Development and evaluation of madecassoside-loaded microemulsion for enhanced topical drug delivery

Abstract

Madecassoside, a major triterpenoid glycoside from Centella asiatica, has potential value for dermatological applications; however, its poor stratum corneum permeability is due to high hydrophilicity and large molecular size. This work focused on designing a microemulsion carrier to enhance the topical delivery of madecassoside. Formulation regions were first screened using pseudo-ternary phase diagrams, followed by preparation of madecassoside-loaded systems with different oil phases, surfactant HLB values, cosurfactant compositions, and drug loadings. The resulting formulations were characterized for droplet size, PDI, viscosity, rat-skin permeation, skin retention, CAM irritation response, and physical stability. All prepared systems were clear and homogeneous, with droplet sizes between 28.00 ± 0.44 and 60.10 ± 0.61 nm and PDI values not exceeding 0.27 ± 0.02. Further optimization identified M10, containing IPM, Brij 30/Brij 35, Transcutol HP, and 2% madecassoside, as the most suitable formulation. M10 produced a high 24 h cumulative permeation amount of 406.28 ± 180.73 μg/cm², corresponding to an approximately 43.2-fold increase compared with the 2% madecassoside control solution (9.41 ± 2.08 μg/cm²). In the HET-CAM assay, M10 showed an irritation score lower than the 1% SLS reference group. The formulation also remained clear after centrifugation, freeze–thaw cycling, and 30 days of storage, with no marked changes in droplet size, PDI, or viscosity. These results suggest that microemulsions are a promising strategy for enhancing topical madecassoside delivery.

Graphical abstract

1. Introduction

Centella asiatica is a medicinal plant widely used across China and Southeast Asia and is well known for its rich content of bioactive constituents. The major bioactive constituents are triterpenoid compounds, including asiaticoside, asiatic acid, madecassoside, and madecassic acid (Bylka et al., 2014). Among these, madecassoside has attracted considerable attention due to its diverse pharmacological activities, including anti-inflammatory (Du et al., 2014), wound-healing (Hou et al., 2016), neuroprotective (Luo et al., 2014), and anti-arthritic effects (Dou et al., 2019), as well as potential anticancer activity (Li et al., 2016b). In addition, the recent study has shown that madecassoside can suppress pro-inflammatory cytokines, including IL-23 and IL-17, regulate keratinocyte proliferation, and promote collagen synthesis, highlighting its therapeutic potential for inflammatory skin disorders such as psoriasis (OuYang et al., 2016). However, because of its high hydrophilicity and relatively large molecular size (975 g/mol), madecassoside shows limited ability to permeate through the stratum corneum, which restricts its topical application. Therefore, advanced delivery systems that can enhance madecassoside permeation are essential to improve its therapeutic potential.

 Microemulsions have been widely explored as promising carriers for topical drug delivery, primarily because of their thermodynamic stability, ease of preparation, and nanoscale droplet size, which is typically below or around 100 nm. These systems consist of an oil phase, surfactant, cosurfactant, and an aqueous phase (Danielsson et al., 1981). Microemulsions can reduce interfacial tension and increase interfacial area, thereby facilitating drug transport. In addition, the presence of surfactants and cosurfactants may alter or fluidize the lipid organization of the stratum corneum, contributing to enhanced permeation (Kogan et al., 2006; Williams et al., 2004). Previous studies have demonstrated that microemulsion systems can improve the topical delivery of poorly permeable compounds (Fang et al., 2020; Peltola et al., 2003; Szumała et al., 2022). Furthermore, these systems can be optimized to balance permeation enhancement and skin tolerability, making them promising candidates for topical applications. Compared with microemulsions, nanoemulsions are kinetically stable systems and commonly require high-energy processing to produce nanoscale droplets (Jaiswal et al., 2015). Liposomes and transfersomes are vesicular carriers that can also enhance topical drug delivery, but their stability and performance may be influenced by vesicle leakage, aggregation, phospholipid oxidation, and processing conditions (Dhiman et al., 2022; Opatha et al., 2020). Therefore, because of its thermodynamic stability and relatively simple preparation, microemulsion was selected in the present study to investigate its potential to improve the skin permeation of madecassoside and to systematically evaluate the effects of oil phase, surfactant HLB, cosurfactant composition, and drug loading.

