Liposomal Gel for enhancing topical efficacy of timolol maleate against proliferating superficial infantile hemangiomas

Abstract

Conventional timolol maleate (TM) eye drops are limited by low transdermal bioavailability and short dermal residence time for the treatment of superficial infantile hemangioma (IH). To address this, a novel 0.5% TM-loaded liposomal gel (TM-lipogel) was developed by embedding optimized liposomes within a carbomer gel matrix. The TM-liposomes exhibited a desirable particle size of 139.5 ± 2.1 nm, a polydispersity index (PDI) of 0.051, a zeta potential of −38.7 mV, and an encapsulation efficiency of 64.76 ± 3.2%. Ex vivo permeation studies revealed a 1.9-fold increase in cumulative transdermal delivery compared to the TM-gel, and 4-fold compared to the TM solution (p < 0.05). Furthermore, confocal laser scanning microscope imaging revealed that TM-lipogel achieved a penetration depth of 400 μm, while the TM-gel and TM-eyedrop only reached 200 μm and 75 μm, respectively, highlighting the superior penetration of the liposomal gel. In a murine xenograft hemangioma model, 0.5% TM-lipogel achieved superior tumor regression (81.09%, p < 0.01 vs.control) compared to 0.5% TM-eyedrop (29.85%) and 0.5% TM-gel (61.69%) by simultaneously downregulating HIF-1α, VEGF, MMP-9 and eNOS. Critically, the optimized formulation exhibited no skin irritation and no systemic toxicity. These findings highlight that the TM-lipogel provides enhanced transdermal delivery with deeper tissue penetration, representing a highly effective and safe transdermal strategy for IH therapy.

Graphical abstract

1. Introduction

Infantile hemangiomas (IH) are among the most common benign tumors of early childhood, characterized by the abnormal proliferation of vascular endothelial cells. The global incidence ranges from approximately 4% to 10%, with a higher prevalence observed in Caucasians, females, preterm infants, and those with low birth weights (Krowchuk et al., 2019; Novoa et al., 2018). IH typically follows a triphasic clinical course: rapid proliferation, stabilization, and spontaneous involution (Drolet et al., 2008). Although many cases resolve without intervention, approximately 10% of untreated or inadequately treated patients develop permanent sequelae, including hyperpigmentation, atrophic scarring, or fibro-adipose tissue overgrowth, which can lead to functional impairment and disfigurement (Mariani et al., 2022; Braun et al., 2024). These risks underscore the need for effective and safe therapeutic strategies.

Current treatment modalities for IH include surgical excision, laser therapy, and pharmacotherapy (Xu and Zhao, 2022; Shah et al., 2023). Given the invasive nature of surgery, pharmacotherapy is generally the preferred option for infants. Propranolol, a non-selective β-adrenergic receptor blocker, was first reported as an effective IH therapy in 2008 (Léauté-Labrèze et al., 2008) and was subsequently approved by the U.S. Food and Drug Administration (FDA) in 2014 as Hemangeol®, the first oral solution specifically indicated for IH. Despite its status as the global first-line treatment, oral propranolol is associated with systemic adverse effects, including bradycardia, hypotension, bronchospasm, hypoglycemia, drowsiness, and sleep disturbances, largely due to its ability to cross the blood-brain barrier (Léauté-Labrèze et al., 2017; Langley and Pope, 2015; Tan et al., 2021). Consequently, topical therapy is considered a safer alternative for superficial IH.

Timolol, another non-selective β-receptor antagonist, was originally introduced for glaucoma management in 1978. In 2010, topical timolol maleate (TM) eye drops were first reported to successfully treat eyelid IH (Guo and Ni, 2010). Since then, numerous studies have confirmed their efficacy against superficial IH with minimal treatment-related adverse effects (Tlougan et al., 2011; Ni et al., 2011). Despite these benefits, the use of TM as eye drops poses substantial limitations: poor transdermal absorption, low viscosity, and short residence time at the lesion site, resulting in suboptimal therapeutic outcomes. In clinical practice, Chinese guidelines recommend wet compresses with TM eye drops to prolong exposure; however, this method is cumbersome and carries a risk of ulceration (Khan et al., 2017; Chinese Society of Dermatology, 2020). Therefore, developing a more convenient and effective topical formulation is essential.

The skin presents a formidable barrier to transdermal drug delivery. The outermost stratum corneum selectively permits the passage of small lipophilic molecules (Jiao et al., 2022). Being highly hydrophilic, TM exhibits poor penetration across this barrier. Furthermore, hemangioma tissue itself often exhibits thickened epidermal layers with elevated keratin and lipid content, further impeding drug penetration (Slavkova et al., 2023). To overcome these challenges, advanced transdermal delivery systems such as gels and liposomes have been explored. Gels offer biocompatibility, adaptability, and the capacity for sustained drug release (Li and Mooney, 2016), while liposomes—nanoscale carriers composed of phospholipid bilayers—can traverse biological barriers, accumulate at target sites, and provide sustained drug release (Gabizon et al., 2003; Dudu et al., 2012). Combining these two systems, liposomal hydrogels integrate the advantages of both: enhanced skin affinity, improved bioavailability, controlled release, ease of application, and patient comfort.

Based on these premises, this study developed a timolol maleate liposomal gel (TM-lipogel) for the treatment of IH. This formulation was designed to address the shortcomings of conventional TM eye drops—specifically poor absorption, short retention time, and inconvenient administration—while further enhancing the transdermal performance and therapeutic efficacy of the gel. This work aims to provide a theoretical basis for TM-lipogel as a practical, child-friendly treatment regimen for superficial IH.

2. Materials and methods

2.1. Materials

Timolol maleate purchased from Hubei Weideli Chemical Reagent Ltd. (China). Carbomer 940 purchased from Shanghai Macklin Biochemical Technology Ltd. (China). Cholesterol purchased from China National Pharmaceutical Group Chemical Reagent Ltd. (China). Soya lecithin purchased from Shanghai Yuanye Biotechnology Ltd. (China). Methanol (chromatographic grade) were obtained from Suzhou Utep Technology Ltd. (China). Propylene glycol, azone and Tween 80 purchased from Shanghai Aladdin Reagent Ltd. (China). Endothelial cell medium (ECM, catalog #1001) purchased from Sciencell (Shanghai, China). Fetal bovine serum (FBS) purchased from Thermo Fisher Scientific (USA) and phosphate-buffered saline (PBS) purchased from Wuhan Servicebio Technology Ltd. (China). All additional reagents and chemicals used throughout the study were of analytical grade quality.

