Initial recognitions of the power of the skin barrier came from chemical warfares used in World War I. Prior to World War I, there was little known about percutaneous penetration. The research of the United States Army, the American University Experiment Station Laboratories, and Western Reserve University, laid the groundwork of our understanding of the effect of foreign chemicals on skin (Jiang and Maibach, 2018). Their work included our foundational knowledge of percutaneous absorption and the effect of various vehicles used on skin.

Nearly 50 years later, Tregear visually described the apparatus used for in vitro experiments (Figure 2). (Tregear, 1966) Tregear’s work, alongside Burch et al., noted that recently dead skin does not affect the permeability of the skin, a major development in our ability to use excised skin in diffusion cells. This was shown experimentally for both water and tri-n-butyl phosphate (Burch and Winsor, 1946; Bronaugh and Stewart, 1984). Thus, excised skin can be used for measuring permeability. Of note, antibiotics are frequently used on excised skin to decrease the chance of bacterial degradation.

The detailed nature of skin is still not fully understood but advances are made frequently in the field of dermatology. In order to understand the depth of percutaneous absorption, it is necessary to have a foundational understanding of its compartments. Skin is composed of three primary layers—epidermis, dermis, and hypodermis. Epidermis (30–80 μm thick) has several sublayers, each with their own distinctive properties and roles—the stratum corneum (8–20 μm thick), stratum lucidum, stratum granulosum, stratum spinosum, and stratum basale. Stratum corneum will be touched on heavily in this paper, as it is recognized as a primary barrier to absorption. Following the epidermis is the dermal-epidermal junction, which contributes greatly to the dynamic nature and integrity of skin. It is made of four parts—basal keratinocytes and hemidesmosomes, lamina lucida, lamina densa, and sub-basal lamina. The higher layers include proteins such as laminin and nidogen. The lamina dense is composed of type IV collagen and heparin sulfate proteoglycan and the sub-basal lamina densa is composed of anchoring fibrils (Zou and Maibach, 2018). Following the dermal-epidermal junction is the dermis, measuring 1–2 mm. The final layer is the hypodermis, measuring 0.1 to a few centimeters (Agache et al., 2017).

When evaluating percutaneous absorption, the topic of “mass balance” must be considered. Mass balance is the physical concept of accounting for mass composition throughout a system. When a topical drug is applied, we must understand where all components are—are they still on the surface, in the stratum corneum, in the bloodstream, or have they been excreted? Typically, in in vivo techniques, a compound is topically applied, either in a volatile or more complex vehicle. Urinary excretion and/or blood is measured over the next several days. Although helpful, important factors are often not considered—namely, its metabolites may not be evaluated (Bucks et al., 1988).

Bucks et al. modified this technique for more definitive results of hydrocortisone, estradiol, testosterone, and progesterone, by evaluating the effects of both occlusion and protection of skin. In the occlusive experiment, a physical intact chamber is placed directly on the skin, which will often increase percutaneous penetration. In the protective experiment, a plastic Hilltop^® chamber was used, which is a chamber not placed directly on the skin, but rather placed above the skin allowing for a level of protection coupled with ventilation. Evaluating for mass balance often is improved by radiochemical tagging, covering the application site to account for sloughing off of particles, washing the skin surface after the dosing period, and sometimes tape stripping the stratum corneum for more complete results (Bucks et al., 1988). Results showed that estradiol, testosterone, and progesterone had much higher absorption rates (measured through higher concentration excreted in urine) under occluded conditions as compared to protected conditions. Tables 1 and 2 compare the distribution of all compounds in both occluded and protected conditions. Hydrocortisone, the least lipophilic drug, was measured at much higher concentrations in the stratum corneum on day 7 compared to the other steroids. This suggests substantivity—specific, strong interactions occurring in the stratum corneum. Other chemicals, similar to hydrocortisone, remain in the stratum corneum for an extended period of time, which may contribute to the toxicity of some drugs when applied topically as opposed to orally.