 In this study, a madecassoside-loaded microemulsion was developed and optimized for topical delivery. The formulation was designed based on pseudo-ternary phase diagram construction and systematically evaluated in terms of physicochemical properties, in vitro skin permeation, irritation potential, and physical and thermodynamic stability. The aim was to identify an optimized formulation with enhanced skin permeation while maintaining acceptable skin tolerability and physicochemical stability, thereby providing a potential platform for the topical delivery of madecassoside.

2. Materials and methods

2.1. Materials

Madecassoside was obtained from TopRhyme (New Taipei, Taiwan). Peceol was supplied by Eligin Corporation (Taipei, Taiwan), and isopropyl myristate (IPM) was purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Brij 30 and Brij 35 were acquired from Acros (Belgium), while Cremophor EL was obtained from Sigma-Aldrich (NY, USA). Transcutol HP was provided by Gattefossé (Nanterre, France), and 1,5-pentanediol was sourced from Alfa Aesar (Massachusetts, USA). Azelaic acid was also obtained from Acros (Belgium). Unless otherwise specified, the chemicals and solvents employed in this study were of analytical grade.

2.2. Pseudo-ternary phase diagram construction

Pseudo-ternary phase diagrams were constructed to identify the composition ranges capable of forming microemulsion systems. The phase diagrams were established using a water titration method (Fang et al., 2025; von Corswant et al., 1997). Briefly, the oil phase and surfactant were combined at weight ratios ranging from 1:9 to 9:1. The cosurfactant was pre-mixed with deionized water to form the aqueous phase, which was then gradually titrated into the oil–surfactant mixtures under continuous vortex mixing to ensure uniform dispersion. After each addition, the resulting systems were visually evaluated to distinguish between transparent and homogeneous systems and those exhibiting turbidity or phase separation. The titration process was continued until the system changed from transparent to turbid or phase-separated, and the corresponding compositions were recorded. These data were subsequently used to construct pseudo-ternary phase diagrams using SigmaPlot software, version 12.5.

2.3. Microemulsion preparation

The compositions of the formulations are presented in Table 1. The oil phase (Peceol and IPM), surfactants (Brij 30 and Brij 35), cosurfactants (Transcutol HP and 1,5-pentanediol), and deionized water were combined at predetermined ratios and mixed until transparent and homogeneous mixtures were formed. For drug incorporation, madecassoside (1% or 2%, w/w) was added to the blank formulations. The mixtures were briefly vortex-mixed and then placed on a horizontal shaker overnight to allow complete dissolution and equilibration. The resulting formulations were visually inspected for turbidity or phase separation and stored in sealed glass containers at room temperature prior to further analysis.

Table 1.

Formulation compositions of madecassoside-loaded systems.

Code	Oil	Oil (%)	Surfactant	Surfactant (%)	HLB	Cosurfactant	Cosurfactant (%)	Madecassoside (%)
M1	Peceol	5	Brij 30	15	9.5	Transcutol HP	32	1
M2	Peceol	50	Brij 30	20	9.5	Transcutol HP	24	1
M3	IPM	5	Brij 30	15	9.5	Transcutol HP	32	1
M4	IPM	5	Brij 30/35	15	10	Transcutol HP	32	1
M5	IPM	5	Brij 30/35	15	11	Transcutol HP	32	1
M6	IPM	5	Brij 30/35	15	12	Transcutol HP	32	1
M7	IPM	5	Brij 30/35	10	10	Transcutol HP	34	1
M8	IPM	5	Brij 30/35	10	10	Transcutol HP	17	1
						1,5-pentanediol	17
M9	IPM	5	Brij 30/35	10	10	Transcutol HP	17	2
						1,5-pentanediol	17
M10	IPM	5	Brij 30/35	10	10	Transcutol HP	34	2

