3.1 Fabrication of dissolving microneedles
We identified four materials-related criteria to make microneedles for self-administration of biotherapeutics from a minimally invasive patch: (1) gentle fabrication to avoid damaging sensitive biomolecules, (2) sufficient mechanical strength for insertion into skin, (3) controlled release for bolus and sustained drug delivery, and (4) rapid dissolution of microneedles made of safe materials. Guided by these criteria, we selected two polysaccharides – i.e., carboxymethylcellulose and amylopectin – because they are biocompatible materials with a history of use in FDA-approval parenteral formulations [18
], are expected to be mechanically strong due to their relatively high Young’s modulus [20
], and are highly water soluble for rapid dissolution in the skin [22
Building off our previous microneedle fabrication methods [11
], we fabricated dissolving microneedles using a micromolding approach that faithfully reproduces microneedle structures in an economical manner suitable for scale up to mass production. Female master-molds were first prepared out of SU-8 photoresist by lithography and used to created PDMS male master-structures, shown in . These master-structures were then molded to make PDMS female molds. PDMS was chosen as the material for master-structures and molds because of its ability to conformally coat microstructures and fill micromolds; its poor adhesion and flexibility to facilitate separation of microstructures from micromolds; and its low cost.
Figure 1 Dissolving microneedles for transdermal drug delivery. (a) Microneedle master-structure (600 μm in height and 300 μm wide at base) used to mold dissolving microneedles made of (b) CMC, (c) amylopectin and (d) BSA. The master-structure (more ...)
These micromolds were used to prepare dissolving microneedles by solvent casting with aqueous solutions of CMC and amylopectin. However, simply filling molds with CMC solution and then drying produced weak needles, probably due to structural voids left within the microneedle matrix after water evaporation (data not shown). To avoid this problem, we developed a modified casting method in which the CMC solution was first concentrated by evaporation under vacuum (i.e., -50 kPa) or heating (i.e., 60–70° C) to produce a highly viscous solution that minimized water content, but was still fluid enough to fill the mold. We found that an aqueous CMC solution with a viscosity of 4.5x105 cP (measured with a Couette viscometer at 1/s shear rate at 23° C) met these criteria. In case of amylopectin, the initial solvent removal was carried out at elevated temperature (i.e., 60–70° C) rather than just under vacuum, because amylopectin has poor water solubility at room temperature.
Concentrated CMC and amylopectin solutions were then cast into micromolds and dried completely during centrifugation at 37° C. The elevated temperature increased the speed of evaporation and the centrifugation continuously compressed the mold contents, which minimized void formation during drying. This modified casting method was effective to reproduce polysaccharide microneedles having the same dimensions as their master-structures, as shown in for CMC and amylopectin microneedles, respectively. A similar approach was used to make microneedles out of BSA in , which is a model for making needles out of pure drug, rather than encapsulating drug within a polysaccharide matrix, as discussed below.
As an alternative approach, we tried using high viscosity CMC (1.5 – 3 x 103 cP for a 1% aqueous solution at 25° C) as the matrix material, but found that it required much more water to be solubilized compared to the ultra-low viscosity CMC used above. As a result, high viscosity CMC took longer to dry and produced deformed microneedles that shrank substantially during drying and were mechanically weak (data not shown).
As discussed below, different drug delivery scenarios were addressed by selectively encapsulating model compounds within microneedles, within the microneedle device backing layer, or within both. To encapsulate within the CMC or amylopectin matrix, we simply mixed the model drug into the polysaccharide solution before casting into the molds. To selectively encapsulate within the microneedles and not in the backing layer, a smaller volume of drug-polysaccharide solution was cast into the holes of the micromold to form microneedles. After wiping off excess solution from the micromold surface, polysaccharide solution without model drug was cast onto the micromold and dried.
To selectively encapsulate within the backing layer and not in the microneedles, a similar two-step process was carried out, in which the model drug was only added to the polysaccharide solution applied to the micromold during the second step. Drying of the complete, integrated system or just the backing layer during the second step required 1–2 h, whereas drying of just the microneedles during the first step took approximately 30 min. These process times varied depending on materials and processing conditions.
3.2 Mechanical properties of dissolving microneedles
The design of dissolving microneedles is governed by a number of interdependent materials and fabrication constraints, one of which is the need for microneedles to have sufficient strength to insert into skin without mechanical failure. We therefore measured and simulated microneedle mechanical properties as a function of microneedle material composition and geometry, and then imaged insertion of optimized microneedles into skin.
