration.

Introduction

This short review will address a number of issues associated with the use of cyclodextrins (CDs) to enhance the formulation and delivery of problematic drugs. The article will take the form of a number of questions that we have been asked by industrial, academic, and regulatory colleagues over the years. The review will be limited; however, it is our hope that the reader will find the material presented and the key references useful.

What Are Cyclodextrins and Why Are They of Interest?

The CDs of biomedical and pharmaceutical interest are cyclic oligosaccharides made up of six to eight dextrose units (α-, β-, and γ-CDs, respectively) joined through one to four bonds. These so-called “parent CDs” have themselves been used in food and pharmaceutical products for many years, although their use in the United States has been more limited than in Japan and Europe. A generalized chemical structure of these CDs is shown in Table 1, which also contains the names and abbreviations of some CDs commonly discussed in CD-related biomedical articles.

The pharmaceutical uses of CDs, which are the focus of this article, have been discussed in numerous reviews and books (Duchéne, 1987; Szejtli, 1988; Frömming and Szejtli, 1994; Uekama et al., 1994; Loftsson, 1995; Albers and Muller, 1995; Loftsson and Brewster, 1996; Rajewski and Stella, 1996; Irie and Uekama, 1997; Stella and Rajewski, 1997; Thompson, 1997; Stella et al., 1999; Szente and Szejtli, 1999; Mosher and Thompson 2002; Stefánsson and Loftsson, 2003; Rao and Stella, 2003; Challa et al., 2005) and are a major subject of a biennial International Cyclodextrin Symposium. In addition, the Society of Cyclodextrins, Japan, symposium is held every year. The reader is directed to these publications and proceedings from these symposia for both a historical perspective and a more comprehensive discussion than can be accomplished here.

CDs are useful pharmaceutically because they can interact with drug molecules to form inclusion complexes. This formation of an inclusion complex, often a 1:1 interaction, is usually described by Equation 1 or as in Scheme 1, although higher order complexes are often seen or postulated.

$${\text{Drug}}_{\text{free}}+{\text{CD}}_{\text{free}}\stackrel{\text{K}}{\rightleftarrows }\text{Drug}/{\text{CD}}_{\text{complex}}$$(1)

In forming the complex, the physicochemical and biological properties of the drug can be altered to effect an advantage.

The renal toxicity of α-CD and β-CD after parenteral administration (Frank et al., 1976) as well as problems with a number of modified CDs have been well documented (Irie and Uekama, 1997; Thompson, 1997; Gould and Scott, 2005). The safety of some CDs will be discussed later in this article. Two modified CDs have been identified as having good inclusion complexation and maximal in vivo safety for various biomedical uses. They are SBE-β-CD, or Captisol, a polyanionic variably substituted sulfobutyl ether of β-CD, and HP-β-CD, a modified CD commercially developed by Janssen. SBE-β-CD and HP-β-CD have undergone extensive safety studies and are currently used in at least six Food and Drug Administration (FDA)-approved products (four for SBE-β-CD and two for HP-β-CD). Recently, there has been interest in two additional derivatives, sugammadex or Org-25969 (Adam et al., 2002), in which the 6-hydroxy groups (eight in all) on γ-CD have been replaced by carboxythio acetate ether linkages, and hydroxybutenyl-β-CD (Buchanan et al., 2002, Buchanan et al., 2006). There has also been a renewed interest in γ-CD itself and modified γ-CDs including HP-γ-CD.

What can Cyclodextrins Do and What can’t they Do?

In forming inclusion complexes, major changes in drug candidate properties, including enhanced solubility, physical and chemical stability, and other physicochemical properties, have been well documented (Szejtli, 1988; Duchéne, 1987; Frömming and Szejtli, 1994; Uekama et al., 1994; Albers and Muller, 1995; Loftsson, 1995; Loftsson and Brewster, 1996; Rajewski and Stella, 1996; Irie and Uekama, 1997; Stella and Rajewski, 1997; Thompson, 1997; Stella et al., 1999; Szente and Szejtli, 1999; Mosher and Thompson 2002; Stefánsson and Loftsson, 2003; Rao and Stella, 2003; Challa et al., 2005). These changes have then resulted in better biological performance, and thus, in the use of CDs in various commercially successful pharmaceutical products.

By far the greatest advantage has been in the area of enhanced solubility of problematic drugs. However, it is important to recognize that aqueous solubility enhancement by CDs is different, mechanistically speaking, than solubility enhanced by the use of cosolvents and surfactants. In the case of cosolvent cocktails, water solubility is enhanced by changes in the bulk properties of the solution. For example, solvents such as dimethylsufo-xide (DMSO), dimethylacetamide (DMA), polyethyelene glycol, propylene glycol, ethanol, and so on can enhance solubility when mixed with water but in a very nonlinear fashion. That is, a drug may be very soluble in these pure solvents, but once the drug–solvent mixture is diluted with an aqueous solvent, the poorly water-soluble drug often precipitates en masse. This nonlinearity is not a problem with the use of CDs, especially those that form 1:1 complexes. Significant solubility enhancement is often seen only with compositions that are >30% to 40% organic. Surfactant and surfactant–cosolvent mixtures tend to be able to solubilize low crystalline, highly lipid-soluble drugs (not all poorly water-soluble drugs are lipid soluble) but suffer from significant toxicity after parenteral administration in some animal species. A good example is the difficulties seen with the use of Cremophor EL in the paclitaxel formulation. Lipid emulsions with the drug dissolved in the lipid phase have seen some use.

