The microbiopsy devices used in our studies were fabricated by laser cutting plates of stainless steel. The in-house fabrication of microbiopsy devices involves laser cutting of two-dimensional designs on 0.05 mm thick medical grade stainless steel (304, Mastercut Technologies, Australia) at 95% power, 30% frequency and speed of 0.9 using a 20 W laser marking system (LaserPro S290, GCC, Taiwan). All of the 2D microbiopsy components were assembled into 3D devices, and sterilized using a glass bead sterilizer (Steri Inotech 350, Sigma-Aldrich, USA) at 250°C for 30 seconds or submerged in 70% ethanol for 1 min and air-dried prior to use. The microbiopsy device was then fitted into a spring-loaded applicator.

A bench top scanning electron microscope (SEM) (JCM-6000 Neoscope, JEOL, USA) was used to acquire high-resolution images of the microbiopsy device. The roughness amplitude of the microbiopsy chamber was obtained by measuring the average distances of the edge to a regression-fitted straight line using MatLab (Mathworks, Australia).

Microbiopsy samples were either obtained in vivo from healthy volunteers (HREC/12/QPAH/082, Metro South Human Research Ethics Committee, Centres for Health Research, Princess Alexandra Hospital) or ex vivo from excised actinic keratosis (AK) lesions. The skin of a healthy volunteer was swabbed with alcohol and the microbiopsy applied. Conventional biopsies e.g. punch ( Figure 1b) or shave biopsies, were performed with informed consent from patients (HREC/08/QPAH/207 for healthy skin or HREC/11/QPAH/477 for AK lesions) prior to being microbiopsied. Samples were collected in sterile RNase and DNase free 1.5 ml microcentrifuge tubes containing either RNALater ^® (Life Technologies, USA) or pH 7.0 phosphate buffered saline (PBS) on ice within 20 minutes after tissue removal from patients. All samples placed in RNALater ^® were kept overnight at 4°C and then stored at -80°C.

Volunteer pain scores were evaluated with Metro South Human Research Ethics Committee, Princess Alexandra Hospital approval (HREC/12/QPAH/082). Each of the 20 volunteers ( Table 1) was presented with an assessment form to grade his/her expected pain score based on a numerical rating 10-point Likert scale, 0 as having no pain and 10 as pain as bad as they can imagine before the start of the experiment. Each microbiopsy application was performed one minute apart and the pain score recorded immediately after application. Five minutes after the final microbiopsy application, the volunteers were asked to rate the level of pain for each microbiopsy site.

Microbiopsy sites were visualized with a reflectance confocal microscope (RCM, Vivascope 1500 ^® Multilaser, Lucid, Inc., USA) in volunteers or in excised skin (HREC/12/QPAH/217). RCM was also used to perform in vivo examination of the skin after microbiopsy. RCM images were collected through the microscope head, integrated with a water immersion objective, at a near-infrared wavelength of 785 nm. Blocks consisting of 7 × 7 mm mosaics of stitched RCM images and 2 µm to 200 µm vertical z-stacks were acquired at the microbiopsy sampling site. Dermatoscopic images were collected with a Canon Power Shot G10 digital camera (Canon, Japan) and a dermatoscope attachment (Dermlite©, 3Gen, USA).

Microbiopsied tissue was embedded (Optimal cutting temperature compound, Sakura Finetek, USA) and cryosectioned (CM1850, Leica Microsystems Pty Ltd, Australia). The 10 µm thick sections were fixed with 100% cold methanol for 10 minutes, air dried and stained with haematoxylin and eosin (Varistain Gemini ES, Thermo Fisher Scientific Inc., USA) to visualize the microbiopsy sites.

The microbiopsy extracted tissue was incubated with DRAQ5 (BioStatus Limited, UK) at 10 µM in PBS for 30 minutes at room temperature. Confocal microscopy images were taken with a Zeiss 510 META confocal microscope (Carl Zeiss Microscopy GmbH, Germany). A 3D projection was then created from the z-stacks (2 μm steps) of confocal images using Imaris (Bitplane Scientific Software, Switzerland). ImageJ (NIH, USA) was used to estimate the nuclei in microbiopsy samples.

