Chemicals

Poly(methylvinylether-co-maelic anhydride) (Gantrez^® AN-139) was a gift from Ashland, (Kidderminster, UK). Poly(ethyleneglycol) 10,000 daltons and the drug substances caffeine, theophylline and 7β– hydroxyethyltheophylline (7-BHT) were obtained from Sigma-Aldrich, Poole, Dorset, UK. Glucose for in vitro and in vivo studies was obtained from VWR^® (Geldenaaksebaan 464, 3001, Belgium) and Thornton & Ross Ltd. (Huddersfield, HD7 5QH, England), respectively. Isoflurane (Isoflo^®) was obtained from Abbott Laboratories, Illinois, USA. All other chemicals were of analytical reagent grade.

Preparation of hydrogel-forming MN

Aqueous blends containing hydrolysed PMVE/MA (11.1%) and PEG 10,000 (5.6%) were utilized to fabricate MN arrays (19×19 array, 600 μm in height, a base width of 300 μm and interspacing of 50 μm) by using laser-engineered silicone micromould templates [20–22]. MN were crosslinked (esterification reaction) by heating at 80°C for 24 h and sidewalls removed using a heated blade [23–25] (Fig 2). Formulation and mechanical properties of such hydrogel-forming MN have previously been reported [20–22, 26].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

(A) Schematic representation of casting and crosslinking of hydrogel film formulations. (B) Diagrammatic representation of steps involved in the preparation of polymeric MNs. Polymer matrix was transferred to the silicone mould (i). The mould was centrifuged at 3000 rpm for 15 minutes (ii). Upon drying and heating for 24 h at 80°C to induce ester-based crosslinking, the silicone mould was carefully peeled away from the polymeric MN array and side walls removed using a hot scalpel blade (iii). Digital photograph image of MN with side walls (iv). Digital image of MN after removing the side walls using hot scalpel blade (v). (C) Illustration of the modified Franz cell apparatus used to investigate MN uptake of analytes across excised dermatomed neonatal porcine skin in vitro.

https://doi.org/10.1371/journal.pone.0145644.g002

Optical coherence tomography

Optical coherence tomography (OCT) was used to visualise the insertion and swelling of the MN in the skin of healthy human volunteers, as described previously [27].

In vitro extraction and detection using MN arrays

In vitro MN-mediated extraction and detection of theophylline, caffeine and glucose was carried out across dermatomed neonatal porcine skin (300–400 μm thickness, shown previously to be a suitable skin model for prediction of in vivo performance of MN) [26] by using a modified Franz-cell setup (Fig 2). MN were applied with a force of 11 N per array, as previously reported [22]. In vitro fluid uptake by MN arrays was studied using Franz cell receptor chambers thermostated to 37°C and containing varying physiologically relevant concentrations of theophylline, caffeine and glucose in phosphate buffered saline pH 7.4, with MN arrays inserted and removed at defined intervals.

In vivo MN extraction and detection of theophylline

Immediately after application of MN to shaved, hairless skin (Smooth Care^®, Hair Removal Cream, Boots Company PLC, Nottingham, UK) on the back of each rat, a theophylline solution of 2 mg/mL concentration was administered to the anaesthetised Sprague Dawley^® rats via oral gavage. Volume of the theophylline solution administered to each rat was based on their individual weight, to achieve a dose of 5 mg/kg or 10 mg/kg. Following application, MN arrays remained in situ for 1 h. Blood samples of 200 μl were collected at pre-defined time intervals by lateral tail vein puncture (Microvette^® CB 300 LH tubes, Sarstedt AG & Co. Nümbrecht, Germany). All animal experiments throughout this study were conducted according to the policy of the Federation of European Laboratory Animal Science Associations (FELASA) and The European Convention for the protection of vertebrate Animals used for Experimental and Other Scientific Purposes, with implementation of the principle of the 3Rs (replacement, reduction, refinement).

Human Volunteer Studies

All work undertaken in this study received approval from the School of Pharmacy Research Ethics Committee in Queen’s University Belfast (Study number: 017PMY2012). Fully-informed written consent was obtained from each volunteer prior to their enrolment in the study.