The solubility of a drug in the presence of a CD that forms a 1:1 inclusion complex can be described by Equation 2,

$${\text{S}}_{\text{total}}={\text{S}}_0+\frac{{\text{KS}}_0[{\text{CD}}_{\text{total}}]}{{\text{KS}}_0+1}$$(2)

where S _total refers to the total drug solubility, S ₀ refers to intrinsic solubility of the drug in the absence of the CD, [CD _total] is the total cyclodextrin added to the solution, and K is the 1:1 binding constant defined by Equation 3.

$$\text{K}=\frac{[\text{Drug}/{\text{CD}}_{\text{complex}}]}{[{\text{Drug}}_{\text{free}}][{\text{CD}}_{\text{free}}]}$$(3)

From Equation 2, it can be seen that there should be a linear increase in the solubility of the drug with increasing CD concentration with a slope equal to KS ₀/(KS ₀ + 1). Therefore, one can immediately see that the ability to increase water solubility is a function of S ₀ and K and the total CD concentration used.

Consider the following example for discussion purposes: if one assumes a CD total of 0.1 M—for example, a 22% solution for SBE-β-CD—and 1:1 complex formation, one can solubilize only up to 0.1 M drug or 30 mg/mL for a drug having a molecular weight of 300. Therefore, one cannot expect to solubilize 50 mg/mL of this drug with this CD concentration. However, getting even a solubility of 30 mg/mL assumes an infinitely large K value. In practice, binding constants of 10 M^–1 to 1 × 10³ M^–1 are not uncommon, values of 1 × 10⁴ M^–1 are seen occasionally, and values >1 × 10⁵ M^–1 are very rare. Therefore, practical solubility increases in the presence of 0.1 M CDs of 10-fold to 1,000-fold are quite often seen, and higher increases are occasionally observed.

A 1,000-fold increase in solubility may sound impressive until one begins to examine the aqueous solubility of many new drug candidates. It is not uncommon these days to be asked to formulate drug candidates with solubilities (S ₀) of <1 μg/mL. One might say “a thousand times zero is still zero,” or practically speaking, one may only be able to solubilize such drugs to >1 mg/mL using CDs for those drugs with very high binding constants or when the S ₀ for the drug is > 1 μg/mL. A neutral drug with a solubility of 0.1 μg/mL may have its solubility increased to 0.1 mg/mL but is unlikely to reach 10 mg/mL.

Rao and Stella (2003) came up with a dimensionless utility number that allows one to play “what if” games when considering whether CDs could be used to effect greater solubility; that is, various scenarios are presented that would allow one to make educated decisions on the use of CDs to solve a specific solubility problem. Their article and others clearly acknowledge that the assumption of 1:1 complex formation is not always valid, and one may achieve very high levels of increased solubility when one combines multiple variables to increase solubility. For example, for basic amine drugs, increased solubility can be achieved by decreasing the pH of the solution to below the pKa of the amine group. However, because of toxicity considerations, one might want to minimize the pH decrease by adding CDs such as SBE-β-CD and HP-β-CD to achieve the same solubility at a pH value closer to physiologic pH (Tinwalla et al., 1993; Johnson et al., 1994; Okimoto et al., 1996; Li et al., 1998; Okimoto et al., 1999; Zia et al., 2001; He and Yalkowsky, 2006). It is interesting to note that the FDA-approved products Vfend (intravenous [IV] voriconazole, SBE-β-CD), Geodon (intramuscular [IM] ziprazidone, SBE-β-CD), Abilify (IM aripiprazole, SBE-β-CD), Cerenia (subcutaneous [SC] maropitant, veterinary use, SBE-β-CD), and Sporanox (IV and liquid itraconazole, HP-β-CD) all contain basic drugs in which pH adjustment and CDs—SBE-β-CD or HP-β-CD—are used to achieve the desired formulation solubility and good in vivo performance. For the two IM-dosed drugs and one SC-dosed drug, pH adjustment at the site of injection could lead to precipitation of the poorly water-soluble free base forms of the drugs at the site. This is prevented by the presence of CDs. The reader is directed to the earlier referenced reviews and references therein for a more extensive discussion on the use of CDs to effect enhanced solubility.

The use of CDs to increase physical and chemical stability of drugs in solution and in other dosage forms has been well documented in the literature (Loftsson and Brewster, 1996). Generally, CDs can enhance the stability or catalyze the degradation of some drug molecules, although there are more examples of the latter than the former. If the reader is interested in a few specific examples, the following articles are suggested: Narisawa and Stella, 1998; Ma et al., 1999; Jarho et al., 2000; Ma et al., 2000; Tonnesen et al., 2002.

CDs have also been used to overcome other issues, such as taste masking of bitter materials, lowering drug volatility, controlled release of drugs for oral and parenteral delivery, and so on. The reader is directed to the earlier referenced reviews for specific examples.

If a Drug is Strongly Bound to a Cyclodextrin, How is it Released?