3.2.1 Measurement of microneedle failure force
We first measured the mechanical behavior of CMC microneedles with a conical shape. As shown in by the black circle data points, the force-displacement curve (which is analogous to a stress-strain curve) exhibited an initial increase in force with displacement, followed by a discontinuity at a force of approximately 0.1 N/needle. This is interpreted as the point of microneedle failure, which is consistent with previous studies [14
]. Moreover, microscopic examination of the microneedles showed little deformation before this failure point and showed microneedles bent up to 90º starting approximately half way up the shaft after this failure point, which is consistent with failure by buckling (data not shown).
Figure 2 Mechanical behavior of dissolving microneedles. Force measured as a function of microneedle displacement while pressing against a rigid surface for (a) CMC and PLA microneedles having conical and pyramidal geometries and (b) pyramidal microneedles made (more ...)
For comparison, we generated a similar curve for PLA microneedles having the same geometry, which demonstrated a five-fold greater failure force of 0.5 N/needle (black diamonds in ). Previous work showed that conical PLA microneedles similar to those used in this study have a failure force more than 3 times greater that the force needed for insertion into the skin, which indicates that these conical PLA microneedles are suitable for skin insertion without breaking [14
]. Given that the conical CMC microneedles are 5 times weaker than their PLA counterparts, this analysis suggests that the conical CMC microneedles are too weak to insert into the skin.
Because microneedle geometry affects mechanical strength, we next examined pyramidal microneedles made of CMC and PLA. In contrast to conical microneedles, pyramidal microneedles did not show a distinct transition point indicating failure over the range of conditions tested. Microscopic examination of pyramidal microneedles showed a progressive deformation of the microneedles, starting near the tip and moving downward with increasing force, but never showed a catastrophic buckling event at a single point of failure (data not shown). This progressive deformation is consistent with the continuous force-displacement curve shown in . The reason for the different behaviors of conical and pyramidal microneedles may have to do with the larger aspect ratio and the smaller cross-sectional area of conical microneedles, as discussed below.
To further study the effect of microneedle composition on mechanical strength, the mechanical behavior of pyramidal microneedles having the same geometry was measured for microneedles made of CMC, PLA, amylopectin, a 20/80 wt% mixture of BSA and CMC, and 100% BSA. As shown in , these five pyramidal microneedles all showed similar mechanical behavior, although the choice of material influenced microneedle strength (i.e., amount of deformation). The materials can be ranked from strongest to weakest as: PLA, amylopectin, CMC/BSA, BSA, and CMC. Amylopectin microneedles were stronger than CMC microneedles, which can be explained by the higher Young’s modulus of amylopectin (4.5 GPa [21
]) compared to CMC (1 GPa, See section 2.2.3). CMC and CMC/BSA microneedles were designed to simulate a CMC microneedle encapsulating a model protein and a microneedle made completely of a model protein, respectively. These two microneedles designs had similar mechanical strength, both of which were greater than for pure CMC microneedles. In this case, encapsulation of protein increased microneedle mechanical strength, but this is unlikely to be true in all cases.
3.2.2 Simulation of microneedle failure force
To better interpret these experimental results, we simulated mechanical behavior of microneedles to predict critical buckling load. As shown in , CMC microneedles with a conical geometry (800 μm length and 200 μm base diameter) have a predicted failure force of 0.10 N and PLA microneedles with the same geometry have a predicted failure force of 0.51 N, which is in excellent agreement with experimental measurements (). The pyramidal microneedles (600 μm length, 300 μm base width) made of CMC and PLA have predicted failure forces of 1.8 N and 8.9 N, respectively (). The 18-fold increase in critical buckling load for these pyramidal microneedles compared to conical microneedles is also consistent with experimental measurements. However, this model accounts only for buckling and does not account for the progressive deformation observed experimentally at smaller forces.
Simulated critical buckling load, Pcr, of CMC and PLA microneedles as a function of microneedle shape, length, and base width/diameter.
The above comparison involved longer and thinner conical microneedles versus shorter and wider pyramidal microneedles. To make a comparison that isolates the effect just of microneedle shape, failure force for microneedles of 600 μm length and 300 μm base width/diameter was predicted to be 0.93 N and 4.7 N for conical microneedle made of CMC and PLA, respectively, which is almost two-fold smaller than the corresponding predictions for pyramidal microneedles (). We therefore conclude that pyramidal microneedles are stronger, probably due to their larger cross-sectional area at the same base width/diameter.
