Customized flexible hollow microneedles for psoriasis treatment with reduced‐dose drug

Abstract Microneedles, especially hollow microneedles (HMNs), play an important role in drug delivery, but most of the current HMNs are manufactured based on silicon microfabrication (lithography, etching, etc.), which are slightly conservative due to the lack of low‐cost, batch‐scale and customized preparation approach, especially for the HMNs with flexible substrate. For the first time, we propose the use of a high‐precision 3D printed master mold followed by a dual‐molding process for the preparation of HMNs with different shapes, heights, and inner and outer diameters to satisfy different drug delivery needs. The 3D printed master mold and negative mold can be reused, thereby significantly reducing the cost. HMNs are based on biocompatible materials, such as heat‐curing polymers or light‐curing resins. The thickness and rigidity/flexibility characteristics of the substrate can be customized for different applications. The drug delivery efficiency of the fabricated HMNs was verified by the in situ treatment of psoriasis on the backs of mice, which required only a 0.1‐fold oral dose to achieve similar efficacy, and the associated side effects and drug toxicity were reduced. Thus, this dual‐molding process can reinvigorate HMNs development.

and the small holes will close within a certain period of time, which will then prevent further drug diffusion, so solid microneedle drug delivery efficiency is low. [23][24][25] Dissolving microneedles are made by mixing a drug with biodegradable polymer and curing so that the resulting microneedles exerts certain mechanical properties when piercing the skin. After piercing into the subcutaneous tissue, the tissue fluid dissolves the polymer, which in turn releases the drug. The needle tip is smaller in size and does not contain much of the drug, so its delivery efficiency is slightly improved. [26][27][28][29] HMNs can be understood as the miniaturization and arraying of traditional syringes and can be administered through an external syringe pump after piercing into the skin, and the microneedle drug delivery volume can theoretically reach the maximum tolerance of the skin, so the HMNs drug delivery efficiency and dose are higher compared to those of the first two, but their development and application are hindered because of the lack of a low-cost, batch-scale and customized manufacturing process. [30][31][32][33] Therefore, a new process is urgently needed to give new vitality to HMNs.
The main HMNs preparation processes are electroplating, 34 laser drilling, 35 deep reactive ion etching (DRIE), 36 etching, 37 evaporation 38 and lithography, 39 which are complex, time-consuming and costly, and the main materials used are silicon 40 and metals. 41 The former is brittle and has a risk of fracturing in the skin, and the biocompatibility of both has yet to be verified for practical application. [42][43][44][45] In addition, the appeal process is mostly a standard process with strict requirements for relevant parameters. Microneedle tip morphology is strongly consistent across different needles, such as conical shapes with all the same parameters, which is rather daunting for certain applications that require different morphologies and heights and lack certain flexibility. Recently, high-precision 3D printing technology has overcome these disadvantages, but its long printing time, high cost, and nonbiocompatibility make it difficult to be practically applied. [46][47][48][49] The dual-molding process is common in the preparation of dissolving microneedles and solid microneedles due to its simple operation. This process is difficult to apply for HMNs because they are hollow inside. To ensure the permeability of HMNs, there must be a vertical column protruding from the grooves of the negative mold. However, this column is a great challenge for the dual-molding process, and it is easy to break or bend the column during operation, so it is necessary to optimize the microneedle structure and negative mold selection.
For the first time, we propose a method using a dual-molding process after high-precision 3D printing, and this method combines the advantages of high-precision 3D printing, and the time-consuming problem is solved. We successfully prepared HMNs with different morphologies, heights and inner and outer diameters. The thickness and rigidity/flexibility characteristics of the substrate can be customized for different applications. When the 3D printed master mold and negative mold were prepared, the total time spent was just tens of minutes to hours depending on the material selected, significantly reducing the cost. To verify the usefulness of this method, psoriasis was generated on the backs of mice. The HMNs required only a 0.1-fold oral dose to achieve similar efficacy, and the associated side effects and drug toxicity were reduced (Figure 1a,b). Therefore, HMNs based on this dual-molding process have great application potential and research value.