Occlusion significantly increases stratum corneum hydration. The complexity of the skin is seen with this profound finding—if the skin was a simple organ, the hydration from occlusion would equally affect penetration coefficients of all drugs as lipophilicity increases. That is not what we see. Figure 4 shows that progesterone, the most lipophilic compound, has a lower dose absorbed than testosterone (Bucks et al., 1988). Bucks et al. hypothesized that hydrating the stratum corneum decreases the partition coefficient of the compound between the stratum corneum and the epidermis. This, in turn, increases the kinetics of transferring the compound from one layer to the next, and thus, is seen more with more lipophilic compounds. But penetration does not progressively increase with increased lipophilicity. The decrease with progesterone is due to a change in the rate-limiting step of the transfer of compounds from the stratum corneum to the epidermis. Thus, there is a limit to increased absorption with more lipophilic compounds—at one point, we see that a more lipophilic compound actually has a slower relative absorption rate (Bucks et al., 1988). The reason for this stark contrast in percutaneous absorption relative to lipophilicity is not fully understood. It has been observed that a certain amount of the aqueous and lipid solubility aids penetration, but extreme lipid solubility decreases it (Dragicevic and Maibach, 2018).

The static cell is an easier method of measuring permeation through skin, although it is scientifically preferred to perform experiments via a flow-through system. Ajjarapu et al. notes that the only two factors that may significantly affect the results seen between static and flow-through systems are time differences in sample collection and the chemicals themselves (Ajjarapu and Maibach, 2022). Namely, more water-soluble chemicals may show similar results with both systems, whereas lipophilic chemicals may not.

During World War II, Tregear used static cells to perform in vitro experiments on skin. Years later, he laid the foundational work for the flow-through system (Figure 2). (Tregear, 1966) Around the same time, Marzulli’s ground-breaking work set the precedent for the future of flow through in vitro skin research (Marzulli, 1962). Bronaugh compared the efficacy of a flow-through system as compared to a static cell. Table 5 details comparison of absorption based on his experiments on water, cortisone, and benzoic acid (Bronaugh and Stewart, 1985). His research showed that a flow rate of at least 5 mL/h is needed for a receptor of volume 0.4 mL to produce accurate results. Moreover, once that minimum flow rate is achieved, increasing the flow rate does not increase absorption (Figure 5). (Bronaugh and Stewart, 1985)

The volume of the receptor is essential to the accuracy of results. The reason for this is that in order to remove the compound that has entered the receptor during a period of time, the volume pumped into the cell must be much larger than the volume of the receptor. Thus, the receptor must have a small volume to compensate for exiting the cell is feasible. Moreover, Bronaugh’s work confirmed that a minimum flow rate must be present in order for proper evaluation to occur. With the minimum flow rate of 5 mL/h for a receptor of volume 0.4 mL, adequate mixing occurs and the absorbed compound is removed from the receptor rapidly, helping decrease the chance for presence of artifact.

However, in cases of using compounds that are poorly water soluble, a different result is seen. Crutcher evaluated the absorption of testosterone and testosterone propionate (two lipophilic compounds) and their results showed that as flow rate increased, absorption also increased. Possible reasons for this could be that they had not reached the minimum rate for a large area of skin they were using or that the compounds they were using were lipophilic, which tend to have increased absorption with increased flow rate. Crutcher pointed out that the increased concentration of compound in the perfusate in their in vitro chamber limits the rate of percutaneous penetration (Crutcher and Maibach, 1969). In human skin, concentration in the dermal capillaries is inconsequential, thus, it is important for in vitro chambers to replicate physiological conditions by adapting a continuous flow of the perfusate, similar to human skin.