This question has been addressed in various articles and reviews. Equation 1 and Scheme 1 often adequately describe the interaction of drugs with CDs. The equilibrium constant (Equation 3) can also be defined as the ratio of the forward and reverse rate constants for this process. For both weakly and strongly bound drugs, we have attempted to measure the kinetics of this process by various fast kinetics techniques; in no case have we been able to estimate either the “on” or “off” rate—that is, the processes are faster than could be measured. Others have noted values for these kinetics using ultra fast kinetics techniques (Cramer et al., 1967; Rohrbach and Wojcik, 1981; Hersey and Robinson, 1984; Hersey et al., 1986) and have found that in most cases, rates approach those of diffusion-controlled limits. In only one case (unpublished work) did we see some evidence of slow kinetics when we observed nuclear magnetic resonance (NMR) line broadening for a drug with a >1 × 10⁵ M^–1 binding constant. However, on further analysis, it was concluded that what we were looking at was actually relatively slow interconversion between two different bound forms of the complex and not broadening caused by slow on/off rates.

If the on/off rates are fast, one can treat the reaction shown in Scheme 1 as just a rapid equilibrium. Therefore, the major driving force for drug release from an inclusion complex is simple dilution (Uekama et al., 1994; Rajewski and Stella, 1996; Stella and Rajewski, 1997; Stella et al., 1999). Consider the parenteral injection of a small volume to a human subject. The volume of distribution (Vd) for both SBE-β-CD and HP-β-CD is said to be that of extracellular water, about 20% of total body weight or 14 L for a 70-kg patient. A 5-mL injection of a drug–CD solution would result in a 1:2,800 dilution. For most drugs, this would be sufficient to completely dissociate the drug from the CD. Only when the K value for the complex is >1 × 10⁵ M^–1 are any issues raised. Other mechanisms can contribute to rapid drug release after administration (Stella et al., 1999). In addition to dilution, they include

drug–protein binding causing a decrease in free drug concentration and thus driving a left shift of the equilibrium described by Scheme 1,

competitive displacement by endogenous and exogenous molecules, and

drug uptake into tissues not accessible by the complex or free CD.

The role of competitive displacement raised some interesting issues with the FDA after studies with sugammadex first surfaced. Sugammadex, a modified γ-CD, was designed to specifically bind to rocuronium, a neuromuscular blocker, and reverse its blockade. The binding constant of rocuronium to sugammadex is said to be about 10⁷ M^–1 (Adam et al., 2002), a value far outside of the normal values of binding of nearly all known drugs to any CD, except for another unusual case—a series of anti-malarials (Perry et al., 2006; Charman et al., 2006) binding to SBE-β-CD. As noted earlier, when binding constants are greater than 1 × 10⁵ M^–1, it is more probable that interactions may persist, even on dilution. Therefore, a question has been raised: whereas a CD might be used to effectively deliver a specific agent, could that same CD, on systemic administration, bind to a second coadministered drug and alter its pharmacodynamics and pharmacokinetics? This has not been observed in any known cases, and one could reasonably argue that this is a nonissue except in the very rare case of a drug’s having an extraordinary interaction with any of the known approvable CDs. Even here, the effect should be fleeting, because the CDs are rapidly renally excreted.

One may not want the complex to dissociate too completely in some cases! For example, Nagase et al. (2002) showed that SBE-β-CD decreased the hemolysis caused by an experimental drug when given IV compared to the non-CD formulation. Three of the commercial products using SBE-β-CD approved by the FDA involve IM or SC administration of acidic solutions of weakly basic drugs. IM or SC administration of acidic solutions of weakly basic drugs often results in precipitation of the free-base form of the drugs, which in turn results in erratic drug release from the injection site. Such injections can also cause local tissue damage at the site of injection as the pH rapidly adjusts to 7.4. CDs such as SBE-β-CD can keep the drug in solution and allow for complete and rapid release from the site of injection, and because the drug is significantly bound, they can also lower any local toxic response. Once the drug–CD combination is released to the systemic circulation through dilution and the other mechanisms discussed above, the complex dissociates and allows the drug to exert its activity. The lowering of local toxic effects has also been seen for some drugs administered ophthalmically (Järvinen et al., 1994; Järvinen, Järvinen, Urtii, et al., 1995; Jarho et al., 1996; Loftsson and Järvinen, 1999).

What if dilution is limited? Oral administration of concentrated, high-volume CD–drug combinations to rodents often results in decreased delivery of drug molecules. The same is seen with some ophthalmic products (Järvinen et al., 1994; Järvinen, Järvinen, Urtti, et al., 1995; Jarho et al., 1996). For example, in the case of ophthalmics, an eyedrop volume of 30 to 45 μL is diluted into only 7 to 10 μL of precorneal fluid. Therefore, minimal dilution occurs. If the CD is used to effect a solution of a poorly soluble drug, a distinct advantage still occurs (Loftsson and Järvinen, 1999), but if, for example, excess CD is used to stabilize the drug, then insufficient free drug is available for corneal permeation. A report by Rajewski and Stella (1996) and a number of papers from the Loftsson’s group have extensively discussed this issue.