| HMNs prepared based on a dual-molding process
The main HMNs structure consists of patches and microchambers ( Figure 1c), both of which are adhered to the same material. The HMNs preparation process involves three components: a highprecision 3D printed master mold, a negative mold with microcolumn arrays and grooves, and biocompatible HMNs. First, the HMNs were designed using Computer 3D software, and the files were exported to printing software, which sliced the files into layers every 10 μm in the Z-direction and printed them layer by layer. The 3D printing equipment used for the 3D printed master mold was provided by Shenzhen MoFang Co., Ltd. The material used was high-temperature resistant HTL yellowish low-viscosity resin, which is sensitive to UV light at 405 nm. The length-width-height of the 3D printed master mold was 19 mm Â 19 mm Â 3.4 mm (3.2 mm), and 25 can be printed at once with a sample interval of 1 mm within 20-30 h. The precision of the equipment was 10 μm, so the layer-by-layer morphology of the needle tip surface appeared, which had no effect on the preparation process and practical applications. Fluoride solution was soaked overnight to ensure that any uncured resin remaining in the pinhole was cleaned, followed by heating at 50 C for 2 h to ensure its mechanical properties. The 3D printed master mold may have hindered resin curing, 50 so a 2 μm thick film of Parylene C was deposited on the mold surface as a protective layer. A 3D printed master mold was placed into plastic petri dishes, and an Ecoflex/PDMS (E/P) mixture with a mass ratio of 10:1 was used as the negative mold material by comparing and calculating parameters, such as Young's modulus. The mold was vacuumed for 30 min until no bubbles emerged, and the liquid level was adjusted to approximately 5 mm above the needle tip and left for 10 min. The samples were heated at 80 C for 2 h to allow the E/P to finish curing. The blade was used to remove the additional E/P and then slowly uncovered along the four sides of the master mold to the edge of the array. Since the negative mold microcolumn array may break in the pinhole, both were immersed in sodium dodecyl sulfate solution (SDSS) for 10 min to reduce the adhesion between them, and slowly separated along the needle tip direction or vertically to form a negative mold with a microcolumn array and grooves. This step is repeated several times to obtain a large number of negative mold for batch manufacturing.
The negative mold was filled with liquid biocompatible materials until the liquid surface completely covered the microcolumn array and vacuumed for 10 min until no visible bubbles emerged. Residual bubbles were removed by a dropper, and then the liquid was slowly removed along the edge of the groove until the microcolumn array penetrated the liquid surface, the aspirated liquid can be reused and this step takes only tens of seconds to complete, and the remained liquid leave for 5 minutes to achieve natural leveling. Due to the surface tension, the liquid level was slightly higher at the edge of the groove and the microcolumn array, but this had no effect on the practical application. For the light-curing polymer, transparent resin with a wavelength of 405 nm, low viscosity and high stiffness was selected and cured by UV light irradiation for 10 min, which was immersed in SDSS solution and uncovered following similar operations to form a light-curing HMNs patch. The back interfaces were prepared in the same way, aligning and gluing them with the same resin, and thin tubes were used as channels to connect the HMNs to the syringe, forming HMNs on a rigid (resin) substrate. For the heat-curing polymers, nonphotosensitive polyimide (PI) with a low viscosity and flexibility was selected, and the curing conditions were 140 C for 5 h followed by 240 C for 5 h. The operations were similar except for the curing conditions, allowing HMNs to form on a flexible (PI) substrate, which reduces the cost compared to the traditional process.
Parameters such as HMNs shape, height, inner and outer diameter,