2.4. In vitro skin permeation and drug deposition study

In vitro skin permeation and drug deposition studies were performed using the Franz diffusion cell (Fang et al., 2025; Franz, 1975; Hsieh et al., 2021; OECD, 2004). The experimental protocol was approved by the Institutional Animal Care and Use Committee of Kaohsiung Medical University (Approval No. 110074). Male Sprague–Dawley rats were humanely euthanized, and the abdominal hair was carefully removed without damaging the skin. Full-thickness abdominal skin was subsequently excised, and the underlying subcutaneous fat and connective tissue were carefully trimmed using surgical scissors. The prepared skin was mounted between the donor and receptor compartments with the stratum corneum facing the donor side. The effective diffusion area was 3.46 cm². The receptor compartment was filled with 20 mL of phosphate buffer (pH 7.4, Na₂HPO₄–NaH₂PO₄) to maintain sink conditions. The system was maintained at 37 ± 0.5 °C with continuous stirring at 600 rpm. An aliquot (1 mL) of the test formulation or control solution was applied to the donor compartment. At predetermined intervals (0.5, 1, 2, 3, 4, 6, 8, 12, and 24 h), 0.5 mL of receptor medium was withdrawn and immediately replaced with an equal volume of fresh medium. The amount of madecassoside in the receptor medium was quantified by HPLC as described in Section 2.5.

At the end of the permeation study (24 h), the skin samples were removed, gently rinsed with deionized water to remove residual formulation, and cut into small pieces. The deposited drug was extracted with methanol under shaking for 24 h at room temperature. The extracts were centrifuged at 9520 ×g for 10 min, and the supernatant was analyzed by HPLC to determine the amount of drug retained in the skin.

2.5. Quantification of madecassoside

Madecassoside was quantified using an HPLC system (Hitachi, Tokyo, Japan) equipped with an L-7100 pump, L-5210 autosampler, and L-4000H UV detector. The chromatographic conditions were adapted from a previously reported method (Li et al., 2016a). The mobile phase consisted of phosphoric acid aqueous solution (pH 3.0) and acetonitrile (72:28, v/v), at a flow rate of 1.0 mL/min. Detection was performed at 210 nm with an injection volume of 20 μL. Azelaic acid (200 μg/mL) was used as an internal standard. The method was validated over a concentration range of 5–250 μg/mL, exhibiting good linearity (R² = 0.9991). The relative error (%RE) did not exceed 10%, and the relative standard deviation (%RSD) was below 4%. The limit of detection and limit of quantification were 0.34 μg/mL and 1.04 μg/mL, respectively.

2.6. Physicochemical characterization

The droplet size distribution and polydispersity index (PDI) of the formulations were analyzed using a particle size analyzer (Otsuka, ELSZ-2000, Japan). Samples were transferred into standard quartz cuvettes and measured under appropriate scattering conditions. Each formulation was analyzed in triplicate, and the results are expressed as mean ± SD. The viscosity of the microemulsion formulations was determined using a cone-and-plate viscometer (Brookfield, Model LVDV-II, USA) (Hung et al., 2021). Approximately 1 mL of each sample was loaded onto the measuring plate and equilibrated at 37 °C for 3 min prior to analysis. Viscosity readings were recorded at a rotational speed of 100 rpm after 30 s of measurement.

2.7. Irritation study

The irritation potential of the formulations was evaluated using the Hen's Egg Test–Chorioallantoic Membrane (HET-CAM) assay (Luepke, 1985; Spielmann, 1995). Fertilized chicken eggs were incubated at 37 ± 0.5 °C with 60–70% relative humidity for 11 days. The eggshells were carefully opened to expose the chorioallantoic membrane (CAM) without damaging the underlying vasculature. Normal saline (0.9% NaCl) was used as the negative control, while 0.1 N NaOH and 1% sodium lauryl sulfate (SLS) served as positive controls (Budai et al., 2021). An aliquot of 300 μL of the test formulation was applied directly onto the CAM surface. Each sample was tested in triplicate (n = 3). The membrane was observed for 5 min, and the onset times (in seconds) of hemorrhage (H), vasoconstriction (V), and coagulation (C) were recorded. The irritation score (IS) was calculated using the following equation (de Araujo Lowndes Viera et al., 2022):

