Microfabricated infuse-withdraw micropump component for an integrated inner-ear drug-delivery platform

Abstract

One of the major challenges in treatment of auditory disorders is that many therapeutic compounds are toxic when delivered systemically. Local intracochlear delivery methods are becoming critical in emerging treatments and in drug discovery. Direct infusion via cochleostomy, in particular, is attractive from a pharmacokinetics standpoint, as there is potential for the kinetics of delivery to be well-controlled. Direct infusion is compatible with a large number of drug types, including large, complex molecules such as proteins and unstable molecules such as siRNA. In addition, hair-cell regeneration therapy will likely require long-term delivery of a timed series of agents. This presents unknown risks associated with increasing the volume of fluid within the cochlea and mechanical damage caused during delivery. There are three key requirements for an intracochlear drug delivery system: (1) a high degree of miniaturization (2) a method for pumping precise and small volumes of fluid into the cochlea in a highly controlled manner, and (3) a method for removing excess fluid from the limited cochlear fluid space. To that end, our group is developing a head-mounted microfluidics-based system for long-term intracochlear drug delivery. We utilize guinea pig animal models for development and demonstration of the device. Central to the system is an infuse-withdraw micropump component that, unlike previous micropump-based systems, has fully integrated drug and fluid storage compartments. Here we characterize the infuse-withdraw capabilities of our micropump, and show experimental results that demonstrate direct drug infusion via cochleostomy in animal models. We utilized DNQX, a glutamate receptor antagonist that suppresses CAPs, as a test drug. We monitored the frequency-dependent changes in auditory nerve CAPs during drug infusion, and observed CAP suppression consistent with the expected drug transport path based on the geometry and tonotopic organization of the cochlea.

Fig. 1

Photograph of our electromagnetic-actuator-based micropump with microfluidics, PCB, and actuator fixture

Fig. 2

Scheme of a 4–port micropump that has a central pump chamber that is fluidically connected to 4 valves, and an integrated drug reservoir. Each valve corresponds to a port, and is numbered

Fig. 3

Photograph of the fluidics portion of the micropump, which comprises several layers of polyimide films (Kapton) that are laminated together

Fig. 4

Scheme of the polyimide, PCB, and 3D-printed actuator fixture layers that make up the 2-port micropump. The numbered layers were fabricated from polyimide, and their thicknesses are indicated

Fig. 5

Scheme of an electromagnetic actuator used for valve control and pumping in its (a) unpowered and (b) powered state

Fig. 6

Volume delivered per stroke of the micropump as a function of load pressure applied at the outlet. Pump scheme 1 (fast infuse) was used to generate these data. DI water was the working fluid

Fig. 7

Flow rate generated by the micropump as a function of time using pump scheme 1 (fast infuse) against an added hydraulic resistance of 1.29 kPa/(μL/min). 6 cm of 75–μm ID PEEK tubing was used to add the resistance, and the average stroke volume for the 7 strokes shown in these data was 0.082 μL. DI water was the working fluid

Fig. 8

Total volume infused as a function of time during a demonstration of reciprocating flow. Pump scheme 2 (slow infuse – 1 min) was used to pump approximately 1.4 μL of DI water over 10 s, every minute for 5 min. Then pump scheme 4 (fast withdraw) was used to withdraw the fluid

Fig. 9

Flow rate generated by the micropump as a function of time using pump scheme 3 (slow infuse – 5 min) during infusion of AP/DNQX into a guinea pig cochlea. 6 cm of 75–μm ID PEEK tubing, 100 cm of 0.25–mm ID Tygon tubing, and the cannula were attached to the pump. One 10–s pump cycle is shown, comprising 8 actuator strokes. AP was the working solution

Fig. 10

Total volume infused as a function of time using pump scheme 3 (slow infuse – 5 min) during infusion of AP/DNQX into a guinea pig cochlea. 6 cm of 75–μm ID PEEK tubing, 100 cm of 0.25–mm ID Tygon tubing, and the cannula were attached to the pump. The inset shows a closeup of the first 15 min of infusion

Fig. 11

(a) Example of a normal CAP for a guinea pig with good hearing, and (b) the (abnormal) CAP for the same guinea pig after 0.6 hours of infusion of 300 μM DNQX in AP using pump scheme 3 (slow – 5 min). The input tone pip was at 16 kHz and 90 dB for both waveforms

Fig. 12

CAP amplitude as a function of time in response to 60-dB tone-pip stimuli during infusion of AP and DNQX into a guinea pig cochlea. Infusion began at t = 0, and amplitudes were normalized to the measurement taken just before t = 0. The gray shaded areas (t = 0 to t = 0.87 hours) represent the times during which artificial perilymph was infused. The yellow shaded areas (t = 0.87 hours to t = 4 hours) represent the times during which DNQX was infused. The asterisks indicate data points that are the mean of either 3 or 4 biological replicates, and the error bars on those points represent the standard error of the mean. All other points have fewer than 3 replicates

Fig. 13

Time at which the CAP amplitude generated in response to a 60-dB tone pip dropped to 20 % of the initial value as a function of frequency. At t = 0 h, pump scheme 3 (slow infuse – 5 min) was initiated to start infusion of AP, and at t = 0.87 h, DNQX infusion began. The CAP amplitude did not drop to 20 % of the initial value at 2.78 kHz during the observational period for any replicate. Error bars indicate the standard error of the mean, and n = 4 biological replicates

Fig. 14

(a) Example of a normal DPOAE spectrum for a guinea pig with good hearing, and (b) the still normal spectrum for the same guinea pig after 1 hour of infusion of 300 μM DNQX in AP using pump scheme 3 (slow infuse – 5 min). f2 was at 16 kHz and 80 dB for both waveforms

Fig. 15

DPOAE (dashed lines, open circles) and CAP (solid lines, closed circles) threshold shifts as a function of time during infusion of AP and DNQX into a guinea pig cochlea. Infusion began at t = 0, and shifts were calculated with respect to measurements taken just before t = 0. The gray shaded areas (t = 0 to t = 0.87 hours) represent the times during which artificial perilymph was infused. The yellow shaded areas (t = 0.87 hours to t = 4 hours) represent the times during which DNQX was infused. The asterisks (CAPs) and hash marks (DPOAEs) indicate data points that are the mean of either 3 or 4 biological replicates, and the error bars on those points represent the standard error of the mean. All other points have fewer than 3 replicates. In cases where there was no CAP response, CAP threshold shifts were set to 120 dB

Fig. 16

Image plots of (a) DPOAE threshold shift, (b) CAP threshold shift, and (c) normalized CAP amplitude as functions of time and frequency. Warmer colors represent increased damage to hearing (larger threshold shifts, smaller CAP amplitudes). Pump scheme 3 (slow infuse – 5 min) was initiated at t = 0, and regions of pre-operation (Pre), artificial perilymph infusion (AP), DNQX infusion (DNQX), and no pumping (Off) are marked. For (c), the color bar is on a log scale, and the CAPs were generated in response to 60–dB tone-pip stimuli. Each pixel represents the mean of up to 4 biological replicates (Figs. 12 and 15 indicate the data points with more or fewer replicates)