Examination of for each microneedle design as a function of base width/diameter also shows that increasing base dimensions (i.e., decreasing aspect ratio) increases needle strength. Thus, using pyramidal microneedles with a small aspect ratio can provide added mechanical strength for mechanically weak biomaterials like CMC. However, microneedles with an aspect ratio that is too small will also have poor insertion due to fabrication difficulties to make a sharp tip and insertion difficulties to force the rapidly widening needle shaft into the small hole made in the skin by the needle tip.
3.2.3 Microneedle insertion into skin
Guided by the expectation that pyramidal CMC microneedles with an aspect ratio of two should be strong, we inserted arrays of these needles into pig cadaver skin and found that 100-needle arrays of microneedles were inserted reliably into the skin using the gentle force of a thumb. The backside of a representative microneedle array made of transparent CMC is shown in with its microneedles embedded in the skin. After removing the microneedles from the skin after just 3 s, the tips had already begun to dissolve (), indicating onset of rapid dissolution in the skin. We next treated the skin with a dye that selectively marks sites of skin penetration and found that typically all microneedles in the array inserted into the skin ().
Figure 3 Imaging microneedle insertion into pig cadaver skin. (a) View of the back side of a CMC microneedle patch applied onto the surface of the skin. (b) CMC pyramidal microneedles after insertion into the skin for 3 s. (c) Skin stained with tissue marking (more ...)
Histological examination of skin pierced with microneedles showed penetration depths of approximately 150 – 200 μm, which corresponded to insertion across the stratum corneum and viable epidermis and into the superficial dermis (). Microneedles used in this experiment measured 600 μm in length, which means that one-fourth to one-third of the microneedle shaft penetrated into skin. This can be explained by deformation of skin’s surface that is known to occur during microneedle insertion due to skin’s viscoelasticity [23
]. The relatively wide base (i.e., 300 μm) and small aspect ratio (i.e., 2) of the pyramidal microneedles contributed to this incomplete insertion. Further optimization of microneedle geometry, such as aspect ratio, tip sharpness, and spacing between microneedles, and microneedle material may increase depth of insertion. However, as discussed below, partial microneedle insertion may be adequate for drug delivery strategies presented in this study.
3.3 Release of model drugs from dissolving microneedle patches
By loading model drug into dissolving microneedles in different ways, we were able to design systems that achieved either bolus or extended release from a microneedle patch. To achieve bolus release, model drug was selectively incorporated into the microneedles themselves and not into the backing layer. In this way, we hypothesize that microneedles can be inserted into skin and release encapsulated drug during their rapid dissolution. The rate of release in this scenario is controlled largely by microneedle dissolution rate. A limitation is that the total dose administered is small, because microneedles each contain about 25–60 μg of matrix material and typically just a fraction of the microneedle matrix can made of drug in order to maintain microneedle mechanical strength. Thus, bolus delivery from a microneedle patch containing a few hundred microneedles is likely to be limited to less than 1 mg of drug.
To administer larger drug doses as an extended release over at least hours, we incorporated model drug into both the microneedles and backing layer or, alternatively, just the backing layer. This permits much larger doses to be administered, because the backing layer can be large (e.g., 10 – 100 mg) and can be loaded with larger fractions of drug, because backing layer mechanical properties have fewer constraints. In this scenario, we hypothesize that drug can diffuse over time from the drug reservoir in the backing layer and into skin through transdermal pathways created by dissolving microneedles. In this way, the backing layer acts as a drug source similar to a conventional matrix-design transdermal patch.
3.3.1 Bolus Release
To test our hypothesis regarding bolus release, we selectively encapsulated a model drug, sulforhodamine B, in pyramidal CMC microneedles. As shown in , red-colored sulforhodamine was encapsulated within each microneedle, but the bottom portion of each microneedle and the backing layer did not contain sulforhodamine. After inserting sulforhodamine-loaded microneedles into pig cadaver skin and then removing them after 5 min, inspection of the skin surface showed an array of red spots corresponding to the sites of each microneedle insertion (). These spots could not be wiped off by cleaning the skin surface and are therefore interpreted as sulforhodamine deposited within skin after microneedle dissolution.
Figure 4 Dissolving microneedles for bolus delivery into skin. (a) CMC pyramidal microneedles encapsulating sulforhodamine B within the microneedle shafts, but not in the backing layer. (b) Skin surface showing sulforhodamine delivered into the skin by insertion (more ...)
This interpretation is confirmed by histological sections, which show deposition of sulforhodamine within skin at sites of microneedle penetration (). Microneedle insertion depth was approximately 150–200 μm, which is in agreement with . The width of each hole was approximately 100 μm (), which is similar to microneedle width at a distance of 150 to 200 μm up the shaft from the tip. To supplement this information, shows a lower-magnification histological section of skin 1 h after insertion of bolus-delivery microneedles. In this case sulforhodamine is not located just at sites of microneedle insertion, but has diffused more extensively within the skin.