$$\mathit{IS}=\frac{\left(301-h\right)\times 5}{300}+\frac{\left(301-v\right)\times 7}{300}+\frac{\left(301-c\right)\times 9}{300}$$

where h, v, and c represent the time (s) of the first appearance of hemorrhage, vasoconstriction, and coagulation, respectively. The irritation potential was classified as follows: non-irritant (IS ≤0.9), slight irritant (0.9 < IS ≤4.9), moderate irritant (4.9 < IS ≤8.9), and severe irritant (IS >8.9). All results were expressed as mean ± standard deviation (SD).

2.8. Stability study

The thermodynamic stability of the developed microemulsions was evaluated by centrifugation and freeze–thaw cycling tests (Park et al., 2021; Vu et al., 2021). The formulations were centrifuged at 11,000 ×g for 5 min and visually inspected for phase separation, creaming, or cracking. For the freeze–thaw test, the samples were subjected to three consecutive cycles between −20 °C and 25 °C, with a duration of 48 h at each temperature. After each cycle, the formulations were examined for any signs of physical instability. In addition, the mean droplet size, polydispersity index (PDI), and viscosity were measured before and after the stability tests to evaluate potential changes in physicochemical properties. The storage stability of the formulations was further evaluated by storing samples in glass bottles at 25 ± 2 °C with 60 ± 5% RH and 40 ± 2 °C with 75 ± 5% RH for 30 days. Samples were withdrawn at predetermined time intervals for visual inspection, droplet size analysis, and viscosity.

2.9. Statistical analysis

Data are presented as mean ± SD from three independent measurements unless otherwise stated. Differences among groups were evaluated by one-way ANOVA with Tukey's post hoc test using GraphPad Prism 11 (GraphPad Software, Inc., CA, USA). A value of p < 0.05 was considered statistically significant.

3. Results and discussion

3.1. Pseudo-Ternary phase diagram and formulation selection

The selection of appropriate oil, surfactant, and cosurfactant systems is critical for microemulsion formation, as phase behavior is highly dependent on interactions among formulation components. In particular, the required hydrophilic–lipophilic balance (HLB) of the surfactant system relative to the selected oil phase can significantly influence the microemulsion-forming region (Syed et al., 2014; Tsai et al., 2023). Therefore, pseudo-ternary phase diagrams were constructed to identify the compositional ranges capable of forming clear and homogeneous microemulsion systems. As shown in Fig. 1, the size of the microemulsion-forming region varied depending on the oil phase and surfactant used, indicating that the phase behavior of the systems was strongly affected by formulation composition. Among the tested systems, the Brij 30-containing systems shown in Fig. 1B and Fig. 1D exhibited broader microemulsion regions than the corresponding Cremophor EL-containing systems shown in Fig. 1A and Fig. 1C. This result suggests that Brij 30 provided a more suitable surfactant environment for microemulsion formation under the present formulation conditions. The larger microemulsion regions also indicate a wider compositional window, which is advantageous for subsequent formulation optimization. The effect of oil phase was also observed from the phase diagrams. When the same surfactant system was used, the microemulsion region differed between IPM- and Peceol-based systems, suggesting that oil phase selection affected the compatibility of the oil/surfactant/cosurfactant system. However, both oils were able to form clear and homogeneous microemulsion regions under the tested conditions. Since the size of the microemulsion region does not necessarily predict skin permeation performance, both Peceol- and IPM-based formulations were selected for the initial formulation screening to further evaluate the effects of oil type and oil proportion on madecassoside delivery.

Fig. 1.

Pseudo-ternary phase diagrams of different oil/surfactant/cosurfactant compositions.