In the quest for finding methods that accurately predict absorption of transdermal compounds, many ethical and financial roadblocks exist. The use of excised human skin may be difficult due to limited availability and cost. Excised animal skin is a method that has been introduced due to its lower cost, however, ethical, and reproducibility concerns may limit their use. Differences in race, age, and sex between animal skins can affect extrapolated results (Neupane et al., 2020).

Due to the various obstacles with using excised human and animal skin as testing models, researchers have used skin surrogates to test compounds. Skin surrogates are broadly separated into three subgroups: artificial human skin models, parallel artificial membrane permeability assays (PAMPAs), and artificial membranes based on polymer/lipid models. The major benefit of using skin surrogates, as compared to animal and human skin, is the consistency in surrogates, leaving minimal room for variations amongst different skin samples.

First, artificial skin models began with a simple design: a de-epidermized dermis was used as the foundation for normal human keratinocytes to grow on (Ponec, 1992). This design was further developed—now, artificial skin samples are recognized in two broad categories: the reconstructed human epidermis model (RHE) and the full thickness model (also known as living skin equivalents, LSE). They differ in that the RHE model allows keratinocytes to grow on the epidermis while the LSE model grows them on both the epidermis and dermis.

Schäfer-Korting et al. evaluated absorption of testosterone across different modalities and showed that RHE models had the highest permeation coefficient, followed by pig skin, bovine udder skin, and human epidermis (Schäfer-Korting et al., 2008). Although RHE had a higher permeation coefficient compared to animal and human skin models, the permeation ranking was similar amongst all samples. Taken together, RHE models can be used as viable alternative methods to human and animal skin for the study of compounds applied as aqueous solutions in in vitro studies. Relevance to human skin in vivo remains to be documented.

There has not been a skin model that perfectly resembles human skin. Namely, a commonly used reconstructed skin model has shown to have similar lipid composition of its stratum corneum to human skin, however the higher permeation coefficient is due to its lack of a comparable barrier. Although the lipid composition is the same to human skin, cholesterol was seen as droplets rather than evenly distributed throughout (Tfayli et al., 2014).

The second category of skin surrogates are known as Skin-PAMPAs (skin parallel artificial membrane permeability assay). The design of PAMPAs involves a donor and a receptor compartment, separated by a liquid membrane. The donor compartment includes the compound to be tested. Niacinamide, when applied to both a Franz diffusion cell (ex vivo) and to a PAMPA model exhibited a linear correlation between the two models (Zhang et al., 2019). This places PAMPAs as a probable candidate for use as an artificial skin surrogate compared to animal and human skin models. Although more studies must be conducted, PAMPAs may show promise for their efficiency in both time and cost.

The final category of skin surrogates includes artificial skin models based on simple polymers and lipids. An example of this is Strat-M™, made of layers of polyester sulfone. The limitation of this is that it is difficult to control permeation through each layer and measure permeability at each layer, as we do with human skin. Studies have reproduced permeability in Strat-M™ models, although as a general rule, hydrophilic compounds tend to show higher permeability than lipophilic compounds (Neupane et al., 2020). Clinical relevance to human in vivo studies has not been documented.

Over recent decades, much work has been done to optimize skin models to be used in vitro to mimic human skin. Current models have an epidermis made of robust models of stratum corneum, stratum granulosum, stratum spinosum, and stratum basale (Bouwstra et al., 2021). The most difficult barrier that has been faced are the differences in skin homeostasis and barrier function. Bouwstra et al. postulates that lipid composition, lipid properties, and cornified envelope composition factor into the complexity of the skin barrier that is difficult to reproduce in vitro (Bouwstra et al., 2021). In 1988, Ponec et al. showed that exposing a de-epidermized dermis to air significantly improved its lipid composition (Ponec et al., 1988). Mak et al. demonstrated that de-epidermized dermis coupled with 1,25-dihydroxy vitamin D3 helps the barrier resemble that of human skin (Mak et al., 1991). Even still, this model lacked covalently bound lipid lamellae that properly extruded into the regions between cells, which is a foundational part of the skin barrier (Bouwstra et al., 1995; Kennedy et al., 1996). Following this, Boyce and Williams showed that adding palmitic acid, essential fatty acids, and antioxidants increased the synthesis of lipids and improved extrusion (Boyce and Williams, 1993). Ponec et al. discovered that the addition of vitamin C to an air-exposed de-epidermized dermis model and epidermis significantly improved the lipid composition and extrusion process (Ponec et al., 1997). Indeed, as was hoped, the number of triglycerides decreased and the most hydrophilic ceramides increased, improving in lipid composition. This finding was confirmed in permeation studies using corticosterone as well (Pasonen-Seppänen, Suhonen, Kirjavainen, Miettinen, Urtti, Tammi, et al., 2001).