CDs have often been used to enhance the solubility and oral delivery of poorly water-soluble drugs. Two good examples are a study by Liversidge and Cundy (1995) showing the enhanced availability of a poorly water-soluble drug, danazol, and a study by Järvinen, Järvinen, Schwarting, et al. (1995) showing the increased availability of cinnarizine from various CD combinations compared to some non-CD formulations. This has been especially successful when the drug–CD combination has been used in larger animal species such as dogs and man. Similar results have also been seen in rodents when small volumes are administered. However, a problem often arises when one attempts to escalate the dose of the drug for toxicology studies. In escalating the dose of drug, one also escalates the CD concentration, and the total volume administered can be as high as a couple of mL/kg. This volume can dominate the gastrointestinal tract (GIT) of the rodent, and if a high concentration of CD is used (often up to 20% to 40%), being hypertonic can result in loose stools and diarrhea (see later comments). Thus, there is minimal dilution, resulting in low free-drug concentration, and the shorter GIT transit time results in an inhibition of GIT absorption of the drug. This has also been seen under similar circumstances when very high concentrations of surfactants are used.

In summary, to create better delivery and drug release of problematic drugs from CDs, one must understand how drugs bind to CDs and recognize that CDs do not behave in the same manner as drugs added to a cosolvent to effect a solution.

Can Cyclodextrins Perturb the Pharmacokinetics Properties of Drugs and Those of Coadministered Drugs After Parenteral Administration?

This is probably the most frequently asked question we have had to address and one that has also been asked by the regulatory agencies. CDs do not cross biological membranes easily, so this question really pertains most to drugs administered by injection rather than orally. One usually uses a CD to effect better solubility and/or less irritation or damage at injection sites. Before answering the question of whether the presence of a CD can alter the pharmacokinetic (PK) and pharmacodynamic (PD) properties of a drug, one might ask the question, what are the alternative formulation strategies for poorly water-soluble drugs? The alternatives are those discussed earlier, namely, the use of cosolvents, surfactants, pH adjustment, or oil-in-water emulsions. So, when one compares the PK and PD properties of a drug from a CD formulation to those of a control formulation, one must ask, what is the proper control, the CD formulation or the alternative? For example, in comparing the PK properties of carbamazepine from a HP-β-CD formulation to a formulation using glycofural as a cosolvent, it was found that glycofural caused a significant change in the PK properties of carbamazepine by inhibiting carbamazepine’s metabolism (Löscher et al., 1995) and showed significant signs of toxicity and alteration in the pharmacologic actions of barbiturates (Yasaka et al., 1978). Surfactants such as Cremophor EL and Tween 80 can cause idiosyncratic histamine release, with patients having to be rescued through the use of antihistamines and steroids. Propylene glycol is a cardiac depressant and has shown renal toxicity (Levy et al., 1995), and the side effects of ethanol are well known. In an early unpublished study involving the use of SBE-β-CD to solubilize a potential drug candidate, an issue of its potential central nervous system (CNS) toxicity was raised after toxicity was seen with the drug candidate relative to a formulation in which the drug was solubilized by a hydro-alcoholic solvent mixture. It turned out that it was not the CD causing the CNS toxicity but the hydro-alcoholic solvent suppressing the intrinsic CNS toxicity of the drug.

As stated earlier in this article, the PK properties of a drug should be unaffected by the use of CDs, provided that the binding constant of the drug for the CD is below 1 × 10⁵ M^–1, and even this strong a binding has been shown to have only a limited effect. For example, Piel et al. (1999) studied the IV PK properties of miconazole from a commercial surfactant solution and two CD formulations that used HP-β-CD or SBE-β-CD. The binding constant of miconazole for both CDs was in the range of 1 × 10⁵ M^–1. No significant differences were seen in the PK parameters among the three formulations. Similar results were seen in the PK properties of methylprednisolone from an SBE-β-CD compared to a cosolvent mixture (Stella et al., 1995a), etomidate from the commercial “Amidate” formulation compared to an SBE-β-CD formulation (McIntosh et al., 2004), and flu-tamide from a HP-β-CD solution (Zuo et al., 2002). In the etomi-date study by McIntosh et al. (2004), although no differences were seen between the CD formulation and the Amidate formulation, very significant hemolysis was seen from the Amidate formulation, whereas none was seen with the SBE-β-CD–based formulation. Similarly, an SBE-β-CD formulation was found to completely release prednisolone from an IM injection in rabbits, whereas a cosolvent formulation gave a slower release and resulted in elevated creatinine kinase levels, presumably because of local site-of-injection irritation or damage (Stella et al., 1995b). There are many other examples, too numerous to reference here, showing that CDs, with two exceptions, do not alter the PK properties of drugs after IV administration.

The two exceptions were mentioned earlier in this article. Sugammadex has been shown to alter the PD properties of rocuronium, although there appears to be some question as to whether the CD actually changed the PK properties. For a series of antimalarials, which were found to bind to SBE-β-CD with very high binding constants, the plasma PK was affected, whereas whole-blood PK properties were not (Perry et al., 2006). The antimalarials were highly taken up into red blood cells, so differences in plasma kinetics were masked by the high concentrations in the red blood cells. In both the rocuronium and the antimalarial cases, the drugs that bind to the CDs were water soluble and did not require the CDs for delivery purposes.