3.2. Physicochemical characterization

The droplet size and polydispersity index (PDI) of the prepared formulations are summarized in Table 2. All formulations appeared transparent and homogeneous, without visible phase separation, indicating successful microemulsion formation within the selected compositional ranges. The mean droplet size ranged from 28.00 ± 0.44 to 60.10 ± 0.61 nm with PDI less than 0.27 ± 0.02, indicating nanoscale dispersion with relatively uniform characteristics. Variations in droplet size were observed among the tested formulations, depending on the composition of the oil, surfactant, and cosurfactant systems. These differences may be associated with changes in interfacial properties and the packing efficiency of surfactant molecules at the oil–water interface. In particular, formulations containing higher proportions of surfactant (M4, M5, M6) tended to exhibit smaller droplet sizes, which may be related to improved interfacial stabilization and enhanced microemulsion formation efficiency (Callender et al., 2017; Chen et al., 2012). Interestingly, increasing the madecassoside loading from 1% to 2% did not lead to a noticeable increase in droplet size in M7–M10. Overall, the developed formulations exhibited nanoscale droplet sizes and low PDI values, supporting their suitability for topical delivery applications. Since smaller droplet sizes provide a larger interfacial area and closer contact with the skin surface, these physicochemical characteristics may contribute to enhanced skin permeation in subsequent permeation studies (Lopes, 2014).

Table 2.

Droplet size and polydispersity index (PDI) of the prepared microemulsion formulations (M1–M10).

Code	Size (nm) Mean ± SD	PDI Mean ± SD
M1	58.17 ± 0.51	0.15 ± 0.01
M2	55.63 ± 0.57	0.27 ± 0.02
M3	60.10 ± 0.61	0.07 ± 0.02
M4	38.67 ± 0.49	0.17 ± 0.02
M5	28.00 ± 0.44	0.26 ± 0.01
M6	30.50 ± 0.87	0.25 ± 0.01
M7	58.30 ± 0.79	0.12 ± 0.01
M8	43.67 ± 0.38	0.22 ± 0.00
M9	45.43 ± 0.15	0.25 ± 0.00
M10	51.33 ± 0.42	0.17 ± 0.02

3.3. In vitro skin permeation

The in vitro skin permeation profiles and skin deposition amounts of madecassoside-loaded microemulsions are shown in Fig. 2. Based on the pseudo-ternary phase diagram results, Brij 30 was selected as the surfactant for the initial formulation screening, and the effects of oil type and oil content were evaluated by comparing Peceol- and IPM-based formulations containing different oil proportions. (Fig. 2 A, B).

Fig. 2.

Results of in vitro skin permeation studies. (A), (C), and (E) show the cumulative permeation profiles of madecassoside through rat skin over 24 h, while (B), (D), and (F) show the corresponding skin deposition amounts after 24 h. Data are expressed as mean ± SD (n = 3). Statistical analysis at 24 h was performed using one-way ANOVA followed by Tukey's multiple comparisons test. *p < 0.05 compared with the corresponding madecassoside control solution; ⁺p < 0.05 compared with M2; ^@p < 0.05 compared with M1 and #p < 0.05 compared with M6. The corresponding control solution contained 1% or 2% madecassoside, depending on the madecassoside content of the tested formulation.

As shown in Fig. 2A, M1 showed a significantly higher 24 h cumulative permeation amount than both C1% (the 1% madecassoside control solution in distilled water) and M2 (p < 0.05), while M2 was not significantly different from C1%. This result indicates that increasing the Peceol content did not improve madecassoside permeation. The lower permeation observed for M2 may be related to its high oil content. M1 contained only 5% Peceol and a higher aqueous proportion, which may provide a more favorable environment for the release and skin partitioning of hydrophilic madecassoside. In contrast, the 50% Peceol content in M2 may have reduced the availability of madecassoside at the skin interface, thereby limiting its permeation. This observation is consistent with previous findings showing that microemulsion microstructure can influence the skin delivery of hydrophilic compounds, with O/W-type systems providing more favorable permeation than W/O systems (Naoui et al., 2011).

The effect of oil type was further evaluated by comparing M1 and M3, which differed only in the oil phase. Both formulations significantly enhanced madecassoside permeation compared with C1% (p < 0.05), but M3 showed significantly greater permeation than M1 (p < 0.05), indicating that the IPM-based formulation provided greater madecassoside permeation than the Peceol-based formulation under otherwise identical formulation conditions. Because the mechanism responsible for this difference was not directly investigated, the specific interactions between the IPM-based formulation and the skin barrier remain unclear. Regarding skin deposition, C1%, M1, and M3 showed significantly higher drug deposition than M2 (p < 0.05), whereas no significant differences were observed among C1%, M1, and M3 (Fig. 2B). Based on the permeation results, IPM was selected as the oil phase for subsequent optimization.

