Rational formulation design


Correspondence: Majella E. Lane, Department of Pharmaceutics, UCL School of Pharmacy, 29-39 Brunswick Square, London WC1N 1AX, U.K. Tel.: +44 207 753 5821; fax: +44 870 1659275; e-mail: majella.lane@btinternet.com


To be efficacious and to satisfy the requirements for claim substantiation, a cosmetic formulation must achieve effective targeting of an active in the skin. Although the basic principles governing the skin permeation and disposition of molecules have been known for many years, attention has been far less focused on the role of the vehicle, particularly at cosmetically relevant doses. In this article, we discuss the necessity to understand the fate of the formulation components as well as the active once applied onto skin. Recent data confirm that the residence time of the formulation constituents can have a profound impact on the fate of the active. Approaches to identify the ideal vehicle for skin delivery are considered critically, specifically the recent work on ‘formulating for efficacy’ (FFE) by the late Johann Wiechers. Essentially, FFE aims to match the active with the optimal vehicle for skin delivery based on matching polarity/solubility values of the trinity of skin, active and vehicle. The emerging importance of techniques that provide insight to how the vehicle distributes in and on skin is highlighted.


Pour être efficace et pour satisfaire aux exigences de justification des revendications, une formulation cosmétique doit assurer un ciblage efficace des ‘actifs’ dans la peau. Bien que les principes fondamentaux qui régissent la diffusion et la disposition des molécules dans la peau soient connus depuis de nombreuses années, beaucoup moins d'attention a été portée sur le rôle du véhicule, en particulier à des doses pertinentes en cosmétique. Dans cet article nous discutons de la nécessité de comprendre le sort des composants de la formulation, ainsi que celui de l’ ‘actif’ une fois appliqué sur la peau. Des données récentes confirment que le temps de séjour cutané des constituants de la formulation peut avoir un impact profond sur le sort de la ‘actif’. Les approches visant à identifier le véhicule idéal pour livrer le ou les ‘actifs’ à la peau sont discutées et analysées, en particulier le travail récent de feu Johann Wiechers sur le concept ‘formulating for efficacy’ (FFE). Essentiellement, FFE vise à faire correspondre l’ ‘actif’ avec le véhicule pour un apport optimal à la peau, en se fondant sur la complémentarité des valeurs de polarité/solubilité du trio ‘peau, actif et véhicule’. L'importance croissante des techniques qui permettent de mieux comprendre la façon dont le véhicule distribue ses composants dans et sur la peau est surlignée ici.


Formulations have developed over the past years in terms of their aesthetic appeal, but have they improved significantly in terms of their ability to deliver an active? In addition, do we fully understand the processes by which an active is delivered and to where? These are important questions that need to be addressed and may be answered, in part, by an understanding of the history of formulation design. It is interesting to look back to the middle ages when the ‘original witches flying ointment’ was described [1]. It was an herbal recipe producing altered states of consciousness. The ointment consisted of belladonna or mandrake, poplar leaves and soot and to hold it all together fat or clove oil. A 50th-century source reads, ‘They anoint a staff and ride on it…or anoint themselves under the arms and in other hairy places’. The belladonna alkaloids were extracted from the leaves and then treated with lime. The alkaline conditions would have created the free base and so the better penetration of the base rather than the salt must have been understood. It was dispersed in fat or clove oil, that is, an occlusive formulation. The producers must have also appreciated the role of an occlusive excipient. It was also applied to certain regions of the body, and so there must have also been a general appreciation that the permeability of skin varies from site to site [1].

Thermodynamic activity and skin penetration

The first real attempts to rationalize skin penetration and formulation effects in terms of the underlying physical chemistry were described by Higuchi in 1960 [2]. In this article, Fick's first law was discussed and hence the role of diffusion, partition and thermodynamic activity. The normal form of Fick's first law that is used to describe the permeation of actives through the skin is

display math(1)

where J is the rate of permeation, D is the diffusion coefficient of the active in the skin, K is the partition coefficient (skin/donor), cdonor is the concentration in the donor phase (formulation), creceptor is the concentration in the receptor phase (blood) and h is the thickness of the skin. In essence, the concentration in the receptor is usually very small compared with the donor and can be regarded as close to zero. In practical terms for an in vitro experiment, this means that the receptor phase concentration is very small and ‘sink conditions’ prevail. The concentration in the donor is large and usually does not deplete significantly; hence, the conditions are said to be ‘infinite dose’.

