Current Advancements in Transdermal Biosensing and Targeted Drug Delivery

sensors

Review Current Advancements in Transdermal Biosensing and Targeted Drug Delivery

Prem C. Pandey 1,* , Shubhangi Shukla 1, Shelby A. Skoog 2, Ryan D. Boehm 2 and Roger J. Narayan 2,*

1 Department of Chemistry, Indian Institute of Technology (BHU), Varanasi 221005, India; [email protected] 2 Joint Department of Biomedical Engineering, University of North Carolina and North Carolina State University, Raleigh, NC 27695, USA; [email protected] (S.A.S.); [email protected] (R.D.B.) * Correspondence: [email protected] (P.C.P.); [email protected] (R.J.N.); Tel.: +1-919-696-8488 (R.J.N.) Received: 9 January 2019; Accepted: 21 February 2019; Published: 28 February 2019

Abstract: In this manuscript, recent advancements in the area of minimally-invasive transdermal biosensing and drug delivery are reviewed. The administration of therapeutic entities through the skin is complicated by the stratum corneum layer, which serves as a barrier to entry and retards bioavailability. A variety of strategies have been adopted for the enhancement of transdermal permeation for drug delivery and biosensing of various substances. Physical techniques such as iontophoresis, reverse iontophoresis, electroporation, and microneedles offer (a) electrical ampliﬁcation for transdermal sensing of biomolecules and (b) transport of amphiphilic drug molecules to the targeted site in a minimally invasive manner. Iontophoretic delivery involves the application of low currents to the skin as well as the migration of polarized and neutral molecules across it. Transdermal biosensing via microneedles has emerged as a novel approach to replace hypodermic needles. In addition, microneedles have facilitated minimally invasive detection of analytes in body ﬂuids. This review considers recent innovations in the structure and performance of transdermal systems.

Keywords: Iontophoresis; transdermal biosensing; drug delivery; electroporation; microfabrication; microﬂuidics; microneedle; luminescent sensors; ﬂuorescent biosensors

1. Introduction Transdermal drug administration and biosensing has become more widely accepted in recent years. The skin typically exhibits a triple-layered structure, with outermost epidermis, middle dermis, and the innermost subcutis. The dermis is fully vascularized and possesses sweat glands, hair follicles, nerve endings, and lymph vessels. This layer of tissue facilitates the local absorption of drugs. The intercellular lipid bilayer of the stratum corneum, the uppermost layer, constitutes the rate-limiting layer for the migration of hydrophilic drug molecules across it [1–3]. Several methods have been investigated to increase the permeation rate of drugs temporarily and locally, including chemical and electrical mechanisms [4,5]. Chemical agents elevate the rate of permeability of drug molecules by improving the diffusion rate through the stratum corneum [6]. Electrical magniﬁcation of biosensing and drug transport involve two methods: electroporation and iontophoresis [4,5,7,8]. Physicochemical techniques like microneedles [9,10], ultrasound [11,12], and nanomaterials act as driving forces in promoting the transdermal migration of molecules. This review summarizes advance technologies that have been developed for transdermal biosensing and drug delivery.

Sensors 2019, 19, 1028; doi:10.3390/s19051028 www.mdpi.com/journal/sensors Sensors 2019, 19, x FOR PEER REVIEW 2 of 37

Sensors 2019, 19, 1028 2 of 36 2. Electrical Amplification Several strategies have been employed to facilitate the drug delivery and sampling of body fluids 2. Electrical Ampliﬁcation for transdermal biosensing applications. These approaches are considered in the following subsections.Several strategies have been employed to facilitate the drug delivery and sampling of body ﬂuids for transdermal biosensing applications. These approaches are considered in the following subsections. 2.1. Iontophoresis 2.1. Iontophoresis Iontophoresis utilizes a small voltage (usually < 10 V) and a small current density (<0.5 mA/cm2) to driveIontophoresis charged drug utilizes molecules a small through voltage the (usually skin, <10utilizing V) and an aelectrode small current with densitythe same (<0.5 charge mA/cm as the2) moleculeto drive charged that is drug being molecules delivered through [4–5,7]. the skin, It increases utilizing an permeation electrode with of ionic the same drugs, charge provides as the programmablemolecule that is being drug delivered delivery, [4 improves,5,7]. It increases patient permeation compliance of ionicwith drugs, electronic provides reminders, programmable allows patient-specificdrug delivery, improves delivery patientcontrolled compliance by current with adjustments. electronic reminders, A few devices allows using patient-specific this approach delivery have beencontrolled commercially by current translated. adjustments. This A process few devices is believed using this to drive approach drugs have through been commercially preexisting pathways; translated. however,This process it should is believed be noted to drive that drugs the throughlow voltage preexisting could result pathways; in new however, pores or it shouldpathways. be noted Another that similarthe low technique voltage could is electroporation, result in new pores which or uses pathways. a high Another voltage similarpulse (>100 technique V) applied is electroporation, for a short whichtime period usesa (microseconds-milliseconds) high voltage pulse (>100 V) appliedto create for pores a short or channels time period in the (microseconds-milliseconds) lipid bilayer membranes into a create reversible pores manner. or channels This inapproach the lipid has bilayer been membranesutilized to transfect in a reversible cells and manner. it can also This be approach used to deliverhas been drugs utilized across to transfectthe skin. cellsIt takes and approximately it can also be used 1 V toper deliver membrane drugs to across cause the pore skin. formation. It takes Thisapproximately technique can 1 V allow per membrane for delivery to of cause large pore molecules, formation. DNA, This oligonucleotides, technique can and allow other for deliveryrelevant molecules.of large molecules, Combining DNA, this oligonucleotides, technique with the and electroporation other relevant process molecules. may improve Combining transdermal this technique drug withdelivery the electroporation as electroporation process opens may up improve channels transdermal [8] and iontophoresis drug delivery drives as electroporation the molecules opens through up themchannels (Figure [8] and 1). iontophoresis drives the molecules through them (Figure1).

Figure 1. SchematicSchematic illustration illustration showing showing the pathways pathways of of topical topical and and transdermal transdermal delivery, delivery, including electrically assisted delivery by iontophoresis, electroporation, andand electroincorporationelectroincorporation [[7].7].

2.2. Iontophoresis in Combination with Electroporation 2.2. Iontophoresis in Combination with Electroporation 2.2.1. Particulate Delivery 2.2.1. Particulate Delivery This approach involves the encapsulation of drug molecules in vesicles or particles; a current or a This approach involves the encapsulation of drug molecules in vesicles or particles; a current or pressure gradient then promotes movement of the encapsulated drug through electrically created pores a pressure gradient then promotes movement of the encapsulated drug through electrically created in the skin. The movement of drugs is believed to be the result of a combination of several mechanisms, pores in the skin. The movement of drugs is believed to be the result of a combination of several governed by the Nernst-Planck equation and Fick’s ﬁrst law of diffusion. Variations/combinations of mechanisms, governed by the Nernst-Planck equation and Fick’s first law of diffusion. these equations have been proposed which explained the combined effects of pore formation, passive Variations/combinations of these equations have been proposed which explained the combined diffusion, and electrophoresis. A variety of in vitro tests have been conducted to study transport, effects of pore formation, passive diffusion, and electrophoresis. A variety of in vitro tests have been including: (a) horizontal or vertical types (e.g., Franz diffusion cells) of diffusion cells and (b) synthetic conducted to study transport, including: (a) horizontal or vertical types (e.g., Franz diffusion cells) of membranes (e.g., Nucleopore®) to mimic skin; due to the high variability of skin tissue, it is difﬁcult

Sensors 2019, 19, 1028 3 of 36 to ﬁnd membrane materials that possess all of the characteristics of skin (e.g., a combination of hydrophobicity and conductivity). The use of chemical enhancers to improve transdermal delivery has been shown to provide synergy with electrotransport of drugs (e.g., iontophoretic delivery of LHRH; use of heparin with electroporation may keep pores open longer). It should be noted that a residual decrease in skin resistance (i.e., increased skin permeability) has been observed in excised skin. Several studies have been conducted on iontophoresis and electroporation using excised skin and porcine models, which generally indicated the low levels of side effects (e.g., erythema) and a good recovery of skin resistance [4,5,7,8]. The problem of drug load dumping observed from a short circuit can be effectively controlled with improved circuitry design.

2.2.2. Clinical Dermatological Applications Iontophoresis has been used in a variety of clinical applications. Shimizu et al. noted that the iontophoresis using direct current causes burns and pain [13]. They used iontophoresis with alternating current in conjunction with anticholinergic drug delivery to reduce perspiration among palmoplantar hyperhydrosis patients. It was noted that alternating current iontophoresis along with anticholinergic drug delivery caused a more rapid reduction in perspiration than alternating current iontophoresis alone.

2.2.3. Protein Delivery Iontophoresis may aid in the pulsed delivery of large peptides, which would otherwise face difﬁculties in transdermal diffusion [14,15] (e.g., thyrotropin-releasing hormone, vasopressin, calcitonin, and luteinizing hormone-releasing hormone (LHRH)). Electroporation has also been investigated to improve transdermal LHRH delivery. Insulin has been previously delivered under the combined action of electroporation and iontophoresis processes. Insulin monomers should be able to penetrate the skin; however, the high molecular weight of synthetic insulin (hexamer) is not compatible with passive transport. Iontophoretic delivery may result in the formation of depots containing the drug in the skin. The combination of electroporation and iontophoresis may allow for controlled transdermal delivery with skin loading and release.

3. Applications of Iontophoresis

3.1. Iontophoresis in Drug Delivery Iontophoresis is known for enhancing drug permeation through the scarred epidermis. Compared with invasive techniques such as laser therapy, radiotherapy, cryotherapy, pressure therapy, and intralesional injections, iontophoresis is less expensive and free of discomfort. Every dermal injury results in scarring, which is a common consequence of the tissue repair process. Extra deposition of collagen tissue during wound healing creates a tough epidermis at the injury site, which restricts the passive transport of drug through it. Iontophoresis supports drug delivery across the scar epidermis [16] and also through the skin, which lacks peripheral circulation pathways. For instance, dead skin cells lack connections with the systemic circulation; iontophoresis assists with drug transport beyond the dead skin layer. In earlier reports, it was established that iontophoresis facilitated the movement of terbinafine across the nail fold epidermis at a much higher rate (150×) than unassisted passage of the drug. In another case, iontophoretic delivery across folded porcine epidermis in which the fending off passages were essentially blocked was thought to occur via imitating shunts. Iontophoresis reportedly enhanced the transport flux of sodium fluorescein (a marker to track the transport of the drug) across the scar skin epidermis to 51.9 ± 8.82 ng/(cm2 h), a rate 46-fold higher than that associated with passive permeation. As expected, iontophoresis also enhanced the transport flux of sodium fluorescein across the normal skin epidermis by approximately 200-fold over normal skin epidermis; a flux of 601.24 ± Sensors 2019, 19, 1028 4 of 36

62.38 ng/(cm2 h) was noted. The iontophoretic motion of sodium ﬂuorescein beyond the scar epidermis was enhanced several fold as compared with the normal epidermis; epidermal drug retention was also

Sensorsincreased 2019, by19, x iontophoresis. FOR PEER REVIEW These results indicate that iontophoresis is an appropriate approach 4 of for 37 the delivery of therapeutic agents across scarred skin for scar treatment. epidermisIn another was enhancedreport, reverse several iontophoresis fold as compared was studied with as the a sensingnormal mechanismepidermis; epidermal for transdermal drug retentiondrug monitoring, was also utilizing increased the “internal by iontophoresis. standard” calibration These results approach indicate [17]. that Various iontophoresis concentrations is an of appropriatesodium valproate approach (21 µ forM–1 the mM) delivery and aof ﬁxed therapeutic concentration agents of across glutamic scarred acid skin (60 µforM) scar were treatment. introduced into theIn another subdermal report, area reverse of full-thickness iontophoresis excised was porcine studied ear as skin. a sensing The recovered mechanism valproate for transdermal ions were drugcompared monitoring, to the glutamateutilizing the ions “internal (serving standard” as the standard) calibration as a approach means of [17]. determining Various concentrations the subdermal ofvalproate sodium concentration. valproate (21 µM–1 mM) and a fixed concentration of glutamic acid (60 µM) were introducedThe results, into the shown subdermal in Figure area2, indicated of full-thickness that the internal excised porcinestandard ear method skin. could The recovered be used valproatesuccessfully ions to determine were compared the valproate to the concentration; glutamate ions the valproate/glutamate(serving as the standard) ratio was as conﬁrmed a means of to determiningbe dependent the on subdermal the subdermal valproate valproate concentration. concentration while the glutamate level remained constant.

Figure 2. ReverseReverse iontophoretic iontophoretic extraction extraction flux flux rates of valproate as a function function of of time time and and subdermal subdermal concentrationconcentration ( (aa)) in in the the range range 209–1050 209–1050 µM,µM, and ( b)) in in the the range range 21–104.5 21–104.5 µM.µM. Reverse iontophoretic extractionextraction flux flux rates rates of of valproate valproate and and glutamate glutamate as as a a function function of of time time relative relative to to their their values values at at 24 24 h. h. Ratio(c) Ratio of the of thereverse reverse iontophoretic iontophoretic extraction extraction flux flux rates rates of of valproate valproate and and glutamate glutamate as as a a function function of timetime and subdermalsubdermal valproatevalproate concentrationconcentration (range (range 21–104.5 21–104.5µ M).µM). (d Each) Each data data point point represents represents the the mean ± standardstandard deviation deviation (n (n = = 6) 6) [17]. [17].

TheSubramony results, shown et al. in described Figure 2, transdermal indicated that drug the delivery internal using standard the method iontophoresis could be process; used successfullythey considered to determine the delivery the valproate of lidocaine concentration; for anesthesia the andvalproate/glutamate fentanyl for post-operative ratio was confirmed pain [18]. toThe be general dependent process on of the iontophoretic subdermal drug valproate delivery concentration involved an electrode while the patch glutamate with positively level remained charged constant.ions driven transdermally by the anode (accepting negative ions such as chloride from the body); the cathodicSubramony counter electrodeet al. described accepts transdermal the positive biologicaldrug delivery ions (e.g.,using sodium the iontophoresis and potassium process; ions) whilethey considereddisplacing negativethe delivery ions fromof lidocaine its reservoir for anesthesia (Figure3). Thisand drugfentanyl delivery for post-operative function or ionic pain ﬂux [18]. is based The generalupon a derivationprocess of ofiontophoretic Faraday’s law: drug delivery involved an electrode patch with positively charged ions driven transdermally by the anode (acceptingJd = tdIM negatived/ZdF, ions such as chloride from the body); the cathodic counter electrode accepts the positive biological ions (e.g., sodium and potassium ions) where Jd is the ionic ﬂux, td is the drug transport number, I is the current density, Md is the molecular while displacing negative ions from its reservoir (Figure 3). This drug delivery function or ionic flux weight of the drug ion, Zd is the charge of the drug ion, and F is Faraday’s constant. is based upon a derivation of Faraday’s law:

Jd = tdIMd/ZdF, where Jd is the ionic flux, td is the drug transport number, I is the current density, Md is the molecular weight of the drug ion, Zd is the charge of the drug ion, and F is Faraday’s constant.

Sensors 2019, 19, x FOR PEER REVIEW 5 of 37

3.2. Iontophoresis in Transdermal Biosensing

3.2.1. Transdermal Glucose Monitoring Use of an iontophoretic approach to extract interstitial fluid for glucose sampling has been considered; it is important to correlate the blood glucose levels with the glucose levels in the interstitial fluid. Approaches include reverse iontophoresis, microporation of the skin, patch Sensorsdelivery 2019 of, 19 permeation, x FOR PEER enhancers,REVIEW ultrasound to increase transdermal flux, and fluorescence tagging 5 of 37 of glucose (Figure 4). Reverse iontophoresis, which uses low levels of current to transport glucose 3.2.from Iontophoresis interstitial fluidin Transdermal to the sensor Biosensing interface, is the approach utilized by the GlucoWatch Biographer (GlucoWatch; Cygnus, Redwood City, CA, USA); this approach allows for acceptable measurements 3.2.1. Transdermal Glucose Monitoring of glucose [19]. Local skin irritation has been noted with this approach; in addition, the device Sensors 2019, 19, 1028 5 of 36 cannotUse be of worn an iontophoreticduring heavy perspiration. approach to extract interstitial fluid for glucose sampling has been considered; it is important to correlate the blood glucose levels with the glucose levels in the interstitial fluid. Approaches include reverse iontophoresis, microporation of the skin, patch delivery of permeation enhancers, ultrasound to increase transdermal flux, and fluorescence tagging of glucose (Figure 4). Reverse iontophoresis, which uses low levels of current to transport glucose from interstitial fluid to the sensor interface, is the approach utilized by the GlucoWatch Biographer (GlucoWatch; Cygnus, Redwood City, CA, USA); this approach allows for acceptable measurements of glucose [19]. Local skin irritation has been noted with this approach; in addition, the device cannot be worn during heavy perspiration.

Figure 3.3. ((a)) AnAn iontophoreticiontophoretic drugdrug deliverydelivery systemsystem comprisingcomprising donordonor andand receptorreceptor compartmentscompartments along withwith aa currentcurrent sourcesource andand controller. controller. D D+:+: cationic cationic drug; drug; M M+:+: biological biological cations;cations; X X−−:: biological biological anions. ( b) Vyteris Inc.Inc. LidoSiteTM topicaltopical system system [18]. [18].

3.2. IontophoresisSpectRX (Norcross, in Transdermal GA, USA) Biosensing described a technique that creates micropores in the skin by laser burning, which facilitates the transdermal transit of ions for several days. TCPI (Fort Lauderdale, FL, 3.2.1. Transdermal Glucose Monitoring USA) described the use of a patch with a permeation enhancer, which allows for glucose readings to be takenUse with of an a iontophoreticglucose meter approachin the affected to extract area. interstitial fluid for glucose sampling has been considered; it is important to correlate the blood glucose levels with the glucose levels in the interstitial Figure 3. (a) An iontophoretic drug delivery system comprising donor and receptor compartments fluid.along Approaches with a current include source reverse and iontophoresis, controller. D+ microporation: cationic drug; ofM the+: biological skin, patch cations; delivery X−: biological of permeation enhancers,anions. ultrasound(b) Vyteris Inc. to LidoSite increaseTM transdermal topical system flux, [18]. and fluorescence tagging of glucose (Figure4). Reverse iontophoresis, which uses low levels of current to transport glucose from interstitial fluid to the sensorSpectRX interface, (Norcross, is the GA, approach USA) described utilized a by technique the GlucoWatch that creates Biographer micropores (GlucoWatch; in the skin Cygnus,by laser burning,Redwood which City, CA,facilitates USA); the this transdermal approach allowstransit forof ions acceptable for several measurements days. TCPI of(Fort glucose Lauderdale, [19]. Local FL, USA)skin irritation described has the been use of noted a patch with with this a permeation approach; in enhancer, addition, which the device allows cannot for glucose be worn readings during to beheavy taken perspiration. with a glucose meter in the affected area.