The effect of surfactant composition and HLB value was further investigated by comparing M3–M6 with C1%. As shown in Fig. 2C, M3 and M4 exhibited significantly higher 24 h cumulative permeation than C1% (p < 0.05), whereas M5 and M6 did not significantly improve permeation compared with the control group. In addition, M3 and M4 showed significantly higher permeation than M6 (p < 0.05). These results indicate that madecassoside permeation was affected by the HLB of the surfactant system, with HLB values of 9.5–10 being more suitable than higher HLB values in the IPM-based microemulsion system. This result may be explained by the requirement for an appropriate balance between the hydrophilic and lipophilic properties of the surfactant blend. Previous work has shown that the HLB value of mixed surfactants affects both the phase behavior of O/W microemulsions and the percutaneous penetration of the incorporated drug (Chen et al., 2012; Fang et al., 2025). Therefore, increasing the HLB value does not necessarily lead to improved permeation, as the optimal HLB depends on the oil phase, surfactant combination, and drug properties. Although M3 and M4 showed comparable permeation performance, M4 was chosen as the starting formulation for further optimization because the Brij 30/Brij 35 mixed surfactant system was more suitable for subsequent surfactant reduction. This design was intended to reduce the amount of surfactant and thereby minimize potential irritation.

The effects of cosurfactant composition and drug loading were further evaluated using M7–M10. M8 was prepared from M7 by replacing part of Transcutol HP with 1,5-pentanediol. Both M7 and M8 showed significantly higher 24 h cumulative permeation than C1% (p < 0.05), while no significant difference was observed between them (Fig. 2E). This indicates that partial replacement of Transcutol HP with 1,5-pentanediol did not further enhance madecassoside permeation. M9 and M10 were prepared by increasing the madecassoside content of M8 and M7, respectively, from 1% to 2%. Both formulations exhibited significantly higher permeation than C2% (the 2% madecassoside control solution in distilled water) (p < 0.05), whereas no significant difference was observed between M9 and M10. These results suggest that increasing drug loading improved the permeation performance of the microemulsion systems, while the two cosurfactant compositions showed comparable effects at 2% madecassoside loading. Additionally, the corresponding skin deposition results are shown in Fig. 2F. No significant differences were observed among C1%, C2%, M7, M8, M9, and M10.

Previous studies have reported enhanced skin permeation of madecassoside using liposomal and nanoemulsion-based delivery systems. The PEG-modified double-emulsion liposomes produced an approximately 5-fold increase in madecassoside permeation using full-thickness dorsal skin from Sprague–Dawley rats. However, cumulative permeation was evaluated over 12 h, with the skin membrane replaced every 4 h (Li et al., 2016a). The paeonol/madecassoside nanoemulsion increased the 24 h cumulative permeation of madecassoside by 273.5%, corresponding to approximately 3.74-fold relative to free madecassoside, using dorsal skin from BAMA miniature pigs (Lu et al., 2023). By comparison, M10 increased the 24 h cumulative permeation of madecassoside by approximately 43.2-fold relative to the corresponding control solution using excised skin from Sprague–Dawley rats. Collectively, these findings demonstrate that liposomal, nanoemulsion, and microemulsion systems are all promising formulation strategies for enhancing the skin delivery of madecassoside under their respective experimental conditions.