2.2. Design and optimization of experiments

Based on preliminary single-factor experiments (data not shown), a 3³ orthogonal design was employed to optimize the TM-gel formulation. The concentrations of carbomer 940 (Factor A), glycerol (Factor B), and propylene glycol (Factor C) served as independent variables. Formulations were evaluated according to the scoring criteria outlined in Table S1, and the specific factor levels are presented in Table S2. The variables, their codes and levels, together with the responses using SPSS 26.0, are listed in Table S3.

The formulation of TM-liposomes was optimized using central composite design (CCD) coupled with response surface methodology (RSM). Preliminary experiments (Fig. S1) were initially conducted to determine the appropriate ranges for the independent variables. Subsequently, the mass ratio of phospholipid to cholesterol (A, 6:1–4:1), the mass ratio of phospholipid to TM (B, 6:1–4:1), the hydration pH (C, 7.0–8.5), and the hydration temperature (D, 25–35 °C) were selected as independent variables. The encapsulation efficiency (%EE) and drug loading (LE%) were designated as dependent variables. A total of 27 experimental runs were generated, and the corresponding design matrix is summarized in Table S5.

2.3. Formulation of TM-liposome

TM-liposomes were prepared using an ether injection-ultrasonication method. The organic phase consisted of soybean phosphatidylcholine (SPC, 170 mg) and cholesterol (34 mg), dissolved in anhydrous diethyl ether (5 mL). Separately, TM (30 mg) was dissolved in phosphate-buffered saline (PBS, 6.7 mL, pH 7.4) to form the aqueous phase. Under continuous magnetic stirring (800 rpm) at 30 °C, the organic phase was slowly injected dropwise into the aqueous phase and hydrated for 45 min. Subsequently, the resulting suspension was ultrasonicated in an ice-water bath using an ultrasonic cell disruptor (JY92-11 N, Ningbo Xinzhi Biotechnology Co., Ltd.) at 15% power for 2 min to obtain the TM-liposome suspension.

2.4. Physicochemical characterization of TM-liposome

2.4.1. Morphology and particle size analysis

The morphology of TM-liposomes was examined using transmission electron microscopy (TEM, Hitachi HT7700, Japan) operated at an accelerating voltage of 100 kV. Samples were diluted 5-fold with deionized water, deposited on carbon-coated copper grids, and negatively stained with 2% (w/v) phosphotungstic acid for 1.5 min prior to imaging (Younes et al., 2024, Yousry et al., 2023). Images were acquired under high-contrast mode (HC-1) at nominal magnifications of ×10,000 and ×80,000. The hydrodynamic particle size, polydispersity index (PDI), and zeta potential were measured using a Zetasizer Nano ZS90 (Malvern Instruments Ltd., UK) at 25 °C. All measurements were performed in triplicate.

2.4.2. Entrapment efficiency (EE)

The EE of TM-liposomes was determined indirectly using an ultracentrifugation method (El Menshawe et al., 2024; Sadek Ahmed et al., 2024). Briefly, 1 mL of the liposomal suspension was centrifuged at 12,000 rpm (15,280 ×g) at 4 °C for 30 min (Beckman Coulter Ltd., BK G165A). The supernatant (100 μL) was collected and diluted to 1 mL with methanol to quantify the free drug content (W₁). To determine the total drug content (W₂), 100 μL of the liposomal suspension was mixed with 900 μL of methanol, vortexed, sonicated (40 kHz, 5 min), and centrifuged (15,280 ×g, 30 min). The resulting supernatant was filtered through a 0.45 μm membrane and analyzed by HPLC. The EE was calculated using the following formula:

$$\mathit{EE}\left(\%\right)=\left(W_2-W_1\right)/W_2\times 100\%$$ (1)

2.4.3. HPLC analytical method

The quantification of TM was performed using a validated HPLC method (Supplementary Material). Chromatographic separation was achieved on a ChromCore 120 C₁₈ column (4.6 mm × 250 mm, 5 μm) with a mobile phase consisting of methanol:water:triethylamine (40,60,0.1, v/v/v, adjusted to pH 5.0 with phosphoric acid) at a flow rate of 1 mL/min. The injection volume was 20 μL, the column temperature was maintained at 25 °C, and the detection wavelength was set at 295 nm.

2.4.4. In vitro release study

In vitro drug release was assessed using the dialysis bag. TM-liposome suspension (5 mL, equivalent to 0.5% TM) and free TM solution (5 mL, 0.5%) were placed in dialysis bags (MWCO = 3500 Da) and immersed in 100 mL of PBS release medium at 37 °C under constant stirring at 300 rpm. At predetermined time intervals (0.5, 1, 2, 4, 6, 8, 12, and 24 h), 2 mL of the dialysate was withdrawn and replaced with an equal volume of fresh, pre-warmed PBS. The cumulative release rate (Er) was calculated using the formula:

$$E_{\mathrm{r}}=\frac{V_e{\sum }_{i=1}^{n-1}{C}_i+V_0{C}_n}{1000\times m_{\mathit{drug}}}\times 100\%$$ (2)

Er: Cumulative drug release rate; V_e: Volume of each sample (mL); V₀: Total volume of dialysis fluid (mL); C_n: Concentration of the n^th sample (μg•mL⁻¹); m_drug: Total drug mass in dialysate (mg); n: Number of times for dialysis fluid replacement.

2.5. Preparation of TM-lipogel and TM-gel

The TM-gel was prepared via a cold-process method. Azone was dissolved in Tween-80 and propylene glycol, followed by the addition of glycerol and distilled water. TM was dissolved separately in water and then incorporated into the mixture. Carbomer 940 was gradually added to the system, allowed to swell overnight, and neutralized with NaOH solution to form a homogeneous gel. The final formulation consisted of 0.5% (w/w) TM, 1% (w/w) carbomer 940, 5% (w/w) glycerol, 9% (w/w) propylene glycol, and 0.25% (w/w) azone.

TM-lipogel was prepared by physically mixing the TM-liposome suspension with an equal volume of the blank gel base. The mixture was homogenized at a low speed to ensure uniform dispersion of liposomes within the gel matrix.

2.6. Physicochemical characterization of TM-lipogel and TM-gel

2.6.1. Rheological property

The rheological properties of TM-gel and TM-lipogel were evaluated using a rotational rheometer (AR2000 DHR-2, TA Instruments, USA) equipped with a Peltier plate system. Samples were equilibrated at 25 ± 0.1 °C prior to measurement.