In addition to the advances that have been made in improving skin surrogates, in order to extrapolate data and translate to compounds used on humans, it is important to evaluate regional variation in human skin. Regional variation is the concept that percutaneous absorption in humans varies in different areas of the body. Consequently, systemic side effects will be different depending on how much of the compound is being absorbed. Feldmann began this quest for understanding the concept of regional variation. Using hydrocortisone, they observed that the scrotum is by far the highest absorbing site tested, followed by regions on the face. The lowest tested absorption site was the forearm (Feldmann and Maibach, 1967). Two decades later, Guy built a table to be referenced by clinicians based off of hydrocortisone and pesticide data for five sites—genitals, arms, legs, trunk, and head (Bronaugh et al., 1985). Table 3 shows these results.

Regional variation has also been tested in animals, which may affect inferences made in humans. Bronaugh showed that rat abdominal skin was more permeable to water, urea, and cortisone than the back skin. This is probably due to the thin skin of the abdomen compared to the back (Bronaugh and Maibach, 1985). As a further example, the rhesus macaque shows regional variation between the forehead and forearm for compounds such as fenitrothion and aminocarb (Moody and Franklin, 1987). Taken together, both animal and human studies documents that variation amongst anatomical sites change the effect of compounds applied to the human skin.

The metabolism of topical compounds can begin in the skin. Phase I compounds, such as cytochrome P450, and phase II compounds, such as N-acetyltransferase, have been seen in skin, although their enzymatic activity, as expected, is significantly lower than their activity in liver (Kazem et al., 2019). Biotransformation effects in the skin can be seen with inactive chemicals becoming active or vice versa, or even a lipophilic compound becoming more hydrophilic. These changes can affect the penetration of the compound. As we saw in Buck et al.’s paper on the penetration of four corticosteroids, very lipophilic compounds like progesterone do not absorb well through skin (Bucks et al., 1988). Conversely, with metabolic changes, a very lipophilic compound like butylparaben metabolizes in the skin to p-hydroxybenzoic acid, which is more hydrophilic than its pro-drug, increasing its absorption capability. We can also determine the effects of metabolism with the effect of UV rays on compounds—many compounds (psoralens, sulfonamides, tetracyclines) have been identified and known to increase photosensitivity when exposed to UV rays (Law et al., 2020). Once compounds have been absorbed through the skin and entered systemic circulation, the more water-soluble drugs typically can be excreted without further metabolic changes (Law et al., 2020). There is still much to be learned about excretion profiles, metabolism, and clearance rates of topical drugs.

Several decades ago, little was known about the power of the stratum corneum. Although it was recognized as a main barrier of the skin, its sheer complexity had not yet been elucidated. Today, we know that stratum corneum serves as a primary barrier, limiting compounds from fully penetrating the skin while simultaneously limiting water from exiting the body through skin. This is in part due to the acid mantle, primarily made of phospholipase A2 (Yosipovitch and Maibach, 1996; Zhai et al., 2008). Indeed, after topical application, many compounds are considered “locked into” the stratum corneum (Bronaugh and Maibach, 2005). Furthermore, when different compounds were evaluated for differences in diffusion through stratum corneum, they all had very similar diffusion coefficients when anatomical site was controlled for (Rougier et al., 1990).

