Rapid, low cost prototyping of transdermal devices for personal healthcare monitoring

The next generation of devices for personal healthcare monitoring will comprise molecular sensors to monitor analytes of interest in the skin compartment. Transdermal devices based on microneedles offer an excellent opportunity to explore the dynamics of molecular markers in the interstitial fluid, however good acceptability of these next generation devices will require several technical problems associated with current commercially available wearable sensors to be overcome. These particularly include reliability, comfort and cost. An essential prerequisite for transdermal molecular sensing devices is that they can be fabricated using scalable technologies which are cost effective. We present here a minimally invasive microneedle array as a continuous monitoring platform technology. Method for scalable fabrication of these structures is presented. The microneedle arrays were characterised mechanically and were shown to penetrate human skin under moderate thumb pressure. They were then functionalised and evaluated as glucose, lactate and theophylline biosensors. The results suggest that this technology can be employed in the measurement of metabolites, therapeutic drugs and biomarkers and could have an important role to play in the management of chronic diseases.


Introduction
The current generation of wearable devices for personal wellness monitoring such as the Fitbit (Fitbit), UP3 (Jawbone) and SIM band (Samsung) comprise a variety of physical sensors that measure parameters such as movement, heart rate, electrocardiogram (ECG), galvanic skin response, bioimpedance, and skin temperature. The next generation of wearable smart devices will also incorporate non-invasive/minimally invasive molecular sensors. The main challenges anticipated for these smart devices would be similar to the ones faced currently by the physical sensors particularly in terms of reliability, cost and the ethical issues associated with ownership and use of continuous bio-signal obtained in real time.
Transdermal devices consisting of microstructures such as hollow (van der Maaden et al. something that could be potentially of great clinical utility. The main application where minimally invasive, continuous measurement is of immense value is in continuous glucose monitoring for diabetes. However with suitable devices this could also be extended to lactate monitoring in athletes for performance training, interstitial therapeutic drug monitoring as well as lifestyle related applications such as help in smoking cessation. Interstitial fluid has been extensively explored for metabolites such as glucose and lactate (Ciechanowska et al. 2013;Miller et al. 2012) . Continuous glucose monitoring (CGM) has been most intensively studied and there is good clinical evidence that effective CGM in Type 1 diabetes leads both to a reduced frequency of hypoglycaemic episodes and lowered HbA 1c (Pickup et al. 2011).
Minimally invasive, continuous monitoring of lactate in the interstitial fluid could also offer a more informative measure of this metabolite that would allow athletes to monitor it during their training rather than taking a single (venous blood) measurement at the end (Goodwin et al. 2007).

ACCEPTED MANUSCRIPT
The ISF microenvironment also offers a compartment in which to measure the dynamics of therapeutic drugs and biomarkers during treatment. There are reports suggesting that there are as many as 80 molecular markers in the skin compartment that show good correlation with venous blood (Paliwal et al. 2013). A recent study in animals has shown a good correlation between the concentration and dynamics of many drugs in ISF and venous blood (Kiang et al. 2012). Therapeutic drug monitoring in the interstitial fluid using minimally invasive devices for continuous or periodical monitoring of drugs with narrow therapeutic index offers a potential route to personalisation of medication regimes.
Despite the promise of minimally invasive continuous monitoring, experience with CGM using some of the approved commercial devices such as those from Medtronic (Enlite), Dexcom (G5) and Abbott (FreeStyle Libre) has revealed that their uptake is still less than 10% despite nearly two and half decades of research and development (Lodwig et al. 2014), (Heinemann et al. 2012). This can be attributed mainly to poor accuracy and precision (giving rise to many false alarms) and high costs of device manufacture. Current CGM devices are mostly subcutaneously implanted and hence the emphasis has been on long term performance. Our approach has been to use a minimally invasive microneedle array format where the sensors can be readily and painlessly inserted and where the cost is sufficiently low that daily replacement is feasible. This could give all the proven benefits of continuous monitoring at a price comparable to the current single measurement strips. We report here on methods used for fabrication of such microneedle arrays.
To achieve the low cost and scalable manufacture injection molding of polycarbonate yielded 300-microneedle structures/hour. Mechanical characterization of these structures is presented and their functionalization to produce glucose, lactate and theophylline biosensors is described. The in vitro characterisation of these, as well as in vivo studies on human skin illustrates their performance.

Fabrication of the base microneedle array:
Copper-Tungsten and stainless steel masters, fabricated using electric discharge milling (EDM), were then used as electrodes for spark erosion of aluminum blocks to create molds.
The spark erosion was repeated up to six times. In one instance, the molds were washed with a pressurized jet to remove the eroded debris. In another example, the aluminum mold created

SEM characterisation:
The bare devices were sputtered with 50nm chromium and imaged using a JEOL 5610 scanning electron microscope.

