Attempts to do in vitro testing were abandoned when properties of human cadaver skin were found to be quite different from in vivo skin. There was considerable sample-to-sample variability in the stratum corneum thickness and the skin impedance in human cadaver skin prior to microscissioning. The variability may have been caused by the preservation techniques, length of time between death and skin harvest and the site of skin on the body. Superficial mechanical state of the skin varied from tough to near-disintegration among samples. As the wide range of possible applications became apparent, a decision was made to demonstrate proofs-of-concept only. Reliable demonstration of drug delivery (lidocaine), analyte sampling (blood glucose) and even electrical impedance measurement were not possible with in vitro testing.

All in vivo results are from experiments done on two adult subjects. We sterilized the aluminum oxide particulate and masks by heating to 200°C for one hour. Nozzles, hoses and the particle reservoir were rinsed in methanol. Sterile gloves were worn and the target areas swabbed with ethanol. Sterile saline solution was used for electrical tests with syringes and needles for handling saline being rinsed with methanol, and nitrogen dried prior to use.

The depth of scized microconduits can be determined approximately by measuring the electrical impedance between an ECG electrode on the subject's unperturbed skin and the mask holding fixture (Figure 6). A SR715 LCR meter (Stanford Research, Sunnyvale, CA, USA) operating at 1 V and 1 kHz was used. The metal mask holder was electrically connected to the microconduit by normal saline placed in the ring to which the mask was bonded. Since the mask was held tight against the skin, the saline did not leak between the mask and skin, thus providing electrical continuity with the microconduit. Typical microconduits generated with a four nozzle, four hole mask are shown (Figure 7, left and right). The data (Table 1) are averages of approximately 100 experiments on two subjects. Microconduit depths were measured using the Vivascope 1000 (Lucid, Rochester, NY, USA), an in vivo near-infrared reflectance-mode confocal microscope, with an illumination wavelength of 830 nm, 30 mW power and a 30 ×, 0.9 N.A. water immersion objective providing a viewing depth of 200 μm. The instrument captures images with a spatial resolution of 0.5 to 1.0 μm in the lateral dimension and 3 to 5 μm in the axial dimension. Further details of this system have been reported recently 17. Standard particle generator parameters were used, that is, N₂ under 138 kPa pressure, reservoir drive of 80 V, 450 μm nozzle diameter.

Microconduit profiles are revealed by infrared confocal microscopy (Figure 8). Microconduits made with a four hole mask and a scission time of 20 seconds are shown on the left. The white lines are an electron microscope grid placed on the skin to provide high contrast for focusing. The 160 μm diameter openings are nearly uniform, with the upper right one corresponding to the upper left image in the confocal views. The confocal images at nine depths clearly illustrate that the microconduit is fully open throughout its depth of approximately 165 μm (lower right image). At its bottom, the diameter is approximately 65 μm.

Usually there are no residual Al₂O₃ microparticles, which are easy to see as bright 30 micrometer particles at full depth. In some cases, up to about ten microparticles were observed. However, a deionized water rinse from a hypodermic needle at low pressure is effective in removing particles. In cases where particles were seen and left, they were not visible in the confocal microscope at the site after full healing. This suggests they were moved out by the healing process, but has not been confirmed by other methods.

Removal of the high electrical resistance stratum corneum takes place during the initial few seconds of scizing a microconduit. If stopped, then scizing is a fast, simple, totally sensation-free method for reducing the electrical impedance through the skin. At 1 kHz and 1V, the impedance between two electrocardiogram electrodes (Type 510–005, Lynn Medical, MI, USA) is approximately 4 kΩ. Placing two of the same type of electrocardiogram electrodes over two 200 μm diameter shallow microconduits reduces the impedance to 500 Ω, measured under the same conditions. This implies that the 1 kHz impedance associated with a single microconduit is only 250 Ω, and, therefore, four microconduits can provide a local skin resistance of the order of 100 Ω each.