Figure 4. Glucose electrode inserted in subcutaneous tissue. Glucose diffuses from the intravasal compartment (G1) into interstitial compartment (G20; it is then taken up by cells if insulin is present. [19].

Figure 4. Glucose electrode inserted in subcutaneous tissue. Glucose Glucose diffuses diffuses from from the intravasal

compartment (G(G11) intointo interstitialinterstitial compartment compartment (G (G20;20; it it is is then then taken taken up up by by cells cells if insulin if insulin is present is present. [19]. [19]. SpectRX (Norcross, GA, USA) described a technique that creates micropores in the skin by laser burning, which facilitates the transdermal transit of ions for several days. TCPI (Fort Lauderdale, FL, USA) described the use of a patch with a permeation enhancer, which allows for glucose readings to be taken with a glucose meter in the affected area. SensorsSensors 20192019,, 1919,, 1028x FOR PEER REVIEW 66 of 36 37

In another report, in vitro reverse iontophoresis was investigated as a mechanism for glucose monitoringIn another in full-thickness report, in vitro skinreverse of hairless iontophoresis mice using was unlabeledinvestigated and as radiolabeled a mechanism (14 forC-labeled) glucose monitoringglucose solutions in full-thickness [20]. The skin experiments of hairless utilized mice using both unlabeledplatinum/glucose and radiolabeled oxidase (Pt-GOD) (14C-labeled) and glucosemodified solutions copper electrodes [20]. The for experiments the delivery utilized of current both (0.36 platinum/glucose mA/cm2) over the oxidase course (Pt-GOD) of two hours. and modifiedThe Pt-GOD copper electrode electrodes was glucose for the deliveryspecific, and of current the modified (0.36 mA/cm copper2 )electrode over the was course able of to two oxidize hours. a Thevariety Pt-GOD of organic electrode species was that glucose contain specific, hydroxyl and the groups. modified The copperresults electrodeof the study was indicated able to oxidize glucose a varietywas able of organicto be electroactively species that contain transported hydroxyl through groups. the The skin results at concentrations of the study indicated proportional glucose to wasthe ablesolution to be bath electroactively to which the transported dermis was through exposed. the skinIt was at noted concentrations that a higher proportional degree of to radiolabeled the solution bathanalyte to whichwas located the dermis at the was anode exposed. than expected; It was noted the thatauthors a higher attributed degree this of radiolabeledobservation to analyte metabolic was locatedbreakdown at the of anodeglucose than into expected; negatively the charged authors metabolites attributed this(e.g., observation lactate and topyruvate) metabolic that breakdown would be ofdrawn glucose to the into anodes negatively of both charged electrodes metabolites and the (e.g., lack lactateof glucose and pyruvate)specificity thatby the would modified be drawn copper to theelectrode. anodes A of correction both electrodes for this and unexpected the lack of signal glucose on specificitythe Pt-GOD by electrode the modified was copperaccomplished electrode. by Aincorporating correction for ascorbic this unexpected acid oxidase signal into on the the process Pt-GOD to electrode remove ascorbic was accomplished acid that was byincorporating drawn to the ascorbicanode (Figure acid oxidase 5). The into study the process demonstrated to remove the ascorbic potential acid that for wasthis drawn approach to the as anode a non-invasive (Figure5). Thetransdermal study demonstrated glucose sensing the potential modality; for this however, approach additional as a non-invasive research transdermal is needed to glucose determine sensing its modality;feasibility however,in vivo. additional research is needed to determine its feasibility in vivo.

Figure 5. a Figure 5. ((a)) ApparentApparent extractionextraction ofof glucoseglucose byby reversereverse iontophoresisiontophoresis inin 22 h.h. ReverseReverse iontophoreticiontophoretic extraction of (b) titrated water and (c) 14C-labeledethanol in 2 h [20]. extraction of (b) titrated water and (c) 14C-labeledethanol in 2 h [20]. A transdermal biosensing approach involving temporary tattoo-based epidermal diagnostic A transdermal biosensing approach involving temporary tattoo-based epidermal diagnostic device combining reverse iontophoretic extraction of interstitial glucose and enzyme-based device combining reverse iontophoretic extraction of interstitial glucose and enzyme-based amperometric biosensor has been recently developed [21]. This reverse iontophoretic biosensing amperometric biosensor has been recently developed [21]. This reverse iontophoretic biosensing system involves a specific type of electrode rather than the typical three-electrode electrochemical system involves a specific type of electrode rather than the typical three-electrode electrochemical biosensing approach. The system involves anodic and cathodic parts, each of which consists of an biosensing approach. The system involves anodic and cathodic parts, each of which consists of an Ag/AgCl reference electrode with reverse iontophoretic working and counter electrodes. Ag/AgCl reference electrode with reverse iontophoretic working and counter electrodes. In this reverse iontophoretic process, extraction of interstitial fluid containing glucose occurs at In this reverse iontophoretic process, extraction of interstitial fluid containing glucose occurs at the cathode. As such, it is modified with the glucose oxidase enzyme for the selective detection of the cathode. As such, it is modified with the glucose oxidase enzyme for the selective detection of glucose in the presence of uric acid, ascorbic acid, or acetaminophen. Issues like skin irritation and glucose in the presence of uric acid, ascorbic acid, or acetaminophen. Issues like skin irritation and biocompatibility were overcome by applying a uniform coating of agarose gel, which also maintains biocompatibility were overcome by applying a uniform coating of agarose gel, which also maintains better contact between the tattoo and the skin. This sensor responded linearly over the concentration better contact between the tattoo and the skin. This sensor responded linearly over the concentration range of 0–100 M, with a limit of detection of 3 M during in vitro studies (Figure6). On body use of range of 0–100 M, with a limit of detection of 3 M during in vitro studies (Figure 6). On body use this reverse iontophoretic-biosensing platform could monitor changes in blood glucose levels without of this reverse iontophoretic-biosensing platform could monitor changes in blood glucose levels entering the skin. without entering the skin.

Sensors 2019, 19, 1028 7 of 36 Sensors 2019, 19, x FOR PEER REVIEW 7 of 37

FigureFigure 6. Tattoo-basedTattoo-based platform platform for noninvasive glucose sensing. ( (11)) Schematic Schematic of of the the printable iontophoretic-sensingiontophoretic-sensing system system displaying displaying the the tattoo-based tattoo-based paper paper (purple), (purple), Ag/AgCl Ag/AgCl electrodes electrodes (silver), (silver),Prussian Prussian Blue electrodes Blue electrodes (black), (black), transparent transparent insulating insulating layer (green), layer (green), and hydrogel and hydrogel layer (blue).layer (blue).(2) Photograph (2) Photograph of a glucose of a glucose iontophoretic iontophoretic sensing sensing tattoo tattoo device device applied applied to a to human a human subject. subject. (3) (Schematic3) Schematic of theof the time time frame frame of a of typical a typical on-body on-body study study and and the differentthe different processes processes involved involved in each in eachphase. phase. (A) Chronoamperometric (A) Chronoamperometric response response of the of tattoo-based the tattoo-based glucose glucose sensor to sensor increasing to increasing glucose µ µ µ glucoseconcentrations concentrations from 0 Mfrom (dashed) 0 µM to(dashed) 100 M to (plot 100 “l”)µM in(plot buffer “l”) in in 10 bufferM increments. in 10 µM increments. (B) Interference (B) µ µ Interferencestudy in the presence study in of the 50 presenceM glucose of (plot 50 µM “a”), glucose followed (plot by subsequent “a”), followed 10 byM additions subsequent of ascorbic 10 µM − additionsacid (plot of “b”), ascorbic uric acid acid (plot (plot “b”), “c”), uric and acid acetaminophen (plot “c”), and (plot acetaminophen “d”). Potential (plot “d”). step toPotential0.1 V step (vs. toAg/AgCl). −0.1 V (vs Medium Ag/AgCl). was Medium phosphate-buffer was phosphate-buffer with 133 mM with NaCl 133 (pH mM 7) NaCl [21]. (pH 7) [21]. Alvarez-Lorenzo and coworkers discussed the use of molecularly imprinted polymers (MIPs), Alvarez-Lorenzo and coworkers discussed the use of molecularly imprinted polymers (MIPs), mainly as drug delivery systems (DDSs); they also mentioned potential applications of MIPs in sensor mainly as drug delivery systems (DDSs); they also mentioned potential applications of MIPs in devices. In feedback-regulated drug delivery systems, the MIPs release drugs based on recognition of a sensor devices. In feedback-regulated drug delivery systems, the MIPs release drugs based on molecule of interest; this approach could also be used in transdermal sensing systems [22]. The highest recognition of a molecule of interest; this approach could also be used in transdermal sensing proﬁle example of this approach is the use of a MIP sensor for glucose sensing in a DDS for insulin. systems [22]. The highest profile example of this approach is the use of a MIP sensor for glucose These devices have been evaluated for determining blood glucose levels; a number of studies have sensing in a DDS for insulin. been conducted to correlate glucose concentrations with pH changes based on (a) metal coordination These devices have been evaluated for determining blood glucose levels; a number of studies due to interactions of functional monomers and cross-linking agents (e.g., [4-(N-vinylbenzyl) have been conducted to correlate glucose concentrations with pH changes based on (a) metal coordination due to interactions of functional monomers and cross-linking agents (e.g., [4-(N-vinylbenzyl)diethylenetriamine) copper(II)] diformate and [(diethylenetriamine) copper(II)]

SensorsSensors2019 2019, 19, 19, 1028, x FOR PEER REVIEW 8 8 of of 36 37

dinitrate), (b) non-polar and polar cross-linking agents for carbohydrate binding discrimination, and diethylenetriamine)(c) insulin releasing copper(II)] gels that diformate react to and glucose [(diethylenetriamine) concentration changes copper(II)] by glucose dinitrate), hydroxyl (b) non-polar group andinteractions polar cross-linking with 3-acrylamidophenylboronic agents for carbohydrate acid binding monomers discrimination, (Figure 7). andIn addition, (c) insulin glucose releasing traps gelshave that been react studied; to glucose for example,concentration a non-covalent changes by glucosenetwork hydroxyl created bygroup ionic interactions associations with of 3-acrylamidophenylboronicpoly(allylamine hydrochloride), acid monomers glucose phosphate (Figure7). In monosodium addition, glucose salt, and traps various have been cross-linking studied; foragents example, has beena non-covalent evaluated. network While these created approaches by ionic associations have been of developed poly(allylamine for capturing hydrochloride), ingested glucoseglucose, phosphate limiting glucose monosodium uptake, salt, and and preventing various spikes cross-linking in blood agents glucose has values, been evaluated. they may Whilealso be theseuseful approaches in sensing havesystems. been developed for capturing ingested glucose, limiting glucose uptake, and preventing spikes in blood glucose values, they may also be useful in sensing systems.

FigureFigure 7. 7.Molecular Molecular imprinting imprinting of of cyclodextrins cyclodextrins as as receptors receptors of of nanometer-scaled nanometer-scaled templates. templates. ( a()a The) The cyclodextrincyclodextrin is is cross-linked cross-linked by by diisocyanate, diisocyanate, in in dimethylsulfoxide, dimethylsulfoxide, inin thethe presence presence of of the the template. template. (b()b The) The vinyl vinyl monomer monomer of of cyclodextrin cyclodextrin is copolymerizedis copolymerized with with methylenebisacrylamide, methylenebisacrylamide, in water, in water, in the in presencethe presence of the of template the template [22]. [22].

Liu et al. investigated transdermal glucose monitoring by reverse iontophoresis using a 3-electrode cell (withLiu aet gold al. investigated working electrode transdermal as well glucose as Ag/AgCl monitoring reference by reverse and counter iontophoresis electrodes) using on a a polycarbonate3-electrode cell sheet; (with a hydrogela gold working membrane electrode served as aswell a reaction as Ag/AgCl area reference [23]. The workingand counter electrode electrodes) was functionalizedon a polycarbonate with an sheet; Os-complex a hydrogel mediator membrane layer (horseradish served as peroxidase a reaction (HRP) area and [23]. poly(ethylene The working glycol)electrode diglycidyl was functionalized ether (PEG-DGE)). with an The Os-complex hydrogel layermediator was composedlayer (horseradish of poly(ethylene peroxidase oxide) (HRP) (PEO) and (4%poly(ethylenew/v), bovine glycol) serum albumin diglycidyl (3% etherw/v), glucose (PEG-DGE)). oxidase The (2 U/ hydrogelµL), and glutaraldehyde layer was composed (0.5% w/ v). of −1 −2 poly(ethyleneAmperometric oxide) characterization (PEO) (4% w/v), of thebovine sensor serum indicated albumin high (3% sensitivity w/v), glucose (14.24 oxidase nA uM (2cm U/µl),) and and aglutaraldehyde good correlation (0.5% coefﬁcient w/v). compared to the calibration standard (r = 0.99). Additionally, since the characterizationAmperometric was completedcharacterization at −0.1 of Vthe (versus sensor Ag/AgCl), indicated high interference sensitivity from (14.24 electroactive nA uM−1cm analytes−2) and ® (e.g.,a good uric correlation acid) was coefficient avoided; the compared GlucoWatch to the ,calibration which operates standard at 0.42 (r = V 0.99). compared Additionally, to its reference, since the cannotcharacterization avoid this complication.was completed Furthermore, at −0.1 V (versus the reverse Ag/AgCl), iontophoretic interference experiments from electroactive on nude analytes mouse 2 skin(e.g., showed uric acid) the was ability avoided; to detect the glucose GlucoWatch in the® 0–18, which mM operates range,with at 0.42 a linear V compared relationship to its (r reference,= 0.99, 2 ® whichcannot is betteravoid thanthis complication. the r = 0.98 of Furthermore, the GlucoWatch the ),reverse and the iontophoretic ability to switch experiments polarity of on the nude electrodes mouse atskin 20 minshowed intervals. the ability It is hypothesized to detect glucose that in reversing the 0 – 18 polarity mM range, periodically with a linear during relationship the sensing (r process2 = 0.99, canwhich help is to better eliminate than irritation the r2 =at 0.98 the of skin-electrode the GlucoWatch interface.®), and the ability to switch polarity of the electrodesIn another at 20 study, min a intervals. reverse-iontophoretic It is hypothesized glucose that sensor reversing for diabetic polarity monitoring periodically of glucose during levels the wassensing considered process [24 can]. Ithelp was to noted eliminate that theirritation only commercially at the skin-electrode available interface. reverse iontophoretic glucose

Sensors 2019, 19, x FOR PEER REVIEW 9 of 37

Sensors 2019, 19, 1028 9 of 36 In another study, a reverse-iontophoretic glucose sensor for diabetic monitoring of glucose levels was considered [24]. It was noted that the only commercially available reverse iontophoretic glucosesensor issensor the GlucoWatch is the GlucoWatch® biographer;® biographer; there arethere reports are reports of skin of irritationskin irritation that wasthat was caused caused by theby thedevice. device. In addition, In addition, there are there inconsistencies are inconsistencies in the ability in of the the device ability to of diagnose the device hypoglycemia to diagnose and hypoglycemiahyperglycemia and in patients hyperglycemia (Figure8 in). patients (Figure 8).

Figure 8. Construction steps for the planar conﬁguration of the screen-printed amperometric transducer. Figure 8. Construction steps for the planar configuration of the screen-printed amperometric (a) PETE support material; (b) printing of conducting silver basal track; (c) printing of insulation layer; transducer. (a) PETE support material; (b) printing of conducting silver basal track; (c) printing of (d) printing of Ag/AgCl pads, the central circular pad is the iontophoresis electrode for reverse insulation layer; (d) printing of Ag/AgCl pads, the central circular pad is the iontophoresis electrode iontophoresis while the other pad is the reference electrode; (e) printing of graphite pads, which serve for reverse iontophoresis while the other pad is the reference electrode; (e) printing of graphite pads, as the working and counter electrodes [24]. which serve as the working and counter electrodes [24]. The authors hypothesized that the cause of patient skin irritation involved the use of glucose oxidaseThe onauthors the electrode hypothesized without that a redoxthe cause mediator, of patient effectively skin irritation trapping involved hydrogen the peroxide use of glucose against oxidasethe skin; on the the peroxide electrode is alsowithout harmful a redox to themediator, glucose effectively oxidase and trapping may limit hydrogen the effectiveness peroxide against of the thesensor. skin; To the combat peroxide these is issues, also harmful the authors to the sought glucose to create oxidase an and amperometric may limit devicethe effectiveness with ferrocene of the as sensor.a redox To mediator combat for these the issues, glucose the oxidase. authors The sought device to wascreate created an amperometric by a series of device screen-printing with ferrocene steps asof Ag,a redox Ag/AgCl, mediator and for graphite the glucose to form oxidase. the electrodes The device on poly(ethylenewas created by terephthalate) a series of screen-printing sheets; one of stepsthe graphite of Ag, electrodesAg/AgCl, and was graphite subsequently to form functionalized the electrodes by successivelyon poly(ethylene drop-casting terephthalate) ferrocene sheets; and oneglucose of oxidase the graphite solutions electrodes (Figure8 ). was The subsequently amperometric functionalized results of the device by successively showed a linear drop-casting range for ferroceneglucose sensing and glucose (0-4 mM) oxidase and asolutions response (Figure time of 8). 10 The s. When amperometric tested on aresults trans-membrane of the device with showed reverse a lineariontophoresis, range for the glucose device was sensing able to (0-4 detect mM) glucose and a levels response with a time good of linear 10 s. relationship When tested (r2 = on 0.99) a trans-membranecompared to the known with reverse concentrations iontophoresis, (3, 5, and the 15 device mM); was these able concentrations to detect glucose mimicked levels hypoglycemic, with a good 2 linearnormal, relationship and hyperglycemic (r = 0.99) conditions. compared The to authors the known concluded concentrations that the ferrocene-mediated (3, 5, and 15 mM); glucose these concentrations mimicked hypoglycemic, normal, and hyperglycemic conditions. The authors sensor may be useful for applications in reverse iontophoretic transdermal sensing; a fast response time, concluded that the ferrocene-mediated glucose sensor may be useful for applications in reverse reproducibility, and good sensitivity were demonstrated. iontophoretic transdermal sensing; a fast response time, reproducibility, and good sensitivity were demonstrated.3.2.2. Transdermal Sampling of Urea and Homocysteine

3.2.2.Ching Transdermal et al. Sampling investigated of Urea non-invasive and Homocysteine transdermal extraction of homocysteine and urea simultaneously by reverse iontophoresis [25]. Earlier reports have described an enhancement in the extractionChing et of al. urea investigated using the non-invasive reverse iontophoresis transdermal approach. extraction An in of vitro homocysteinemodel consisting and urea of simultaneouslythree layers that by resemble reverse theiontophoresis triple-layered [25]. pattern Earlier ofreports skin (epidermis,have described dermis, an enhancement and subcutis), in wasthe extractionused to detect of urea urea using and the homocysteine. reverse iontophoresis The diffusion approach. cell imitated An in vitro the function model consisting of the dermis of three and layerssubcutis that layers; resemble porcine the skintriple-layered with a thickness pattern of of 240 skinµ m(epidermis, replicated dermis, the epidermis. and subcutis), Electromigration was used to of detecturea, which urea and is an homocysteine. ionic molecule, The and diffusion electro-osmosis cell imitated of homocysteine, the function whichof the dermis is an uncharged and subcutis and layers;hydrophilic porcine molecule, skin with facilitate a thickness transport of 240 across µm thereplicated layers. the epidermis. Electromigration of urea, whichDiffusion is an ionic in combination molecule, and with electro-osmosis reverse iontophoresis of homocysteine, ampliﬁed the which extraction is an of uncharged homocysteine; and hydrophilicthere was much molecule, higher facilitate extraction transport of homocysteine across the layers. for direct current (dc) and four symmetrical biphasicDiffusion direct currentsin combination (SBdc) thanwith forreverse samples iontophoresis that relied solely amplified on diffusion. the extraction Further, of ithomocysteine; was observed therethat the was transfer much of higher homocysteine extraction was of lower homocysteine at the dc-anode for direct and current dc-cathode (dc) andthan four for the symmetrical four SBdc

Sensors 2019, 19, 1028 10 of 36 approach. This result is attributed to polarization of the porcine skin associated with the usage of direct current, which retarded the migration of homocysteine (neutral molecule) during reverse iontophoretic operation. Furthermore, similar behavior was observed in the case of the extraction of urea; reverse iontophoretic operation facilitated greater levels of extraction than the diffusion process. Urea is negatively charged; as such, its extraction is more reinforced at the dc anode than at the dc cathode. Urea also followed phase-duration dependent extraction in case of SBdc. The extraction of urea was found to be optimum when 60 s PDSBdc was employed. The reverse iontophoretic operation served as an appropriate approach for transdermal extraction of both substances.