Steady-state flow behavior. Flow curves were recorded by measuring the viscosity as a function of shear rate ranging from 0.001 to 100 s⁻¹. This test was performed to characterize the flow behavior and viscosity of the formulations under simulated application conditions.

Dynamic oscillatory measurements. To assess the viscoelastic properties, strain sweep tests were first conducted at a fixed frequency of 1 Hz over a strain range of 0.01%–100%. The linear viscoelastic region (LVR) was identified for both formulations, defined as the range where the storage modulus (G′) remained independent of strain. Subsequently, frequency sweep tests were performed from 0.01 to 100 Hz at a constant strain of 10%, which was confirmed to lie within the LVR. The storage modulus (G′) and loss modulus (G′′) were recorded to evaluate the structural stability and elasticity of the gels.

2.6.2. Preliminary stability

The stabilities of TM-gel and TM-lipogel were evaluated using centrifugation, freeze–thaw cycling, and accelerated stress tests. Samples were subjected to centrifugation at 10,610 ×g for 60 min to assess physical stability. Additionally, freeze–thaw stability was evaluated by subjecting 1 g of each formulation to three consecutive cycles of freezing at −20 °C for 48 h, followed by thawing at room temperature and subsequent incubation at 60 °C for 48 h. Accelerated stability studies were conducted under the following conditions for 10 days: (i) high temperature (40 °C and 60 °C), (ii) high humidity (25 °C/75% RH and 25 °C/92.5% RH), and (iii) photostability (4500 ± 500 lx). Samples were withdrawn on Days 5 and 10 and analyzed for appearance, pH, and drug content. The results were compared with initial values to assess stability.

The pH values of TM-gel and TM-lipogel were determined using a calibrated PB-10 pH meter (Sartorius, Germany) equipped with a glass electrode. Samples (1 g) were dispersed in 10 mL of deionized water and equilibrated at 25 ± 0.1 °C prior to measurement. Data are expressed as mean ± standard deviation (SD) (n = 3).

2.6.3. In Vitro drug release

In vitro drug release was evaluated using Franz diffusion cells. Cellophane membranes and 0.22/0.45 μm filter membranes were mounted between the donor and receptor compartments. The receptor chamber was filled with PBS (pH 7.4), ensuring the complete removal of air bubbles.

Samples (0.1 g) of TM-eyedrop, TM-gel and TM-lipogel were applied uniformly to the membrane surface. The system was maintained at 32 ± 1 °C with magnetic stirring at 300 rpm. At predetermined time intervals (0.5, 1, 2, 4, 6, 8, 12, and 24 h), 2 mL of the receptor solution was withdrawn and immediately replaced with an equal volume of fresh PBS. The collected samples were filtered through a 0.45 μm membrane and analyzed by HPLC.

The cumulative amount released per unit area (Q) was calculated using Formula (3):

$$Q=\frac{C_n\times V+{\sum }_{i=1}^{n-1}{C}_i\times V_{\mathit{sample}}}{A}$$ (3)

Q: Cumulative amount released per unit area at time t (μg•cm⁻²); V: Total volume of the receiving solution (mL); Cₙ: Concentration of the n^th sample (μg•mL⁻¹); Ci: Concentration of the i^th sample (μg•mL⁻¹); V_sample: Volume of each sample (mL); A: Effective permeation area of the diffusion cell (cm²); n: Number of sampling times.

The cumulative permeation rate (Er) was then calculated by normalizing Q to the total applied dose (m_drug):

$$\mathit{Er}=\frac{Q\times A}{m_{\mathit{drug}}}\times 100\%$$ (4)

Where m_drug is the total mass of TM applied in the donor compartment (μg).

Release profiles were plotted as Er versus time (h). The release kinetics were fitted to first-order, Higuchi, and Peppas diffusion models to determine the best-fit model, which was identified as the one with the highest coefficient of determination (R²) (Ahmed et al., 2020).

2.6.4. In Vitro Transdermal Permeability

All animal experimental protocols were approved by the Research Ethics Committee of the Faculty of Pharmacy, Soochow University (Approval No. 202406A0123). Male Sprague-Dawley (SD) rats were sacrificed by cervical dislocation. The abdominal hair was carefully removed using electric clippers, ensuring the stratum corneum remained intact. The abdominal skin was then excised, and the underlying subcutaneous tissue and adipose tissue were carefully trimmed away using surgical scissors to obtain full-thickness skin pieces.

In vitro transdermal permeation studies were performed using Franz diffusion cells. Excised skin samples were mounted between the donor and receptor compartments, with the receptor chamber filled with degassed PBS (pH 7.4). Samples (0.1 g) of TM-gel or TM-lipogel were evenly applied to the epidermal surface of the skin. The system was maintained at 32 ± 1 °C with continuous magnetic stirring at 300 rpm.

At predetermined time intervals (0.5, 1, 2, 4, 6, 8, 12, and 24 h), 2 mL of the receptor solution was withdrawn and immediately replaced with an equal volume of fresh PBS. The collected samples were filtered through a 0.45 μm membrane prior to HPLC analysis. The cumulative amount permeated per unit area (Q) was calculated according to Formula (3).

2.6.5. In vivo skin permeation

Preparation of fluorescent probes. A Rhodamine B (Rhod B) solution (0.5 mmol/L) was mixed with an equal volume of TM solution and stirred in the dark for 72 h to allow tight binding between Rhod B and TM. Unbound Rhod B was subsequently removed by dialysis (MWCO = 500 Da) to obtain the Rhod B-TM complex. Rhod B-TM-gel and Rhod B-TM-lipogel were then prepared using this complex and stored at 4 °C, protected from light.

Animal administration and imaging. Female Kunming (KM) mice were randomly divided into three groups based on sampling time (0.5, 4, and 12 h; n = 3 per group). The abdominal hair was shaved 12 h prior to administration. Mice were administered 0.5 g of Rhod B-TM solution, Rhod B-TM-gel, or Rhod B-TM-lipogel over a 1 cm² area. For the solution group, administration mimicked the clinical eyedrop regimen: a 1 cm² gauze pad was completely soaked with the drug solution, applied to the shaved area, and secured with a waterproof patch.