For HP-β-CD or SBE-β-CD, the only route of elimination is renal excretion via glomerular filtration. Because of their fairly high clearance and small Vd values, their half-lives (t _½) are quite short. This means that immediately after IV administration, HP-β-CD or SBE-β-CD appear in the proximal tubules of the kidney. As water is reabsorbed, the CDs are concentrated. Therefore, for a short period of time, drugs that undergo glomerular filtration, that are passively well reabsorbed, and that also have a significant binding to HP-β-CD or SBE-β-CD may be retained in the urine at these early time points. Many of these drugs are excreted only to a small extent in the urine, a few percent, but this can be increased by the presence of the CDs. For example, when given IV with HP-β-CD, the renal excretion of both unchanged carbamazepine (Brewster et al., 1997) and dexamethasone (Dietzel et al., 1990) is increased. However, this is insufficient to alter their plasma PK properties, as the renal excretion only increases from about 0.75% to about 2.3% for carbamazepine. For dexamethasone, about 6% unchanged dexamethasone was renally excreted from the HP-β-CD formulation, but this was only compared to an aqueous dexamethasone phosphate formulation.

What is the Safety Record of Cyclodextrins?

Safety is a primary concern when new excipients are considered for use in pharmaceuticals. Here, we focus primarily on recent developments in the safety evaluation of pharmaceutically useful CDs, including pharmacokinetics and toxicological issues.

Pharmacokinetics:

On oral administration, only insignificant amounts of intact CDs are absorbed from the gastrointestinal tract because of their bulky and hydrophilic nature. Any absorption is by passive diffusion. The parent α-CD and β-CD are practically resistant to stomach acid and salivary and pancreatic enzyme digestion, whereas γ-CD is digested partly by amylases in the GIT. The absorbed CDs are essentially excreted in the urine without undergoing significant metabolism. The absorption of the parent CDs and their derivatives following oral administration in the rat is summarized in Table 2. The absorption of methylated derivatives of β-CD in rats has been reported as 6.3% to 9.6% for DM-β-CD and 0.5% to 11.5% for M-β-CD (Mosher and Thompson, 2002). The oral bioavailability of HP-β-CD in humans was estimated to be less than 1% following administration of HP-β-CD as an oral solution of itraconazole (de Repentigny et al., 1998; Stevens, 1999).

IV-administered CDs disappear rapidly from systemic circulation and are renally excreted intact. The steady-state volume of distribution (Vd_ss) for β-CD and most of its chemically modified derivatives in all animal species tested corresponds well with the respective extracellular fluid volume (Table 3). This indicates that systemically absorbed CDs distribute mainly in the extracellular compartments, and no deep compartments or storage pools are involved. However, methylated β-CD has a larger Vd_ss and longer t _½ compared to other CD derivatives. This may be related to its ability to interact with cellular membranes and its hemolytic activity (Thompson, 1997). The total plasma clearance for HP-β-CD and SBE-β-CD in all species tested is similar to the glomerular filtration rate of individual species (Davies and Morris, 1993), and essentially 100% of a given dose is recovered in the urine within 6 to 12 hours following IV administration (Zhou et al., 1998; CyDex, unpublished). The elimination of CDs strongly depends on renal clearance; thus, renal insufficiency could result in accumulation of HP-β-CD (Slain et al., 2001), SBE-β-CD (von Mach et al., 2006), and other CDs.

Following absorption, CDs distribute to various tissues including the kidney, urinary bladder, liver, adrenal gland, and others (Gerloczy et al., 1990; Monbaliu et al., 1990; Antlsperger and Schmid, 1996; Kubota et al., 1996; De Bie et al., 1998; Van Ommen et al., 2004). The kidney has the highest level of CDs of all tissues. The retention of M-β-CD in the kidney persists for at least 6 days following a single IV administration (10 mg/kg) at almost the same level, mainly in the renal cortex (Antlsperger and Schmid, 1996). Most CDs disappear from tissues dramatically within the first several hours of administration. For example, the level of SBE-β-CD in most tissues decreased by more than 90%; even in the kidney, SBE-β-CD was reduced by at least 60% from 0.1 hour to 1 hour following IV bolus injection of 600 mg/kg SBE-β-CD in the rat (Figure 1).

In summary, CDs have very limited bioavailability and distribute among extracellular compartments on absorption. Systemically absorbed CDs are eliminated rapidly from the body and appear in the urine unmetabolized.

Toxicological Studies, Cellular Toxicity:

The hemolytic effect of CDs has been reported in several in vitro studies; however, the toxicological implication in vivo is considered negligible. The in vitro hemolytic activity of CDs is reported in the order β-CD > α-CD > HP-β-CD (TDS 4.2) > γ-CD >> HP-γ-CD ≥ HP-α-CD in erythrocytes freshly collected from human and P388 murine leukemic cells (Leroy-Lechat et al., 1994). Irie and Uekama (1997) reported that hemolytic effects of various CD derivatives decreased in the order DM-β-CD >> TM-β-CD > β-CD > HP-β-CD (total degree of substitution, TDS, 5.8) ≥ α-CD >>> γ-CD > DHP-β-CD (TDS 5.9), whereas S-β-CD (TDS 10.7) was virtually nonhemolytic. The sulfobutylether derivatives of β-CD are less hemolytic than β-CD, but the effect of TDS is quite dramatic. The order of hemolysis to human erythrocytes follows β-CD > SBE1-β-CD >> SBE4-β-CD, whereas no hemolysis is observed with SBE7-β-CD (Mosher and Thompson, 2002). Several CDs also have been demonstrated to cause cell lysis in different types of cells, indicating that the effect is not cell-type specific (Irie and Uekama, 1997).