Extraction of theophylline and caffeine from hydrogel-forming MN arrays

Following in vitro and in vivo application and removal, all MN were transferred to a sealed, sterile glass vial. Extraction was performed by adding 1 mL HPLC water to each bottle, vortexing for 1 min, then allowing to stand for a further 1 min, at which point all solution in the vial was removed by pipette and transferred to a 1 mL Eppendorf tube. Following centrifugation at 13,000 rpm for 10 min by a SIGMA^® 2-16K Centrifuge (SciQuip Ltd. Shropshire, UK) 100 μl of supernatant was removed by pipette and transferred into an auto-sample HPLC vial from which 20 μl was injected into the HPLC column for analysis.

Preparation and extraction of theophylline from rat blood

For HPLC method validation, blank rat blood samples were obtained and, following centrifuge separation, 20 μl of a theophylline solution of increasing concentration was added to 180 μl of blank plasma. Two consecutive steps were carried out using acetonitrile (ACN) for theophylline extraction. Specifically, following addition of ACN, samples were vortex mixed for 10 min, centrifuged at 13,000 rpm for 10 min at 4°C, and the supernatant collected in disposable glass culture vials. Extract were dried under a stream of nitrogen at 35°C for 50 min using a Zymark TurboVap^® LV Evaporator Workstation (Biotage, Uppsala, Sweden) and the residue reconstituted in 200 μl of water, vortex mixed for 3 s and centrifuged at 14,000 rpm at room temperature for 10 min. The supernatant was filtered using a 0.2 μm cellulose acetate filter coupled to a 1 mL syringe and transferred into an auto-sample vial from which 20 μl was injected on to the HPLC column for analysis.

Preparation and extraction of caffeine from dried blood spots

A stock solution of caffeine was prepared by dissolving 50 mg in 100 mL of HPLC grade water to produce a 500 μg/mL concentration. This was further diluted with water to produce serially diluted working solutions of 160, 120, 80, 60, 40, 20, 10, 5, 2.5 μg/mL, 50 μl of which was added to 950 μl of fresh whole human blood, obtained from volunteers with fully-informed consent. These were then mixed using a Stuart^® SB2 fixed speed rotator with Stuart^® tube holders (Bibby Scientific, Staffordshire, UK) for 45 min to give final blood concentrations of 8, 6, 4, 3, 2, 1, 0.5, 0.25, 0.125 μg/mL. Drug free blood without caffeine was used as blank. DBS were prepared by pipetting 15 μl of blood onto Guthrie^® cards. These were left to dry at room temperature overnight before extraction. Validation was confirmed using spiked concentrations of 7, 2.5, 0.3 μg/mL. For the extraction process, an 8 mm disc was punched out from the centre of DBS samples and transferred to an Eppendorf tube containing 500 μl of a solution of 0.5 μg/mL internal standard (IS) 7-BHT in 100% methanol. The tubes were then vortexed using a Fisons Whirlmixer (Rigal Bennett, Whitley, UK) for 60 min. Following centrifugation at 13,000 rpm for 10 min by a SIGMA^® 2-16K centrifuge the supernatant was transferred to a glass vial and evaporated under a stream of nitrogen at 50°C for approximately 60 min using a Zymark TurboVap^® LV Evaporator Workstation (Biotage, Uppsala, Sweden). Following evaporation, the residue was reconstituted in 100 μl of HPLC water and dissolved via slow vortexing (400 rpm) for 1 min. The resulting solution was then transferred into an auto-sample HPLC vial from which 20 μl was injected onto the HPLC column for analysis.

Analysis of analyte uptake by MN

Theophylline and caffeine MN samples were quantified using an isocratic HPLC method (Agilent Technologies 1200 Series, Stockport, UK). The separation was performed on an XSelect^® HSS C18 (4.6 mm x 100 mm, 3.5 μm) analytical column (Waters Ireland, Dublin) protected with a 20 mm XSelect^® guard cartridge of identical chemistry. The column was thermostated at 25°C. The mobile phase consisted of 72.5% 10 mM ammonium acetate buffer and 27.5% methanol at a flow rate of 1.0 mL/min. Sample injection volume was 20 μl. UV detection was performed at 273 nm. The chromatograms obtained were analysed using Agilent ChemStation^® Software B.02.01. Least squares linear regression analysis and correlation analysis were performed on the calibration curve produced, enabling determination of the equation of the line, its coefficient of determination and the residual sum of squares (RSS). To determine lower limit of detection (LoD) and lower limit of quantification (LoQ), an approach based on the standard deviation of the response and the slope of the representative calibration curve was employed, as described in the guidelines from ICH [31]. Specifically, the linearity of the method was established by constructing calibration curves for both theophylline and caffeine on three consecutive days. Plots of peak area against analyte concentration were used. The slope, the intercept and the correlation coefficient of each calibration curve were determined. The following equation was then used to estimate the LoD for each compound [31]:

Similarly, the LoQ was estimated using the following equation: where σ = the standard deviation of the response and S = the slope of the calibration curve. The accuracy and precision of the method were determined from analysis of samples at three concentrations representing the low, medium and high portions of the standard curves (LQC, MQC and HQC). The mean accuracy (% RE) and precision (% CV) being within 15% of the actual value, while the LoQ which should not deviate by more than 20% [31]. ISF glucose uptake by MN arrays was determined using an Abcam^® Assay Kit (Cambridge, UK) and read by a FLUOstar OPTIMA micro-plate reader (Allmendgruenresor, Ortenberg, Germany) at 544 nm and 590 nm excitation and emission wavelengths, respectively.

Analysis of blood samples

Theophylline and caffeine blood samples were analysed using the same conditions as MN extract analysis, with the exception of variation of the mobile phase to 85% 10 mM ammonium acetate buffer and 15% methanol. The adjustment was made to allow better separation of the chromatographic peaks.

Statistical analysis

Where appropriate, statistical analyses to compare results were performed using a two-way analysis of variance (ANOVA). In all analyses, p < 0.05 denoted statistical significance. Post-hoc statistical comparisons of the means of individual groups were performed using Bonferroni’s post-test. The Statistical Package GraphPad Prism^® Version 5.03 (GraphPad Software Inc., CA, USA) was used for all statistical analysis.

Validation of Analytical Methods

A HPLC method was optimised for the detection of both theophylline and caffeine in both water and plasma samples by simple variation of mobile phase previously used in similar HPLC procedures [44, 45]. The validation of theophylline and caffeine analytical methods was performed in accordance with ICH guidelines [31]. Upon validation, the LoD and LoQ of theophylline in water was found to be 0.009 and 0.028 μg/mL, respectively, while the LoD and LoQ values of theophylline in plasma samples was 0.19 μg/mL and 0.57 μg/mL, respectively. As the therapeutic range of theophylline in human plasma is 10–20 μg/mL, the LoD and LoQ values reported using these methods display a high degree of sensitivity. Furthermore, these values compare favourably to previous HPLC studies analysing theophylline in plasma, in which higher detection and quantification limits around 0.50 μg/mL were reported [46, 47]. Using ICH guidelines, the LoD and LoQ of caffeine in water was found to be 0.013 μg/mL and 0.039 μg/mL, respectively while, following extraction from DBS, caffeine LoD and LoQ of 0.24 μg/mL and 0.73 μg/mL, respectively, were obtained. These findings are similar to those previously reported in a study investigating DBS therapeutic caffeine monitoring in neonates, where the LoQ was found to be 0.50 μg/mL [48].

For glucose measurements, the fluorescence intensity was measured using the fluorescence plate reader and glucose concentration calculated by comparison with a standard curve for glucose ranging from 0–1600 nmol/mL. The results obtained using this analytical method are in keeping with those reported in similar studies which analysed caffeine [48], theophylline [49] and glucose [50] using HPLC and liquid chromatography-mass spectrometry analysis.

OCT Analysis

Previous studies have demonstrated the potential of OCT as a tool to study MN penetration across the stratum corneum and to subsequently visualise the dissolving/swelling pattern of MN [22, 51, 52]. The sharpness of our MN allowed them to penetrate the stratum corneum of the forearm of human volunteers, as shown in the OCT images at the time of insertion (Fig 3C). However, 1 h later, the MN remained inserted and had swollen with ISF, losing their sharpness (Fig 3D). Despite being soft in the swollen state, MN were always removed intact in all of the in vivo studies, demonstrating the inherent safety of our MN technology.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

Swollen MN arrays after removal from (A) the back of a rat and (B) the forearm of a human volunteer, both following 1 h insertion. OCT images showing MN inserted in the forearm of a human volunteer at t = 0 h (C) and in the swollen state after 1 h (D).