4. Microfabrication Technique for Transdermal Sensing and Drug Delivery Microfabrication techniques have been used to create micro- and nano-scale features; for example, the techniques may be used to: (a) impart desirable surface properties, (b) create circuitry for electronic components, and (c) facilitate high-throughput and large-scale production of devices with medical relevance. These techniques can aid in the creation of micro-scale devices and biosensors for transdermal sensor applications [26–28]. For instance, microfabrication techniques can be employed to fabricate microneedle devices for penetration of the skin into the epidermis or dermis. Previous studies have created arrays with up to thousands of microneedles (noted diameters of 10–100 µm) out of polymers and silicon for delivery of therapeutic agents, vaccines, or DNA. Furthermore, micromolding the master structures of the microneedles can enable processing of microneedles out of biodegradable materials (e.g., carboxymethylcellulose, amylopectin, polylactic acid, polyglycolic acid, and poly(lactic-co-glycolic) acid). Furthermore, these microfabrication techniques can be used in the development of biosensors (e.g., electrochemical or mechanical loading sensors). We have recently demonstrated a screen- printed electrode in the three-electrode configuration, including working (WE), reference (RE) and counter (CE) electrodes, as shown in Figure9. The working electrode was modified with a catalytic layer for detection of hydrogen peroxide either generated at the electrode surface or in the sample. The active catalytic layer may include enzymes, palladium nanoparticles with controlled nanogeometries, or quantum dots [26]. The electrodes were printed over a flexible material, which can be integrated with a hollow microneedle assembly for transdermal sensing [26]. Similarly, a solid-state ion sensor in a two-electrode configuration assembled with a hollow microneedle was designed for transdermal sensing of electrolyte (shown in Figure 10)[27]. The all solid-state microneedle assembled potentiometric sensor was comprised of a PVC matrix membrane ion sensing layer over a non-specific ion carrier impregnated membrane made from a suspension of siloxane-polyindole-gold nanoparticles [27] to maintain a constant dipolar potential during potentiometric transdermal sensing. The reference electrode was an Ag/AgCl electrode covered with a p-toluenesulfonate impregated polyndole-gold nanoparticle-organically modified silicate membrane [27], which was combined with the microneedle-assembled electrode to yield a solid-state ion sensor. Glucose oxidase is used in this manner in glucose meters for diabetic patients; microfabrication techniques can be utilized to (a) pattern the interfaces of these types of biosensors with the biological environment and (b) create the circuitry needed to transmit electronic signals produced by the enzymatic interactions. However, currently these devices have been investigated mainly for implantation (e.g., an implant in the rat peritoneal cavity for ethanol detection or a glucose sensitive sensor that releases antibiotics when glucose produced by bacteria at an orthopedic implant site changes the viscosity of the solution surrounding a microcantilever) and lab-on-a-chip applications [27,28]. Sensors 2019, 19, x FOR PEER REVIEW 10 of 36

biphasic direct currents (SBdc) than for samples that relied solely on diffusion. Further, it was observed that the transfer of homocysteine was lower at the dc-anode and dc-cathode than for the four SBdc approach. This result is attributed to polarization of the porcine skin associated with the usage of direct current, which retarded the migration of homocysteine (neutral molecule) during reverse iontophoretic operation. Furthermore, similar behavior was observed in the case of the extraction of urea; reverse iontophoretic operation facilitated greater levels of extraction than the diffusion process. Urea is negatively charged; as such, its extraction is more reinforced at the dc anode than at the dc cathode. Urea also followed phase-duration dependent extraction in case of SBdc. The extraction of urea was found to be optimum when 60s PDSBdc was employed. The reverse iontophoretic operation served as an appropriate approach for transdermal extraction of Sensors 2019, 19, 1028 11 of 36 both substances.

FigureFigure 9. Construction9. Construction of microneedle of microneedle assembled assemble screen-printedd screen-printed electrode (SPE)-basedelectrode (SPE)-based amperometric sensoramperometric for hydrogen sensor peroxide. for hydrogen (A) Electrodeperoxide. assembly(A) Electrode connected assembly to aconnected dedicated to electrochemical a dedicated detector;electrochemical (B) Microneedle detector; assembled (B) Microneedle SPE in three assembled electrode SPE conﬁguration; in three electrode (C) Electrode configuration; connector (C ﬁxed) toElectrode SPE: WE =connector working fixed electrode, to SPE: RE WE = = reference working electrodeelectrode, andRE = CE reference = counter electrode electrode) and CE [26 =]. counter electrode) [26].

Figure 10. Construction of a microneedle assembly-based screen-printed electrode (SPE)-based potentiometric potassium ion sensor [27].

5. Nanotechnology in Transdermal Biosensing The review by Cintenza on the biomedical applications of quantum dots (QDs) touches on a few areas that are relevant to transdermal sensing [29]. Quantum dots are semiconductor nanocrystals in the size range of 0.2 to 100 nm (crystals in the size range of 2 to 10 nm are typically used). Due to the small size of these nanoparticles, they exhibit discrete energy levels with the behavior of the electrons and holes in their orbitals governed by quantum mechanics; as a result, small changes in their size (which can be controlled by fabrication techniques) cause changes in their ﬂuorescence properties [30–32]. These nanostructures have tunable ﬂuorescence properties by controlling their size, material composition, and fabrication temperature; consequently, they have the ability to outperform

Sensors 2019, 19, 1028 12 of 36 Sensors 2019, 19, x FOR PEER REVIEW 12 of 37

the ability to outperform organic fluorophores due to their strong signal, tunable emission organic fluorophores due to their strong signal, tunable emission wavelengths, narrow emission wavelengths, narrow emission profiles, and potential for use with only one light source for profiles,excitation and potential for a variety for use of withemission only signals. one light While source these for nanocrystals excitation forhave a variety gained ofattention emission for signals. a Whilevariety these ofnanocrystals biomedical haveapplications gained (e.g., attention in vivo for imaging, a variety drug of biomedical delivery systems), applications they also (e.g., havein vivo imaging,applications drug delivery in sensing systems), technology they [31]. also The have excellent applications optical inproperties sensing of technology QDs have [allowed31]. The them excellent opticalto propertiesfunction in various of QDs FRET have assays allowed for themthe detection to function of biologically in various relevant FRET assaysmolecules for (e.g., the detectionnucleic of biologicallyacids, proteins, relevant and molecules antibodies). (e.g., nucleic acids, proteins, and antibodies). The FRETThe FRET characteristics characteristics of of the the QDsQDs have led led to to the the development development of sensors of sensors for the for detection the detection of of specificspecific nucleic nucleic acidacid hybridization reactions reactions (e.g., (e.g., green-emitting green-emitting CdSe/ZnS CdSe/ZnS QD donors QD coupled donors with coupled Cy3 and rhodamine RhR acceptor dyes) (Figure 11), detection of glucose levels in human serum with Cy3 and rhodamine RhR acceptor dyes) (Figure 11), detection of glucose levels in human serum (e.g., conjugated CdTe-conclavin A QD donors and gold nanoparticles with thiolated b-cyclodextrin (e.g., conjugated CdTe-conclavin A QD donors and gold nanoparticles with thiolated b-cyclodextrin modification as the acceptors), and detection of metabolic protease activities (e.g., CdSe/ZnS with a modificationrhodamine-labeled as the acceptors), tetrapeptide and RGDC detection coating of as metabolic the acceptor protease molecule). activities (e.g., CdSe/ZnS with a rhodamine-labeled tetrapeptide RGDC coating as the acceptor molecule).

FigureFigure 11. Size-dependent 11. Size-dependent absorption absorption and and emission emission spectra spectra of of CdSe CdSe semiconductor semiconductor nanocrystals nanocrystals with variouswith sizes. various (a) sizes. Size-dependent (a) Size-dependent luminescence luminescence color color and and(b) schematic (b) schematic presentation presentation of of size, size, color, color, and emission wavelength of CdSe–ZnS QDs. (c) Absorption (solid lines) and emission (broken and emission wavelength of CdSe–ZnS QDs. (c) Absorption (solid lines) and emission (broken lines) lines) spectra of CdSe QDs with various sizes [33]. spectra of CdSe QDs with various sizes [33].

6. Microneedles6. Microneedles Technique Technique in in Transdermal Transdermal Sensing Sensing and Drug Delivery Delivery 6.1. Polymer-Based Microneedles for Drug Delivery 6.1. Polymer-Based Microneedles for Drug Delivery 2 Frisk et al. described the fabrication and testing of a microneedle array patch device (1 cm ) for2 Frisktransdermal et al. describeddrug delivery, the fabricationfeaturing 200 and µm testing tall microneedles of a microneedle that contained array patchout-of-plane device hollow (1 cm ) for transdermalbores, using drug a novel delivery, deep featuring reactive-ion 200 etchingµm tall (DRIE) microneedles process [34]. that The contained array was out-of-plane coupled with hollow a bores,heat-activated/printed using a novel deep circuit reactive-ion board-controlled etching (DRIE)(PCB) drug process delivery [34 ].fluid The actuator array wasmade coupled of silicone with a heat-activated/printedrubber, which expelled circuit the drug board-controlled load from a reservoir (PCB) through drug delivery the microneedles fluid actuator when madeheated. of This silicone rubber,concept which was expelled tested with the Fast drug Green load dye from on the a hands reservoir of human through subjects the and microneedles radio-labeled when inulin heated.in Thisan concept in vivo wasrat model. tested The with results Fast of Greenthis study dye indicated on the that hands the offluid human delivery subjects rate in each and case radio-labeled could inulinbe in controlled an in vivo byrat the model. input power The results (150–450 of thismW). study A rate indicated of 1 µL/min that thewas fluid applied delivery in the ratehuman in each case couldsubjects, be allowing controlled for successful by the input delivery power of the (150–450 Fast Green mW). into A the rate tissue; of 1however,µL/min it waswas appliednoted that in the this flow rate was too high for tissue absorption of all of the liquid. Residual dye remained on the human subjects, allowing for successful delivery of the Fast Green into the tissue; however, it was skin surface after patch removal. In the rat study, a rate of 1.8 µL/h was utilized, with no observed notedinulin that this residue flow on rate the was skin too following high for patch tissue removal. absorption Sampling of all of of the the liquid. rat urine Residual following dye remainedfluid on thedelivery skin surface indicated after successful patch removal. delivery of In the the radio-labeled rat study, a3H-inulin; rate of 1.8 samplingµL/h was was utilized,conducted with at no observed inulin residue on the skin following patch removal. Sampling of the rat urine following fluid delivery indicated successful delivery of the radio-labeled 3H-inulin; sampling was conducted at 20-min intervals, with steadily increasing levels of the inulin observed. This result indicated that the drug was being delivered and absorbed into the circulatory system of the rats. SensorsSensors2019 2019, 19, ,19 1028, x FOR PEER REVIEW 13 of13 37 of 36

20-min intervals, with steadily increasing levels of the inulin observed. This result indicated that the drugTargeted was being drug delivered release and using absorbed the microneedle into the circulatory technology system has of the evolved rats. recently as a second generationTargeted of minimally drug release invasive using means the microneedle of transdermal technology drug delivery. has evolved Chen recently et al. reviewedas a second NIR lasergeneration irradiation-triggered of minimally releaseinvasive of means a drug of from transdermal the microneedles drug delivery. at a directed Chen et site al. [35 reviewed–38]. When NIR the microneedleslaser irradiation-triggered were irradiated release with of NIR a drug light, from the the mobility microneedles of polymeric at a directed chains site in the [35–38]. microneedles When enabledthe microneedles drug release; were the release irradiated was with terminated NIR light, when the laser mobility light was of removed. polymeric The chains microneedles in the microneedles enabled drug release; the release was terminated when laser light was removed. The offer an effective means for drug administration in the body; the system is repeatedly switchable, microneedles offer an effective means for drug administration in the body; the system is repeatedly which enables release of a consistent dosage of the drug. switchable, which enables release of a consistent dosage of the drug. In another study, solid and robust R6G-loaded microneedles were replicated from a master In another study, solid and robust R6G-loaded microneedles were replicated from a master structure [37]. The quantity of drug liberated by the formulated microneedles could be altered structure [37]. The quantity of drug liberated by the formulated microneedles could be altered by bychanging changing the the number number of of on/off on/off laser laser cycles. cycles. In In vitro vitro drugdrug insertion insertion was was studied studied using using porcine porcine cadavercadaver skin skin tissues tissues and and R6G R6G as as model model drug;drug; anan applicationapplication force force of of 10 10 N/patch N/patch was was used used in inthis this studystudy (Figure (Figure 12 12).). Balanced Balanced forces forces facilitatedfacilitated the movement of of microneedles microneedles during during the the puncture puncture by by subsumingsubsuming the the skin skin resistance. resistance. In general, In general, a larger a larger insertion insertion force force was associated was associated with a with greater a greater insertion depthinsertion and a depth higher and level a higher of drug level permeation. of drug permeation.In vivo applicability In vivo applicability of NIR absorptive of NIR microneedles absorptive weremicroneedles evaluated bywere loading evaluated the devicesby loading with the an devices anti-cancer with an drug anti-cancer known asdrug doxorubicin known as (DOX)doxorubicin into rat skin(DOX) with intofour on/offrat skin laser with cycles. four on/off The NIR-light laser cycles. responsive The NIR-light microneedles responsive are a promising microneedles transdermal are a drugpromising delivery transdermal system as theydrug offerdelivery controlled system andas they on-demand offer controlled drug release. and on-demand drug release.

FigureFigure 12. 12.(a )(a Schematic) Schematic illustrations illustrations ofof on-demandon-demand transdermal drug drug delivery delivery using using near-infrared near-infrared (NIR)(NIR) light-responsive light-responsive microneedles microneedles (MNs), (MNs), composed of of polycaprolactone polycaprolactone and and NIR NIR absorbers absorbers as as 6 2 wellwell as as silica-coated silica-coated lanthanum lanthanum hexaboride hexaboride (LaB(LaB6-SiO 2)) nanostructures. nanostructures. (b ()b Temperature) Temperature changes changes of of R6G-loadedR6G-loaded microneedles microneedles after after continuous continuous exposureexposure to 808 808 nm nm NIR NIR laser laser light light at at an an output output power power of of −2 7 W7 W cm cm−2.(. c(c––hh)) Infrared Infrared thermalthermal images of of microneedles microneedles after after continuous continuous irradiation irradiation for for 10, 10, 30, 30,90, 90, 120,120, 240, 240, and and 300 300 s. s. The The insets insets in in ( b(b)) show show thethe light-triggeredlight-triggered melting melting behavior behavior of of microneedles microneedles [37]. [37 ].

SensorsSensors 20192019,, 1919,, 1028x FOR PEER REVIEW 1414 of 36 37

Ashraf et al. described a method for designing and fabricating a transdermal drug delivery systemAshraf consisting et al. describedof out-of-plane a method hollow for silicon designing microneedles, and fabricating which a are transdermal coupled to drug a valve-less delivery systempiezoelectric consisting actuated of out-of-plane micropump hollow [39]. They silicon conducted microneedles, theoretical which analyses are coupled to understand to a valve-less the piezoelectricmechanical function actuated micropumpof the microneedles, [39]. They the conducted piezoelectric theoretical mechanism analyses of to the understand pump, and the mechanicalcomputational function fluid ofdynamics; the microneedles, a multifield the computational piezoelectric mechanism examination of the of pump,the entire and working computational device fluidwas also dynamics; performed. a multifield The microneedles computational were examinationfabricated using of the a combination entire working of deep device reactive was alsoion performed.etching (DRIE), The wet microneedles etching, and were gas fabricatedetching of usingmasked a combinationsilicon wafers; of microneedles deep reactive with ion an etching inner (DRIE),diameter wet of etching,60 µm, an and outer gas etchingdiameter of of masked 150 µm, silicon a center-to-center wafers; microneedles spacing with of 1000 an inner µm, diameterand high ofaspect 60 µ m,ratio an tips outer were diameter created of in 150 thisµm, manner. a center-to-center The theoretical spacing and ofnumerical 1000 µm, analyses and high of aspect the system ratio tipsindicated were createdthat skin in penetration this manner. of The these theoretical devices andcould numerical occur at analyses3.18 MPa; of under the system these indicated conditions, that a skinmaximum penetration stress ofon these the devices3.26 MPa could was occurobserved at 3.18 with MPa; minimal under thesedeflection conditions, of the aluminal maximum structure stress on(these the 3.26 values MPa are was below observed the yield with minimalstrength deflection of the material). of the luminal When structure coupling (these the piezoelectric values are belowmicropump the yield to strengththe system, of the the material). theoretical When maximum coupling pumping the piezoelectric rate was micropump determined to tothe be system, 83.99 theµl/min theoretical at 250 Hz maximum and 100 pumpingV; it was also rate noted was determined that this value to be corresponds 83.99 µL/min to pulsatile at 250 Hz flow, and which 100 V; itis wasappropriate also noted for thatpharmacologic this value corresponds treatment of to some pulsatile cardiovascular flow, which disorders is appropriate (e.g., hypertension). for pharmacologic treatment of some cardiovascular disorders (e.g., hypertension). 6.2. Microneedles for Chemical and Electrochemical Biosensing 6.2. Microneedles for Chemical and Electrochemical Biosensing Goud et al. fabricated an electrochemical microneedle-based biosensor and characterized the glucoseGoud sensor et al. fabricatedfunctionality an electrochemical of the minimally-invasive microneedle-based transdermal biosensor device and characterized[40]. The device the glucosecontained sensor an amperometric functionality ofbiosensor, the minimally-invasive a working electrode transdermal of carbon device nanotubes/platinized [40]. The device contained glassy ancarbon amperometric within a sol-gel biosensor, matrix a working of zirconia electrode and Nafion; of carbon the nanotubes/platinizedelectrode was functionalized glassy carbon with glucose within aoxidase. sol-gel The matrix electrodes of zirconia were and housed Nafion; within the a electrodemicrofluidic was chamber functionalized made of with SU8 glucosephotoresist oxidase. and Thesealed electrodes with polydimethylsiloxane were housed within a microfluidic (PDMS). Conceptually, chamber made the of SU8 analytes photoresist were and drawn sealed to with this polydimethylsiloxanemicrofluidic electrode (PDMS).chamber Conceptually, through organically the analytes modified were silicates drawn tomicroneedles this microfluidic that electrodehad been chamberfabricated through by two-photon organically polymerization modified silicates onto the microneedles external surface that had of the been sensor fabricated unit (Figure by two-photon 13). The polymerizationmicroneedles exhibited onto the base external diameters surface of of100 the µm sensor and heights unit (Figure of a few 13). hundred The microneedles micrometers. exhibited Cyclic basevoltammetric diameters and of amperometric 100 µm and heightsexaminations of a few of the hundred sensor electrodes micrometers. in glucose Cyclic solutions voltammetric (0–20 mM) and amperometricshowed nearly examinations instantaneous of responses the sensor times; electrodes the measured in glucose current solutions values (0–20 ranged mM) from showed 1.0–2.5 nearly µA instantaneousas the glucose responses concentration times; in the solution measured changed current from values 5–20 ranged mM. from This 1.0–2.5 study µ demonstratedA as the glucose the concentrationpotential to use in an solution integrated changed biosensor from 5–20device mM. with This microneedle study demonstrated arrays for transdermal the potential sensing to use anof integratedglucose levels. biosensor device with microneedle arrays for transdermal sensing of glucose levels.