| Characterization of HMNs fabricated with a dual-molding process and biocompatible materials
Different needle tip shapes may be suitable for different scenarios, for example, square/cylindrical + cone has sharp tip with slightly smaller hole, which is less likely to be blocked after penetration and has less penetration force, and may be more suitable for surface drug delivery.
Rhombic and cylinder has a gentle tip with a slightly larger hole and slightly more penetration force, and may be more suitable for implant drug delivery. 51 Cylindrical and rhombic-shaped HMNs have slightly large internal diameters, gentle tips, and large penetration forces that may be more suitable for implantable drug delivery applications. The tip size of the 3D printed master mold, which has a high flexibility, can be designed according to the desired application. The material of the negative mold was E/P, which is commonly used for the preparation of various molds and has a certain heat resistance. For the negative molds, the out-of-plane height of the microcolumn array was 0.6 mm, which can be flexibly changed according to requirements. Due to the uniqueness of the process, the 3D printed master mold and negative mold can be used repeatedly to meet low-cost and batch require-  HMNs with less penetration force, and four types of HMNs with a height of 1.2 mm could be successfully penetrated by mutation point (penetration force), because psoriasis causes the skin surface to harden and thicken, HMNs with less penetration force is required to ensure adequate penetration. The penetration force of square/ cylindrical + cone is less than that of rhombic/cylinder because of its sharp tip and small penetration angle, and the penetration force of cylindrical + cone is slightly less than that of square + cone because the tip surface is probably smoother, but the difference is not significant. Psoriasis thickens the skin and makes it difficult for the needle tip to penetrate, so a shape with less penetration force should be chosen for the experiment (Figure 3a). Therefore, cylindrical + cone HMNs were prepared and used for all subsequent experiments (Figure 3b). The HMNs were used to perform  HMNs +0.2 mg/kg had the best therapeutic effect because of the moderate drug amount. HMNs +0.4 mg/kg is less effective than 0.2 mg/kg, possibly because the drug is already saturated and excess drug may not be effective and remain in the subcutaneous tissue, which in turn may cause further skin damage. The damaged skin was more likely to induce psoriasis after modeling the same night, resulting in slightly poorer efficacy. In the oral +0.5, 1, and 2 mg/kg MTX groups, the efficacy was better with increasing dosage. Because a portion of the drug will be degraded by the gastrointestinal tract after oral administration, the remaining portion will be absorbed into the body and transported throughout the whole body in circulation, and only a small portion of the drug will arrive at the onset site, so an efficacy similar to that of HMNs +0.2 mg/kg was achieved by oral +2 mg/kg. In addition, assuming a psoriasis onset area of 4 cm 2 , the HMNs required only 1 μg/cm 2 MTX to achieve a good therapeutic effect, much lower than the approximately 3.5 μg/cm 2 dissolving microneedles, 52  H&E staining, mastocyte counting, and kidney and liver sampling were performed in mice after euthanasia. H&E (Figure 5a 1 ,b) was used to measure the epidermal thickness, and it was seen that HMNs + 0.2 mg/kg and oral + 2 mg/kg had similar results to the normal group. Mastocyte counting (Figure 5a Table S1), which needed to be rated by some uninvolved person, was rated according to the visual changes of the skin, and the integer score of healthy to the most severe condition was from 0 to 4. This paper used three uninvolved persons for scoring, and their results were consistent. Relevant physiological analyses demonstrated the good efficacy of small HMNs doses, which also had fewer side effects, such as biological toxicity. Serum results (SI Appendix, Figure S5) were used as a quantitative analysis of liver and kidney function, and it can be seen that in terms of ALT, AST, BUN, CREA and O/P, HMNs+0.2 mg/kg always remained almost the same as the normal group, i.e., there was less of an impact on the liver and kidney function and good health. The kidney diagram (SI Appendix, Figure S6) shows the surface whitening in the model, reference and oral groups, but that of the oral group was the most obvious, which also indicated large amounts of body damage.

| DISCUSSION
With the rise in microneedle drug delivery, the microneedle preparation process has gradually become important. Compared with the preparation processes of solid microneedles and dissolving microneedles fabricated with a dual-molding process, the preparation process of HMNs has disadvantages of a higher production cost and time consumption and low flexibility because of processes including lithography, etching, DRIE and complex plating, which makes it difficult to realize low-cost and batch-scale production. Among the categories of microneedle drug delivery that have been industrialized, dissolving microneedle drug delivery has dominated due to the simple preparation process of dissolving microneedles.
For the first time, we prepared HMNs with different morphologies, heights, inner and outer diameters, substrate thicknesses and rigidity/flexibility values using a dual-molding process, which greatly enhances HMNs flexibility. The 3D printed master mold and negative mold are reusable, and the whole preparation process takes only tens of minutes to hours depending on the material selection with a wellprepared 3D printed master mold and negative mold, which greatly reduces the cost and time and has great application prospects.
The HMNs were made with biocompatible materials, such as light-curing and heat-curing materials, and the thickness and rigidity/ flexibility characteristics of the substrate can be customized for different applications. PI was chosen as the heat-curing flexible material, whose flexibility can be verified by bending and subsequent recovery at different angles. Resin was chosen as the light-curing rigid material.
The above two materials were used to validate the unique advantages of the process. Psoriasis hardens and thickens the skin, so a lightcuring rigid material was used for the experiments. This type of material was originally used for dental repair and has a high hardness that can easily pierce pig skin. To verify the recovery of the pinholes after puncture, the arms of three volunteers were used as experimental subjects, and the recovery of the pinhole within 30 min was observed by the change in DC resistance. In vivo fluorescence imaging verified that the drug could diffuse throughout the backs of mice within 10 h, demonstrating extremely high drug utilization. Water flow uniformly and consistently exited from each pinhole, verifying the stability of the process. Mechanical testing quantitatively verified that the HMNs were undamaged before and after use. HMNs based on the dualmolding process have the advantage of low cost and batch size, and can be prepared in different morphologies and array sizes depending on the site of drug delivery, and materials can be selected according to different degrees of damaged skin surfaces. The feasibility of the process was demonstrated by the use of HMNs based on light-curing resin for the treatment of psoriasis in mice, which has a large area of onset and causes the skin to thicken and harden until silver flakes appear. Psoriasis was generated on the backs of mice, and a treatment drug was administered in situ using HMNs for a 7-day experimental period, with treatment in the morning and modeling in the evening each day until day 8. Ten groups were used for the experiment after taking into account the effects of all variables. The efficacy of HMNs + MTX 0.2 mg/kg and oral + 2 mg/kg was similar and assessed by daily visual skin recovery, bilayer skin thickness measurements, body weight changes, PASI scores and so on, and the mice in the HMNs group had a higher body weight and better health than the mice in the oral group.