Madecassoside has a molecular weight of 975.13 Da, which exceeds the generally recognized 500-Da threshold and is therefore intrinsically unfavorable for passive diffusion through the intact stratum corneum (Bos et al., 2000). Nevertheless, previous studies have suggested that skin transport from topical microemulsions is influenced not only by molecular size but also by drug partitioning between the formulation and the skin and by formulation-mediated modulation of the stratum corneum barrier (Szumała et al., 2022). Microemulsion components may enhance skin permeation by interacting with or perturbing intercellular stratum corneum lipids, while the aqueous phase of water-rich systems may increase stratum corneum hydration and thereby reduce its diffusional resistance (Hathout et al., 2010). Accordingly, the enhanced permeation observed with M10 may reflect the combined effects of the formulation on madecassoside partitioning and stratum-corneum barrier properties, which may partly offset its unfavorable molecular size. However, because these processes were not directly investigated in the present study, the proposed mechanism requires further experimental verification.

Although skin deposition did not differ significantly from that of the control solutions, the markedly increased cumulative permeation indicates that the microemulsions facilitated the passage of madecassoside across the stratum corneum. Previous studies have shown that skin permeation and skin deposition are distinct outcomes that do not necessarily change in parallel, indicating that increased permeation does not invariably lead to increased skin retention (Fang et al., 2025; Naoui et al., 2011). Therefore, the present rat skin permeation study was used primarily as a formulation-screening model to determine whether the microemulsions could improve the delivery of madecassoside across the skin barrier. However, rat skin is generally thinner and more permeable than human skin, and the present findings cannot be directly extrapolated to human topical application (Abd et al., 2016; Todo, 2017). Further studies using human skin or other clinically relevant skin models, together with layer-specific drug distribution and pharmacodynamic evaluations, are needed to determine the applicability of the optimized formulation.

3.4. Irritation potential

The irritation potential of the formulations was evaluated using the HET-CAM assay, and the results are shown in Fig. 3. Normal saline, used as the negative control, showed no apparent hemorrhage, vasoconstriction, or coagulation during the observation period, with an irritation score of 0. In contrast, 0.1 N NaOH induced severe vascular damage within approximately 20 s, resulting in an irritation score of 17.60 ± 0.29. The 1% SLS group showed a moderate irritation response with an irritation score of 8.92 ± 0.69. In this study, 1% SLS was used as a reference irritant rather than a negative control. Previous safety assessment has indicated that SLS concentrations in products intended for prolonged skin contact should not exceed 1% (Cosmetic Ingredient Review, 1983). Therefore, formulations with irritation scores lower than or comparable to the 1% SLS group were considered to have an acceptable irritation profile under the present experimental conditions. For the microemulsion formulations, Blank-M9 and M9 showed higher irritation scores (14.55 ± 1.37, 13.50 ± 1.19) than Blank-M10 and M10 (7.21 ± 0.88 and 5.98 ± 0.47). The higher irritation observed in M9 may be associated with the presence of 1,5-pentanediol in the cosurfactant system. Compared with M9, M10 showed a milder CAM response and an irritation score lower than that of the 1% SLS reference group. Therefore, M10 was considered to have acceptable irritation potential and was selected as the optimized formulation.

Fig. 3.

Representative images of the HET-CAM assay after treatment with different formulations.

3.5. Stability study

The stability of the optimized madecassoside-loaded microemulsion (M10) was evaluated under thermodynamic, short-term, and accelerated conditions. As shown in Table 3, M10 remained clear after centrifugation and freeze–thaw cycling, with no visible phase separation, creaming, or turbidity. The droplet size remained within a narrow range of 48.60 ± 0.10 to 50.00 ± 0.46 nm, and the PDI values remained below 0.20, indicating that the microemulsion maintained a uniform droplet distribution after stress testing. During storage at 25 ± 2 °C with 60 ± 5% RH and 40 ± 2 °C with 75 ± 5% RH for 30 days, M10 also remained clear, and no marked increase in droplet size, PDI, or viscosity was observed. The good stability of M10 is consistent with the reported characteristics of microemulsions, which are generally described as clear, isotropic, and thermodynamically stable systems. This stability is mainly associated with low interfacial tension and the formation of a flexible interfacial film, allowing the system to maintain nanoscale droplets and resist phase separation during storage (Lawrence et al., 2012; Lopes, 2014). Overall, these results indicate that M10 maintained good physical stability under the tested conditions.

Table 3.

Physical and thermodynamic stability of the optimized madecassoside-loaded microemulsion formulation (M10).