DNA was extracted from the microbiopsy samples using QIAamp DNA Micro purification kits (QIAGEN GmbH, Germany) using a protocol modified to accommodate the unique microbiopsy sample. The microbiopsy was removed from the applicator housing after application. The presence of a tissue sample was quickly confirmed by opening the microbiopsy device and visualizing with a stereo microscope (Carl Zeiss Microscopy GmbH, Germany). The opened device was then immersed in 180 μl lysis buffer (Buffer ATL, QIAGEN GmbH, Germany) in the presence of proteinase K (20 μl) in a 1.5 ml tube. Fifteen seconds of pulse-vortexing was applied to ensure that the sample was well-mixed. The sample was then placed in a thermomixer (Eppendorf, Australia) at 56°C with shaking at 800 rpm overnight. The tube containing the device was briefly centrifuged for 10 seconds to remove the solution adhering to the cap of the tube before transferring the lysate into a new tube. The tube containing the device was centrifuged again at 6000 g for 30 seconds to remove any lysate adhering to the device. The lysates were combined and the remaining procedure followed the manufacturer supplied instructions.

The modified sample lysing processing described in DNA isolation above was also used with the Arcturus ^® PicoPure ^® RNA Isolation Kit (Life Technologies, USA) to obtain total RNA from microbiopsy samples. RNA isolation of lesional samples was performed using RNeasy Mini Kit (QIAGEN GmbH, Germany) according to the RNeasy Mini Handbook (version 2010).

A quantitative fluorometer-based assay (Qubit ^® 2.0, Life Technologies, USA) was used to determine the concentration of DNA and RNA with the protocol provided by the manufacturer.

Agilent DNA 12000 DNA kit (Agilent Technologies, USA) was used to determine the integrity and quality of DNA after whole genomic amplification of isolated DNA from microbiopsy and matched conventional shave biopsy samples. Agilent RNA 6000 Nano and Pico kits (Agilent Technologies, USA) were used to determine the RNA integrity number (RIN) of lesional and microbiopsy samples after RNA isolation. The supplied protocol was followed.

Identical amounts of total DNA (1.85 ng) for both lesional and microbiopsy samples were subjected to whole genomic amplification. The amplification procedure was carried according to manufacturer's instructions (REPLI-g Single Cell Kit, QIAGEN, Australia).

Identical amounts of total RNA (14 ng) for both lesional and microbiopsy samples were subjected to whole transcriptome amplification. The procedure was carried out using the supplied instructions (QuantiTect Whole Transcriptome Kit, QIAGEN, Australia). The cDNA obtained from the amplification process was used as a template in a polymerase chain reaction (PCR) reaction using human beta actin primers (forward: ATC TGG CAC ACC TTC TAC AAT GA; reverse: CGT CAT ACT CCT GCT TGC TGA TCC AC) (Integrated DNA Technologies, NSW, Australia). There was an initial denaturation for 2 minutes at 98°C, followed by 30 cycles of 98°C for 30s (denaturation), 67°C for 30s (annealing), and 72°C for 1 minute (extension). The cDNA and PCR products were run on a 1% agarose gel (Bio-rad Laboratories, Inc., Australia) and visualized with RedSafe (ChemBio, UK). HyperLadder™ 1 kb (HyperLadder I) (Bioline, UK) was used to determine the mass of cDNA and human beta actin amplicons.

Statistical analysis was performed using PRISM 6 for Windows (GraphPad Software, Inc., USA). Channel width, velocity and roughness amplitude data were presented as mean ± SD. Pain score for channel width and velocity were presented as minimum to maximum box-and-whiskers plot. One-way ANOVA combined with a Tukey’s multiple comparison post-test was performed to determine the statistical significance.

We conducted a volunteer study in 20 healthy individuals to determine the optimal channel width and application velocity for this device. Extracted DNA mass was chosen to be the primary indicator for sample-to-sample comparisons. Interestingly, we observed tissue collection (4.5 ± 1.5 ng DNA) around the rough edges of a microbiopsy device without a chamber ( Figure 2a–b, channel width of 0 mm). After applying the microbiopsy, the device was opened up and visualized under a dissecting microscope. Successful collection was achieved when tissue was evident within or around the device and unsuccessful if no tissue was present. Tissue was collected from all volunteers (n=20) when a 0.15 mm channel width microbiopsy device was used. Only 13 successful collections were achieved from 20 applications when a 0.20 mm channel width microbiopsy device was used. This indicated that the collection rate decreased from 100% to 65% when channel width was increased by 0.05 mm. The device without a channel (0 mm channel width) captured tissue around the edges of the tapered plates in all replicates (n=20). This experiment showed that a channel width of 0.15 mm obtained the highest average amount of DNA (5.9 ± 3.4 ng), which was significantly higher than 0.25 and 0.30 mm channel widths (p < 0.0001). There was no significant difference in the total DNA extracted between 0–0.20 mm channel widths ( Figure 2a).