At each predetermined time point, the mice were euthanized. The residual formulation was gently wiped off, and the treated skin tissues were excised for confocal laser scanning microscopy (CLSM) analysis. The scanning area was set to 800 μm × 800 μm, with a scanning depth of 400 μm and a step size of 100 μm. ZEN 2 software (Zeiss, Germany) was used to generate 2.5D simulated mappings to visualize the spatial distribution of the fluorescent probe.

2.7. In vivo safety studies

2.7.1. Histopathological examination

In addition to efficacy, the safety profile of the formulations was rigorously evaluated. Twelve female KM mice were randomly assigned to four groups (n = 3 per group): normal control, blank gel, TM-gel, and TM-lipogel. Following hair removal 12 h prior to treatment, 0.5 g of the respective formulations was topically applied to a 1 cm² area on the abdomen daily for one week. After the treatment period, residual formulations were gently washed off, and the animals were euthanized via cervical dislocation. Skin tissues were harvested and fixed in 4% paraformaldehyde, followed by hematoxylin and eosin (H&E) staining to evaluate histological changes and structural integrity.

2.7.2. Skin irritation studies

New Zealand rabbits and Sprague-Dawley (SD) rats were used to establish intact and abraded skin models (n = 3 per group). Following hair clipping 12 h prior to dosing, intact skin was visually confirmed. For abraded skin models, the shaved area was gently abraded with 400-grit sandpaper until punctate bleeding was observed.

Single-dose irritation test: As an internal control, each animal received TM-gel on the left flank and TM-lipogel on the right flank (0.2 g/cm²), and a control area on the head. The application sites were covered with gauze and a waterproof patch to prevent ingestion. After 24 h, residual formulations were gently washed off. Skin reactions (erythema and edema) were scored at 0, 24, 48, and 72 h post-application according to standard Draize scoring criteria (Table S15 and S16).

Multiple-dose irritation test: Animals received daily topical applications (0.2 g/cm²) for one week. To simulate persistent lesions, the abraded skin was gently re-abraded with 400-grit sandpaper prior to each administration. Following the final dose, skin sites were observed at 0, 24, 48, and 72 h post-treatment. Mean irritation scores were calculated to evaluate the severity and reversibility of skin reactions.

2.8. In Vivo animal experiments

2.8.1. In vivo efficacy

Female BALB/c nude mice (SPF grade, 3–4 weeks old, 18–20 g) were obtained from Cavens Animal Center (Changzhou, China). To establish the IH xenograft model, HemECs (10⁷ cells/mL) and HUVECs (10⁶ cells/mL) were mixed at a 10:1 ratio (v/v) with ice-cold Matrigel (Corning Incorporated, USA). Subsequently, 0.2 mL of the cell suspension was subcutaneously injected into the dorsum of each mouse. In this model, HemECs serve as the proliferating hemangioma-derived endothelial cell population responsible for forming the hemangioma-like vascular network, while HUVECs are co-injected at a lower density to provide stromal support through paracrine signaling (including VEGF, angiopoietin, and PDGF), which facilitates vascular stabilization and maturation.

When the tumor volume reached approximately 80 mm³, the mice were randomly assigned to six treatment groups (n = 6 per group): (1) PBS control, (2) 0.5% TM eyedrops, (3) 0.5% TM-gel, (4) 0.25% TM-lipogel (low dose), (5) 0.5% TM-lipogel (medium dose), and (6) 1% TM-lipogel (high dose). Formulations were topically administered three times daily at a dose of 0.2 g·cm⁻² (per application) for 30 days, starting from day 15 post-injection. Tumor volume and body weight were measured every 3 days. Tumor volume (V) was calculated using the formula:

$$V=0.5\times a\times b^2$$ (5)

where a is the length (mm) and b is the width (mm) of the tumor.

It should be noted that this dose is intentionally higher than typical clinical thin-layer applications in order to fully evaluate the efficacy potential and safety margin of the formulation within the constrained timeframe. At the end of the study, mice were euthanized, and tumor tissues were excised and weighed. The tumor inhibition rate (TIR) was calculated as:

$$\mathit{TIR}\left(\%\right)=\frac{\mathit{Mc}-\mathit{Mt}}{\mathit{Mc}}\times 100\%$$ (6)

Where Mc is the average tumor weight of the control group (mg) and Mt is the average tumor weight of the treatment group (mg).

Major organs (heart, liver, spleen, lungs, and kidneys) were harvested, rinsed, blotted dry, and weighed to calculate the organ index. Tissue samples were fixed in 4% paraformaldehyde and processed for HE staining to assess histopathological toxicity. The organ index was calculated as:

$$\mathit{Organ\; index}=\left(\text{Weight of organ}/\text{Body weight of mice}\right)\times 100\%$$ (7)

2.8.2. Immunohistochemical Analysis of HIF-1α, VEGF, and CD34

Harvested IH tissues were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at 4 μm thickness. Immunohistochemical staining was performed using primary antibodies against VEGF, HIF-1α, and CD34 (1:200), followed by HRP-conjugated secondary antibodies (1200). Protein expression was visualized via DAB staining. VEGF and HIF-1α expression levels were quantified using Image-Pro Plus 6.0 software by calculating the integrated optical density (IOD).

CD34, a specific marker for vascular endothelial cells, was used to evaluate microvessel density (MVD). Briefly, the areas with the highest neovascularization (hotspots) were first identified under low-power magnification (40×). Subsequently, the number of tubular structures was manually counted in five hotspots under high-power magnification (200×) (Yugawa et al., 2021). The average count was recorded as the MVD value. For counting criteria, individual endothelial cells or clusters forming tubular structures that appeared as brown, cord-like, or cleft-like shapes were counted as a single vessel. However, larger vessels with a lumen area greater than eight red blood cells or those surrounded by smooth muscle were excluded from the count to focus specifically on microvessels.

2.8.3. Quantification of MMP-2, MMP-9, and eNOS

Tumor tissues (100 mg) were rinsed three times with ice-cold PBS and homogenized rapidly in 1 mL of ice-cold PBS. After centrifugation of the homogenates (4 °C, 5000 rpm, 10 min), supernatants were collected. The concentrations of MMP-2, MMP-9, and eNOS were quantified using commercial ELISA kits (Hunan Aifang Biotechnology Co., Ltd.) according to the manufacturer's instructions.

2.9. Statistical analysis

All data are expressed as mean ± standard deviation (SD). Data normality was assessed using the Shapiro-Wilk test. Statistical comparisons were performed using one-way analysis of variance (ANOVA) followed by Tukey's post hoc test for multiple comparisons. GraphPad Prism 9.0 software was used for all analyses. Statistical significance was set at *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001 indicating increasing levels of significance.