https://doi.org/10.1371/journal.pone.0145644.g003

In vitro MN detection of theophylline, caffeine and glucose

Hydrogel-forming MN are strong enough to puncture the stratum corneum when dry and, following skin insertion, they absorb moisture from the skin ISF and swell, allowing substances present in the ISF to permeate into them. Excised neonatal porcine skin has extensively been employed in previous studies as a recognised substitute for human skin [20, 24, 32, 52]. As such, it was selected for in vitro use in this study. As shown in Fig 4(A) the amount of theophylline recovered from the MN inserted (5, 30 and 60 min) across neonatal pig skin reflects the increasing concentrations of theophylline (5, 10, 15, 20 and 35 μg/mL) in what is usually the Franz cell receiver compartment. Specifically, after 1 h of MN insertion at the lowest Franz cell concentration (5 μg/mL), a mean recovered concentration of 0.33 μg/mL was reported, whilst at the highest Franz cell concentration (35 μg/mL), a mean theophylline concentration of 1.65 μg/mL was reported. Importantly, these findings also demonstrate that, after only 5 min of insertion, the MN were able to reflect the varying theophylline concentrations in the Franz cell chamber. In particular, concentrations of 0.16 μg/mL ± 0.04 μg/mL and 0.85 μg/mL ± 0.14 μg/mL were detected for the lowest and highest concentrations of theophylline, respectively, after 5 min. These results are in line with work by Kendall et al., where MN coated with capture molecules were able to detect the presence of biomarkers within 10 min of insertion [53]. In the present study, the noted variability observed in the release measurements for MN inserted into neonatal porcine skin for 5 min could potentially reflect slight differences in the rates of initial diffusion of fluid from the Franz cell into the hydrogel matrix [54, 55], since swelling is necessary for substance capture.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4.

(A) Quantities of theophylline (μg) taken up in vitro by MN arrays 5, 30 and 60 min after insertion into dermatomed excised neonatal porcine skin mounted on modified Franz cells and bathed on the underside by phosphate buffered saline pH 7.4 thermostated to 37°C and containing defined concentrations of theophylline (Means ± SD, n = 6). (B) Quantity of theophylline (μg) taken up in vivo by MN arrays inserted into the skin on the back of Sprague Dawley^® rats for 1 h after administering theophylline via oral gavage at doses of 5 and 10 mg/kg of rat’s body mass (Means ± SD, n = 6). (C) Exemplar chromatogram showing HPLC detection of theophylline following extraction from MN after insertion into the skin on a rat’s back for 1 h. Dose administered to rat via oral gavage = 10 mg/kg. (D) Exemplar chromatogram showing HPLC detection of theophylline from a plasma sample obtained from a rat via lateral vein tail puncture 1 h after administration of theophylline (10 mg/kg) via oral gavage.

https://doi.org/10.1371/journal.pone.0145644.g004

Statistical analysis of values shown in Fig 4(A) confirm a highly significant difference (p < 0.01) in theophylline levels detected after 5, 30 and 60 min at all concentrations, with the exception of the 15 μg/mL against 20 μg/mL Franz cell receptor compartment. Furthermore, there was a very highly significant difference (p < 0.001) at the 5 min time point when theophylline levels detected from the 5 μg/mL and 10 μg/mL Franz cell receptor compartment were compared with levels from the 35 μg/mL compartment. Similarly, there was a significant difference (p < 0.05) in theophylline levels detected at the 5 min time point when the 15 μg/mL and 35 μg/mL concentrations were compared. These findings suggest that, for theophylline TDM, shorter periods of MN insertion may actually be preferable, since any difference in concentration can be detected at a relatively early stage. This, in turn, could facilitate patient compliance and, importantly, provide clinically-relevant drug or biomarker data to the clinician in a timely fashion.