Fick's first law is most simply interpreted when the formulation has no impact on the barrier properties of the skin and it is ‘non-interactive’. Under this circumstance, if the formulation is changed, it is likely that the partition coefficient is altered. However, a direct corollary of this is that the solubility in the formulation will be reduced. This was demonstrated in some classic experiments conducted by Katz and Poulsen [3] in which they showed that an optimum formulation was obtained when the product Kcdonor was maximum. In simple terms, they compared penetration from water/propylene glycol gels. When the water concentration is high, the steroid ‘likes’ partitioning into the skin, but the solubility of the steroid in the water is low. Conversely, when the propylene glycol concentration is high, the steroid ‘does not like’ portioning into the skin as much, but the solubility is higher than in pure water. Depending on the nature of the steroid, the optimum formulation will be an appropriate water – propylene glycol mix when the product of the partition and solubility is at a maximum.

The other important physicochemical property described in the Higuchi equation is the thermodynamic activity. Hadgraft et al. [4] tested the effects of thermodynamic activity by examining the in vivo response to methyl nicotinate. Water/glycerol mixtures containing methyl nicotinate at different concentrations but at the same thermodynamic activity gave similar rates of permeation confirming the importance of activity rather than concentration. Coldman et al. [5] used an interesting approach to enhance penetration of a steroid. They used a mixture of a volatile and non-volatile solvent. As the volatile solvent evaporated on the skin surface, it left the steroid above its solubility limit in the non-volatile (residual) phase. The concept of producing formulations that are above their saturated concentration limit has been developed further, for example, by Davis and Hadgraft [6] and bioequivalent steroid formulations produced in which the concentration can be reduced by a factor of 50 [7]. Driving the drug into and across the stratum corneum using a high activity state is an obvious advantage in delivery to the skin from topical formulations.

Solubility and partition coefficient

For transdermal delivery, it is often stated that the best permeation is for compounds that have a logP around 2, this is because there is a balance between their ability to partition and their inherent solubility. It is also the reason why free acids or free bases are used in transdermal delivery rather than the equivalent salts. However, for topical delivery and in the application of cosmetics, the dose applied is not ‘infinite’. To put this in perspective, a cream or lotion is usually applied to the skin at a rate of about 2 mg cm−2. This is equivalent to a thickness of 20 μm, that is, a similar volume equivalent to the stratum corneum. This has implications for the role of solubility, and partition as an ‘infinite dose’ is not applied. Steady-state diffusion is not obtained, and to gain some perspective on what occurs, solutions to Fick's second law are needed.

Figure 1 shows a simulation of the concentration profiles in the formulation and across the stratum corneum for actives that have partition coefficients of 10 and 1000.

Figure 1.

A simulation of the concentration profiles in an applied formulation at the skin for an active that has a partition coefficient of 10 (A) and 1000 (B). The development of the profiles with time is shown. Significant depletion in the formulation is seen, and steady-state diffusion across the stratum corneum is never reached.

If an infinite dose had been used after a period of time, the concentration in the outer layers of the skin would be 10 times and 1000 times the concentration in the applied formulations. The simulations show that this is not the case, and for the instance when K = 10, only just over twice as much is seen in the skin; when K = 1000, it is slightly higher at approximately 3. Perhaps, it is necessary to reconsider what physicochemical properties dominate when clinical doses are applied to the skin. Clearly, there is not a simple linear relationship between permeation and partition coefficient. This can also be seen in a simulation in which the amount permeating through the skin is followed as a function of time. Because a finite dose is applied, a sigmoidal profile is obtained as shown in Fig. 2.

Figure 2.

The amount of active permeating as a function of time. Two actives are shown with a partition coefficient of 10 and 1000. The lines are coincident.

Even though there is a 100-fold difference in the partition coefficient, there is no apparent difference in the amount of drug permeated. Simulations of this nature show that 100% of what is applied to the skin permeates through. However, this is far from what is seen in practice. Feldmann and Maibach [8] demonstrated that only about 2% of an applied solution of a steroid penetrated the skin in vivo. The important question that arises from this is: where does the rest go to? If this can be addressed, it should be possible to produce very much more effective formulations.

Formulating for efficacy (FFE)

The major route of penetration through the skin involves diffusion through the intercellular spaces. When a formulation is spread thinly on the skin surface, only a small fraction will be in direct contact with the intercellular spaces, and if lateral diffusion across the corneocytes is not possible, active deposited here will not penetrate. The experiment conducted by Feldmann and Maibach [8] used an acetone solution. The acetone will rapidly penetrate but even more will evaporate from the skin surface. It is likely that the very poor bioavailability seen in this publication is a result of solid steroid forming on the corneocytes from which further transport is unlikely. This means that the nature in which the active is applied is critical, and the more it allows the active to reach the intercellular channels, the more effective it will be. It also indicates that most vehicles that are applied to the skin are not ‘non-interactive’. They will penetrate into the skin and change the properties of the lipids in the skin.