3. Results and discussion

3.1. Formulation of TM liposome and TM-gel

3.1.1. Optimization of TM-liposome

The formulation of TM-liposomes was optimized using CCD coupled with RSM. Quadratic polynomial models were fitted to correlate the independent variables with the dependent variables (EE and LE). The regression equations (provided in Supplementary Material) demonstrated high statistical significance (p = 0.0003 and 0.0004, respectively) with coefficients of determination (R²) of 0.91 and 0.90, indicating good model fit.

Response surface plots revealed that EE and LE initially increased and then decreased with increasing levels of the four factors (Fig. S2 and S3). The analysis of variance (ANOVA) results indicated that the order of influence on EE was: phospholipid to cholesterol ratio > lipid to drug ratio > temperature > pH. A significant interaction between temperature and phospholipid to cholesterol ratio was observed for LE. Based on the desirability function, the optimal conditions were determined as: phospholipid to cholesterol ratio of 4.94:1, lipid to drug ratio of 5.95:1, hydration temperature of 30 °C, and pH 7.5. Under these conditions, the predicted EE and LE were 67.43% and 28.86%, respectively. Validation experiments (n = 6) yielded actual values of 64.76% and 27.35%, with relative errors of 2.00% and 1.77% compared to the predictions, confirming the reliability of the model.

3.1.2. Optimization of TM-gel

A 3³ full factorial design was employed to optimize the TM-gel formulation. The compositions and responses for the nine experimental runs are summarized in Table S3. ANOVA results (Table S4) demonstrated that Factor A (carbomer 940 concentration) significantly affected the overall gel score (p < 0.05), whereas Factors B (glycerol) and C (propylene glycol) did not. Thus, the order of influence was A > B > C. Although propylene glycol did not significantly affect the physical scores of the gel, it was incorporated at a high concentration (9%) to act as a penetration enhancer, synergistically facilitating TM transdermal delivery in conjunction with azone (0.25%). The optimal formulation was determined to be A1B1C3, corresponding to 1% carbomer 940, 5% glycerol, and 9% propylene glycol.

The optimized TM-gel was colorless, transparent, homogeneous, and smooth, devoid of visible particles or impurities, meeting the standards for topical gels outlined in the United States and Chinese Pharmacopoeias (Fig.S5). The TM-lipogel appeared semi-transparent with a pale-yellow hue and a fine, smooth texture. It maintained a stable semi-solid state without phase separation or suspended particles, indicating the uniform dispersion of liposomes within the gel matrix.

3.2. Physicochemical characterization of TM-liposome

The optimized TM-liposome suspension appeared as a light-yellow, translucent liquid with slight opalescence (Fig. 1A). TEM revealed uniformly sized vesicles with intact spherical morphology and a predominantly large unilamellar vesicle (LUV) structure. This LUV architecture, featuring a spacious internal aqueous core, facilitates the efficient encapsulation of the hydrophilic drug. The mean encapsulation efficiency (EE) was 64.76% (n = 6). While achieving high EE for hydrophilic drugs is inherently challenging due to drug leakage into the external aqueous phase during preparation, this result surpasses the 39.96% EE reported for timolol maleate transfersomes using similar methods (Morsi et al., 2016). This moderate EE was deemed acceptable to avoid introducing charged lipids or complex gradients that might compromise the safety profile for infant skin.

Fig. 1.

Physicochemical characterization of TM-liposome. A digital image of TM liposome. B ∼ C representative TEM images of TM-liposome. D size distribution of TM-liposome. E Zeta potential of TM-liposome. FIn vitro release profiles of TM-liposome and TM solution.

Dynamic light scattering (DLS) analysis indicated a narrow size distribution, with a mean diameter of 139.5 nm, a polydispersity index (PDI) of 0.051, and a zeta potential of −38.7 mV. Particle size is critical for transdermal delivery; nanoparticles smaller than 200 nm penetrate the stratum corneum more efficiently, whereas ultra-small particles (< 50 nm) may be trapped by the epidermal barrier. Conversely, larger liposomes (>300 nm) offer greater stability but are hindered by poor penetration and potential immunogenicity. The TM-liposomes were controlled within 130 ± 20 nm, a range reported to optimize follicular transport while balancing penetration and retention (Xu et al., 2023). The highly negative zeta potential (−38.7 mV) suggests a stable colloidal dispersion, likely attributed to the adsorption of anionic timolol molecules or hydrolysis products of phospholipids onto the vesicle surface.

The sustained-release profile of TM-liposomes was confirmed by in vitro release studies (Fig. 1F). While the free TM solution achieved ∼90% release within 4 h, the TM-liposomes exhibited a controlled-release pattern, releasing only ∼50% at 4 h and reaching ∼80% cumulative release after 16 h. This demonstrates the ability of the liposomal carrier to prolong drug release and potentially reduce dosing frequency.

3.3. Physicochemical characterization of TM-lipogel and TM-gel

3.3.1. Rheological property

The rheological behavior of TM-gel and TM-lipogel was evaluated at 25 ± 0.1 °C. As shown in Fig. 2, both formulations exhibited a gradual decrease in viscosity with increasing shear rate, demonstrating typical shear-thinning behavior characteristic of pseudoplastic non-Newtonian fluids. This property is favorable for topical application, as the gels become less viscous upon rubbing (high shear) and recover viscosity once applied (static state).

Fig. 2.

Rheological property and in vitro release profiles. Viscosity curves of TM-gel (A) and TM-lipogel (D); Dynamic strain sweep curves of TM-gel (B) and TM-lipogel (E); Dynamic frequency sweep curves of TM-gel (C) and TM-lipogel (F); Cumulative release profiles (%) of TM-eyedrop, TM-gel and TM-lipogel. (G) in cellophane, (H) in 0.22 μm filter membrane, (I) in 0.45 μm filter membrane (n = 6).

Oscillatory stress sweeps were performed to determine the linear viscoelastic region (LVR). As shown in Fig. 2, the storage modulus (G′) and loss modulus (G′′) remained relatively constant within the strain range of 0.01% to 1%, indicating the structural integrity of the gel network within this range. Frequency sweep tests were subsequently conducted at a constant strain of 0.1% (within the LVR). Over the frequency range of 0.01 to 100 Hz, the G'values for both TM-gel and TM-lipogel were consistently higher than their corresponding G′'values. The dominance of the elastic modulus (G′ > G′′) indicates that the formulations possess a robust gel network structure, indicative of good physical stability and resistance to deformation during storage and handling.