The potential of MN as a means of minimally-invasive means of TDM and diagnosis has recently become the focus of considerable interest [32, 56, 57]. Although previous studies have detected certain compounds using MN in combination with other techniques [57], no study has demonstrated the ability of MN alone to both detect and quantify analytes of clinical interest, thereby facilitating both diagnosis and TDM. In addition to theophylline, this study investigated the ability of hydrogel-forming MN to detect caffeine in vitro, using the previously-described Franz cell setup. Fig 5(A) shows the extracted caffeine levels from MN inserted into neonatal porcine skin at defined time points and Franz-cell concentrations. As expected, caffeine levels detected from 5 and 60 min MN insertion times appropriately reflect the different concentrations present in the respective Franz cell compartments, with mean caffeine levels detected from the 15 μg/mL Franz cells being higher than mean caffeine levels from the 5 μg/mL Franz-cells at both time points. As Fig 5(A) clearly shows, we have demonstrated that our hydrogel-forming MN possess the ability to uptake sufficient ISF after only 5 min insertion time to allow quantification of caffeine. Despite no statistically significant difference (p > 0.05) identified between in vitro caffeine levels detected at 5 and 60 min time points in the 5 μg/mL and 15 μg/mL Franz-cell receptor compartments, the levels detected at both time points had increased. Specifically, the mean concentration of 0.10 μg/mL obtained from the 5 μg/mL Franz-cell after 5 min had increased to 0.23 μg/mL after 60 min, while the mean concentration obtained from the 15 μg/mL Franz cell increased from 0.33 μg/mL to 0.38 μg/mL over the same application time. With respect to MN-mediated TDM, specifically relating to drugs with a narrow therapeutic index, these findings further suggest that hydrogel-forming MN may have the potential to rapidly differentiate between high and low drug concentrations. Crucially, this ability to distinguish between a therapeutic drug concentration and a sub-therapeutic or toxic concentration could allow hydrogel-forming MN to provide invaluable guidance on dosing of drugs with a narrow therapeutic range in which rapid measurement is recognised as a major factor in optimising clinical outcome [58].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5.

(A) Quantity of caffeine (μg) taken up in vitro by MN arrays 5 and 60 min after insertion into dermatomed excised neonatal porcine skin mounted on modified Franz cells and bathed on the underside by phosphate buffered saline pH 7.4 thermostated to 37°C and containing defined concentrations of caffeine (Means ± SD, n = 6). (B) Comparison of caffeine (μg) detected in plasma versus levels detected from MN arrays inserted in the forearm of human volunteers at defined time points (0–1, 1–2, 2–3 and 0–3 h) following administration of defined caffeine doses (Mean ± SD, n = 9). (C) Exemplar chromatogram showing HPLC detection of caffeine following extraction from MN after insertion in the forearm of a human volunteer for 1 h. Dose administered = 100 mg oral caffeine in the form of two Proplus^® tablets. (D) Exemplar chromatogram showing HPLC detection of caffeine from a plasma sample obtained from a human volunteer 1 h after the administration of 100 mg of caffeine. 7-BHT was used as an internal standard.

https://doi.org/10.1371/journal.pone.0145644.g005

Similar to the above studies, in vitro analysis was performed to investigate the ability of the hydrogel-forming MN to extract glucose through neonatal porcine skin using a modified Franz-cell setup, as described previously. As shown in Fig 6(A), the mean detected glucose concentration after 5 min from the Franz cell containing 4 mmol/L was 19.46 nmol/L. Although not statistically significant (p > 0.05), the mean detected glucose concentration after 60 minutes application had approximately doubled to 35.67 nmol/L. These results successfully demonstrate the in vitro ability of hydrogel-forming MN to detect glucose in neonatal porcine skin. The promising nature of the obtained results, which quantify glucose after only 5 minutes in vitro, led us to proceed to human volunteer studies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6.

(A) Quantity of glucose (μg) taken up in vitro by MN arrays 5 and 60 min after insertion into dermatomed excised neonatal porcine skin mounted on modified Franz cells and bathed on the underside by phosphate buffered saline pH 7.4 thermostated to 37°C and containing defined concentrations of of glucose (Means ± SD, n = 6). (B) Comparison of glucose (μg) detected in plasma versus levels detected from MN arrays inserted in the forearm of human volunteers at defined time points preceding (-1-0 h) and following (0–1, 1–2, 2–3 and 0–3 h) oral administration of 75 g of glucose (Means ± SD, n = 9).