It is important to understand the proteins involved in the stratum corneum, which are collectively contributing to this strong barrier. Stratum corneum is made of corneocytes, which are keratin threads weaved together in an organized fashion, holding a large amount of water amongst them. There are about 12–16 layers of these corneocytes, each layer with an average thickness of 1 μm. Variations are seen based on anatomical site and UV exposure (Nemes, Steinert, 1999). Crystalline lamellar lipid regions envelop around corneocytes. These lamellar bodies are generated by keratinocytes in the underlying stratum granulosum and stratum spinosum. Maturation pushes these lamellar bodies higher into the stratum corneum, where they are simultaneously degraded, causing release of ceramides and free fatty acids. The release of these substances results in their fusion, which forms the aforementioned lamellar lipid bilayer (Norlén, 2006; Proksch et al., 2008). In addition to corneocytes being surrounded by this lamellar lipid bilayer, specific proteins, namely, loricrin and involucrin, form a tight envelope around them. Attached to them are ceramides that hold onto water. These proteins and lipids simply lay the groundwork of the skin’s barrier. Other proteins, including Natural Moisturizing Factor and filaggrin, amongst many others, further trap water and contribute to the meshwork of the stratum corneum (Hui et al., 2013). Note that permeation through the stratum corneum not only depends on the proteins that make up its core, but also depends on the compound’s solubility, pH, concentration, and partitioning between the stratum corneum and the vehicle.

The further research that is performed, the further we learn about the sheer complexity of the human skin, much of which has yet to be clarified. Law et al.’s recent review is considered up-to-date on the factors affecting percutaneous absorption of a compound in humans. Table 4 details twenty pertinent factors that have been studied up to this point. Law et al. summarizes all the relevant factors, a few of which will be detailed here. First, the compound being applied to the skin has its own chemical properties that affect its absorption. Smaller molecules are more permeable, lipophilic drugs are more easily absorbed due to the lipophilic nature of the stratum corneum (although highly lipophilic drugs are not as easily absorbed), and basic drugs have an easier time penetrating the skin as well. The reason for this is that the skin’s proteins are slightly basic and unionized forms of drugs pass through membranes. Basic drugs are more likely to be un-ionized, thus, it follows that basic drugs are more readily absorbed. It is also important to discuss the partition coefficient of a drug, which is the relative concentration ratio of the compound between two immiscible solvents. It can help us infer the solubility differences of the compound between the two phases. It is important to focus on the octanol:water partition coefficient (also known as oil to water coefficient) and the formulation:skin partition coefficient (also known as the cream or ointment relative to the stratum corneum). Hydrophobic compounds have a high octanol:water partition coefficient and are more likely to be found near lipid bilayers. On the other hand, hydrophilic compounds have a low octanol:water partition coefficient and are more likely to be found in aqueous areas. The formulation:skin partition coefficient is based on the actual formulation of the drug (Law et al., 2020).

The vehicle used in the compound plays a role in the latters absorption. More occlusive vehicles aid in absorption, whereas water-based vehicles do not aid all types of drug absorption, but do help increase absorption of hydrophilic drugs. An example of an occlusive vehicle is betamethasone dipropionate 0.1% ointment being used as opposed to a cream—this change to an oil-based vehicle helps increase permeability (Law et al., 2020).

Application frequencies, sites of application, and population variability are also important factors to consider. Application frequency is an area of minimal research, however, with corticosteroids a high concentration single dose may have a greater impact on absorption than multiple doses of a lower concentration (Wester et al., 1977). Other factors that are vital to consider include skin hydration and skin health/integrity.