Figure 13. Fabrication process of hollow silicon out-of-plane microneedlesmicroneedles [[41].41].

Sensors 2019, 19, 1028 15 of 36

Sensors 2019, 19, x FOR PEER REVIEW 15 of 36 Miller et al. created polymer microneedle arrays with integrated carbon fiber electrodes for electrochemicalMiller et sensingal. created of polymer biological microneedle analytes (e.g.,arrays ascorbic with integrated acid and carbon hydrogen fiber electrodes peroxide) for [42 ]. Theelectrochemical microneedles sensing were triangular of biological pyramids analytes in (e.g., shape, ascorbic with acid base and lengths hydrogen of ~1120 peroxide)µm, heights[42]. Theof ~1030microneedlesµm, and circular were triangular bores of ~375pyramidsµm (Figurein shape, 14 ).with Penetration base lengths by theof ~1120 microneedle µm, heights arrays of into~1030 the stratumµm, and corneum circular layer bores of of excised ~375 µm porcine (Figure skin 14). was Penetration confirmed by by the the microneedle delivery of trypanarrays into blue the dye. Electrochemicalstratum corneum experiments layer of excised involving porcine the carbon-fiberskin was confirmed electrodes by the within delivery the hollowof trypan microneedles blue dye. confirmedElectrochemical that ascorbic experiments acid and involving hydrogen the peroxide carbon-fiber values electrodes couldbe within determined. the hollow The microneedles carbon fibers confirmed that ascorbic acid and hydrogen peroxide values could be determined. The carbon fibers were functionalized to enable detection of the analytes. The fibers were activated in KOH at 1.3 V (vs. were functionalized to enable detection of the analytes. The fibers were activated in KOH at 1.3 V Ag/AgCl); they were subsequently exposed to 2-amino-4-nitrophenol and sodium nitrite to create (vs. Ag/AgCl); they were subsequently exposed to 2-amino-4-nitrophenol and sodium nitrite to diazonium salt in situ. They were then scanned with cyclic voltammetry (CV) (0.4 V to −0.8 V, vs. create diazonium salt in situ. They were then scanned with cyclic voltammetry (CV) (0.4 V to −0.8 V, Ag/AgCl) to graft the 2-nitrophenol and reduce it to aminophenol; for hydrogen peroxide detection, the vs. Ag/AgCl) to graft the 2-nitrophenol and reduce it to aminophenol; for hydrogen peroxide − fibersdetection, were exposed the fibers to palladiumwere exposed (II) to chloride palladium at a potential(II) chloride of at0.8 a potential V (vs. Ag/AgCl). of −0.8 V (vs. Electrochemical Ag/AgCl). detectionElectrochemical indicated detection dose-response indicated behavior dose-response of the sensor behavior for hydrogen of the sensor peroxide, for hydrogen with a linear peroxide, range betweenwith a 100–500 linear rangeµM andbetween a predicted 100–500 detection µM and limita predicted of 15 µ detectionM; for the lim ascorbateit of 15 µM; analyte, for the detection ascorbate of a 1 mManalyte, concentration detection wasof a 1 possible mM concentration at 195 mV. was possible at 195 mV.

FigureFigure 14. 14.Schematic Schematic showing showing steps steps (a(a–)f–)(f used) used for for assembly assembly ofof thethe microneedlemicroneedle device. device. Images Images of of microneedlemicroneedle array array and and cadaveric cadaveric porcine porcine skin skin after after microneedlemicroneedle insertion.insertion. ( (gg) )Optical Optical micrograph micrograph showingshowing delivery delivery of trypanof trypan blue blue into into microneedle-fabricated microneedle-fabricated pores pores within within cadaveric cadaveric porcine porcine skin skin scale barscale 1 = mm.bar 1 ( h=) Opticalmm. (h) micrograph Optical micrograph showing showing hollow microneedles hollow microneedles before insertion before insertion into cadaveric into porcinecadaveric skin. (porcinei) Optical skin. micrograph (i) Optical showing micrograph hollow showing microneedles hollow after microneedles insertion into after cadaveric insertion porcine into skincadaveric [42]. porcine skin [42].

In addition,In addition, the the e-shell e-shell 200 200 acrylate-based acrylate-based polymer polymer underwentunderwent MTT MTT assay assay testing testing with with neonatal neonatal epidermalepidermal keratinocytes keratinocytes and and human human dermal dermal fibroblasts fibroblasts (Figure(Figure 1515).). The The results results showed showed a asignificant significant decreasedecrease in the in viabilitythe viability of the of cellsthe cells grown grown on the on polymer the polymer compared compared to the control;to the control; however, however, the decrease the decrease in viability was not at levels that would be of concern for in vivo transdermal sensing in viability was not at levels that would be of concern for in vivo transdermal sensing applications. applications. Windmiller et al. fabricated triangle-based pyramidal microneedle arrays with hollow pores, Windmiller et al. fabricated triangle-based pyramidal microneedle arrays with hollow pores, and packed the hollow bores with metalized carbon paste (functionalized with rhodium and lactate and packed the hollow bores with metalized carbon paste (functionalized with rhodium and lactate

Sensors 2019, 19, 1028 16 of 36

Sensors 2019, 19, x FOR PEER REVIEW 16 of 37 oxidase); the ability of these electrochemical sensors to detect lactate was assessed. Three-by-three oxidase); the ability of these electrochemical sensors to detect lactate was assessed. Three-by-three arrays of the pyramidal microneedles (1500 µm tall, 425 µm bore) were fabricated out of the e-shell 200 arrays of the pyramidal microneedles (1500 µm tall, 425 µm bore) were fabricated out of the e-shell acrylate-based polymer using a dynamic light micro stereolithographic rapid prototyping system [43]. 200 acrylate-based polymer using a dynamic light micro stereolithographic rapid prototyping Electrochemical characterization of the electrodes in hydrogen peroxide solutions showed good system [43]. Electrochemical characterization of the electrodes in hydrogen peroxide solutions µ 2 µ linearityshowed in good the 0–500 linearityM in range the 0 (r – 500= 0.999) µM range with a(r lower2 = 0.999) detection with a limitlower of detection ~20 M; limit highly of ~20 repeatable µM; resultshighly were repeatable also observed results were with also repacking observed ofwith the repacking carbon pasteof the intocarbon bores paste of into the bores microneedles of the followingmicroneedles characterization. following characterization.

FigureFigure 15. 15.MTT MTT viability viability data data for for cells cells growngrown onon e-shell 200 acrylate-based acrylate-based polymer polymer and and glass glass cover cover slip.slip. (a ) ( MTTa) MTT viability viability of human of human epidermal epidermal keratinocytes keratinocytes grown grown on e-Shell on e-Shell 200 acrylate-based 200 acrylate-based polymer comparedpolymer to compared glass cover to glass slip. A cover and slip. B denote A and statistical B denote differences statistical differencesp < 0.05 between p 0.05 the between polymer the and thepolymer control. and(b) MTT the viabilitycontrol. ( ofb) human MTT viability dermal ﬁbroblasts of human grown dermal on fibroblasts e-Shell 200 grown acrylate-based on e-Shell polymer 200 comparedacrylate-based to glass polymer coverslip. compared A and Bto denoteglass coverslip. statistical A differences and B denotep < statistical 0.05 between differences the polymer p < 0.05 and thebetween control the [42 ].polymer and the control [42].

ExaminationExamination of of lactate lactate detection detection showed showed high high linearity linearity over over (r2 (r=2 0.990)= 0.990) a rangea range of 0–8of 0 mM– 8 mM lactate solutions,lactate solutions, with anestimated with an estimated detection detection limit of 0.42 limit mM of lactate; 0.42 mM this lactate; value this is well value below is well physiologic below levels.physiologic In addition, levels. lactate In addition, was able lactate to be was detected able to under be detected challenge under conditions challenge with conditions uric acid, with citric uric acid, andacid, acetaminophen citric acid, and as acetaminophen competing analytes as competing with minimal analytes deviation with minimal from expected deviation values. from expected The results ofvalues. this study The show results progress of this towardstudy show an amperometric progress toward sensor an for amperometric transdermal sensor sensing for of transdermal bioanalytes. sensingElectrochemical of bioanalytes. characterization of the electrodes showed linearity (r2 = 0.995) over a range of 2 glutamateElectrochemical concentrations characterization (0–140 µM) inof buffer,the electrodes with a showed sensitivity linearity of 7.129 (r = nA/ 0.995)µM over and a a range predicted of glutamate concentrations (0–140 µM) in buffer, with a sensitivity of 7.129 nA/µM and a predicted limit-of-detection of 3 µM; this limit is well-below the value observed physiologically. Additionally, limit-of-detection of 3 µM; this limit is well-below the value observed physiologically. Additionally, linearity (r2 = 0.992) was observed over the same concentration range (0–140 µM) when the electrodes linearity (r2 = 0.992) was observed over the same concentration range (0–140 µM) when the were tested in human serum; the limit-of-detection under these conditions was noted to be 21 µM with electrodes were tested in human serum; the limit-of-detection under these conditions was noted to a sensitivity of 8.077 nA/µM. Glucose testing in buffer was also successful with linearity (r2 = 0.996) be 21 µM with a sensitivity of 8.077 nA/µM. Glucose testing in buffer was also successful with over the range of 0–14 mM; a sensitivity of 0.353 nA/µM and a limit-of-detection of 0.1 mM were linearity (r2 = 0.996) over the range of 0–14 mM; a sensitivity of 0.353 nA/µM and a limit-of-detection noted.of 0.1 In mM addition, were noted. interference In addition, studies interference involving uricstudies acid, involving ascorbic uric acid, acid, cysteine, ascorbic and acid, acetaminophen cysteine, duringand acetaminophen glutamate and during glucose glutamate testing showed and glucose minimal testing interference showed by minimal these competinginterference analytes. by these competing analytes. 7. Transdermal Alcohol Sensors 7. TransdermalDumett et al. Alcohol described Sensors a mathematical modeling approach for determining blood alcohol concentrationsDumett (BAC) et al. described from transdermal a mathematical alcohol concentrations modeling approach (TAC); for they determining compared bloodthis information alcohol toconcentrations breathe alcohol (BAC) concentration from transdermal (BrAC) data alcohol [44]. Since concentrations ethanol can (TAC); be detected they in compared perspiration, this a noninvasiveinformation transdermal to breathe alcohol sensor [ concentration45–49] for ethanol (BrAC) known data [44].as the Since WrisTAS ethanolTM (Giner, can be Inc.,detected Newton, in MA,perspiration, USA) has a been noninvasive developed. transdermal This device sensor serves [45–49] essentially for ethanol asknown a sobriety as the WrisTAS check andTM (Giner, does not provideInc., Newton, quantitative MA, USA) data onhas thebeen blood developed. alcohol This concentration device serves of essentially the individual as a sobriety wearing check the device.and Thedoes authors not provide described quantitative a mechanism data on for the comparing blood alcohol TAC concentration measures obtained of the individual clinically fromwearing this the type ofdevice. device The to BrAC authors values described in order a mechanism to determine for comparing quantitative TAC BAC measures levels fromobtained TAC clinically measurements from (Figurethis type 16). of This device process to BrAC took into values account in order (a) tomovement determine of quantitative alcohol through BAC the levels skin from and TAC to the

Sensors 2019, 19, x FOR PEER REVIEW 17 of 37

measurements (Figure 16). This process took into account (a) movement of alcohol through the skin and to the sensor, (b) calibration procedures for both the individual subject and the sensor being worn, and (c) deconvolution of the BAC value from the TAC measurement. An inversion scheme was used to test the ability of the model to accurately predict BrAC values based on TAC values. The Sensorsapproach2019, 19, 1028was a good predictor of BrAC values from TAC measurements. Advances in this approach17 of 36 may allow BAC values to be obtained from TAC measurements. sensor,8. Ultrasound (b) calibration Technique procedures for Transdermal for both the Biosensing individual subject and the sensor being worn, and (c) deconvolution of the BAC value from the TAC measurement. An inversion scheme was used to test Lee et al. examined transdermal sensing of glucose on the abdominal skin of rats, following the abilitypermeablization of the model of the to accuratelyskin with ultrasound predict BrAC [50]. valuesA cymbal based ultrasound on TAC array values. was The applied approach to the was a goodabdominal predictor area of BrAC of the values rats, delivery from TAC a low measurements. frequency (20 Advances kHz) signal; in this approach aided may allowin the BAC valuespermeablization to be obtained of fromthe skin TAC for measurements. sampling of interstitial fluid.

FigureFigure 16. Breath16. Breath and and transdermal transdermal datadata for for (a ()a )Subject Subject 1 and 1 and (b) Subject (b) Subject 2. (c) Two 2. (c )phase Two design phase of design the of the datadata analysisanalysis system system (left) (left) and and the the blood/lung/breath blood/lung/breath analyzer analyzer and blood/skin/TAS and blood/skin/TAS systems (right) systems (right)[44]. [44 ].

8. UltrasoundFollowing Technique ultrasonication, for Transdermal a three-electrode Biosensing (platinum as working electrode, carbon as counter electrode, silver/chloride as reference electrode) cell with a glucose oxidase functionalized working Leeelectrode et al. and examined a poly(ethylene transdermal glycol) sensing hydrogel of application glucose on pad the was abdominal applied to skin the skin of rats, to detect following permeablizationglucose levels of in the the skin interstitial with ultrasound fluid of the [ 50 permeablized]. A cymbal area. ultrasound Comparison array of was measurements applied to the abdominalobtained area using of the ultrasound/biosensor rats, delivery a low and frequency a commercial (20 kHz) glucose signal; meter this in approachhyperglycemic aided rats in the permeablization(Figure 17) showed of the skinno statistically for sampling significant of interstitial differences fluid. between the measurements (356.0 ± 116.6 Followingmg/dl and ultrasonication,424.8 ± 59.1 mg/dl, a three-electrode respectively); glucose (platinum could as notworking be accurately electrode, detected carbon by as the counter electrode,biosensor silver/chloride without ultrasonication. as reference Histology electrode) indicated cell with tissue a glucose damage oxidase at high functionalizedintensities (200 workingand 300 mW/cm2, 20% duty cycle) and no tissue damage at low duty cycles (100 mW/cm2, 20% duty electrode and a poly(ethylene glycol) hydrogel application pad was applied to the skin to detect cycle); this problem can be addressed by reducing the duty cycle to 15% or lower for higher glucose levels in the interstitial fluid of the permeablized area. Comparison of measurements obtained intensities. using theSmith ultrasound/biosensor et al. investigated ultrasound and a commercial permeabilization glucose of meterrabbit skin in hyperglycemic for transdermal ratsdelivery (Figure of 17) showedinsulin no statisticallyand transdermal significant biosensing differences of glucose between levels in the the measurements interstitial fluid (356.0 [51]. ±2 x116.6 2 and mg/dL 3 x 3 and 424.8cymbal± 59.1 ultrasound mg/dL, respectively); arrays were used glucose to ultrasonicate could not be(I = accurately 100 mW/cm detected2, 20% duty by thecycle, biosensor 20 kHz) the without ultrasonication.abdominal skin Histology of hyperglycemic indicated New tissue Zealand damage White at rabbits. high intensities (200 and 300 mW/cm2, 20% duty cycle) and no tissue damage at low duty cycles (100 mW/cm2, 20% duty cycle); this problem can be addressed by reducing the duty cycle to 15% or lower for higher intensities. Smith et al. investigated ultrasound permeabilization of rabbit skin for transdermal delivery of insulin and transdermal biosensing of glucose levels in the interstitial fluid [51]. 2 × 2 and 3 × 3 cymbal ultrasound arrays were used to ultrasonicate (I = 100 mW/cm2, 20% duty cycle, 20 kHz) the abdominal skin of hyperglycemic New Zealand White rabbits. Sensors 2019, 19, 1028 18 of 36 Sensors 2019, 19, x FOR PEER REVIEW 18 of 37