The hemolytic activity of CDs correlates well with their ability to solubilize cellular membrane lipids rather than their intrinsic solubility or surface activity. This is supported by the fact that a positive correlation exists between the hemolytic activity of several CDs and their capacity to solubilize cholesterol, a main component of lipid bilayers, irrespective of their quite different physicochemical properties (Irie and Uekama, 1997; Thompson, 1997).

It is reported that β-CD and M-β-CD induce caspase-dependent apoptotic cell death in human keratinocytes on depletion of membrane cholesterol, whereas α-CD, γ-CD, and HP-β-CD are not apoptotic to this type of cell (Schönfelder et al., 2006).

Toxicological Studies, General Toxicology Studies in Laboratory Animals:

Extensive reviews of early toxicological studies of CDs have been published elsewhere (Irie and Uekama, 1997; Thompson, 1997; Mosher and Thompson, 2002). An updated toxicology review on HP-β-CD was published recently (Gould and Scott, 2005). Toxicological studies of CDs and some of their derivatives have been conducted following various administration routes in various species. In addition to the above reviews, the safety of SBE-β-CD has been evaluated following oral administration, subcutaneous injections, inhalation, and ocular administration. Some studies with SBE-β-CD are listed in Table 4.

The major findings following oral treatment with various CDs or their derivatives are soft feces or diarrhea, cecal enlargement on oral administration, and renal effects subsequent to systemic absorption. Other observed effects when administered by IV, subcutaneous, intramuscular injections at very high doses or during a long period of time include vacuolation of hepatocytes and urinary bladder and foamy macrophages in the liver, lung, and lymph nodes. Vacuolation is minimal or mild in all affected tissues except for the kidney and urinary bladder. No evidence of inflammatory responses, cell death, and cell degeneration or regeneration is observed in studies reported so far, except for α-CD, β-CD, and methylated-β-CD, in which renal damage and dysfunction occur (Irie and Uekama, 1997; Thompson, 1997). Vacuolation is reversible following cessation of the CD, such as γ-CD, HP-β-CD, and SBE-β-CD treatment.

Toxicological Studies, Effects on Gastrointestinal Tract:

The prominent GIT effects of CDs when administered orally to rats and dogs—soft feces or diarrhea and cecal enlargement—are reversible. These effects are observed not only with HP-β-CD (Gould and Scott, 2005) and SBE-β-CD but also with α-CD (JECFA, 2001; Lina and Bar 2004a, 2004b), β-CD (Bellringer et al., 1995), and γ-CD (Lina and Bar 1998; Til and Bar 1998). The parent CDs all are accepted as food additives and “generally recognized as safe” (GRAS). The dog seems to be more susceptible than the rat to CD-induced diarrhea, as the effective doses are reported to be lower in dogs than in rats. These effects are reversible on withdrawal of treatment as demonstrated in studies with SBE-β-CD. All CDs but γ-CD have very low digestibility. The nature of effects of CDs on the GIT is similar to those of low digestible carbohydrates, such as lactose, and other widely used pharmaceutical excipients (Baldrick and Bamford, 1997). It is generally accepted that diarrhea and cecal enlargement observed in experimental animals treated with oral CDs represent physiologically adaptive responses to a large load of poorly digestible carbohydrates and other osmotically active nutrients, and the relevance to humans is minimal (Baldrick and Bamford 1997; Grice and Goldsmith, 2000).

Toxicological Studies, Effects on Kidney:

The kidney is the organ most affected following systemic absorption of CDs. The primary effect in the kidney is vacuolation of proximal tubular epithelium as a result of CD treatment. As manifested by a series of changes in the kidney, parenteral administration of α-CD, β-CD, or methylated-β-CD results in renal toxicity (Frank et al., 1976; Irie and Uekama, 1997; Thompson, 1997). These changes begin with an increase of apical vacuoles and the appearance of giant lysosomes. Prominent acicular microcrystals are observed in lysosomes. These changes are followed by disruption of mitochondria and other organelles, and eventually, irreversible cell injury and kidney dysfunction occurs. The effective dose of M-β-CD to induce kidney damage is even lower than that of β-CD (Antlsperger and Schmid, 1996; Irie and Uekama, 1997). The greater toxicity of methylated-β-CD to the kidney could be related to its greater ability to interact with cellular lipids and to persist in this organ for a longer time, leading to cell-membrane damage (Antlsperger and Schmid, 1996). It was proposed that the acicular crystals in kidney tubular cells observed under microscopy after administration of β-CD were likely to be complexes formed between cholesterol or its esters and the small amount of β-CD that remains in the kidney (Frijlink et al., 1991). Both β-CD and HP-β-CD form a complex with cholesterol or its esters in blood, but only β-CD forms an insoluble complex. It was proposed that complexes of cholesterol/CDs enter primary urine by glomerular filtration and are then taken up by the kidney tubular cells. The insoluble cholesterol in kidney aggregates as intracellular needle-like crystals that cause kidney damage (Frijlink et al., 1991).