Wiechers et al. [9] introduced the concept of ‘formulating for efficacy’ (FFE). They wrote ‘From a rather theoretical approach based on the polarity of the active ingredient, the stratum corneum and the oil phase, the Relative Polarity Index was established that enables the selection of a suitable emollient for ensuring skin penetration of the active ingredient.’ In essence, the formulation needs to have the correct polarity to contain enough of the active to have a biological effect, but it also must create an environment in which the active is at a high thermodynamic activity. Furthermore, it is recognized that the vehicle components will interact with the skin lipids. The polarity of the vehicle should aid uptake of the active into the skin lipids by creating a vehicle-skin lipid mix that is favourable for dissolving the active.

A model penetrant, oxybutynin, was used to determine differences between simple solvent systems [10]. The oxybutynin was formulated such that it was applied to the skin at a finite dose at 1, 2 and 5 times saturated in either propylene glycol or octyl salicylate. The amount of oxybutynin that penetrated into and through the skin was measured for the two solvent systems. Figure 3 shows the amount of oxybutynin extracted from the skin after 24 h.

Figure 3.

The amount of oxybutynin extracted from the skin after the application of a finite dose at 1, 2 and 5 times saturated from octyl salicylate or propylene glycol (adapted from reference [10]).

The amount that gets into the stratum corneum is in direct proportion to the degree of saturation and is greater for octyl salicylate as the solvent. The oxybutynin is about 1.5 times more soluble in octyl salicylate than in propylene glycol which would account for the higher levels in the skin. However, when the amount permeated is considered, a different picture emerges. The penetration profiles are shown in Fig. 4.

Figure 4.

The amount of oxybutynin permeated through the skin after the application of a finite dose at 1, 2 and 5 times saturated from octyl salicylate or propylene glycol (adapted from reference [10]).

The amount permeated at 24 h for octyl salicylate behaves as expected; there is a linear relationship between the amount and the degree of saturation. However, for propylene glycol, the amount is invariant with degree of saturation. This can be explained in terms of the properties of the two different solvents. Octyl salicylate is very lipophilic and permeates slowly through the stratum corneum; its residence time in the skin is long compared with propylene glycol, which can be seen in Fig. 5.

Figure 5.

The amount of octyl salicylate and propylene glycol permeated through the skin (adapted from reference [10]).

The majority of the propylene glycol has permeated through the skin in 8 h; the dissolved oxybutynin will no longer be held in solution and will crystallize and therefore not be free to permeate as rapidly. It is possible to estimate the amount of oxybutynin that does not permeate, and it is about 80% of that applied. Therefore, the choice of vehicle/formulation excipient can be critical in controlling the amount of active that will permeate. The polarity [9] is important but so is the amount of solvent that remains in the skin for a sufficient time to allow the active to permeate.

The role of the solvent can also be exemplified by some unpublished data from our laboratory that focuses on ibuprofen. The amount permeated over 24 h from a variety of solvents was determined together with the solubility. Finite dose saturated solutions were used. If these experiments had been conducted at infinite dose and if the solvents do not interact with the skin, the same amount of ibuprofen should be delivered. The ibuprofen would be at the same thermodynamic activity. However, the data shown in Fig. 6 show that the amounts delivered are significantly different and are related to the solubility of the ibuprofen in the solvents. Again, under finite dose conditions, it seems that the solubility is a major determinant rather than the partition coefficient.

Figure 6.

The relation between the amount of ibuprofen permeated at 24 h through human skin (in vitro) and the solubility of ibuprofen in the applied solvent (R. Vieira, unpublished data).

As Wiechers et al. [9] postulated in FFE, the polarity is important. One way of estimating the polarity of the active, vehicle and skin is to use the Hansen Solubility Parameter (HSP). In this concept, ‘like dissolves like’ and the closer the HSP values, the more soluble. This has been explained in a recent publication by Abbott [11]. The concept for ibuprofen can be tested because the HSP values for both ibuprofen and the solvents can be estimated. Figure 7 shows the correlation between the experimental solubilities and the HSP difference (ibuprofen – solvent).

Figure 7.

The relationship between ibuprofen experimental solubility (unpublished data) and the HSP difference between ibuprofen and solvent.