Figure 17. (a) Rat on its back; 1-mm thick water-tight stand-off between the shaved abdomen and array. Figure 17. (a) Rat on its back; 1-mm thick water-tight stand-off between the shaved abdomen and Reservoir inside the stand-off filled with saline. US (ISPTP = 100 mW/cm2) applied to exposed group array. Reservoir inside the stand-off filled with saline. US (ISPTP = 100 mW/cm2) applied to exposed for 20 min; US not applied to the control. (b) In presence of glucose oxidase, glucose becomes gluconic group for 20 min; US not applied to the control. (b) In presence of glucose oxidase, glucose becomes acid via glucono-δ-lactone: hydrogen peroxide is generated as byproduct. Electrochemical biosensor gluconic acid via glucono--lactone: hydrogen peroxide is generated as byproduct. Electrochemical uses the signal caused by oxidation of hydrogen peroxide. One glucose molecule generates an oxygen biosensor uses the signal caused by oxidation of hydrogen peroxide. One glucose molecule generates molecule, a water molecule, and two electrons in the reaction. (c) Voltage between working electrode an oxygen molecule, a water molecule, and two electrons in the reaction. (c) Voltage between and reference electrode, 0.7 V, applied from potentiostat controlled by computer. At the same time, working electrode and reference electrode, 0.7 V, applied from potentiostat controlled by computer. current between two electrodes is determined by potentiostat and recorded in the computer [50]. At the same time, current between two electrodes is determined by potentiostat and recorded in the Thecomputer insulin [50]. delivery process involved ultrasonicating a reservoir of 50 U/mL insulin in saline into the skin of the animal. Blood glucose measurements obtained using a standard glucose meter The insulin delivery process involved ultrasonicating a reservoir of 50 U/mL insulin in saline from the 2 × 2 cymbal array groups showed that after 60 min the blood glucose levels decreased into the skin of the animal. Blood glucose measurements obtained using a standard glucose meter ± − ± − ± atfrom 30, 60,the and2 × 2 90 cymbal min time array points groups (-20.0 showed20.8 that mg/dL, after 60 min132.6 the blood35.7 mg/dL, glucose levels208.1 decreased29 mg/dL, at ± respectively)30, 60, and from90 min the time initial points measured (-20.0 value ± 20.8 of mg/dL, 245.4 −132.645.5 mg/dL; ± 35.7 similarmg/dL, results −208.1 were ± 29 observed mg/dL, inrespectively) the groups treatedfrom the with initial the measured 3 × 3 cymbal value arrayof 245.4 (Figure ± 45.5 18mg/dL;). However, similar theresults saline-only were observed sonication in groupthe groups and the treated insulin with without the 3 sonication× 3 cymbal group array saw(Figure increase 18). However, in blood glucose the saline-only levels over sonication the same timegroup period. and the Noninvasive insulin without glucose sonication sensing ofgroup the hyperglycemicsaw increase in rabbits blood glucose was also levels conducted over the following same ultrasonication;time period. Noninvasive measurements glucose obtained sensing using of this the method hyperglycemic were compared rabbits to wasthose also obtained conducted using a standardfollowing glucose ultrasonication; meter (Figure measurements 17). obtained using this method were compared to those obtainedThese using experiments a standard were glucose conducted meter with (Figure a three-electrode 17). (platinum as working electrode, carbon as counterThese electrode, experiments silver/chloride were conducted as reference with aelectrode) three-electrode cell using (platinum a glucose as oxidase-functionalized working electrode, workingcarbon electrode as counter and electrode, a poly(ethylene silver/chloride glycol) hydrogel as reference application electrode) pad. The cell results usingindicated a glucose that theoxidase-functionalized measured values of working blood glucose electrode from and the a poly(ethylene noninvasive glycol) ultrasound hydrogel method application were comparable pad. The toresults the conventional indicated that glucose the measured meter (395 values± 42mg/dL of blood and glucose 425 ± 35 from mg/dL, the noninvasive respectively); ultrasound the results alsomethod indicated were comparable that the noninvasive to the conventional glucose sensor glucose was meter unable (395 to± accurately42 mg/dL and detect 425 glucose± 35 mg/dL, levels withoutrespectively); ultrasonication. the results also indicated that the noninvasive glucose sensor was unable to accurately detectLi glucose et al. proposed levels without a minimally ultrasonication. invasive technique to determine glucose levels that coupled ultrasoundLi et al. application proposed a to minimally permeabilize invasive the technique skin, vacuum to determine extraction glucose of interstitial levels that fluid, coupled and determinationultrasound application of the glucose to permeabilize concentration the in the skin, interstitial vacuum fluid extraction by surface of plasmon interstitial resonance fluid, and [52 ]. Thedetermination authors proposed of the glucose permeabilizing concentration the in the skin interstitial by breaking fluid by down surface the plasmon stratum resonance corneum [52]. with low-frequencyThe authors proposed ultrasound permeabilizing (55 kHz), enabling the skin transport by breaking of interstitial down fluid. the stratum corneum with low-frequencyThe interstitial ultrasound fluid could (55 kHz), then beenabling extracted transport by vacuum of interstitial at <10 in. fluid. Hg in 20-min intervals (15 min needed per extraction) for up to 15 h (the projected length of time over which the skin maintains enhanced permeability). They developed a method to predict blood glucose levels based on glucose levels in the interstitial fluid with a prediction error of 16%. Following collection of the interstitial

Sensors 2019, 19, 1028 19 of 36 glucose, the samples could be analyzed with surface plasmon resonance, calibrated to detect the changes in the reﬂective index of the solution as due to glucose. Using this technique, the authors established the ability to detect glucose at 10 mg/dL. They are also investigating the coupling of glucose to proteins in a protein-glucose binding method (e.g., D-glucose/D-galactose-binding protein) for enhancing the sensitivity of the technique; this approach involves binding glucose onto the sensor ofSensors the surface 2019, 19 plasmon, x FOR PEER resonance REVIEW detector. 19 of 37

FigureFigure 18. 18.The The cymbal cymbaltransducer transducer made made of of piezoelectric piezoelectric material material lead lead zirconate-titanate zirconate-titanate (PZT)-4 (PZT)-4 operatedoperated at at a frequencya frequency of of 20 20 kHz. kHz. TheThe cymbalcymbal disk was placed placed between between two two titanium titanium caps caps with with air air cavitiescavities beneath beneath the the caps, caps, which which convert convert thethe radialradial oscillations of of the the disk disk into into flexure ﬂexure motions motions of ofthe the caps.caps. Graph Graph of of the the change change in in blood blood glucoseglucose overover the 90-min transdermal transdermal insulin insulin delivery delivery experiment experiment duration,duration, comparing comparing both both controls controls (insulin(insulin onlyonly and ultrasound only) only) to to the the insulin insulin and and ultrasound ultrasound

resultsresults generated generated by by the the 2 2× ×22 array array (insulin (insulin and and ultrasound ultrasound 2 2 × 2)2) and and 3 3 ××3 3 array array (insulin (insulin and and ultrasound 3 × 3) [51]. ultrasound 3 × 3) [51].

9. MicrofluidicsThe interstitial Approach fluid could in Transdermal then be extracted Sensing by vacuum at < 10 in. Hg in 20-min intervals (15 min needed per extraction) for up to 15 h (the projected length of time over which the skin maintains enhancedGadre etpermeability). al. demonstrated They developed a novel a fabrication method to predict strategy blood to create glucose an levels integrable based on transdermal glucose (B-FIT)levelsmicrosystem in the interstitial for detectionfluid with ofa prediction biomolecules error in of interstitial 16%. Following fluid [collection53]. The of B-FIT the interstitial system was incorporatedglucose, the intosamples a robust could dermalbe analyzed patch with with surface complex plasmon microfluidics, resonance, acalibrated heater element, to detect andthe a colorimetricchanges in detectionthe reflective membrane. index of the The solution fabrication as due process to glucose. involved Using depositionthis technique, of multilayers the authors of SU-8established epoxy-based the ability photoresist to detect onto glucose a glass slideat 10 withmg/dl. a TeflonThey arerelease also layerinvestigating and a sacrificial the coupling aluminum of layerglucose for preparing to proteins microfluidic in a protein-glucose capillaries and binding the fluid method reservoir (e.g., (Figure D-glucose/ 19). Sputtering,D-galactose-binding lithographic patterning,protein) for and enhancing chemical the wet sensitivity etching of were the technique; utilized to this produce approach the involves chromium/gold binding glucose electrodes onto of thethe micro-heater. sensor of the surface The B-FIT plasmon heater resonance elements detector. enable thermal ablation of micropores in the upper layers of the skin for diffusion of biomolecules to the transdermal sensor. Nisar et al. applied a three-dimensional9. Microfluidics Approach model with in Transdermal coupled field Sensing effects for finite element analysis of the piezoelectric and fluidicGadre performance et al. demonstrated of a proposed a novel micropump fabrication for strategy transdermal to create drug an delivery-based integrable transdermal treatment of cardiovascular(B-FIT) microsystem disorders for [54 detection]. The valveless, of biomolecules PDMS pump in interstitial design contains fluid [53]. diffuser The B-FIT and nozzle system elements was andincorporated is driven by into a piezoelectric a robust dermal actuator. patch Piezoelectric with complex actuator microfluidics, deflection a and heater micropump element, flow and rate a werecolorimetric evaluated detection at various membrane. operational The parameters fabrication withprocess varying involved voltage deposition and excitation of multilayers frequency of SU-8 both experimentallyepoxy-based photoresist and theoretically. onto a glass slide with a Teflon release layer and a sacrificial aluminum layerThe for highest preparing flow rate microfluidic of the micropump capillaries was and measured the fluid experimentally reservoir (Figure using deionized 19). Sputtering, water; it waslithographic approximately patterning, 8.9 µL/min and chemicalat 160 Vp-p wet with etching an excitation were utilized frequency to produce of 250 Hz. the chromium/goldThe experimental andelectrodes numerically of the predictive micro-heater. values The ofB-FIT micropump heater elements flow rates enable were thermal similar, ablation validating of micropores the transient in multifieldthe upper analysis layers of model. the skin The for results diffusion demonstrated of biomolecules that theto the flow transdermal rate is more sensor. dependent Nisar et on al. the excitationapplied voltage a three-dimensional than the excitation model frequency;with coupled furthermore, field effects a for high finite excitation element voltage analysis at constantof the piezoelectric and fluidic performance of a proposed micropump for transdermal drug frequency was shown to be important for an increased flow rate. Other microfluidic systems have also delivery-based treatment of cardiovascular disorders [54]. The valveless, PDMS pump design been designed for efficient and targeted drug delivery [55,56] contains diffuser and nozzle elements and is driven by a piezoelectric actuator. Piezoelectric actuator deflection and micropump flow rate were evaluated at various operational parameters with varying voltage and excitation frequency both experimentally and theoretically.

Sensors 2019, 19, 1028 20 of 36 Sensors 2019, 19, x FOR PEER REVIEW 20 of 37

FigureFigure 19.19.( a(a)) A A single single B-FITB-FIT devicedevice cellcell withwith allall maskingmasking layers. ( (bb-(-(ii)))) Cross-section Cross-section of of the the released released device,device, ((bb-(-(iiii)))) SEMSEM micrographsmicrographs ofof SU-8SU-8 dry etching using masking masking layer. layer. ( (cc)) Fully Fully functional functional B-FIT B-FIT picturedpictured on on human human armarm [[53].53].

10. OpticalThe highest Transdermal flow rate Sensors: of the micropump A Condensed was Overviewmeasured experimentally using deionized water; it was approximately 8.9 µL/min at 160 Vp-p with an excitation frequency of 250 Hz. The experimental 10.1. Fluorophore Based Glucose Sensor and numerically predictive values of micropump flow rates were similar, validating the transient multifieldLakowicz analysis et al. model. adopted The the results polarization demonstrated approach that for the glucose flow sensingrate is more using dependent a stretch-oriented on the referenceexcitation film voltage of Hoechst than the dye excitation in polyvinyl frequency; alcohol furthermore, and a fluorophore-based a high excitation sensorvoltage containingat constant a polarizerfrequency [57 was]. The shown polarizer to be important located in for the an sensor increased was orientedflow rate. perpendicularly Other microfluidic to the systems sample have and thealso excitation been designed source; for this efficient approach and ensuredtargeted sampledrug delivery emission [55,56] added only to the perpendicular total intensity. The glucose-sensitive fluorescence signal was emitted by mutant glucose/galactose binding protein10. Optical (GGBP) Transdermal from E. coli Sensors: with aA single Condensed cysteine Overview residue inserted at position 26 and labeled with 2-(40-iodoacetamidoanilino)naphthalate-6-sulfonic acid (I-ANS). Two excitation sources were utilized, including10.1. Fluorophore a UV hand Based lamp Glucose for Sensor excitation of the sample and a spectrofluorometer for excitation of the referenceLakowicz film. Polarization et al. adopted sensing the polarization of glucose approach was demonstrated for glucose with sensing an accuracy using a ofstretch-oriented 0.2 µM glucose concentration;reference film polarizationof Hoechst increased dye in polyvinyl with increasing alcohol glucoseand a fluorophore-based levels. sensor containing a polarizerBallerstadt [57]. The et al. polarizer described located a fluorescence in the sensor hollow was fiber oriented glucose perpendicularly sensor with porous to the dyed sample Sephadex and beadsthe excitation containing source; immobilized this approach glucose ensured residues sample and Alexa488-labeled emission added only concanavalin to the perpendicular A (Figure 20 total)[58 ]. intensity.The hollow The glucose-sensitive fiber dialysis membrane fluorescence contains signal 20–50 was um emitted Sephadex by beads mutant dyed glucose/galactose with safranin O andbinding pararosanilin. protein (GGBP) In the from presence E. coli ofwith glucose, a single Alexa-488-ConA cysteine residue inserted molecules at position are displaced 26 and fromlabeled the beadswith 2-(4’-iodoacetamidoanilino)naphthalate-6-sulfonic and exposed to excitation light, resulting in an increase acid (I-ANS). in fluorescence Two excitation emission sources at 520 were nm. Theutilized, sensor including demonstrated a UV a rapidhand response lamp for time, excitation a glucose of the detection sample range and from a spectrofluorometer 0.15 to 100 mM, and for a dynamicexcitation signal of the change reference from film. 0.2 toPolarization 30 mM; however, sensing stabilityof glucose studies was demonstrated indicated a 40–50% with an reduction accuracy in sensorof 0.2 µM intensity glucose over concentration; three months polarization [58]. Ballerstadt increased et al. with incorporated increasing aglucose near-infrared levels. chromophore linkedBallerstadt to ConA within et al. an described immobilized a fluorescence bead matrix hollow [59]. The fiber fluorophore glucose sensor Alexa with 647 was porous utilized dyed to extendSephadex the fluorescence beads containing emission immobilized of the glucose glucose sensor residues to the andnear-infrared Alexa488-labeled region (670 concanavalin nm), enabling A transmission(Figure 20) [58]. through the skin for transdermal glucose monitoring. A six-month study showed that the glucose sensitivity of the sensor gradually decreased over time; it should be noted that the sensor response amplitude for a glucose concentration of 2.5–10 mM remained stable. Ballerstadt et al. further

Sensors 2019, 19, 1028 21 of 36 evaluated the long-term stability of the near-infrared affinity sensor under physiological conditions [60]. The sensor was subjected to in vitro testing using continuous glucose cycling with alternating exposures to glucose concentrations of 2.5 and 20 mM for three-hour segments at 37 ◦C. The long-term performance of the sensors showed an initial increase in fluorescence during the first 3–4 weeks of the study, followed by a gradual and linear decrease. The authors suggest that leakage of Alexa6470-ConA is the underlying cause for the 60% loss in the fluorescence signal over the course of the long-term study. They also performed acute and chronic in vivo tests in hairless rats using a cellulose hollow fiber containing a suspension of Cy7-succinimidyl ester-ConA-Sepharose, Alexa647 dextran, and TransFluoSpheres with 2 µm diameter [61]. The sensor demonstrated extremely high sensitivity for a glucose concentration around 5 mM and between 20–25 mM. At two weeks post-surgery, the implanted sensors exhibited 10–15 min delays but were able to detect variations in blood glucose levels. Minimal host response was observed for the implanted sensors. In addition, intradermal injections of the sensor components in rats showedSensors no2019 toxicity, 19, x FOR compared PEER REVIEW to the control group. 21 of 37

FigureFigure 20. 20.(a ()a Schematics) Schematics illustratingillustrating the the principles principles of of the the fluorescence fluorescence affinity affinity hollow hollow fiber fibersensor. sensor. In In the absence of of glucose, glucose, fluorochrome-labeled fluorochrome-labeled Concanavalin Concanavalin A A is isbound bound to to the the fixed fixed glucose glucose residues residues insideinside the the porous porous beads beads (A-i (A().i)). The The beads beads are are colored colored withwith dyes that that prevent prevent the the excitation excitation light light from from inducinginducing Con Con A A to to fluoresce, fluoresce, keeping keeping the the fluorescence fluorescence emission at at 520 520 nm. nm. After After diffusion diffusion of ofglucose glucose throughthrough the the hollow hollow fiber fiber membrane membrane (molecular (molecular weightweight cutoff, cutoff, 10 10 kDa), kDa), Con Con A A is is displaced displaced from from the the beadsbeads and and diffuses diffuses out out of of them. them. Fluorochrome-labeledFluorochrome-labeled Con Con A A becomes becomes exposed exposed to toexcitation excitation light, light, resultingresulting in in a a strong strong increaseincrease in fluorescence fluorescence (A (A-ii(ii)).). (b ()b Spectra) Spectra of fluorescence of fluorescence and absorption and absorption of the of thevarious various chromophoric chromophoric components components of of the the bead-based bead-based affinity affinity sensor. sensor. (1) (1) Excitation Excitation spectrum spectrum of of fluorescein,fluorescein, (2) (2) emission emission spectrum spectrum of offluorescein, fluorescein, (3)(3) absorption spectrum spectrum of of dye-labeled dye-labeled Sephadex Sephadex beads.beads. (c) Light(c) Light microscopy microscopy picture picture of a small of a smallsection section of the hollow of the fiber hollow that fiber was that filled was with filled dye-colored with dye-colored Sephadex G150 beads. A bead fraction having a bead diameter of less than 25 µm was Sephadex G150 beads. A bead fraction having a bead diameter of less than 25 µm was used in this used in this study; the beads were obtained by sieving the original material (mesh size of 25 µm) [58]. study; the beads were obtained by sieving the original material (mesh size of 25 µm) [58].