Whereas the parent α-CD or β-CD induce irreversible kidney damage and dysfunction, parent γ-CD and many substituted CDs cause only vacuolation in the kidney without any manifestation of cell injury, degeneration, or alterations of renal functions. A no-observed-effect level (NOEL) was demonstrated as at least 200 mg/kg and 120 mg/kg for γ-CD was in the rat following daily IV injection for 1 month and 3 months, respectively (Donaubauer et al., 1998). The NOEL in the rat is at least 50 mg/kg for HP-β-CD receiving daily IV injections for 3 months (Thompson, 1997; Gould and Scott, 2005) and 80 mg/kg for SBE-β-CD daily IV injections for 1 month, respectively. Parenteral administration of SBE-β-CD at a single IV dose of 2,000 mg/kg or a daily dose of up to 15,000 mg/kg for 14 days produced only vacuolation of the kidney tubular cells without loss of kidney function. Electron microscopy studies showed that the vacuoles were membrane-bound and contained finely granular and some electron-dense membranous profiles, indicating lysosomal origin, but no needle-like crystals or other abnormal organelles as seen in β-CD–treated animals were observed. Unlike the results with β-CD, cell necrosis was not observed by light or electron microscopy even following high doses of γ-CD (Donaubauer et al., 1998), HP-β-CD, or SBE-β-CD.

The vacuolation observed in animals treated with γ-CD, HP-β-CD, or SBE-β-CD is reversible following withdrawal of treatment, whereas the effect of parent β-CD progressed into cellular injury and kidney dysfunction (Frank et al., 1976). The length of period for complete recovery of vacuolation may depend on the rapidity of cell turnover, severity of vacuolation, and period and dosage of treatment. Vacuolation in cells is generally regarded as a physiological response, presumably for sequestration of materials and fluids taken up by the cells (Henics and Wheatley, 1999). The reversibility of vacuolation and lack of evidence of kidney damage or functional impairment indicate that the vacuolation observed following treatment with these CDs is a physiological adaptation and has few toxicological implications.

Toxicological Studies, Reproductive and Developmental Effects:

The parent CDs and some of their derivatives have been evaluated for reproductive and developmental toxicity in the rat and/or rabbit (Table 5). These studies demonstrated that neither α-CD nor γ-CD was embryotoxic or teratogenic. Transient neonatal-growth retardation occurred with 5% β-CD; a similar but equivocal effect was also seen with 2.5%. The dietary level of 1.25% was observed to be a no-observed-adverse-effect level (NOAEL) of β-CD for developmental toxicity based on neonatal growth retardation (Barrow et al., 1995). At high doses of HP-β-CD and SBE-β-CD, maternal body weight and food consumption were decreased, suggesting maternal toxicity. No teratogenic effects at any dose levels were observed following treatment with either HP-β-CD or SBE-β-CD. Compared to the vehicle controls in the segment III studies, HP-β-CD treatment at the highest tested dose levels (5 g/kg oral and 400 mg/kg IV administration) decreased survival of pups at birth and during the early lactation period. A similar result was observed for SBE-β-CD after IV injection of 3 g/kg but not at 600 mg/kg or 100 mg/kg. These effects may be associated with nutritional deficiency as a consequence of maternal toxicity.

Toxicological Studies, Genotoxicity, and Carcinogenicity:

A standard battery of genotoxicity tests for CDs has been performed with the parent CDs, α-CD, β-CD, and γ-CD, as well as with HP-β-CD and SBE-β-CD. Negative genotoxicity has been demonstrated in these studies, indicating that none of the tested CDs are genotoxic or mutagenic.

The potential of CDs for carcinogenicity has been evaluated with β-CD and HP-β-CD administered orally in both mice and rats. No treatment-related neoplastic lesions were observed in carcinogenicity studies with β-CD administered in the diet at dose levels from 25 to 675 mg/kg/day in either species (WHO Food Additives Series 35). In the mouse study with orally administered HP-β-CD by diet at dose levels of 500, 2,000, and 5,000 mg/kg/day, no evidence of primary carcinogenic potential was demonstrated. In rats, however, an increase was observed in exocrine acinar cell neoplasia at 500, 2,000, and 5,000 mg/kg. A slight increase in neoplasms in the large intestine at 5,000 mg/kg was also observed but may be part of an adaptive hypertrophy of the large intestines after high doses of osmotically active nutrients (Janssen Research Foundation, NDA 20-966, 1999). Changes in the pancreas of rats following HP-β-CD oral administration were initially seen within 12 months as exocrine pancreatic hyperplasia, which developed to neoplasia by 24 months. It is believed that the mechanism of pancreatic carcinogenicity is secondary to an increase in cholecystokinin (CCK) as a consequence of facilitated fecal elimination of bile acids by HP-β-CD (Gould and Scott, 2005). As a mitogen, CCK stimulates hyperplasia in pancreatic acinar cells, a phenomenon seen specifically in the rat (Irie and Uekama, 1997).

Numerous studies suggest that exocrine acinar tumors of rats may be irrelevant to human risk. The histopathological type of tumors of the exocrine pancreas in rats is dramatically different from that in humans. Both spontaneous and chemical-induced tumors in the exocrine pancreas of rats occur exclusively in the acinar cells but not in the ductal parts of the pancreas (Scarpelli et al., 1984; Longnecker and Millar, 1990). In contrast to the acinar origin of pancreas tumors in the rat, exocrine pancreatic tumors are ductal in origin, and acinar tumors are exceptionally rare in humans (Solcia et al., 1997). A class of lipid-lowering drugs called fibrates, such as clofibrate and fenofibrate, are well known to induce exocrine acinar tumors in rats but not in mice. A large, multicenter, randomized double-blind clinical study with clofibrate treatment for 4 to 8 years followed by 13 years of prospective observations has not reported any increase in incidence of human pancreatic tumors throughout the study period (Report of the Committee of Principal Investigators, 1984). This class of drugs has been used clinically for almost 40 years worldwide, and no human pancreatic tumors related to these drugs have been reported. The pancreas exocrine tumors induced by orally administered HP-β-CD in the rat are histopatho-logically identical to those observed in the above studies. The exocrine tumors of the rat possess histopathological characteristics that are very distinct from those of common human pancreas tumors. In addition, cyclodextrins including HP-β-CD are neither mutagenic nor genotoxic. Therefore, HP-β-CD–induced acinar pancreatic tumors are considered rat specific and have little relevance to human carcinogenic risk.