https://doi.org/10.1371/journal.pone.0145644.g006

This study demonstrates the ability of hydrogel-forming MN to detect and quantify theophylline, caffeine and glucose following only 5 min insertion in neonatal porcine skin. In addition, it demonstrates the ability of hydrogel-forming MN arrays to rapidly detect various drugs with only 5 min exposure in a fresh water solution as opposed to the overnight exposure required in previous studies to detect glucose and sodium ions from poly(vinyl alcohol) hydrogels [59]. Notably, this previous study involved a time-consuming two-step approach, in which volunteer forearms were stamped with a plastic MN. This was then followed by application of a hydrogel patch placed on the stamped area, with the aim of accumulating ISF glucose [59]. Significantly, in this present study, hydrogel-forming MN were able to differentiate between in vitro theophylline concentrations of 5 μg/mL and 15 μg/mL after 60 minutes MN application (p < 0.001) and 5 μg/mL and 35 μg/mL after only 5 minutes application time (p < 0.001) but without the assorted problems related to needle insertion and associated processing [38–41].

In vivo MN detection of theophylline

Theophylline is a xanthine bronchodilator typically used in the treatment of asthma and chronic obstructive pulmonary disease [60]. It possesses a narrow therapeutic range, with a plasma concentration of 10–20 mg/L usually providing a satisfactory therapeutic response. Importantly, the desired therapeutic range in neonates is typically lower at 5–10 mg/L [61, 62]. As the metabolism of theophylline is subject to several variable factors, plasma concentration often requires monitoring in individual patients to ensure both optimal clinical response and avoidance of undesirable toxic effects [63, 64]. Typical effects of toxic levels of theophylline (> 25 mg/L) in both adults and neonates include tachycardia and central nervous system excitation, whilst in severe cases, potentially fatal theophylline-associated seizures can occur [65]. Previous studies in rats have reported a linear pharmacokinetic profile of theophylline for doses not exceeding 10 mg/kg of rat body mass [66]. Therefore, this was chosen as the higher dose administered to the rats in this study. The average concentration of theophylline extracted from MN inserted for 1 h in the rats’ back, and the average theophylline concentration extracted from blood between 0 h (pre-theophylline administration) and 1 h are shown in Fig 4(B). This allowed the comparison of mean blood analyte values over a 1 h time period and total analyte extracted using hydrogel-forming MN in the 1 h time period. The insertion period employed was chosen conservatively in order to ensure sufficient uptake of fluid for accurate theophylline quantification, thus minimising the number of rats used, bearing in mind the requirements of the 3 Rs.

Theophylline was successfully extracted by MN arrays and subsequently detected from the ISF, as depicted in the chromatogram obtained from the extraction of theophylline from an individual MN array (Fig 4C). The highest theophylline concentration detected from a hydrogel-forming MN from a rat dosed with 10 mg/kg was of 0.363 μg/mL, whilst a maximum of 0.063 μg/mL was detected from a rat dosed with 5 mg/kg theophylline. Statistical analysis of the values shown in Fig 4(B) demonstrated that, following oral theophylline doses of 5mg/kg and 10mg/kg, a very highly significant difference (p < 0.001) in theophylline levels between both concentrations administered is detectable via both blood and MN. These findings demonstrate that hydrogel-forming MN have the in vivo ability to differentiate between significantly different (p < 0.001) ISF concentrations of a drug with a narrow therapeutic range, thus raising the very real possibility of hydrogel-forming MN as a means to determine whether a drug with a narrow therapeutic range is within its therapeutic window.

Notably, no theophylline was detected from MN-free baseplates (prepared using identical formulae to the MN, as described in Fig 2A) applied to the rats’ backs. This suggests that the swelling of MN was due to uptake of ISF containing theophylline rather than any theophylline potentially present in rat perspiration. In addition, our polymeric MN systems were removed macroscopically intact after skin insertion, leaving no discernible polymer behind.