There does not appear to be a relationship between the solubility and the HSP gap, and the concept of using the HSP in topical formulation design needs further refinement. The concepts of Wiechers et al. are correct, but fine tuning needs to be performed to create an algorithm that can be used in the optimization of formulations.

Selection of salt form of active versus free acid or free base

In transdermal delivery, and when infinite doses are selected, it is common practice to select the free acid or free base rather than the salt form of an active. This can be seen in the transdermal systems for scopolamine, clonidine, fentanyl, etc., where the free base has been formulated. However, where solubility dominates, it is possible to deliver either the ionized or the unionized form. This can be exemplified by the work of Hirata [12]. A complex mixture of Transcutol, propylene glycol, isopropyl myristate and propylene glycol laurate was formulated such that the solubility of carbenoxolone and its sodium salt were similar. The solvents were also selected as it was known that they would intercalate into the skin lipids. Permeation across human skin was measured, and the profiles are shown in Fig. 8. The permeation of the sodium salt was as good as the free acid because of the nature of the vehicle.

Figure 8.

Permeation of carbenoxolone and carbenoxolone sodium though human skin in vitro. Data adapted from reference [12]).

Real-time monitoring of vehicle effects in vivo

The profound effect of formulation on concentration of active in the skin can be seen in the data generated by Mohammed et al. [13]. In this study, 5% solutions of niacinamide in various solvents were applied to the human forearm, and the niacinamide concentration as a function of depth into the skin was determined using Laser Confocal Raman (River Diagnostics). The solvents selected were dimethyl isosorbide (DMI), miglyol (MG), mineral oil (MO), propylene glycol (PG), propylene glycol monolaurate (PGML) and N-methyl pyrrolidone (NMP). The data are shown in Fig. 9.

Figure 9.

Concentration profiles of niacinamide in vivo (human) across the stratum corneum from various vehicles (data adapted from reference [13]).

Another element of this work was to determine the in vitro flux from the same vehicles. It is then possible to establish any in vitroin vivo correlations. Figure 10 shows that there is a good in vitroin vivo correlation.

Figure 10.

The relationship between the in vitro flux and the amount of niacinamide measured at a depth of 4 mm in vivo. Data adapted from reference [13]).


The above data suggest a number of ways of optimizing formulations the roots of which have been described in FFE by Wiechers et al. [9]. A typical formulation will contain volatile and non-volatile components. Often the volatile component will be water or a low-molecular-weight alcohol. This will evaporate from the skin surface leaving a ‘residual phase’.

The components of the residual phase need to be chosen such that:

  1. They are capable of solubilizing an appropriate amount of active at as high a thermodynamic activity as possible. They may even be supersaturated with respect to the active. However, if they contain a high mole fraction of the active, the activity state of the solvent may be reduced as exemplified by Santos et al. [10]. Appropriate anti-nucleant polymers (e.g. hydroxypropyl methyl cellulose) may be added to prevent crystallization if necessary.
  2. Lowering of the solvent activity may be important if the solvent intercalates into the skin lipids and improves the solubility of the active in the skin. The solvent/solvents should be selected such that they can alter the polarity of the skin lipids in the direction of the active.
  3. The solvents should be selected such that they remain in the skin for a sufficient time to allow complete permeation of the active. If their residence time is too short, they can ‘strand' the active as it crystallizes in the skin lipids.
  4. If the residual phase remains on the skin surface, they should promote the lateral diffusion of the active to the intercellular spaces otherwise the active will crystallize on the corneocyte surface and become unavailable.

At the present time, we conclude that it is not possible to design an algorithm for selecting the most appropriate solvent or solvent choice. Some indication may be provided by solubility parameter, but even selection on this criterion does not automatically provide a series of solvents in which the active is most soluble. There may be the possibility of refining this approach in the future, but the algorithm will still have to incorporate a measure of the relative permeation rate of the active and the solvent through the skin. In the immediate future, it would be beneficial to have data that show the effects of simple solvents on solubility of actives in the skin. Also, there is a paucity of information on the rate of diffusion of excipients through the skin. It is also important to conduct in vitro skin permeation experiments with finite doses of the formulation. Infinite dose experiments may give incorrect or misleading results. With the advent of sensitive biophysical techniques such as confocal raman spectroscopy, it should be possible to monitor the permeation, in vivo, of the active and the excipient through the skin, and this potential has already been explore [14]. These advances will enable us to make much more rational formulation designs in the future.


The help and assistance of Dr. Jonathan Crowther with the collation of the Raman data is gratefully acknowledged.