3.3.2. Preliminary stability

As shown in Fig. S6, both formulations remained physically stable without phase separation or precipitation following centrifugation (10,610 ×g, 30 min) and three freeze–thaw cycles (−20 °C for 2 d ↔ 60 °C for 2 d). Furthermore, no significant changes in gel matrix integrity, pH, or drug content were observed after exposure to high temperatures (40 °C and 60 °C), high humidity (25 °C/75% RH and 25 °C/92.5% RH), or illumination (4500 ± 500 Lx) for 10 days (Tables S12-S14). These results indicate robust short-term stability of both formulations under stressed conditions.

The pH values of TM-gel and TM-lipogel were measured as 7.21 ± 0.04 and 7.29 ± 0.04, respectively. Although this pH is slightly alkaline relative to the established acid mantle of older infants, the skin's substantial buffering capacity rapidly modulates and restores pH homeostasis upon topical application (Soujanya et al., 2025).

3.3.3. In Vitro Drug Release

As shown in Fig. 2, all three formulations demonstrated favorable release profiles across the three tested permeable membranes (cellophane, 0.22 μm, and 0.45 μm filters). However, their release kinetics exhibited distinct differences. As a free solution without a polymeric matrix, TM-eyedrop exhibited the fastest release rate. It reached approximately 90% cumulative release within the first 4 h and quickly plateaued, reflecting the free diffusion characteristic of solution. TM-lipogel showed slower release than TM-gel during the first 6 h due to drug encapsulation within lipid bilayers, which delayed the initial burst. Between 6–12 h, TM-lipogel exhibited faster release, as liposome disintegration combined with gel matrix swelling enhanced drug diffusion. By 24 h, The cumulative release ratios of the two formulations converged, indicating comparable total release. However, TM-lipogel's more gradual kinetic profile is advantageous for maintaining sustained drug levels in transdermal delivery.

To clarify the release kinetics, the relationship between cumulative release (Q) and time (t) was fitted to the first-order, Higuchi, and Peppas diffusion models (Table S17). The best-fit model was determined by the correlation coefficient (R²): For TM-gel, the release curve showed the highest correlation with the Peppas model (highest R² value), indicating a release mechanism involving the combined effects of diffusion and gel erosion. For TM-lipogel, the release curve showed the best fit with the Higuchi model, suggesting that its drug release is governed by controlled slow diffusion (regulated by the lipid bilayer and gel matrix).

3.3.4. In Vitro Transdermal Permeability

The in vitro transdermal permeation profiles of TM-lipogel, TM-gel, and TM-eyedrop were evaluated using excised SD rat skin, a model characterized by a relatively thin stratum corneum and abundant appendages (Jung and Maibach, 2015). As shown in Fig. 3A, TM-lipogel exhibited significantly higher cumulative transdermal delivery over 24 h compared to other formulations. Specifically, the 24 h cumulative permeation of TM-lipogel (85.3%) was approximately 1.9-fold and 4-fold greater than that of TM-gel (45.1%) and TM-solution (21.3%). These findings demonstrate that liposomal incorporation significantly enhances timolol maleate penetration across the skin barrier. The enhanced permeation is likely attributed to the liposomal carrier's ability to fuse with or penetrate the stratum corneum, thereby facilitating drug transport. Both TM-gel and TM-lipogel showed comparable cumulative skin retention (∼18–23 μg·cm⁻²) across all four skin types, with no significant difference (p > 0.05). The comparable skin retention suggests that both formulations establish a similar reservoir of drug in the upper skin layers (stratum corneum + viable epidermis). However, the liposomal carrier enables greater partitioning into deeper dermal layers, likely via follicular targeting and the biocompatibility of the lipid bilayer, allowing more drug to reach the dermal–subcutaneous junction where the hemangioma is located.

Fig. 3.

(A) In vitro cumulative transdermal rates of 0.5% TM-solution, TM-gel and TM-lipogel in SD rat (n = 6). (B) Cumulative retention of TM-gel and TM-lipogel in the skin of animals (n = 6). (C) CLSM images of TM solution, TM-gel and TM-lipogel labeled with Rhod B at different times in KM mice (n = 3).

To elucidate the release kinetics, the relationship between cumulative permeation amount (Q) and time (t) was fitted to zero-order, first-order, and Higuchi models (Table 1). Based on the coefficient of determination (R²), the permeation profiles of both TM-gel and TM-lipogel adhered most closely to zero-order kinetics. This indicates a constant release rate, characteristic of a controlled-release matrix system, where the release is governed by the diffusion of liposomes through the gel network and subsequent drug release from the vesicles upon contact with the skin. Notably, the enhanced solubility and biocompatibility of TM-lipogel with the stratum corneum likely facilitated faster initial dissolution, thereby promoting efficient transdermal permeation.

Table 1.

Model fits for penetration of TM-gel and TM-lipogel in rat skin.

Permeation model	TM-gel			TM-lipogel
	Fitting equation	Rate constant (cm2·h)	R2	Fitting equation	Rate constant (cm2·h)	R2
zero-order	Q = 1.49 t - 1.59	1.49	0.99	Q = 3.46 t + 3.93	3.46	0.99
first-order	lnQ = 0.61 t − 1.27	0.61	0.90	lnQ = 0.10 t + 2.31	0.10	0.81
Higuchi	Q = 3.29 t1/2–3.78	3.29	0.97	Q = 20.49 t1/2–20.62	20.49	0.96

The difference in apparent release kinetics between the synthetic membrane and skin permeation studies reflects the change in rate-limiting barrier (Table S18). In the synthetic membrane studies, drug release is governed by diffusion within the formulation matrix, yielding Fickian (TM-gel, n ≈ 0.15) or near-Fickian (TM-lipogel, n ≈ 0.43) behavior that is well described by the Higuchi or Peppas models. In contrast, skin permeation is controlled by the stratum corneum barrier, which establishes a steady-state concentration gradient and results in an approximately constant flux (zero-order kinetics, R² = 0.99). The Korsmeyer-Peppas n-exponent for skin permeation (TM-gel: n ≈ 1.06; TM-lipogel: n ≈ 0.84) confirms Case-II transport characteristics, where relaxation and partitioning processes within the stratum corneum dominate the overall rate rather than matrix diffusion. This transition in rate-limiting step — from matrix diffusion to barrier penetration — is a well-recognized phenomenon in transdermal delivery research and explains the seemingly different kinetic models that provide the best fit in the two experimental systems (Siepmann and Peppas, 2011).

