**1. Introduction**

Personalized medicine requires healthcare customization with medical practices, treatments, decisions, or products specific for each patient. It is most commonly associated with diagnostic testing to detect the early onset for a change in health state. When applied to personalized therapy, it involves choosing medication based on the patient's genetic content to identify the correct drug for treatment [1] and therapeutic drug monitoring (TDM) to ensure the correct dosing period and optimum concentration to minimize toxicity and improve efficacy [2,3]. TDM requires the monitoring of blood/plasma concentrations and is important when there is a narrow therapeutic range. TDM has been proven to progress anticancer therapies, such as methotrexate [4], as well as in new targeted anticancer agents, such as nilotinib, imatinib, sunitinib, dasatinib, lapatinib, and sorafenib [5], and is critical for immunospuressants [6]. TDM also shows benefits for a wide array of drugs, particularly antibacterials, anticonvulsants, antidepressants, antiretrovirals, antipsychotics, and β-lactam antibiotics [7]. There are a range of methods which can be used to conduct TDM, but it is typically carried out in a laboratory by using liquid chromatography mass spectrometry and immunoassays [8].

For some applications of TDM, there is a desire to either be able to test at-home, such as for patients receiving treatment for chronic conditions, or for a rapid and quick response to enable quicker and more efficient treatment. The most well-known example of TDM is measuring blood glucose levels such that a diabetic can self-dose the correct amount of insulin. There are now many different blood glucose analyzers available on the market for rapid on-site testing; however, these are predominantly based on either an antibody or an immunoassay. For targets where there is insufficient specificity in these approaches, they are usually analysed with a high resolution separation, which is much harder to miniaturise than an immunoassay. This capability has been recently demonstrated through the introduction of a portable electrophoretic device, Medimate [9], which can measure lithium levels in blood. The Medimate is an example of a point-of-care (POC) device based on the micro total analysis system (μTAS) concept [10]. A μTAS offers the possibility to integrate multiple procedures in a portable, low-cost platform which is simple to utilize without affecting the results. TDM of lithium with Medimate has been studied and recently deemed to be clinically useful [11] and is appropriate for self-testing [12]. While the Medimate device is an impressive and important development for the field, lithium is well-separated from other matrix components in the blood, and has a high therapeutic range, making detection relatively simple. Other μTAS devices applicable to TDM have been developed for glutathione [13], creatinine [14], and β-hydroxybutyrate (βHB) [15].

One of the main impediments in developing a μTAS is the complexity and size of the instrumentation required for the on-site analysis, and for at-home monitoring, it needs to be fully automated, simple to use, easy, and low-cost to make. Recently, Shallan et al. [16] developed a microfluidic device featuring two nanojunctions with different pore sizes to create a size and mobility trap (SMT) to purify and concentrate small molecules prior to an integrated electrophoretic separation. Significantly, the nanojunctions were created by dielectric breakdown after the device containing easier-to-fabricate micron-sized channels was bonded, providing a simpler pathway to commercial manufacture. The potential of the device for TDM was demonstrated with the determination of the antibiotic ampicillin from whole blood within 5 min. However, this device was fabricated in Polydimethylsiloxane (PDMS) where large-volume production is not possible through hot embossing or injection molding. Li et al. [17] developed a similar device incorporating two commercially manufactured nanoporous membranes for the concentration and monitoring of albumin in urine as a marker of albuminurea. While functional, the pore sizes were too large for small molecule pharmaceuticals, and the additional fabrication steps to incorporate the two different membranes would significantly increase the cost of the device.

Here, a microfluidic device with an SMT feature created by dielectric breakdown was fabricated in PMMA instead of PDMS because of the well-known ability of hot embossing or injection molding in this material [18]. We demonstrate that similar nanojunctions can be created in PMMA using controlled dielectric breakdown and that a device with two different sized nanojunctions can be used for the analysis of aminoglycoside antibiotic drugs in whole blood.

## **2. Experimental Section**

#### *2.1. Materials and Chemicals*

All solutions were prepared using Milli-Q water (18 M Ω, Millipore, North Ryde, Australia). PDMS (Sylgard 184) elastomer and curing agen<sup>t</sup> were purchased from Dow Corning (Michigan, MI, USA). Potassium thiocyanate (KSCN) was purchased from AJAX Chemicals (Sydney, Australia). 5(6)-Carboxy-X-rhodamine (ROX), R-phycoerythrin (RPE), fluorescein, fluorescamine, bovine serum albumin (BSA), Ferric chloride (FeCl3), aminoglycoside antibiotics (gentamicin, amikacin and tobramycin), hexadimethrine bromide (HDMB), hydroxypropyl methylcellulose (HPMC), sodium tetraborate, di-sodium hydrogen phosphate, and sodium phosphate monobasic for buffer preparation were purchased from Sigma-Aldrich (Sydney, Australia).