To generate a better understanding of the kinetics of bolus release from dissolving microneedles, we imaged the microneedle dissolution process after insertion into skin for different times. The tips of microneedles dissolved within 10 s (), half of the microneedle height disappeared within 1 min (), and two-thirds disappeared within 15 min (). After 1 h, microneedles were fully dissolved (). This kinetics could be altered by changing microneedle geometry and matrix material. For example, we observed that similar microneedles made of amylopectin dissolved more slowly and ones made of polyvinylpyrolide dissolved more quickly based on their different levels of water solubility (data not shown). It is also worth noting that even though microneedles did not penetrate to their full length into the skin, they were nonetheless able to fully dissolve, probably due to transport of interstitial fluid from the skin up the needle shaft, as discussed below.
Dissolution kinetics of microneedles after insertion in skin. (a) CMC pyramidal microneedles imaged by brightfield microscopy before insertion and (b) 10 sec, (c) 1 min, (d) 15 min, and (e) 1 h after insertion into pig cadaver skin.
3.3.2 Sustained Release
To test our hypothesis regarding sustained release, we encapsulated sulforhodamine in the backing layer and shafts of pyramidal CMC microneedles (). The microneedle device contained 1 mg of sulforhodamine at a concentration of 10 wt% (on a dry basis). These microneedles designed for sustained release could be inserted into skin () and histological examination showed release of sulforhodamine throughout the skin ().
Figure 6 Dissolving microneedles for sustained release. (a) CMC pyramidal microneedles encapsulating sulforhadamine only in the backing layer. (b) Skin surface showing sulforhodamine delivered into the skin by insertion of the microneedles shown in part (a) for (more ...)
To quantify sustained release properties in greater detail, we inserted microneedle patches into human cadaver skin and measured transdermal flux. Sulforhodamine release from CMC microneedle patches exhibited an initial lag time of a few hours, followed by steady release for approximately one day (). Similar behavior was seen for microneedle patches made of amylopectin, but with slower kinetics. In this case, lag time was longer and release took place over a few days ().
Figure 7 Transdermal release profile from dissolving microneedles patches. (a) Cumulative release of sulforhodamine encapsulated at 10 wt% in the pyramidal microneedles and the backing layer of patches made of CMC and amylopectin. (b) Cumulative release during (more ...)
These data validate the hypothesis that drug encapsulated within the backing layer of a microneedle patch can diffuse out of the patch and into skin. Moreover, they show that changing microneedle patch matrix material can alter release kinetics. It is important to be able to vary release kinetics based on patch design, because different drugs administered for different indications require different release patterns.
Release rate should also depend on sulforhodamine concentration in the patch. Consistent with this expectation, the drug release rate from a patch containing 30 wt% sulforhodamine was approximately three times greater than a patch containing 10 wt% sulforhodamine ().
3.4 Protein stability after encapsulation in dissolving microneedles
Dissolving microneedles were designed to encapsulate sensitive biomolecules using a gentle fabrication process. To assess success of this design, we used lysozyme as a model protein and measured changes in secondary structure and enzymatic activity after encapsulation and storage in CMC microneedle patches.
Circular dichroism (CD) analysis of untreated lysozyme compared to lysozyme encapsulated within a microneedle patch and then released by dissolution in water showed no detectable change in protein secondary structure (). Even after storage of microneedle patches containing lysozyme for 2 months at room temperature, protein structure was unchanged (). As a positive control, the CD spectrum showed extensive degradation of secondary structure after thermal denaturation ().
Figure 8 Protein stability after encapsulation and release from dissolving microneedles. (a) Circular dichroism spectrum of untreated lysozyme (negative control); lysozyme encapsulated in CMC microneedles and released by dissolution in PBS; lysozyme encapsulated (more ...)
To further test lysozyme integrity, enzymatic activity of lysozyme was measured. To make sure that the presence of dissolved CMC after microneedle dissolution did not create an artifact, a CMC microneedle containing no lysozyme was dissolved in PBS and then mixed with untreated lysozyme. This resulted in no change in lysozyme activity (Student’s t-test, p=0.51). To test the effect of encapsulation, microneedles containing encapsulated lysozyme were dissolved in PBS and found to have no loss of enzymatic activity compared to untreated enzyme (Student’s t-test, p=0.28). After two months of storage, lysozyme released from microneedles retained 96% enzymatic activity, indicating a small loss of activity (Student’s t-test, p=0.03).