McShaneThe hollow et al. fiber developed dialysis membrane a custom contains fiber optic 20-50 probe um Sephadex and optical beads system dyed with for deliverysafranin O and and pararosanilin. In the presence of glucose, Alexa-488-ConA molecules are displaced from the collection of light to an implanted fluorescence-based sensor for glucose monitoring in interstitial beads and exposed to excitation light, resulting in an increase in fluorescence emission at 520 nm. fluid [62]. The system was compared with two commercially-available fluorometers and demonstrated The sensor demonstrated a rapid response time, a glucose detection range from 0.15 to 100 mM, and a higher peak response and a faster time per scan than both commercial instruments. Solutions a dynamic signal change from 0.2 to 30 mM; however, stability studies indicated a 40-50% reduction of FITC-dextran and TRITC-Con A at different ratios were measured under similar conditions in sensor intensity over three months [58]. Ballerstadt et al. incorporated a near-infrared chromophore linked to ConA within an immobilized bead matrix [59]. The fluorophore Alexa 647 was utilized to extend the fluorescence emission of the glucose sensor to the near-infrared region (670 nm), enabling transmission through the skin for transdermal glucose monitoring. A six-month study showed that the glucose sensitivity of the sensor gradually decreased over time; it should be noted that the sensor response amplitude for a glucose concentration of 2.5-10 mM remained stable. Ballerstadt et al. further evaluated the long-term stability of the near-infrared affinity sensor under physiological conditions [60]. The sensor was subjected to in vitro testing using continuous glucose cycling with alternating exposures to glucose concentrations of 2.5 and 20 mM for three-hour segments at 37 °C. The long-term performance of the sensors showed an initial increase in fluorescence during the first 3–4 weeks of the study, followed by a gradual and linear decrease. The authors suggest that leakage of Alexa6470-ConA is the underlying cause for the 60% loss in the fluorescence signal over the course of the long-term study. They also performed acute and chronic in vivo tests in hairless rats using a cellulose hollow fiber containing a suspension of Cy7-succinimidyl

Sensors 2019, 19, 1028 22 of 36 upon incremental addition of glucose. The system observed high fluorescence measurements from implanted fluorescence particles in both excised porcine skin and live animals; however, quantitative measurements were not possible due to fluctuations in the implantation depth. Long et al. fabricated microsphere glucose sensors that contained encapsulated glucose oxidase, catalase, and platinum octaethylprophine (PtOEP) in silica-alginate microspheres, which were coated with an ultrathin polyelectrolyte film [63]. These luminescence-based microparticle sensors were produced with various concentrations of enzymes and evaluated using a flow through system with three different glucose concentrations (0, 25, and 75 mg/mL). The particle sensors were embedded in a PEG hydrogel. The glucose levels were maintained for a duration of 8 h and repeated in triplicate. Theoretical optical modeling of the sensors suggested the emission signals from the particles are adequate for high signal-to-noise ratio measurements. Experimental evaluation of the microparticle sensors revealed no significant difference in sensor stability with increasing catalase concentration; in addition, trace amounts (≤100 µM) of catalase in commercial glucose oxidase were sufficient for stabilization of the enzyme. Kermis et al. evaluated the permeability properties of optically transparent ethylene glycol dimethacrylate (EGDMA) crosslinked poly(2-hydroxyethyl methacrylate) (pHEMA) membranes for transdermal glucose detection (Figure 21)[64]. The transport properties of the flat sheet and cylindrical membranes were determined using a diffusion cell at 37 ◦C. The creatinine and IgG-Fab fragment concentrations were determined using a UV-Vis diode array spectrophotometer and a digital fluorometer. The permeability experiments indicated that an increase in crosslinker concentration decreased the degree of membrane swelling, which resulted in lower creatinine permeability and increased selectivity for the protein. In vivo studies in rats showed that subcutaneous implantation of the 4% EGDMA cylindrical membranes up to six weeks resulted in minimal host response with a thin fibrotic capsule and negligible loss of membrane integrity. Kholodnykh et al. demonstrated glucose monitoring using optical coherence tomography (OCT); glucose-induced changes in the optical signal attenuation could be detected with an implantable turbidity sensor [65]. The sensor contained cellulose membranes encapsulating a suspension of ConA and hydrogel particles ranging from 1–4 µm in diameter; the sensor also included a highly specularly reflective Mylar film. They measured changes in light attenuation as a function of glucose concentration; they determined that an increase in glucose level was associated with a decrease in the optical turbidity of the sensor. In vitro testing using excised skin tissue demonstrated the advantages of measuring specular reflection over backscattering light attenuation. When testing the sensor covered by tissue phantoms, the specular reflection decreased almost linearly with increasing implantation depth; the lowest acceptable sensitivity of 1 bD per mM glucose was estimated at an optical depth of 12 mean free paths (mfp) which translates to approximately 1.2 mm of skin tissue and a scattering coefficient of 10 mm−1. Khan et al. developed five different mutants of glucose/galactose-binding protein (GBP), fluorescently labeled the mutants with the fluorophore 6-bromoacetyl-2-dimethyl aminonaphthalene (Badan), and evaluated the GBP mutant response to glucose [66]. An emission wavelength of 550 nanometers and an excitation wavelength of 400 nanometers were used for steady-state fluorescence measurements of the labeled proteins. The triple mutant H152C/A213R/L238S-Badan demonstrated the highest binding constant of 11 mM in PBS and 14 mM in serum; the change in fluorescence intensity upon the addition of glucose was 180% in PBS and a 200% in serum. This mutant was capable of glucose sensing within the physiologically relevant range of 1–100 mM. Potential applications of the GBP mutant-based fluorescence sensor including use for continuous in vivo monitoring of glucose concentrations in diabetic patients. Sensors 2019, 19, 1028 23 of 36 Sensors 2019, 19, x FOR PEER REVIEW 23 of 37

FigureFigure 21. 21.(a ()a Photomicrograph) Photomicrograph ofof thethe hosthost response to to a a pHEMA pHEMA membrane membrane implanted implanted for for 3 weeks. 3 weeks. ◦ ◦ (b)(b Dependence) Dependence of of equilibrium equilibrium water water content, content, q qww,, for for flat flat sheet sheet pHEMA pHEMA membranes membranes at at 37 37 C C on on crosslinkingcrosslinking ratio. ratio. Experiments Experiments were were performed performed in triplicate in triplicate at each atcrosslinking each crosslinking ratio. (c) Dependence ratio. (c) ofDependence creatinine permeability of creatinine at permeability 37 ◦C in flat at sheet 37 °C pHEMA in flat sheet membranes pHEMA on membranes equilibrium on water equilibrium content. Forwater reference, content. the For permeability reference, the of permeability creatinine in of a creatinine 1000 MWCO in a 1000 cellulosic MWCO dialysis cellulosic membrane dialysis is −7 2 9.0membrane× 10−7 cm is2 /s9.0 [×64 10]. cm /s [64].

10.2. Hydrogel-BasedKholodnykh etFluorescent al. demonstrated Glucose Sensors glucose monitoring using optical coherence tomography (OCT); glucose-induced changes in the optical signal attenuation could be detected with an Heo et al. applied a fluorescent polyacrylamide (PAM) hydrogel and polyethylene-glycol implantable turbidity sensor [65]. The sensor contained cellulose membranes encapsulating a (PEG)-bonded PAM hydrogel fibers for long-term in vivo glucose sensing [67]. The fibrous structure suspension of ConA and hydrogel particles ranging from 1–4 µm in diameter; the sensor also ofincluded the fibers a permittedhighly specularly good adhesion reflective toMylar the tissuefilm. They with measured minimal migration,changes in light enabling attenuation the sensor as a to remainfunction in theof glucose implant concentration; site. The high they surface determined area to that volume an increase ratio ofin theglucose fibers level with was approximately associated 1 mmwith diameter a decrease enabled in the high optical diffusion turbidity for ofrapid the response sensor. In times. vitro For testingin vitro usingstudies, excised the skin fibers tissue were exposeddemonstrated to various the advantages glucose concentrations of measuring specular covering reflection the physiological over backscattering normal, hypoglycemic,light attenuation. and hyperglycemicWhen testing ranges the sensor (0–500 covered mg/dL). by Fluorescent tissue phantoms, images the were specular obtained reflection at a wavelength decreased of almost 488 nm; bothlinearly fiber types with demonstratedincreasing implantation increasing depth; fluorescence the lowest intensity acceptable with increasing sensitivity glucose of 1 bD concentration per mM (Figureglucose 22 ). wasIn vivo estimatedevaluation at an in optical the transparent depth of ear 12 skin mean of free mice paths indicated (mfp) 75% which of the translates PEG-bonded to PAMapproximately hydrogel fibers 1.2 mm responded of skin tissue to blood and a glucose scattering concentration coefficient of fluctuations. 10 mm−1. Only 25% of the PAM hydrogelKhan fibers et responded al. developed to the five glucose different changes. mutants A ten-minute of glucose/galactose-binding delay was observed protein between (GBP), blood glucosefluorescently concentration labeled change the and mutants fiber response. with the fluorophore 6-bromoacetyl-2-dimethyl aminonaphthaleneShibata et al. developed (Badan), highly and evaluated sensitive andthe GBP injectable mutant microbeads response containingto glucose a[66]. glucose-responsive An emission fluorescentwavelength monomer of 550 nanometers for in vivo andcontinuous an excitation glucose wavelength monitoring of (Figure 400 nanometers 23) [68]. were The fluorescentused for monomersteady-state was derived fluorescence from a glucose-responsive measurements of dye, the which labeled was comprised proteins. of The a diboronic triple acid mutant moiety andH152C/A213R/L238S-Badan an anthracene moiety; polyethylene demonstrated glycol the andhighest polyacrylamide binding constant groups of 11 served mM in as PBS spacers. and 14 mM in serum; the change in fluorescence intensity upon the addition of glucose was 180% in PBS and a 200% in serum. This mutant was capable of glucose sensing within the physiologically relevant

Sensors 2019, 19, x FOR PEER REVIEW 24 of 36

PEG-bonded PAM hydrogel fibers responded to blood glucose concentration fluctuations. Only 25% of the PAM hydrogel fibers responded to the glucose changes. A ten-minute delay was observed between blood glucose concentration change and fiber response. Shibata et al. developed highly sensitive and injectable microbeads containing a glucose-responsive fluorescent monomer for in vivo continuous glucose monitoring (Figure 23) [68]. The fluorescent monomer was derived from a glucose-responsive dye, which was comprised of a Sensorsdiboronic2019, 19 acid, 1028 moiety and an anthracene moiety; polyethylene glycol and polyacrylamide groups24 of 36 served as spacers.

FigureFigure 22. 22.( A(A)) Schematic Schematic illustration illustration ofof thethe fluorescentfluorescent hy hydrogeldrogel fiber fiber designed designed for for long-term long-term inin vivo vivo glucoseglucose monitoring. monitoring. The The fiberfiber cancan bebe injectedinjected into subcutaneous tissues. tissues. The The implanted implanted fiber fiber remains remains atat the the implantation implantation site site for for a longa long period period and and transmits transmits fluorescent fluorescent signals signals transdermally transdermally depending depending on bloodon blood glucose glucose concentration. concentration. The implanted The implanted fiber canfibe ber can easily be removedeasily removed from the from implantation the implantation site after use.site ( Bafter) Fluorescent use. (B) Fluorescent hydrogel fibers hydrogel in a glass fibers vial in witha glass a 50% vial glucosewith a 50% solution. glucose The solution. fibers are The excited fibers by ultravioletare excited light. by ultraviolet The fluorescent light. The image fluorescent indicates im thatage the indicates glucose-responsive that the glucose-responsive monomer is immobilized monomer withinis immobilized the hydrogel within fibers. the hydrogel (C) Inflammation fibers. (B) Inflammation indices of the indices mouse of ears the mouse with the ears fibers with one the fibers month afterone implantation.month after implantation. Inflammation Inflammation was evaluated was basedevaluated on reddening, based on reddening, swelling andswelling scab formationand scab forformation a month. for The a month. inflammation The inflammation index was index obtained was byobtained summing by summing the reddening, the reddening, swelling, swelling, and scab formationand scab scores.formation If the scores. ear skinIf the showed ear skin any showed reddening, any reddening, reddening reddening scores 1 sco point;res 1 similarly, point; similarly, swelling andswelling scab formation and scab were formation each also were scored each 1 point.also scored PEG-bonded 1 point. PAM PEG-bonded induced less PAM inflammation induced less than PAMinflammation only. (D) Numbersthan PAM of only. mice showing(C) Numbers transdermal of mice transmission showing tran ofsdermal fiber fluorescence transmission [67 ].of fiber fluorescence [67]. Glucose responsiveness testing of the microbeads indicated that the fluorescence intensity increasedGlucose with responsiveness increasing glucose testing concentration of the microbea (0–1000ds indicated mg/dL); thethat association the fluorescence was found intensity to be reversibleincreased since with the increasing intensity glucose curves concentration matched with (0 an–1000 increase mg/dL); and decreasethe association in glucose was concentration.found to be Inreversible vivo implantation since the intensity of the microbead curves matched sensors with in the an ear increase skin of and mice decrease showed in that glucose the microbeads concentration. were In vivo implantation of the microbead sensors in the ear skin of mice showed that the microbeads able to monitor fluctuations in blood glucose concentration with a lag time of 11 min. were able to monitor fluctuations in blood glucose concentration with a lag time of 11 min.

Sensors 2019, 19, x FOR PEER REVIEW 26 of 37

10.3. Fluorescence-Based Ion Sensors Geddes reviewed the medical relevance of halide ion sensing in a variety of diagnostic processes [69]. Monitoring of fluoride levels can diagnosis high fluoride levels in the blood; while fluoride is particularly present in the teeth and bone, its slow clearance from the body following excessive intake can result in poisoning. Additionally, determining toxic levels of bromide (an Sensorsunnecessary2019, 19, or 1028 misunderstood ion of the trace element in the body) and biological levels of iodide25 of 36 (which is important in production of certain hormones) have medical significance.

Figure 23. (A) Fluorescent microbeads fabricated by a microfluidicmicrofluidic system.system. (A) Schematic diagram of the AFFD.AFFD. TheThe pregelpregel solution solution flows flows into into the the inner inner channel channel of theof the AFFD, AFFD, and and the siliconethe silicone oil flows oil flows into theinto outer the outer channel. channel. The droplets The droplets in silicone in silicone oil are oil collected are collected from the from solution the solution with TEMED with TEMED at 37 ◦C; at they 37 turn°C; they into microbeads turn into microbeads after gelation. after (B) Glucose gelation. responsiveness (B) Glucose of responsiveness GF-beads. (i–iii) of Fluorescent GF-beads. images (i)-(iii) ofFluorescent GF-beads images at a glucose of GF- concentrationbeads at a glucose of 0 mg concentration·dL−1, 250 mg of·dL 0− mg·dL−1,1, and 1000 250 mg mg·dL−1,·dL−1, respectively, and 1;000 (iv)mg·dL−1, The fluorescence respectively, intensity (iv) The of fluorescence GF-beads (n intensity = 19) changes of GF-beads according (n=19) to thechanges increase according and decrease to the inincrease glucose and concentration. decrease in The glucose overlapping concentration. curves are The evidence overlapping of the curves reversible are reaction evidence between of the GF-beadsreversible and reaction glucose. between (C) The GF-beads fluorescence and intensity glucose. traces(C) The blood fluorescence glucose concentration. intensity traces The blood time lagglucose mainly concentration. comes from theThe timetime response lag mainly between comes the from interstitial the time fluid response glucose between concentration the interstitial and the bloodfluid glucose glucose concentration concentration and [68]. the blood glucose concentration [68].

10.3. Fluorescence-Based Ion Sensors Levels of fluoride, chloride, bromide, and iodide ions in blood, urine, sweat, and serum have been Geddestested for reviewed diagnostic the purposes medical using relevance a variety of of halide testing ion methods; sensing these in a methods variety include of diagnostic use of processesion selective [69 ]. electrodes, Monitoring ion of chromatography, fluoride levels can clinical diagnosis analyzers, high fluoride microvolume levels in methods, the blood; titrator while fluoridestrips, gas is chromatography, particularly present anion-exchange in the teeth and chromatography, bone, its slow high-purity clearance from liquid the chromatography body following excessive(HPLC), head-space intake can flow result injection, in poisoning. x-ray fluorescence, Additionally, and determining HPLC with toxicAg electrode. levels of However, bromide (anthe unnecessaryfluorescence quenching or misunderstood behavior ion of offluorophores the trace element by halides in the has body) been and investigated biological as levels an alternative of iodide (whichsensing is method important to in those production previously of certain listed; hormones) this property have may medical be incorporated significance. into transdermal sensingLevels techniques. of fluoride, Fluorescence chloride, bromide,occurs due and to iodidethe rapid ions emission in blood, of urine,a photon sweat, when and an serum electron have in beenan excited tested orbital for diagnostic is allowed purposes to return using to the a varietyground of state testing quickly methods; (the electron these methods is “spin include allowed,” use by of ionhaving selective the opposite electrodes, spin ion to chromatography, the ground state clinical electron analyzers, with which microvolume it is paired). methods, Quenching titrator of strips, this gasfluorescence chromatography, signal is anion-exchangethe basis for optical chromatography, halide sensing. high-purity liquid chromatography (HPLC), head-spaceFluorescence flow injection, from different x-ray fluorescence, marker agents and (e.g., HPLC quinine, with Ag methoxyquinoline, electrode. However, rhodamines) the fluorescence may quenchingbe quenched behavior in the ofpresence fluorophores of halides, by halides with the has decrease been investigated in intensity as anor alternativelength of the sensing fluorescence method tosignal those being previously correlated listed; to halide this property concentration. may be The incorporated process of into dynamic transdermal fluorophore sensing quenching techniques. by Fluorescencehalides was first occurs described due to the by rapid the Stern-Vollmer emission of a kinetics;photon when more an recently, electron the in an kinetics excited have orbital been is allowedextended to to return determine to the the ground effects state of additional quickly (the quenching electron ishalides “spin (e.g., allowed,” a solution by having containing the opposite I−, Cl−, spinBr−) toby the adding ground terms state to electron the equations. with which Similarly, it is paired). the QuenchingStern-Volmer of this calculations fluorescence can signal be slightly is the basis for optical halide sensing. Fluorescence from different marker agents (e.g., quinine, methoxyquinoline, rhodamines) may be quenched in the presence of halides, with the decrease in intensity or length of the fluorescence signal being correlated to halide concentration. The process of dynamic fluorophore quenching by halides was first described by the Stern-Vollmer kinetics; more recently, the kinetics have been extended to determine the effects of additional quenching halides (e.g., a solution containing I−, Cl−, Br−) by Sensors 2019, 19, 1028 26 of 36 adding terms to the equations. Similarly, the Stern-Volmer calculations can be slightly modified to account for scenarios where there is static quenching of the fluorophore or limited accessibility to the fluorophores due to multiple fluorophore populations (e.g., the halide can quench one fluorophore on a molecule, but cannot reach the others). Another aspect of fluorophore quenching that can be harnessed for detection is fluorescence resonance energy transfer (FRET); in this detection technique, the energy of an electron in an excited state is donated to an acceptor molecule; this molecule has an excitation spectrum that overlaps with the emission spectrum of the donor. By monitoring this dipole-dipole interaction fluorescence in the presence and absence of the solvated halide-ion of interest, detection of the halide ion can occur. Other factors to consider with the halide ion detection via fluorescence quenching are dissolved oxygen concentration (which in itself can act as a quenching agent) and the electrostatic charge of the environment surrounding the fluorophore (which can affect charged quenching agents). By utilizing these properties of fluorescence quenching by halide ions, optical sensors can be created to gauge the concentrations of halide ions of interest by immobilizing fluorophore molecules onto or within the sensor. The quenching of the fluorophore molecules when interacting with the halide ions allows for the detection of the halide ions. Fluorophores such as acridinium, quinolinium, and rhodamines have been utilized in this manner as sensor detection molecules; similar sensors can be created by immobilization of FRET donor and acceptor molecules, allowing for the changes in fluorescence quenching in the presence of the halide ions. The potential for multiphoton detection of halide ion concentrations may also be possible for transdermal sensors. By utilizing an implanted halide ion sensor under the skin, excitation by multiphoton laser light (e.g., femtosecond pulses at λ = 800 nm), fluorescence detection may be accomplished within the “therapeutic range” of 600–1000 nm; the auto-fluorescence interference associated with tissues at physiologically relevant concentrations of the halide ions of interest can be avoided. This technique harnesses the energy of multiple photons of light at a focal point during a short time period (10−15–10−16 s) to excite the fluorophore that would otherwise need the energy of a shorter wavelength of light to emit a photon during relaxation of its excited electron. By utilizing the combined energy of the excitation photons at the focal point, a longer wavelength of light can be used that is less damaging to, and avoids interference from, biological fluids and tissues. Badugu et al. demonstrated the cyanide detection up to physiologically relevant levels using water-soluble fluorescent probes (Figure 24). Three isomeric boronic acid derivative probes, ortho-, meta-, and para-(4-[4-(N,N-dimethylamino)styryl]-1-(x-boronobenzyl)pyridinium bromide (DSPBA), were bound to cyanide resulting in a reduced intramolecular charge transfer for enhanced fluorescence sensing. The ortho-DSPBA isomer probe demonstrated the best response to cyanide compared to the other isomeric probes due to improved electrostatic interactions. Cyanide concentrations of 1–30 µM were detectable in the presence of physiologically relevant interferants, such as aqueous chloride up to 100 nM. The emission of the probes fell within the optical window of tissue and blood thus demonstrating the sensor applicability in transdermal cyanide monitoring. Sensors 2019, 19, x FOR PEER REVIEW 27 of 37