Human volunteer studies for caffeine

With the aim of investigating the ability of hydrogel-forming MN to detect caffeine in ISF, we applied MN to the ventral forearm of healthy male and female human volunteers with no existing skin conditions. Fig 3(B) shows MN swelling after 1 h insertion in a human volunteer, demonstrating that hydrogel-forming MN are able to imbibe ISF and swell over a relatively short period of time in vivo. This was in line with what was seen in the rat studies. The levels of caffeine detected by both MN and DBS are shown in Fig 5(B). As with the in vivo theophylline study, this allowed the comparison of mean blood analyte values and total analyte extracted using hydrogel-forming MN over the same time period. As demonstrated, our novel hydrogel-forming MN were able to quantify caffeine taken up by MN arrays at each time point. Fig 5(C) shows a chromatogram demonstrating caffeine detection via hydrogel-forming MN using the above method, and as outlined in Fig 1. Significantly, our MN readily detected caffeine after only 1 h post-consumption. The mean detected level of 73.05 μg/mL demonstrates that TDM using hydrogel-forming MN is a feasible concept in human subjects and can provide rapid analyte quantification soon after administration. These findings further emphasise the advantages of our hydrogel-forming MN technology, which include ease of sampling and storage, reduced problems associated with time-consuming analysis of blood-based matrices and the avoidance of hypodermic-needle use, when compared with traditional blood sampling. The highest mean concentration of caffeine detected using our MN was 91.31 μg/mL over the period from 1 to 2 h post-consumption of 100 mg Proplus^®. This result is similar to previous findings which reported caffeine absorption from the small intestines within 45 minutes of consumption and peak blood concentration occurring after 1–2 h[67]. Furthermore, in keeping with the pharmacokinetics of caffeine, MN arrays applied during the 2–3 h interval post-caffeine consumption demonstrated the lowest mean caffeine level of 49.56 μg/mL, suggesting that, by this time, the absorbed caffeine is likely to be undergoing hepatic metabolism to its three primary metabolites; paraxanthine, theobromine and theophylline. Of further interest is the level of caffeine detected in the MN arrays applied continuously for 3 h. As shown in Fig 5(B), these arrays detected a much lower level of caffeine than the MN applied in the 1–2 h post-caffeine interval and a similar level to that detected in the 0–1 h interval. A possible explanation for this is that the detected level in the MN reflects the established pharmacokinetics of caffeine. Specifically, after 3 h, significant hepatic caffeine metabolism has already occurred and, as a result of their continuous application, these MN arrays may have reached an equilibrium with the ISF, resulting in a lower detected level of caffeine than that reported in the peak interval of 1–2 h. Given these findings it would therefore, be appropriate, as well as advantageous, to apply the MN arrays for a maximum period of 1 h.

In addition to 100 mg caffeine as Proplus^®, Fig 5(B) also shows caffeine levels detected from both MN and DBS following volunteer consumption of a Starbucks^® double-shot espresso coffee. Interestingly, despite a purported caffeine content of 150 mg [28, 29], both MN and DBS sampling following coffee consumption demonstrated caffeine levels lower than those obtained following administration of 100 mg caffeine (as Proplus^®) at identical time points. Despite reference sources indicating a caffeine content of 150 mg present in the coffee, the results would indicate this not to be the case. It is possible that each coffee had a different caffeine content, due to the variability associated with human preparation. Alternatively, the formulation of the coffee itself could have impaired absorption. However, the latter would appear unlikely, as coffee, being in liquid form would, in all likelihood, actually improve absorption.

As shown in Fig 5(D), caffeine was also readily detected after 1 h in DBS sampled concurrently with MN. Caffeine detected via DBS is reported at higher concentrations than MN concentrations. However, it is notable that, whilst the values detected are statistically different due to the higher detected DBS concentrations (p > 0.05), the pharmacokinetic profile of MN and DBS sampling displays general agreement over the 0–1 h and 1–2 h time intervals. As shown in Fig 5(B), both MN and DBS sampling show peak caffeine levels in the 1–2 h interval post-caffeine consumption. However, as above, due to the increased concentration of caffeine detected using the DBS method, the values reported are statistically different (p > 0.05). Similarly, MN and DBS sampling following the purported 150 mg dose of caffeine both indicate a lower detected caffeine level than that following 100 mg caffeine (as Proplus^®). Despite the differences in concentrations reported between DBS and MN sampling, the similar concentration profile demonstrated for DBS and MN displayed in Fig 5(B) suggests that, with a validated mathematical method, it may well be possible to correlate the detected MN concentration to that detected from DBS sampling. Such a conversion could, therefore, make hydrogel-forming MN a realistic means of TDM.