Assessment of microconduit efficacy for delivering a drug into the dermis and epidermis was carried out using the topical anaesthetic, lidocaine. The presence of lidocaine was evaluated by using non-scarring pulses from a 585 nm pulsed-dye laser, employed as a pain inducer 18. Also, the 'pin stick' on skin around the microconduit tests for sensation was used to map lateral anesthesized distance from the microconduit. Masks with four 160 μm diameter holes on 450 μm centers were used. Testing was done on the inner, left wrist, approximately 10 cm away from the palm center.

The first test determined lidocaine uptake and level of anesthesia. After scission for 7 seconds, a subject reported a barely perceptible 'pricking' sensation. The microconduits had an impedance of 23 kΩ, implying a depth of 140 μm. A slight discharge of clear fluid was evident. Both the microconduit and control sites were exposed to a 50% lidocaine solution (5 gm lidocaine hydrochloride in 5 ml normal saline) in saturated filter paper pads with Finn chambers taped over each. In prior studies, we used 40% and 50% lidocaine solution and found that the subjective degree of anesthesia was easily quantifiable by the subjects enrolled in those studies, so we selected 50% lidocaine concentration for this proof-of-concept demonstration 18. These two sites were exposed to a lidocaine 'soak' for 5 minutes. A third 'normal' control site, receiving no lidocaine exposure, was marked and tested. These sites were separated by 5–6 cm. Sensation testing was done with a Coherent Palomar (Burlington, MA, USA) Light Sheer Laser emitting a 30 ms, 35 mJ pulse. The sites were lazed randomly three to five times each, with and without power on, while the subject's arm was extended and the subject looked away. The subject verbally reported sensation effects. The testing was carried out 15, 30 and 75 minutes after lidocaine exposure (Tables 2 and 3). The sensation data strongly suggest that the area around microconduits was anesthetized.

A number of additional tests were done using the 'pin stick' test for sensation, to determine the minimum time for full anaesthetic effect. Most testing was done on one subject, with verification tests done on two other subjects. These tests were done on a single 160 μm diameter microconduit, 100–150 μm deep on the left inner wrist. Using 50% lidocaine saturated cotton pads as the anesthetic source, anesthesia of a control area without a microconduit required 60–90 minutes. Anesthesia became discernable (sensation diminishes perceptibly) after 1 minute. Maximum anesthetic effect and anesthetized radius occurred between 2.5 and 3.5 minutes. The anesthetized radius was 1,800–2,200 μm. When the lidocaine was applied to the microconduit under a pressure of 30 inches of water, the full anesthetic effect occurred in 1.5–2 minutes and the anesthetized radius increased to 2,800–3,000 μm.

The onset of anesthesia takes longer in microconduits deep enough to yield blood than in shallower, non-blood producing microconduits. Possibly the blood outflow impedes inflow of the externally-applied lidocaine to the sub-stratum corneum tissues, or the clotting blood partially obstructs the microconduit.

Experiments involving creation of deep microconduits provided blood samples for glucose testing and assessment of the sensation (pain) level involved. Testing was carried out with a FastTake glucose monitor (Lifescan, Milpitas, CA, USA). The instrument's disposable test strips required a minimum blood volume of 1.5 μl.

These experiments involved one subject using a subjective pain sensation scale of one to ten ('none' to 'sharp'). The test site was always on the side of the left hand fourth finger or the left hand middle finger between the finger end and outer joint. The standard mask-nozzle-generator conditions were used. Opening a microconduit to the blood capillary level consistently took 14–18 seconds. Within 3–6 seconds after removing the scizing device, 1–3 μl of blood formed in a drop (Figure 9, left and right). The FastTake test strip end was held against the drop and a sample was drawn into the sensing chamber by capillary action. Results are summarized (Table 4) comparing accessing blood with the Penlet II Automatic Blood Sampler (a lancet system supplied with the FastTake instrument) and through microconduits made by scizing. The tests were done two days apart, with the two lancet tests first, followed by the scizing tests. Time for the lancet test is zero due to its rapid plunge in/withdraw action. Times for scizing were intentionally varied slightly to test sensation and blood flow effects. The glucose readings were all in a 'normal' range. These experiments show that scized microconduits may provide painless acquisition of samples for established blood glucose tests.