modified to account for scenarios where there is static quenching of the fluorophore or limited accessibility to the fluorophores due to multiple fluorophore populations (e.g., the halide can quench one fluorophore on a molecule, but cannot reach the others). Another aspect of fluorophore quenching that can be harnessed for detection is fluorescence resonance energy transfer (FRET); in this detection technique, the energy of an electron in an excited state is donated to an acceptor molecule; this molecule has an excitation spectrum that overlaps with the emission spectrum of the donor. By monitoring this dipole-dipole interaction fluorescence in the presence and absence of the solvated halide-ion of interest, detection of the halide ion can occur. Other factors to consider with the halide ion detection via fluorescence quenching are dissolved oxygen concentration (which in itself can act as a quenching agent) and the electrostatic charge of the environment surrounding the fluorophore (which can affect charged quenching agents). By utilizing these properties of fluorescence quenching by halide ions, optical sensors can be created to gauge the concentrations of halide ions of interest by immobilizing fluorophore molecules onto or within the sensor. The quenching of the fluorophore molecules when interacting with the halide ions allows for the detection of the halide ions. Fluorophores such as acridinium, quinolinium, and rhodamines have been utilized in this manner as sensor detection molecules; similar sensors can be created by immobilization of FRET donor and acceptor molecules, allowing for the changes in fluorescence quenching in the presence of the halide ions. The potential for multiphoton detection of halide ion concentrations may also be possible for transdermal sensors. By utilizing an implanted halide ion sensor under the skin, excitation by multiphoton laser light (e.g., femtosecond pulses at λ = 800 nm), fluorescence detection may be accomplished within the “therapeutic range” of 600–1000 nm; the auto-fluorescence interference associated with tissues at physiologically relevant concentrations of the halide ions of interest can be avoided. This technique harnesses the energy of multiple photons of light at a focal point during a short time period (10−15–10−16 s) to excite the fluorophore that would otherwise need the energy of a shorter wavelength of light to emit a photon during relaxation of its excited electron. By utilizing the combined energy of the excitation photons at the focal point, a longer wavelength of light can be used that is less damaging to, and avoids interference from, biological fluids and tissues. Badugu et al. demonstrated the cyanide detection up to physiologically relevant levels using water-soluble fluorescent probes (Figure 24). Three isomeric boronic acid derivative probes, ortho-, meta-, and para-(4-[4-(N,N-dimethylamino)styryl]-1-(x-boronobenzyl)pyridinium bromide (DSPBA), were bound to cyanide resulting in a reduced intramolecular charge transfer for enhanced fluorescence sensing. The ortho-DSPBA isomer probe demonstrated the best response to cyanide Sensorscompared2019, 19to, 1028the other isomeric probes due to improved electrostatic interactions. 27 of 36

Sensors 2019, 19, x FOR PEER REVIEW 28 of 37

Figure 24. (a) Complexation of DSPBA probes with aqueous free cyanide. (b) Ratiometric response of the probes in water with increasing cyanide concentrations. (c) Fluorescence emission of o-DSPBA in water with time when excited at 475 and 375 nm [70].

FigureCyanide 24. (a )concentrations Complexation of of DSPBA 1-30 µM probes were with detectable aqueous freein the cyanide. presence (b) Ratiometric of physiologically response of relevant

interferants,the probes in such water as with aqueous increasing chloride cyanide up concentrations. to 100 nM. The (c) Fluorescence emission of emission the probes of o-DSPBA fell within in the opticalwater windowwith time whenof tissue excited and at 475blood and thus 375 nm demonstrating [70]. the sensor applicability in transdermal cyanide monitoring. AnotherAnother report report demonstrates demonstrates the the detection detection of cyanideof cyanide anions anions in in solution solution has has been been accomplished accomplished byby monitoring monitoring changes changes in ﬂuorescence in fluorescence of indium of indium phosphide phosphide quantum quantum wires (InPwires QWs) (InP thatQWs) have that been have coatedbeen withcoated mercaptohexanoic with mercaptohexanoic acid (Figure acid 25(Figure)[71]. 25) [71].

FigureFigure 25. 25.(a) ( FLa) FL intensity intensity for for functionalized functionalized InP InP QWs QWs at differentat different cyanide cyanide concentration. concentration. (b) ( Relativeb) Relative intensityintensity for for functionalized functionalized InP InP QWs QWs for for various various metal metal anions anions at 50at pM50 pM concentration concentration [71 ].[71].

TheseThese quantum quantum wires wires were were modified modified to to have have a peaka peak emission emission wavelength wavelength of ofλ = λ 750= 750 (± 10)(± 10) nm nm (excitation(excitationλ =λ 375= 375 nm), nm), which which is opticallyis optically compatible compatible with with blood blood and and tissue; tissue; these these devices devices have have potentialpotential use use for transdermalfor transdermal detection detection of cyanide of cyanide ions. The ions. InP The QWs InP exhibited QWs exhibited dose-dependent dose-dependent linear decreaseslinear decreases in fluorescence, in fluorescence, potentially potentially due to Förster due resonanceto Förster energyresonance transfer energy between transfer the between InP QWs the andInP the QWs cyanide and ions, the cyanide upon exposure ions, upon of the exposure QWs to cyanideof the QWs ions to down cyanide to picomolar ions down concentration to picomolar −12 −1 levelsconcentration (range of 0–110 levels× (range10 ofmol 0 – L110). × The10−12 InP mol QWs L−1). alsoThe showedInP QWs a also limited showed response a limited to potentially response to − interferingpotentially ions, interfering indicating ions, specificity indicating to CN specificity. to CN−.

10.4.10.4. Luminescent Luminescent Sensors Sensors LongLong et al.et al. further further predicted predicted the the performance performance of of these these implantable implantable luminescent luminescent sensors sensors using using threethree dimensional, dimensional, multiwavelength multiwavelength Monte Monte Carlo Carlo modeling modeling to to determine determine the the distribution distribution of of the the emittedemitted luminescence luminescence at different at different implantation implantation depths, depths, excitation excitation light source light characteristics, source characteristics, particle particle properties, and particle packing efficiency [72]. The modeling system enabled realistic simulation of light propagation in a four-layer skin tissue model with embedded microparticle sensors hexagonally packed at different densities. The spatial distribution of the emitted luminescence induced by an excitation beam with a diameter of 10 mm can be contained within an 18 mm circle, allowing a sensor design utilizing a matched optoelectronic system to effectively deliver the excitation source and subsequently collect and analyze the luminescence. The simulation study indicates a strong dependency of signal attenuation on the depth of implantation, with the predicted ratio of output to input power ranging from 10−3 to 10−6 for sensors at the proximal and distal regions of the dermis. Analysis of the effect of sensor packing on absolute output showed a higher luminescence intensity with tightly packed microspheres, requiring fewer particles, suggesting a single site implantation with a high concentration of sensors.

10.5. Nanomaterial-Based Optical Sensors

Sensors 2019, 19, 1028 28 of 36 properties, and particle packing efﬁciency [72]. The modeling system enabled realistic simulation of light propagation in a four-layer skin tissue model with embedded microparticle sensors hexagonally packed at different densities. The spatial distribution of the emitted luminescence induced by an excitation beam with a diameter of 10 mm can be contained within an 18 mm circle, allowing a sensor design utilizing a matched optoelectronic system to effectively deliver the excitation source and subsequently collect and analyze the luminescence. The simulation study indicates a strong dependency of signal attenuation on the depth of implantation, with the predicted ratio of output to input power ranging from 10−3 to 10−6 for sensors at the proximal and distal regions of the dermis. Analysis of the effect of sensor packing on absolute output showed a higher luminescence intensity with tightly packed microspheres, requiring fewer particles, suggesting a single site implantation with a high concentration of sensors.

10.5. Nanomaterial-Based Optical Sensors Aslan et al. utilized dextran-coated gold nanoparticle colloids to measure wavelength-ratiometric resonance light scattering for transdermal dynamic glucose sensing [73]. Addition of glucose to the complex caused disassociation of the aggregates, resulting in an intensity decrease and a blue shift in the scattering spectrum. The ratio of scattered light intensities at different wavelengths (560 and 680 nm) by the nanogold aggregates was independent of aggregate concentration and light excitation. Absorption measurements were performed using a Varian UV/VIS 50 spectrophotometer; scattering measurements were conducted using a white LED light and a simple scanning monochromator. Based on measurements obtained by dextran-nanogold plasmon resonance, glucose concentrations were able to be determined in a sensing range of 1–60 mM.

10.6. Raman Spectroscopy Based Transdermal Biosensors The use of Raman spectroscopy for detection of biochemicals via transdermal biosensing has been explored [74]. The monitoring of blood glucose and interstitial glucose levels in diabetic patients is one example of this type of noninvasive monitoring. A glucose-specific Raman signature may be obtained from irradiance of the skin by a laser. Skin irradiance with an 830 nm wavelength laser light for 10 s at 0.36 W/cm2 is sufficient to obtain a Raman signal while remaining within safe operating parameters (ANSI 2000). Analysis of the vibration stretching band of the COH group at 2900 cm−1 or the stretch bands of COO and COC in the range of 900–1200 cm−1 can allow for detection of glucose [75]. Furthermore, the use of 830 nm near-infrared light can minimize biological auto-fluorescence that interferes with specific molecular signal recognition [76]. Subtraction of the interfering background fluorescence caused by biological fluids and skin can improve the detection of the target molecule (e.g., glucose) [76]. Avoidance of the interfering signal was attempted by Shao et al.; they focused the laser on a blood vessel for Raman detection of glucose and hemoglobin [77]. The hemoglobin served as an internal standard, and background contributions were determined from the tissue surrounding the in-focus blood vessel. The Raman signal peak at 1549 cm−1 for hemoglobin was compared to the most intense peak observed for glucose (1125 cm−1); the ratio of these intensities was used to determine the glucose concentration. While the authors were able to observe a concentration-dependent linear relationship for the intensity changes in the mouse model, the limit of detection for the system was 50 mmol/dl for testing the glucose solutions. This value is well above physiologically relevant blood-glucose levels (3–10 mmol/dL and <30 mmol/mL for health individuals and diabetic patients, respectively), indicating a need for an improvement in sensor detection capabilities. Incorporation of the surface-enhanced Raman scattering (SERS) effect, which is observed on nanostructured metallic materials, may also prove beneficial in transdermal Raman detection schemes. The SERS effect resulting from a target analyte adsorbed onto or in close proximity to a SERS substrate may increase the intensity of the signal by a factor of 103–108 [74,78]. Sensors 2019, 19, 1028 29 of 36

The increased Raman signal is attributed to localized surface plasmon resonance effects that are caused by an electromagnetic enhancement of the Raman scattering [78–80]. Lyandres et al. demonstrated quantitative detection of glucose using the first in vivo surface-enhanced Raman spectroscopy (SERS) sensor [78]. A film over nanospheres (FON) structure was fabricated; an aqueous suspension of polystyrene nanospheres was deposited on a titanium substrate and coated with a 200 nm layer of silver. This structure was chemically functionalized with a self-assembled monolayer of 1-decanethiol and 6-mercapto-1-hexanol or benzenethiol (BZT). These SERS-active sensors demonstrated rapid and reversible interactions with glucose and high stability up to ten days with only a 2.08% decrease in intensity over time. In vitro studies using the Clarke error grid showed 85% accuracy for validation testing in the physiologically relevant range of 10–450 mg/dL glucose; results fell within the clinically acceptable range in the presence of interfering analytes. Additional studies showed that the sensors were able to track changes in glucose over time in rats; the trends measured with the SERS sensor were similar to those indicated by a commercial electrochemical sensor; the difference in measurements between the two approaches emphasizes the need for careful calibration and determination of accuracy. This design utilized subcutaneous implantation of the sensor, which was optically accessible through a glass window that was implanted above the sensor; this approach limited interference caused by skin tissue. This design was further improved by eliminating the glass window and capturing the signal from the SERS sensor transcutaneously with spatially offset Raman spectroscopy (SORS) [81]. SORS collects the scattered Raman signal from the concentric area surrounding the central laser excitation point rather than the signal at the excitation point [82], eliminating some of the interference caused by the skin surface; this approach also facilitates capture of the SERS signal from a deeper location. The device was able to track changes in interstitial glucose levels in a rat model as a proof-of-concept. A later iteration of the device showed improved accuracy and reliability to 17 days during in vivo testing in a rat model; only an initial calibration was required [83]. The measurements indicated that the root-mean-square error for calibration (RMSEC) and prediction (RMSEP) values for the device were below the ISO/DIS 15197 standard for the hypoglycemic range (<80 mg/dL); it is typically difficult to achieve accuracy over this range [84–87]. The accuracy and stability of the device indicate the potential use of this approach for continuous glucose monitoring. Despite the challenges posed by various physiologic conditions that complicate in vivo Raman signal acquisition, there are a number of other target analytes for transdermal detection with Raman spectroscopy. Non-invasive detection of lactic acid levels in the blood or muscles of athletes is one possibility. Following calibration of a near-IR Raman setup for detection of lactic acid levels in human serum and Wistar rat blood, an optical fiber catheter was utilized to detect lactic acid levels transcutaneously in the ileac vein of Wistar rats after intraperitoneal injections of aqueous lactic acid solution (0.12 mol/L) [88]. An increase in Raman intensity was observed 30 s after injection, suggesting that lactic acid can be measured in this transdermal manner. Skin cancer screening is another potential application of Raman spectroscopy. Optical identification of biomarkers in malignancies can (a) reduce the time needed for diagnosis and (b) minimize the pain and damage caused by unnecessary biopsies. Raman spectroscopy has been utilized in efforts to distinguish the differences between malignant and benign tissues. Basal cell carcinoma (BCC) lesions, squamous cell carcinoma (SCC) lesions, and inflamed scar tissues have been investigated in such a manner. Raman spectra obtained from scar tissue, BCC lesions, and SCC lesions in 19 patients (21 lesions) were compared to healthy sites on the patient skin [89]. A Raman probe was used to obtain 30 s signals from the lesions and the perilesional healthy tissue sites. Spectral analysis was able to correctly classify 100% (21/21 lesions) of the abnormal tissue sites as inflamed scar tissue, BCC, or SCC; it also identified 91% (19/21) of the healthy tissue sites. The changes in the spectral peaks associated with different lipid and protein structures caused by various disease states can be used for identification of the lesions and surrounding tissue. Sensors 2019, 19, 1028 30 of 36

Another study, which spanned a little more than eight years, investigated in vivo Raman diagnosis of 518 skin lesions from 453 patients [90]. In addition to BCC and SCC lesions, the study investigated diagnosis of melanomas, actinic keratoses, atypical nevi, melanocytic nevi, blue nevi, and seborrheic keratoses. The Raman integration was completed within one second and utilized differences in the peak intensities in the spectral range of 500–1800 cm−1 in order to make a diagnosis; the 1055–1800 cm−1 wavelength range was noted as being useful for evaluation of melanoma. The study indicated that through use of principal component with generalized discriminant analysis (PC-GDA) and partial least-squares (PLS) analysis of the acquired Raman data, differences between malignancies and pre-malignancies from benign conditions, melanomas from nonmelanoma pigmented lesions, and melanomas from seborrheic keratoses can be identiﬁed with accuracies similar to those of clinical and other optical-based characterization methods. In a similar manner, the biochemical signatures of breast cancer lesions have been investigated via Raman spectroscopy.