Microbiopsy application velocity was evaluated by using defined applicator springs to achieve velocities between 0–20.2 m/s ( Figure 2d). Only negligible amounts of DNA were recovered when the device was applied at less than 9.2 m/s. However, there was a 7.5-fold increase (0.8 ± 0.8 to 6.0 ± 3.0 ng) in DNA recovered when the application velocity was increased from 9.2 m/s to 16.6 m/s (p < 0.0001). An additional increase to 20.2 m/s increase in application velocity did not result in significant increases in DNA collection.

There was no change in the level of pain reported when microbiopsies with varying channel widths were applied ( Figure 2c). The level of pain significantly increased and variation increased when microbiopsies were applied at increasing velocities ( Figure 2e). All of the volunteers had a pain score of 0 at 5 minutes after the final microbiopsy application (data not shown). The volunteers scored pain between 0 to 10 with an average score of 1.5 ± 1.1, when the 0.15 mm channel width microbiopsy was applied at 20.2 m/s.

We observed DNA collection without a centre chamber ( Figure 2a, 4.5 ± 1.4 ng from channel width 0 mm) and hypothesized that surface roughness could be key to successful sample collection. We compared microbiopsy devices with varying roughness amplitudes (R _A) and observed an increasing trend in total DNA extracted with increasing R _A ( Figure 3a) (n=20 for 5.36 R _A and n=5 for 0.92 and 1.32). The SEM images of the inner chamber edge were used to measure the R _A. These measurements ranged from relatively smooth (R _A 0.92) to rough (R _A 5.36) ( Figure 3b, side view).

RCM was used to visualise the microbiopsy application site. RCM images revealed a microbiopsy site similar to the size of a small hair follicle in the dermal papillary post microbiopsy application ( Figure 4a and 4b). The inset shows that microbiopsy application resulted in a puncture site that was approximately 0.10 × 0.50 mm in dimension ( Figure 4b). Microbiopsy samples were stained with DRAQ5 (BioStatus Limited, UK) to highlight the nuclei. Confocal images were used to generate a 3D model of the microbiopsy sample. The usual skin strata were apparent in this model. We estimated that there were 1634 nuclei in Figure 4c. We observed 0.22 ± 0.12 mm wide and 0.26 ± 0.09 mm deep puncture sites in excised abdominal skin from 10 histological sections ( Figure 4d).

One of our goals was to compare AK lesions to microbiopsy samples since many of these lesions can be present in patients but only a few progress ¹² to squamous cell carcinoma and the molecular mechanism behind this transition is a focus of intense research. Microbiopsy samples were taken from freshly excised AKs. The RNA from these matched samples had comparable quality with an average RIN difference of -0.85 ± 0.85 for n=4. Representative Bioanalyzer (Agilent Technologies, USA) results for the conventional biopsy (RIN 6.50) and microbiopsy (RIN 5.10) are shown in Figure 5, left panel. The results showed that there are similarities and differences in the RNA bands, which may be due to the large amount of tissue sampled with conventional biopsy and the relatively small number of cells sampled with the microbiopsy.

The small sample size is an inherent limitation of the microbiopsy device. To mitigate this, a pilot experiment was conducted where the microbiopsy sample was subjected to whole transcriptome amplification for RNA analysis. We included a skin sample with the same amount of starting material (14 ng) as an amplification control in the experiment. We observed a 2000-fold increase in cDNA from both samples based on total RNA and cDNA measurements. The cDNAs were of comparable quality and quantity ( Figure 5, Transcriptome Amplification to cDNA). Subsequently, we used PCR to amplify human beta-actin mRNA. We observed 2 identical bands at 800 bp with comparable PCR product quality both lanes ( Figure 5, Actin PCR).

Immediately after microbiopsy application we observed local erythema that resolved within 24 hours. The tiny excision site from the microbiopsy healed quickly and was invisible to the naked eye after 24 hours ( Figure 6). We used dermoscopy and RCM to monitor the wound healing process at regular intervals ( Figure 6, center and right columns).