| Fabrication of HMNs systems
After the HMNs patch and the back connection structure were prepared, the two were aligned, and a layer of biocompatible resin was applied at the junction as glue. The two were placed under a UV lamp of 405 nm for 5 min to cure, and the excess was cut off with scissors and polished slowly with sandpaper. A TYD02-02 syringe pump was purchased from Baoding LEADFLUID Technology Company, and the speed was set to 5 μL/min. An infusion tube was attached to the HMNs system in the same way and used to connect the syringe pump. The total elapsed time for single mice was less than 5 min, with good efficiency.

| Recording and analyzing in vivo experiments
Vernier calipers were used to perform daily double skin thickness measurements on the dorsal surfaces of the mice, and the average was taken after measuring three different locations. An in vivo fluorescence imaging system comprising PerkinElmer's IVIS Lumina SRMS Series product line was used to observe the subcutaneous distribution of rhodamine B in mice. The mice were subjected to an orbital extraction of 1 mL of blood into anticoagulation tubes after removal of the right eyeball. The blood samples were centrifuged at 4 C and 3000 rpm, and serum (300 μL) was separated at high speed for 15 min before liver and kidney function indexes were measured. After blood collection, the mice were fixed in the supine position on a mouse plate, the thorax was quickly opened, excess connective tissue was separated and removed, the spleen was dissected, and the residual blood was gently aspirated from the surface with filter paper. The mice were weighed and photographed and then stored in a À80 C refrigerator. The redundant connective tissues of the mice were separated and removed, the liver was dissected, the residual blood was gently aspirated from the surface with filter paper, the largest lobe of the liver was fixed with 4% paraformaldehyde (PA) to prepare pathological sections, and the rest of the liver tissues were stored at À80 C in the refrigerator. The right kidney was dissected, and the residual blood was gently aspirated from the surface with filter paper. The kidney was cut crosswise, half of the kidney tissue was fixed with 4% PA, and pathological sections were prepared. The whole layer of skin on the dorsum of mice (2 cm Â 2 cm) was cut off and fixed with 4% PA to prepare pathological sections. The mouse spleen organs were removed intact and weighed on an electronic balance, and the spleen organ coefficients were calculated according to the formula: spleen organ coefficient = spleen mass (g)/body mass (kg). Pathological tissue changes: liver, kidney, skin behind the ear, and ear tissues were taken and fixed with 4% PA and then made into paraffin sections after dehydration, wax immersion, embedding, and the morphological changes of the tissues were observed by H&E staining. H&E staining was performed by taking paraffin sections of tissues and sequentially performing dewaxing, debenzene, hydration, hematoxylin staining, water washing, and color separation. After the sections appeared pink, were water washed and dyed, the sections were stained with ethanol and dehydrated from low to high concentration, after which the xylene was transparent, and finally the sections were sealed with neutral gum. Pathological sections were observed by randomly selecting five microscopic fields and photographing them at 4Â and 10Â magnification by technicians who were unaware of the experimental grouping to determine the damage to the liver and kidney tissues and the degree of skin keratinization and inflammatory cell infiltration. The PASI score is an important index for evaluating the severity of psoriasis in clinical and pharmacological studies and includes four aspects: erythema, scaling, infiltration and total score (i.e., erythema score + scaling score + infiltration score). Erythema, scaling and infiltration were scored as 0, 1, 2, 3, and 4, respectively, according to the severity of psoriasis development, which corresponded to the symptomatology of none, mild, moderate, severe and extremely severe. This scoring was evaluated by three students with different specialties, and the mean value of each index was obtained from the results. A trend line was drawn, thus monitoring the severity of psoriasis in each group of mice.

| Statistical analysis
Differences between two independent groups were calculated using t test. p values less than 0.05 were considered statistically significant and are denoted as follows: ns, no significance, *p < 0.05, **p < 0.01, and ***p < 0.001. All data are plotted as the means ± SDs. All statistical analyses were performed using Origin 2020b (Origin Lab Corporation).