Condition		Appearance	Droplet size (nm)	PDI	Viscosity (cP)
M10 Before stability test		Clear	48.60 ± 0.44	0.18 ± 0.01	9.51 ± 0.15
After centrifugation		Clear	48.60 ± 0.10	0.18 ± 0.01	9.30 ± 0.32
After freeze–thaw cycle		Clear	50.00 ± 0.46	0.19 ± 0.01	10.23 ± 0.12
25 ± 2 °C 60 ± 5% RH stability	Day 0	Clear	48.60 ± 0.44	0.18 ± 0.01	9.51 ± 0.15
	Day 7	Clear	49.17 ± 0.61	0.18 ± 0.01	10.60 ± 0.10
	Day 30	Clear	49.80 ± 0.26	0.17 ± 0.01	10.63 ± 0.15
40 ± 2 °C 75 ± 5% RH Accelerated stability	Day 0	Clear	48.73 ± 0.23	0.19 ± 0.01	9.51 ± 0.18
	Day 7	Clear	49.53 ± 0.47	0.18 ± 0.02	10.93 ± 0.15
	Day 30	Clear	49.23 ± 0.45	0.17 ± 0.01	10.80 ± 0.10

Previous studies have demonstrated that different madecassoside-containing delivery systems can maintain favorable physical stability under their respective experimental conditions. The PEG-modified double-emulsion liposomes maintained a homogeneous appearance for 5 months, although storage-related changes in particle size and polydispersity index were not reported (Li et al., 2016a). The paeonol/madecassoside nanoemulsion remained stable for 30 days, but the relative humidity during storage was not specified (Lu et al., 2023). In the present study, M10 showed no substantial changes in appearance, droplet size, or polydispersity index following centrifugation, freeze–thaw cycling, and 30 days of storage under specified temperature and relative-humidity conditions. Collectively, these findings demonstrate the potential of different formulation platforms to maintain the physical stability of madecassoside-containing delivery systems.

4. Conclusions

In this study, madecassoside-loaded microemulsions were successfully developed and optimized for topical delivery. The formulation screening results indicated that both the type and proportion of oil phase, as well as the HLB of the surfactant system, were important factors affecting madecassoside permeation. In contrast, reducing the surfactant concentration from 15% to 10% did not significantly impair permeation, and partial replacement of Transcutol HP with 1,5-pentanediol did not further enhance drug permeation under the tested conditions. The prepared formulations showed clear and homogeneous appearances, with nanoscale droplet sizes ranging from 28.00 ± 0.44 to 60.10 ± 0.61 nm and low PDI values. Among the tested formulations, M10, composed of IPM, Brij 30/Brij 35, Transcutol HP, and 2% madecassoside, showed the most favorable permeation performance. The 24 h cumulative permeation amount of M10 reached 406.28 ± 180.73 μg/cm², corresponding to an approximately 43.2-fold increase compared with the 2% madecassoside control solution. In the HET-CAM assay, M10 showed a lower irritation score than the 1% SLS reference group, indicating an acceptable irritation profile under the present experimental conditions. M10 also maintained a clear appearance, nanoscale droplet size, low PDI, and stable viscosity after thermodynamic stress testing and 30-day storage. In addition, M10 could be prepared by simple mixing without the need for polymer synthesis, organic-solvent removal, membrane extrusion, or high-pressure homogenization, indicating a comparatively straightforward preparation process. Overall, the optimized microemulsion improved madecassoside skin permeation while maintaining acceptable irritation potential and physical stability. These findings suggest that microemulsion-based delivery is a promising strategy for enhancing the topical application of hydrophilic compounds with limited skin permeability. Building on the present findings, future studies incorporating in vitro release testing, zeta-potential characterization, mechanistic evaluation of formulation–skin barrier interactions, and appropriate in vivo efficacy assessments would provide a more comprehensive evaluation of the optimized formulation.

CRediT authorship contribution statement

I-Hui Chiu: Writing – original draft, Investigation, Formal analysis, Data curation. Quoc Lam Vu: Validation, Project administration. Pao-Chu Wu: Writing – review & editing, Supervision, Project administration, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.