3.3.5. In vivo skin permeation

Confocal laser scanning microscopy (CLSM) was utilized to evaluate the transdermal profiles of TM formulations. As shown in Fig. 3C and Table 2, while all formulations exhibited strong fluorescence in the stratum corneum, the penetration depth and retention varied significantly among the three groups. The TM solution showed minimal dermal penetration, with fluorescence largely confined to the upper stratum corneum (≤75 μm) and negligible signal at 12 h. In contrast, the TM-gel facilitated rapid penetration into the dermis within 30 min, with dermal fluorescence intensity peaking at 4 h, but declining sharply by 12 h, likely due to metabolic clearance. The TM-lipogel exhibited a distinctly different profile. Although its dermal fluorescence intensity at 4 h was lower than that of the TM-gel, it continued to increase over time, reaching the highest intensity among all groups at 12 h. This sustained-release profile indicates that the liposomal carrier significantly prolongs drug retention in the dermis compared to conventional formulations.

Table 2.

Integrated optical density (IOD) of KM mice skin.

Groups	IOD/×106
	0.5 h	4 h	12 h
Rhod B-TM	0.09 ± 0.03	0.51 ± 0.02	1.37 ± 0.21
Rhod B-TM-gel	0.57 ± 0.16⁎⁎	8.13 ± 1.02⁎⁎⁎	1.87 ± 0.23⁎
Rhod B-TM-lipogel	1.03 ± 0.23⁎⁎	2.30 ± 0.43⁎⁎	7.80 ± 0.82⁎⁎⁎

Note: vs. Rhod B-TM group, *p < 0.05, **p < 0.01, ***p < 0.001.

2.5D reconstruction revealed distinct penetration mechanisms: The TM-gel formed a “plateau-like” distribution limited to approximately 200 μm, whereas the TM-lipogel exhibited a continuous gradient extending to approximately 400 μm. Furthermore, intense fluorescence was observed in hair follicles and sweat glands across all groups, confirming skin appendages as key shunt pathways.

3.4. In vivo safety studies

3.4.1. Histopathological examination

Histopathological evaluation of mouse skin was performed at 7 days post-administration to assess the biosafety of the formulations. As shown in Fig. 4, no overt histopathological changes were observed in the skin sections treated with TM-gel or TM-lipogel relative to both the untreated normal skin and the blank gel treated controls.

Fig. 4.

Representative H&E staining of KM mice skin after 7 days' administration. (A) Untreated skin (×200); (B) Untreated skin (×100); (C) Blank gel (×100); (D) TM-gel (×100); (E) TM-lipogel (×100). (①Stratum corneum; ②Stratum granulosum; ③Stratum spinosum; ④ Stratum basalis; ⑤Stratum dermatosum; ⑥Hair follicle).

Microscopic examination revealed no signs of stratum corneum hyperplasia, epidermal spongiosis or vacuolar degeneration, or dermal inflammatory cell infiltration. These findings indicate that the transdermal delivery systems maintained the structural integrity of the skin barrier and that the penetration enhancement process was not accompanied by significant tissue damage. This provides robust morphological evidence supporting the biosafety of the developed formulations.

3.4.2. Skin irritation studies

Single-dose test: An intra-animal comparative design was employed. Irritation was scored according to standard criteria at 0, 24, 48, and 72 h post-administration (Fig. 5). TM-lipogel exhibited negligible irritation on intact skin. On abraded skin, the observed mild erythema was attributable to mechanical trauma rather than chemical irritation, and the erythema resolved completely upon wound healing within 48 h.

Fig. 5.

Skin irritation of New Zealand rabbits and SD rats after single and multiple administration of TM-gel (A) and TM-lipogel (B).

Multiple-dose test: Repeated daily application resulted in exacerbated irritation compared to the single-dose test, presumably due to cumulative trauma from daily abrasion. Notably, rabbits exhibited greater sensitivity than SD rats, consistent with their known higher responsiveness to topical irritants (Jung and Maibach, 2015). Compared to the blank gel control, no irritation was observed on intact skin at any time point. On abraded skin, initial mild erythema was noted, but scores decreased to <0.49 (indicating negligible irritation) following wound healing.

3.5. In Vivo Antitumor activity

Murine xenograft hemangioma models were successfully established via subcutaneous injection of HemECs and HUVECs into the dorsal region of nude mice. All mice survived throughout the experimental period. The therapeutic efficacy was primarily assessed by the color change of tumor tissues (Wu et al., 2019): initial pink lesions progressively faded, and by day 30, many became pale or skin-colored, indicating regression characterized by the replacement of proliferated vessels with connective or adipose tissue (Fig. 6A).

Fig. 6.

In Vivo Antitumor Activity. (A) Changes in hemangioma size and color in model mice at different treatment points. (B) Body weight changes in model mice (⁻x ± s, n = 6). (C) Tumor volume changes every 3 days of each group (⁻x ± s, n = 6) (^⁎⁎⁎⁎p < 0.0001). (D) Weights of isolated tumors after administration of each group (⁻x ± s, n = 6) (^⁎p < 0.05, ^⁎⁎p < 0.01). (E) Organ index of model mice in each group (^⁎p < 0.05).

No behavioral abnormalities or skin irritation were observed in any group, and body weight increased steadily across all cohorts (Fig. 6B). In contrast to the rapid tumor expansion observed in the control group (>1000 mm³ by day 15), all treatment groups exhibited dose-dependent tumor growth inhibition (p < 0.0001) (Fig. 6C). The 0.25% TM-lipogel group showed only moderate inhibition, likely due to an inadequate intratumoral drug concentration. Quantitative analysis revealed that TIR for 0.5% TM-lipogel, 0.5% TM-gel, and 0.5% TM-eyedrop were 81.09% (p < 0.01), 61.69% (p < 0.01), and 29.85% (p < 0.05), respectively (Fig. 6D). The superior efficacy of 0.5% TM-lipogel over the conventional gel highlights the distinct advantage of the liposomal carrier in enhancing transdermal delivery.