Measurement of axial compression force
The effect of applying an axial compression load to the microneedle array was assessed using an Instron 5866 instrument with a 500 N load cell. Microneedle arrays were placed on a fixed metal plate with the microneedles facing upwards before applying the desired force through the movable probe of the Instron compression system. The Instron instrument applied pressure on the microneedle arrays using an axial force (parallel to the microneedles' axes) at a rate of 1mm/s until the required force was exerted. Forces ranging from 50N to 400N were A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
tested. Scanning electron microscopy of microneedle arrays were obtained before and after application of the compression load. The height of each microneedle was measured after testing and the percentage change in microneedle height calculated.

Measurement of transverse fracture force:
The transverse failure forces of microneedle arrays were measured with the same force station as described above. A thin probe (Agar Scientific) (2 mm thickness at the tip) was slotted in the clamp and used for this purpose. It was adjusted to ensure that it pressed orthogonally against a row of eight microneedles. The probe was moved at a speed of 1 mm/min. The force required to fracture a single microneedle was determined by dividing the transverse force required to fracture one row by the number of microneedles in each row (four). The microneedle arrays were examined by scanning electron microscopy prior to and after fracture testing.

Functionalisation:
Bare microneedle arrays were sputtered with chromium (15 nm)/Platinum (50nm) to obtain the working electrodes. One of the microneedle arrays was sputtered with Ag (150nm), which was modified to an Ag/AgCl reference electrode by treating with a saturated solution of 3.0 Results:

SEM Characterisation:
The SEM images were analysed for geometry. Each injection molded part was 20x20mm with a 2 mm thick base and comprising 4 arrays each of which had 16 microneedles in a 4x4 subarray. The individual microneedles were pyramidal in shape; 1000 microns tall, with a 600 microns square base and a tip diameter of ~40microns. The pitch between the microneedles is 1200 microns.

Mechanical Evaluation
In axial compression tests (n=4) the microneedles tolerated large forces without fracture. The reduction of microneedle height increased non-linearly with increasing applied force. This ranged from 1% for 50 N to 17 % for 400 N axial forces. Transverse fracture tests (n=3) showed that the force required to fracture one row of four microneedles was 80 ± 3 N (20N per microneedle).

Functional evaluation:
The functionalised devices were tested with varying glucose, lactate and theophylline concentrations using chronoamperometry. From the chronoamperometry studies the dose response curves were fitted to the Michealis-Menten equation.

(b) (c)
As seen from the dose response curves, glucose and lactate concentration as low as 0.5mM could be easily detected. For the glucose biosensors, the three microneedle electrode arrays within a single device exhibited Km values of 11.4+2.7mM, 13.9+5.9mM and 15.5+9.6mM.

ACCEPTED MANUSCRIPT
The corresponding I max values for the three electrodes were 20.3+2.4µA, 23.6 +5.4µA and 25.5 +8.7µA.
Insert Table 1 The Imax values were found to be low for the lactate and theophylline biosensors. This can be Whilst the increase in linear range is expected for a mass transport limiting membrane, it would also be expected to reduce the maximum current. A possible explanation for the increased limiting currents is that the membrane reduces the rate of diffusion of hydrogen peroxide away from the electrode surface, resulting in an increased local concentration quickly consequently leading to a higher current.

Insertion Test on human skin:
The insertion ratio, as determined from the tests (n=16) in vivo (human skin) was 100% for very moderate forces (<10 Newton). Whilst the SEM of the microneedle arrays following insertion tests confirmed the structural integrity of the device the OCT images confirm the penetration of the stratum corneum.
Insert Figure 2

(a), (b), (c), (d) & (e)
It is observed that the channels created by penetration of microneedles collapse on account of the highly elastic nature of the skin layers therefore OCT was done with device inserted.

Conclusions:
Next generation wearable devices will carry minimally invasive, continuous monitoring systems for metabolites, biomarkers and drugs and will have implications for the better management of both lifestyle and chronic diseases. We have demonstrated here a scalable (300/hr) and cost effective approach to fabricating polycarbonate-based (material cost of 2 pence) microneedle structures with desired geometry and mechanical properties. The sterilized structures were tested on human skin and it was observed that the geometry of the tips allowed penetration of the stratum corneum and access to the dermal interstitial fluid by insertion under moderate thumb pressure. The microneedle arrays could subsequently be A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
functionalised to perform as electrochemical biosensors.

1(b) Lactate biosensor
With membrane no membrane