Although previous studies have studied the collection of ISF using a variety of MN designs [68–70], to the best of our knowledge, this is the first MN study to demonstrate detection and quantification of a model drug substance (caffeine) in human subjects following oral administration. The simplicity of the methodology employed also overcomes problems associated with other methods employed for ISF-mediated TDM, such as RI and CM, while Fig 3(B) clearly demonstrates the minimally-invasive nature of the technology. These findings unequivocally demonstrate the potential of our hydrogel-forming MN arrays as a means of simple and rapid TDM, whilst avoiding the aforementioned problems associated with conventional blood sampling [32]. In addition, the analytical process illustrates how TDM using hydrogel-forming MN could be incorporated into a conventional clinical laboratory setting, thus further emphasising the potential of our novel MN technology as a means monitoring clinically-important drugs and biomarkers.

Human Volunteer Studies for Glucose

Despite numerous proposed methods for ISF extraction from the skin, including RI, sonophoresis and laser ablation for creation of micropores, Glucowatch^®, (Cygnus Redwood City, CA) remains the only method endorsed by the US FDA. However, the Glucowatch^® system uses an electric current for ISF extraction, which can cause pain, irritation and damage to the treated skin [71]. Therefore, the development of a minimally-invasive or non-invasive, and pain free method remains the main focus in production of new systems for glucose monitoring.

Over recent years, MN have become one of the most promising minimally-invasive and pain-free approaches for ISF glucose extraction [71]. However, lack of sufficient data regarding their practical application has led to a need to establish if ISF glucose can be extracted using MN and, if so, can the extracted glucose concentration be quantified and/or correlated to that found in blood [71]. To date, previous studies have investigated both dissolving and hollow MN as means of minimally-invasive glucose monitoring [57, 71]. However, unlike these previous studies, which based their conclusions on a two-step process in a rat model and uptake of simulated ISF by hollow MN in a petri dish respectively [57,71], our study focussed solely on in vivo glucose ISF monitoring in human volunteers. Specifically, we evaluated the potential of MN-mediated transdermal glucose monitoring in healthy volunteers and compared the obtained results with those from finger-prick samples using the Accu-Check^® Aviva glucometer. Glucose concentrations at the sampled time points are shown in Fig 6(B). The highest mean BGL was 7.89 nmol/L nmol/L detected 1 h following ingestion of 75 g of glucose, while the highest mean glucose concentration extracted from MN was 4.29 nmol/L detected after 3 hours. This 2 hour difference between MN levels and BGL may be explained by a lag-time phenomenon, which has been previously described in several studies [72] and would explain the statistical difference between MN and finger-prick BGL at identical time points (p > 0.05). Briefly, the lag-time is due to the delay in transport of glucose from plasma, via the vasculature, to the interstitial space. It is also likely that there is a high degree of variability between individuals, due to individual metabolism. Previous studies in both diabetic and non-diabetic individuals have reported this effect may take between 4 to 50 minutes to occur [73]. The appearance of the peak BGL after 2 hours is in keeping with other studies that investigated the glucose profile of healthy volunteers following an oral glucose tolerance test (OGTT) [74]. Despite the longer application time, the mean detected glucose from the MN applied for a full 3 hours was 3.13 nmol/L, lower than the concentration detected after 2 hours. This may be explained by the equilibrium effect between ISF and MN, as outlined above for caffeine.

These findings demonstrate that ISF glucose measurements using hydrogel-forming MN tracked the fluctuation in BGL, with an approximate 1 hour lag time. As discussed previously, this may allow development of mathematical correlation or algorithm to compensate for the time difference between BGL and ISF glucose concentrations at a given time point [75], which in turn could facilitate rapid insulin delivery in conjunction with existing technologies [76]. In addition, several other factors may influence differences between ISF glucose and BGL. Firstly, the altered pharmacokinetic and pharmacodynamic parameters of glucose resulting from its movement in and out of the ISF during the period of MN application [77]. Secondly, the small ISF volume contained within the epidermis may delay migration of ISF glucose in a sufficient quantity to imbibe the hydrogel-forming MN [72]. Despite these challenges, this study demonstrates the exciting potential of hydrogel-forming MN to extract glucose from ISF in human subjects in vivo. Therefore, future work will build upon the findings in this study prior to clinical use. In addition, studies will be conducted in a larger population to facilitate statistical correlation of ISF glucose values to those which are present in volunteer blood samples.