Toxicological Studies, Human Clinical Experience:

The tolerance of orally administered γ-CD was examined in a double-blind, placebo-controlled, crossover study in 24 healthy volunteers. It was found that a single dose of 8 g γ-CD is well tolerated (Koutsou et al., 1999). Many clinical studies have been conducted with IV and oral itraconazole solution, a formulation containing HP-β-CD (1% itraconazole and 40% HP-β-CD). In clinical studies, HP-β-CD has been well tolerated when administered orally with itraconazole at doses equivalent to HP-β-CD up to 200 mg/kg, twice a day (400 mg/kg/day; Lass-Flörl et al., 2003; Stevens, 1999). Although HP-β-CD containing oral solution is blamed for diarrhea, the incidence and severity of this effect is comparable to that occurring with other treatments in most clinical studies (Stevens, 1999; Harousseau et al., 2000). Clinical studies have been conducted or are under way with SBE-β-CD-containing drugs by oral administration. The highest dose of SBE-β-CD that has been administered orally to humans is at least 12 g, and no adverse effects attributable to SBE-β-CD were reported.

Intravenous administration of HP-β-CD in healthy volunteers has been studied at a single dose up to 3 g and found to be well tolerated, without any evidence of renal toxicity by analysis of urinary excretion of N-acetyl-β-glucoamidase (NAG), creatinine, and γ-glutamyl transpeptidase (Seiler et al., 1990). Administration of HP-β-CD at a dose of 470 mg/kg/day by IV infusion in a patient with severe hypervitaminosis A did not produce evidence of renal or liver damage, although this dose occasionally caused agitation and pulmonary edema in rabbits and dogs, respectively (Carpenter et al., 1995). Normally, IV infusion of itracona-zole contains an HP-β-CD–equivalent dose of 8 g twice a day (16 g/day) for the first 2 days, followed by 8 g/day for 14 days (Slain et al., 2001).

Intravenous voriconazole (Vfend IV) contains 16% SBE-β-CD (1 mg voriconazole to 16 mg SBE-β-CD). In a single-blind, placebo-controlled study with healthy volunteers, SBE-β-CD was administered by IV infusion for 1 hour at a dose of 96 mg/kg, every 12 hours for the first two doses, followed by 48 mg/kg every 12 hours for 2 to 9 days (Purkins et al., 2003). All subjects tolerated the treatment. Routine clinical laboratory tests, hema-tology, and clinical chemistry were conducted on blood samples at screening and before the morning dose on days 1, 6, 7, 10, and 12 and at 48 hours or 2 weeks after the final dose. No significant abnormalities were observed throughout the study. No clinically significant changes were observed in blood pressure, pulse rate, or ECG data. Renal-function parameters were measured for urinary excretion of NAG, creatinine, total protein, microalbu-min, and β2-microglobulin. All of these parameters were within normal range and remained at similar levels through the period of study. In clinical trials, SBE-β-CD has been administered at 96 mg/kg every 12 hours for the first day followed by 80 mg/kg every 12 hours for days 2 to 7. No treatment-related adverse events were observed in these studies (Purkins et al., 2002). Limited case reports have indicated that several patients have been exposed to SBE-β-CD safely by IV infusion at doses up to 144 mg/kg every 12 hours, or 96 mg/kg every 8 hours, for more than a week (Muldrew et al., 2005). The safety following high doses of SBE-β-CD IV administration in humans is continually being investigated.

What is the Regulatory Status of Cds?

There is still skepticism within the regulatory agencies concerning the use of CDs, although this seems to be a bigger issue with the FDA than agencies in many other countries. The use of CDs in products to treat acute and life-threatening diseases will continue to meet less resistance than their use in chronic treatments. The simple fact that there are now numerous approved products containing CDs bodes well for the future. As more products containing CDs undergo rigorous evaluation and their strengths and weaknesses are understood, agencies will become more comfortable with them and lower the barriers to their use.

Conclusions

CDs offer an additional tool for the formulator to overcome some of the formulation and delivery limitations of problematic drugs. As with any new tool, they have both strengths and weaknesses. Their major strengths are the specific nature of how they interact with drug molecules and their ability to deliver safely a number of intractable drug molecules. The specific nature of their interaction is also a weakness in that only molecules with the right size, geometry, and intrinsic solubility properties benefit from their use. Whereas both α-CD and β-CD and a number of alkylated CDs are known to be renally toxic and disruptive of biological membranes, γ-CD and some of its derivatives, as well as HP-β-CD and SBE-β-CD, appear to be much safer. As more products using CDs are approved or undergo evaluation by regulatory agencies for use in food and pharmaceuticals, concerns about their safety will be clarified.