As previously discussed, skin compartments and protein binding play a role in permeation of a compound. An important term to know here is “substantivity”, which can be seen as the power of the drug to remain in the stratum corneum. When undergoing topical formulation, it is important to ask ourselves that during activities such as sweating, exercise or water exposure—does the compound remain effective? And for how long? Thus, the higher the substantivity, the greater chance that the drug has a prolonged effect both topically and systemically (Law et al., 2020).

This leads to our discussion of systemic absorption. There are three major factors to consider here—substantivity, binding affinity to skin substances, and binding activity to blood components. Substantivity, as previously discussed, is the concept of if the drug remains in the stratum corneum and/or other skin compartments, potentially increasing its duration of action. Binding affinity to skin substances includes proteins, lipids, hair follicles, and sweat glands. As chemicals descend down the skin, and enter the dermis, they encounter vascular blood supply. Affinity for blood substances such as albumins and plasma proteins is important to take into account here. Moreover, some drugs have vasodilatory properties, increasing their systemic effect (Law et al., 2020).

Other environmental factors such as exfoliation, washing, massaging, and skin to skin contact are included in the complexity of percutaneous absorption. Furthermore, metabolic factors, discussed previously, play an important role in the effect of the drug. Aging skin can also pose its own complexities, which may affect the absorption of a drug. The only in vivo study that evaluated this was done in 1989 by Roskos et al., who evaluated the effects of testosterone, estradiol, hydrocortisone, benzoic acid, acetylsalicylic acid, and caffeine through urinary excretion profiles. Their results showed a significantly lower permeation of hydrocortisone, benzoic acid, acetylsalicylic acid, and caffeine in older subjects (greater than 65 years old, compared to the 18–40 years old group). Testosterone and estradiol had similar systemic uptake in both age groups. They concluded that aging skin has a relatively lower lipid content which made substances with lower lipid solubility (e.g., hydrocortisone) more susceptible to permeation changes (Roskos et al., 1989).

Finally, in addition to the twenty factors listed in Table 4, it is important to note the effect of drugs on damaged skin, particularly in the realm of occupational skin hazards. Gattu’s review showed different effects of compounds on diseased skin—some had an increase in penetration through damaged skin, while others had an equal or decreased penetration. Their damaged skin results showed that hydrophilic compounds generally show higher penetration than lipophilic compounds. This can be explained by patterns already observed in healthy skin. As a general rule, lipophilic compounds are more absorbed through the skin’s lipid bilayer, while hydrophilic compounds have a more difficult time being absorbed. Thus, the absorption of hydrophilic compounds may be more reliant on damage to epidermis (Gattu and Maibach, 2011). Others postulate that severity of disease plays an important role in percutaneous absorption. There is still much to be learned in the area of damaged skin (Turpeinen, 1988).

We are aware of the difficulty of obtaining human skin. Human and animal studies also introduce ethical concerns. The goal is to extrapolate data from in vitro and ex vivo studies for the approval of compounds to be used on human skin for a variety of dermatologic concerns. Lehman et al. compared several in vitro and in vivo studies in order to evaluate their similarity. Figure 3 graphically depicts their results. They split their data sets into all data (Figure 3A) and harmonized data (Figure 3B), with harmonized data being more trustworthy as several confounding variables were accounted for such as anatomical site, drug concentration, vehicle used, and experiment length, amongst others. Ratios were created between in vitro and in vivo studies, referred to as the IVIV ratio, which would ideally be a value close to or equal to 1.0. Amongst all 92 data sets, the mean IVIV ratio was 1.6. There were anomalies that had close to a 20-fold difference between in vitro and in vivo studies, but 85% of data sets had a difference of less than 3-fold. Once data was harmonized, which excluded 81 of the datasets, the IVIV ratio was 0.96 with less than a 2-fold difference between all in vitro vs. in vivo studies. Conclusively, using 11 datasets that were harmonized and matched to minimize confounding factors, we see an almost perfect 1:1 ratio between evidence obtained from in vitro to in vivo studies (Lehman et al., 2011). Although prospective studies must be done, this is an encouraging start.