Notably, the 1% TM-lipogel group exhibited a significantly reduced liver index compared to the control (p < 0.05)(Fig. 6E). According to established toxicological principles, an isolated liver index decrease without evidence of liver toxicity are regarded as adaptive rather than adverse per European Society of Toxicological Pathology consensus. Never the less, 0.5% was selected as the suitable formulation, offering a balance between maximal therapeutic efficacy and systemic safety. Consistent with the in vivo safety profile, H&E staining revealed no pathological changes in major organs (heart, liver, spleen, lung, or kidney) across treatment groups compared to the control (Fig. 7), confirming the systemic safety of the TM formulations.

Fig. 7.

H&E staining of heart, liver, spleen, lung and kidney of TM formulations.

3.6. Immunohistochemical Analysis of HIF-1α, VEGF and CD34

Immunohistochemical analysis was performed to elucidate the anti-angiogenic mechanism of TM. As shown in Fig. 8, TM inhibited hemangioma growth primarily through the downregulation of the HIF-1α/VEGF signaling pathway.

Fig. 8.

Results of immunohistochemical analysis. (A) Fluorescence microscopy images of HIF-1α and VEGF. (B) Fluorescence microscopy images of CD34-labled microvessels.

Quantitative analysis of relative expression levels revealed that HIF-1α and VEGF in the control group were (6.13 ± 2.18) × 10⁵ and (4.55 ± 1.96) × 10⁵, respectively. Treatment with 0.5% TM-gel significantly reduced these levels to (0.69 ± 0.55) × 10⁵ and (0.42 ± 0.13) × 10⁵, respectively. A more pronounced reduction was observed in the 0.5% TM-lipogel group, with expression levels of (0.45 ± 0.38) × 10⁵ for HIF-1α and (0.30 ± 0.19) × 10⁵ for VEGF (p < 0.01 vs. control).

To evaluate neovascularization, CD34 was used to label vascular endothelial cells and quantify microvessel density (MVD). The control group exhibited well-defined and densely distributed vascular structures, with an MVD of 709.09 ± 72.73 vessels/mm². In contrast, MVD was markedly suppressed in the treatment groups, decreasing to 178.79 ± 41.22 vessels/mm² in the 0.5% TM-lipogel group and 187.88 ± 68.57 vessels/mm² in the 1% TM-lipogel group (p < 0.001 vs. control). Notably, no significant difference in MVD was observed between the 0.5% and 1% TM-lipogel groups (p > 0.05), indicating that the 0.5% represents the appropriate therapeutic concentration.

3.7. Quantitation of MMP-2, MMP-9 and eNOS expression

Matrix metalloproteinases (MMPs), particularly MMP-2 and MMP-9, are zinc-dependent endopeptidases that degrade the extracellular matrix and vascular basement membrane, thereby facilitating angiogenesis and tumor invasion. Notably, these enzymes are highly expressed during the proliferative phase of hemangiomas (Zhao et al., 2022). Concurrently, endothelial nitric oxide synthase (eNOS) produces nitric oxide (NO), which mediates vasodilation and supports endothelial cell proliferation, migration, and survival (Storch and Hoeger, 2010). By amplifying pro-angiogenic signals such as VEGF, NO drives the rapid formation of vascular networks during hemangioma progression (Mabeta and Steenkamp, 2022).

ELISA quantification (Fig. 9) demonstrated that the protein concentrations of MMP-2, MMP-9, and eNOS were significantly downregulated in all treatment groups compared to the control. Notably, the 0.5% TM-lipogel group exhibited the most pronounced suppression of these targets, showing a significantly greater reduction compared to the TM-eyedrop and TM-gel groups (p < 0.01). This confirms that the liposomal gel formulation leads to superior inhibition of the proteolytic and vasodilatory pathways driving hemangioma growth. The concurrent downregulation of eNOS and MMP-9 suggests that TM disrupts both the signaling and remodeling phases of angiogenesis. Consistent with the reduction in microvessel density, no statistically significant difference was observed between the 0.5% and 1% TM-lipogel groups.

Fig. 9.

Expression levels of (A) HIF-1α, VEGF. (B) Microvessel density of all administrated groups. (C) Expression levels of MMP-2. (D) Expression levels of MMP-9. (E) Expression levels of eNOS. (ns = no significance, ^⁎⁎p < 0.01, ^⁎⁎⁎p < 0.001, ^⁎⁎⁎⁎p < 0.0001).

4. Limitations

Several limitations of the present study warrant consideration. First, while the HemEC/HUVEC xenograft model is a widely accepted infantile hemangiomas model, it lacks the full pathophysiological complexity of human infantile hemangiomas, particularly the GLUT-1-positive endothelial phenotype (North et al., 2000, North et al., 2001). Therefore, the observed therapeutic effects primarily reflect the relative pharmacodynamic potency of the formulations rather than definitive clinical cure rates. Second, the stability assessment was limited to short-term accelerated conditions. Although these stress tests demonstrated robust physicochemical stability, long-term stability data are still required to fully validate the shelf-life and storage conditions for clinical translation. Third, regarding the in vivo imaging, the absence of a covalent bond between Rhod B and timolol maleate raises the possibility of dissociation during transdermal penetration. Consequently, the fluorescence signal should be interpreted as a qualitative indicator of formulation distribution rather than a strict quantitative proxy for the drug itself. Nevertheless, since all formulations were prepared and imaged identically, the observed differences in fluorescence intensity and penetration depth remain valid for comparing the relative enhancement effects of the liposomal gel carrier.

5. Conclusion

This study successfully developed a 0.5% TM-lipogel as a promising transdermal delivery system for the treatment of superficial infantile hemangiomas. Compared to conventional TM eye drops, the liposomal gel demonstrated significantly enhanced transdermal delivery with deeper tissue penetration, and superior antitumor efficacy via the downregulation of key pro-angiogenic markers (HIF-1α, VEGF, eNOS, and MMP-9). Importantly, the optimized formulation exhibited an excellent safety profile, with no observable organ toxicity or skin irritation at the overloaded dose. The simple, ready-to-use gel design is expected to improve patient compliance, representing a child-friendly and effective alternative for the clinical management of superficial IH.

CRediT authorship contribution statement

Xiaoxing Hang: Writing – original draft, Visualization, Validation, Methodology, Investigation, Data curation. Taiwei Jin: Validation, Investigation, Data curation. Jun Hua: Resources, Investigation. Xuenong Zhang: Supervision, Resources, Investigation, Conceptualization. Wenlin Song: Writing – review & editing, Validation, Project administration, Funding acquisition, Formal analysis, Data curation.

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.