11. Miscellaneous Transdermal Biosensors Lin et al. evaluated a continuous blood glucose monitoring system in female white rabbits [91]. The system contains three main components, including (a) a blood sampling and mixing mechanism with a peristaltic pump and an air stirring pump, (b) a glucose oxidase-based sensing probe (YSI Inc., Yellow springs, OH, USA.) for blood glucose measurements, and (c) a data acquisition network. An in vivo analysis methodically evaluated the effect of various parameters on the system such as (a) glucose concentration, (b) sampling pumping rate and duration, and (c) diameter of the withdrawing tube for blood and buffer samples. The glucose concentration measurement increased with increased sampling time until the measurement reached the concentration of the standard after a sample pumping time of 90 s for all of the glucose standard concentration values. Higher pumping speeds and larger diameter sample withdrawing tubes independently reduced the time required to reach the peak glucose concentration measurement; a sampling time of 0.5 min or less could not be measured in any testing setup. The in vivo study showed a good correlation between the glucose concentration measured by the system and the actual standard concentration. The system measurements remained stable throughout a seven-hour continuous testing duration, and the concentrations obtained using the system were identical to those obtained using conventional intermittent blood sampling techniques. Schiaffini et al. evaluated a commercial continuous glucose monitoring system (CGMS) for reduction of hypoglycemia episodes in children with type 1 diabetes mellitus [92]. The MiniMed. Inc. CGMS (Medtronic MiniMed, Inc., Northridge, Los Angeles, CA, USA) was used to monitor glucose levels in twenty-seven diabetic children who obtained standard 4–5 registrations of capillary glycemia per day using the Glucotrend 2 (Roche Diagnostics, Risch-Rotkreuz, Switzerland) glucometer. The average glycemic values measured by CGMS were similar to those obtained using a standard monitoring system (10.58 + 1.92 mmol/L and 10.74 + 1.58 mmol/L). The CGMS system showed a significantly higher number of asymptomatic hypoglycemic events compared to the standard approach (3.6 + 2.3 compared with 0.7 + 0.9) over 72 h, with 26% of patients experiencing hypoglycemic events with a duration less than 30 min, 44% of patients experiencing hypoglycemic events lasting between 30–60 min, and 30% of patients experiencing hypoglycemic events longer than 60 min (Figure 26). The six-week follow-up study indicated that changes in insulin therapy did not significantly affect the incidence of hypoglycemic events; however, a reduction in fructosamine levels was observed. Sensors 2019, 19, x FOR PEER REVIEW 31 of 37

An in vivo analysis methodically evaluated the effect of various parameters on the system such as (a) glucose concentration, (b) sampling pumping rate and duration, and (c) diameter of the withdrawing tube for blood and buffer samples. The glucose concentration measurement increased with increased sampling time until the measurement reached the concentration of the standard after a sample pumping time of 90 s for all of the glucose standard concentration values. Higher pumping speeds and larger diameter sample withdrawing tubes independently reduced the time required to reach the peak glucose concentration measurement; a sampling time of 0.5 min or less could not be measured in any testing setup. The in vivo study showed a good correlation between the glucose concentration measured by the system and the actual standard concentration. The system measurements remained stable throughout a seven-hour continuous testing duration, and the concentrations obtained using the system were identical to those obtained using conventional intermittent blood sampling techniques. Schiaffini et al. evaluated a commercial continuous glucose monitoring system (CGMS) for reduction of hypoglycemia episodes in children with type 1 diabetes mellitus [92]. The MiniMed. Inc. CGMS (Medtronic MiniMed, Inc, Northridge, Los Angeles, California, USA) was used to monitor glucose levels in twenty-seven diabetic children who obtained standard 4–5 registrations of capillary glycemia per day using the Glucotrend 2 (Roche Diagnostics, Risch-Rotkreuz, Switzerland) glucometer. The average glycemic values measured by CGMS were similar to those obtained using a standard monitoring system (10.58 + 1.92 mmol/l and 10.74 + 1.58 mmol/l). The CGMS system showed a significantly higher number of asymptomatic hypoglycemic events compared to the standard approach (3.6 + 2.3 compared with 0.7 + 0.9) over 72 h, with 26% of patients experiencing hypoglycemic events with a duration less than 30 min, 44% of patients experiencing hypoglycemic events lasting between 30–60 min, and 30% of patients experiencing hypoglycemic events longer than 60 min (Figure 26). The six-week follow-up study indicated that changes in insulin therapy did Sensorsnot significantly2019, 19, 1028 affect the incidence of hypoglycemic events; however, a reduction in fructosamine31 of 36 levels was observed.

FigureFigure 26. 26.Number Number of of hypoglyceamia hypoglyceamia events events (A), (A (B),) median (B) median number number of hypoglyceamic of hypoglyceamic events events within 72within h per 72 patient h per beforepatient (CGMS-1) before (CGMS-1) and after and (CGMS-2) after (CGMS-2) insulin insulin therapy therapy corrections corrections [92]. [92].

IchimoriIchimori et et al. al. describeddescribed aa needle-typeneedle-type glucose sensor for for real-time real-time monitoring monitoring of of interstitial interstitial and and bloodblood glucose glucose concentrationsconcentrations forfor aa wearablewearable artificialartificial endocrine pancreas (WAEP) (WAEP) [93]. [93]. The The sensor sensor consistedconsisted ofof aa polyimide core core with with a aplatinum platinum anode anode in inthe the distal distal region region and and a silver a silver cathode cathode in the in theproximal proximal region, region, which which was was encased encased in in a a 6% 6% polyurethane polyurethane and and poly(2-methacryloyloxyethyl phosphorylcholine-co-phosphorylcholine-co-nn-butyl-butyl methacrylate methacrylate (MPC-co-BMA) (MPC-co-BMA) membrane. membrane.In vitro In vitroanalysis analysis of the of sensor the insensor a 0.9% in NaCla 0.9% solution NaCl solution with varying with varying glucose glucose concentrations concentrations (0–500 (0–500 mg/100 mg/100 mL) showed mL) showed a linear a relationshiplinear relationship between between the output the current output of the current sensor of and the the sensor glucose and concentration. the glucose Aconcentration. residual current A ofresidual 2.8 + − current1.2 nA wasof 2.8 measured + −1.2 nA in was 0.9% measured NaCl without in 0.9% glucose; NaCl without a mean glucose; output currenta mean ofoutput 58.0 +current 7.2 nA wasof 58.0obtained + 7.2 against nA was a 200 obtained mg/100 against mL glucose a 200 solution. mg/100 In mL vivo glucose studies solution. in diabetes-induced In vivo studies beagles in showeddiabetes-induced a linear relationship beagles showed (Y = 1.03X a linear + 7.98, relationship r = 0.969) between (Y = 1.03X the interstitial + 7.98, r = glucose 0.969) concentration between the measuredinterstitial by glucose the sensor concentration (Y) and the measured whole blood by glucose the sensor concentration (Y) and the measured whole with blood a HemCue glucose B-glucose analyzer for 30–350 mg/100 mL (X) (Figure 24). Sensors implanted subcutaneously and intravenously in dogs were used to determine the time delay between interstitial glucose and blood glucose levels. By analyzing the continuous glucose curves for the two anatomical locations, a 6.6 + 1.2 min time delay was determined. In order to make continuous glucose monitoring more practical, the use of implantable analyte detection sensors that are coupled to transdermal sensor control units has been proposed [94]. The sensors may be implanted into a vein or artery, subcutaneously, or interstitially; they would be connected to a sensor control unit on the skin of the patient. Detection of glucose levels in blood or interstitial fluid can be monitored through the sensing unit directly or transmitted to another device to provide real-time feedback on blood glucose levels and/or trigger alerts for hypo- or hyperglycemic conditions. Transmitted data can be utilized directly by the patient (e.g., displayed on a telephone, computer, and/or alarm clock) or monitored remotely by caregivers or a watchdog circuit, which triggering controls in a drug delivery system to provide effective and efficient patient care.

12. Conclusions This article summarizes the development of minimally invasive techniques for transdermal drug delivery and sensing of biologically important analytes. Technologies such as electrical stimulation (e.g., reverse iontophoresis and electroporation), micro-fabrication (e.g., microneedles and micro-fluidics), fluorescence-based biosensing, tattoo-based biosensing, and drug delivery have been considered. Iontophoresis has been successfully implemented for enhancing the rate of drug delivery across the skin. The localized drug delivery and minimally invasive biosensing based on microdevices such as microneedles is the recent evolution in the area. The nanomaterial (LaB6, gold nanoparticles) loaded robust microneedles offer an effective means for drug administration in the body; the system is modified to be repeatedly switchable, which enables release of a consistent dosage of the drug. Further the scope of ultrasound directed sensing and drug transport has been thoroughly Sensors 2019, 19, 1028 32 of 36 reviewed, the ultrasound based portable devices are a part of routine healthcare services now-a-days. The microfluidic systems like piezoelectric micropumps often serve as efficient route for targeted drug delivery. In addition, the role of optical sensors in detection of various probe molecules and glucose monitoring, by means of light emission and scattering techniques like fluorescence, luminescence and surface enhanced Raman spectroscopies were discussed thoroughly. Overall, various scientific advancements in the area enable the targeted disruption of uppermost layer of skin without harming the sensitive tissues have added the additional appendages with new level of potentials, which renders the transdermal drug delivery with marked influence in medical field.

Acknowledgments: The authors wish to acknowledge support from SERB DST Grant Number VJR-2017-000034. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Ita, K. Transdermal delivery of drugs with microneedles: Strategies and outcomes. J. Drug. Deliv. Sci. Technol. 2015, 29, 16–23. [CrossRef] 2. Ita, K.B.; Du Preez, J.; Lane, M.E. Dermal delivery of selected hydrophilic drugs from elastic liposomes: Effect of phospholipid formulation and surfactants. J. Pharm. Pharmacol. 2007, 59, 1215–1222. [CrossRef] [PubMed] 3. Anselmo, A.C.; Gokarn, Y.; Mitragotri, S. Non-invasive delivery strategies for biologics. Nat. Rev. Drug Discov. 2018, 18, 19–40. [CrossRef][PubMed] 4. Lipani, L.; Dupont, B.G.; Doungmene, F.; Marken, F.; Tyrrell, R.M.; Guy, R.H.; Ilie, A. Non-invasive, transdermal, path-selective and speciﬁc glucose monitoring via a graphene-based platform. Nat. Nanotechnol. 2018, 13, 504–511. [CrossRef][PubMed] 5. Djabri, A.; Guy, R.H.; Delgado-Charro, M.B. Transdermal iontophoresis of ranitidine: An opportunity in paediatric drug therapy. Int. J. Pharm. 2012, 435, 27–32. [CrossRef][PubMed] 6. Lundborg, M.; Wennberg, C.L.; Narangifard, A.; Lindahl, E.; Norlen, L. Predicting drug permeability through skin using molecular dynamics simulation. J. Control. Release 2018, 283, 269–279. [CrossRef][PubMed] 7. Banga, A.K.; Bose, S.; Ghosh, T.K. Iontophoresis and electroporation: Comparisons and contrasts. Int. J. Pharm. 1999, 179, 1–19. [CrossRef] 8. Shim, G.; Kim, D.; Park, G.T.; Jin, H.; Suh, S.K.; Oh, Y.K. Therapeutic gene editing: Delivery and regulatory perspectives. Acta Pharmacol. Sin. 2017, 38, 738–753. [CrossRef][PubMed] 9. Kaur, M.; Ita, K.B.; Popova, I.E.; Parikh, S.J.; Bair, D.A. Microneedle-assisted delivery of verapamil hydrochloride and amlodipine besylate. Eur. J. Pharm. Biopharm. 2014, 86, 284–291. [CrossRef][PubMed] 10. Ita, K. Transdermal delivery of drugs with microneedles—Potential and challenges. Pharmaceutics 2015, 7, 90–105. [CrossRef][PubMed] 11. Tachibana, K. Transdermal delivery of insulin to alloxan-diabetc rabbits by ultrasound exposure. Pharm. Res. 1992, 9, 952–954. [CrossRef][PubMed] 12. Azagury, A.; Khoury, L.; Enden, G.; Kost, J. Ultrasound mediated transdermal drug delivery. Adv. Drug Deliv. Rev. 2014, 72, 127–143. [CrossRef][PubMed] 13. Shimizu, H.; Tamada, Y.; Shimizu, J.; Ohshima, Y.; Matsumoto, Y.; Sugenoya, J.I. Effectiveness of iontophoresis with alternating current (AC) in the treatment of patients with palmoplantar hyperhidrosis. J. Dermatol. 2003, 30, 444–449. [CrossRef][PubMed] 14. Kalluri, H.; Banga, A.K. Transdermal delivery of proteins. Aaps Pharmscitech. 2011, 12, 431–441. [CrossRef] [PubMed] 15. Cázares-Delgadillo, J.; Naik, A.; Ganem-Rondero, A.; Quintanar-Guerrero, D.; Kalia, Y.N. Transdermal Delivery of Cytochrome C—A 12.4 kDa protein—Across Intact Skin by constant–current Iontophoresis. Pharm. Res. 2007, 24, 1360–1368. [CrossRef][PubMed] 16. Manda, P.; Angamuthu, M.; Hiremath, S.R.; Raman, V.; Murthy, S.N. Iontophoretic drug delivery for the treatment of scars. J. Pharm. Sci. 2014, 103, 1638–1642. [CrossRef][PubMed] 17. Delgado-Charro, M.B.; Guy, R.H. Transdermal reverse iontophoresis of valproate: A noninvasive method for therapeutic drug monitoring. Pharm. Res. 2003, 20, 1508–1513. [CrossRef][PubMed] Sensors 2019, 19, 1028 33 of 36

18. Subramony, J.A.; Sharma, A.; Phipps, J.B. Microprocessor controlled transdermal drug delivery. Int. J. Pharm. 2006, 317, 1–6. [CrossRef][PubMed] 19. Koschinsky, T.; Heinemann, L. Sensors for glucose monitoring: Technical and clinical aspects. Diabetes Metab. Res Rev. 2001, 17, 113–123. [CrossRef][PubMed] 20. Rao, G.; Glikfeld, P.; Guy, R.H. Reverse iontophoresis: Development of a noninvasive approach for glucose monitoring. Pharm. Res. 1993, 10, 1751–1755. [CrossRef][PubMed] 21. Bandodkar, A.J.; Jia, W.; Yardımcı, C.; Wang, X.; Ramirez, J.; Wang, J. Tattoo-based noninvasive glucose monitoring: A proof-of-concept study. Anal. Chem. 2014, 87, 394–398. [CrossRef][PubMed] 22. Alvarez-Lorenzo, C.; Concheiro, A. Molecularly imprinted polymers for drug delivery. J. Chromatogr. B 2004, 804, 231–245. [CrossRef][PubMed] 23. Liu, J.; Liu, C.; Liu, H.; Jiang, L.; Yang, Q.; Cai, X. Study of noninvasive sampling of subcutaneous glucose by reverse iontophoresis. In Proceedings of the 2nd IEEE International Conference on Nano/Micro Engineered and Molecular Systems (NEMS’07), Bangkok, Thailand, 16–19 January 2007. [CrossRef] 24. Ching, C.T.S.; Sun, T.P.; Huang, S.H.; Shieh, H.L.; Chen, C.Y. A mediated glucose biosensor incorporated with reverse iontophoresis function for noninvasive glucose monitoring. Ann. Biomed. Eng. 2010, 38, 1548–1555. [CrossRef][PubMed] 25. Ching, C.T.S.; Chou, T.R.; Sun, T.P.; Huang, S.Y.; Shieh, H.L. Simultaneous, noninvasive, and transdermal extraction of urea and homocysteine by reverse iontophoresis. Int. J. Nanomed. 2011, 6, 417–423. [CrossRef] [PubMed] 26. Pandey, P.C.; Pandey, G.; Narayan, R.J. A Process for Making Microneedle-Based Transdermal Biosensor Involving Poyethyleninmine-Modified Prussian Blue-Gold Nanohybrids Enzyme Ink for On-Chip Electrochemical Biosensing in Body Fluid. India Patent 201811044930, 28 November 2018. 27. Pandey, P.C.; Pandey, G.; Narayan, R.J. A Process for Making Solid-State Ion Sensor for On-Chip Determination of Potassium Ion in Body Fluid. India Patent 201811041978, 6 November 2018. 28. Wei, C.; Dong, X.; Zhang, Y.; Liang, J.; Yang, A.; Zhu, D.; Liu, T.; Kong, D.; Lv, F. Simultaneous fluorescence imaging monitoring of the programMed. release of dual drugs from a hydrogel-carbon nanotube delivery system. Sens. Actuators B Chem. 2018, 273, 264–275. [CrossRef] 29. Cinteza, L.O. Quantum dots in biomedical applications: Advances and challenges. J. Nanophotonics 2010, 4, 042503. [CrossRef] 30. Michalet, X.; Pinaud, F.F.; Bentolila, L.A.; Tsay, J.M.; Doose, S.J.J.L.; Li, J.J.; Sundaresan, G.; Wu, A.M.; Gambhir, S.S.; Weiss, S. Quantum dots for live cells, in vivo imaging, and diagnostics. Science 2005, 307, 538–544. [CrossRef][PubMed] 31. Hsu, C.P.; Hejazi, Z.; Armagan, E.; Zhao, S.; Schmid, M.; Zhang, H.; Guo, H.; Weidenbacher, L.; Rossi, R.M.; Koebel, M.M.; et al. Carbon dots and fluorescein: The ideal FRET pair for the fabrication of a precise and fully reversible ammonia sensor. Sens. Actuators B Chem. 2017, 253, 714–722. [CrossRef] 32. Wang, P.; Fan, Y.; Lu, L.; Liu, L.; Fan, L.; Zhao, M.; Xie, Y.; Xu, C.; Zhang, F. NIR-II nanoprobes in-vivo assembly to improve image-guided surgery for metastatic ovarian cancer. Nat. Commun. 2018, 9, 1–10. [CrossRef][PubMed] 33. Biju, V.; Itoh, T.; Anas, A.; Sujith, A.; Ishikawa, M. Semiconductor quantum dots and metal nanoparticles: Syntheses, optical properties, and biological applications. Anal. Bioanal. Chem. 2008, 391, 2469–2495. [CrossRef][PubMed] 34. Hanks, D.F.; Lu, Z.; Sircar, J.; Salamon, T.R.; Antao, D.S.; Bagnall, K.R.; Barabadi, B.; Wang, E.N. Nanoporous membrane device for ultra high heat flux thermal management. Microsyst Nanoeng. 2018, 4, 1–10. [CrossRef]

35. Chen, C.J.; Chen, D.H. Preparation of LaB6 nanoparticles as a novel and effective near-infrared photothermal conversion material. Chem. Eng. J. 2012, 180, 337–342. [CrossRef]

36. Lui, B.H.; Chen, D.H. LaB6 nanoparticles with carbon-doped silica coating for ﬂuorescence imaging and near-IR photothermal therapy of cancer cells. Acta Biomater. 2013, 9, 7556–7563. [CrossRef][PubMed] 37. Chen, M.C.; Ling, M.H.; Wang, K.W.; Lin, Z.W.; Lai, B.H.; Chen, D.H. Near-infrared light-responsive composite microneedles for on-demand transdermal drug delivery. Biomacromolecules 2015, 16, 1598–1607. [CrossRef][PubMed] 38. Jiang, F.; Leong, Y.K.; Saunders, M.; Martyniuk, M.; Faraone, L.; Keating, A.; Dell, J.M. Uniform dispersion of lanthanum hexaboride nanoparticles in a silica thin ﬁlm: Synthesis and optical properties. ACS Appl. Mater. Interfaces 2012, 4, 5833–5838. [CrossRef][PubMed] Sensors 2019, 19, 1028 34 of 36

39. Bollella, P.; Sharma, S.; Cass, A.E.G.; Antiochia, R. Microneedle-based biosensor for minimally-invasive lactate detection. Biosens. Bioelectron. 2019, 123, 152–159. [CrossRef][PubMed] 40. Ranamukhaarachchi, S.A.; Padeste, C.; Dubner, M.; Hafeli, U.O.; Stoeber, B.; Cadarso, V.J. Integrated hollow microneedle-optofluidic biosensor for therapeutic drug monitoring in sub-nanoliter volumes. Sci. Rep. 2016, 6, 29075. [CrossRef][PubMed] 41. Sanaee, Z.; Mohajerzadeh, S. Ultrafine hollow needle formation on silicon. J. Appl. Phys. 2010, 107, 083711. [CrossRef] 42. Miller, P.R.; Gittard, S.D.; Edwards, T.L.; Lopez, D.M.; Xiao, X.; Wheeler, D.R.; Monteiro-Riviere, N.A.; Brozik, S.M.; Polsky, R.; Narayan, R.J. Integrated carbon fiber electrodes within hollow polymer microneedles for transdermal electrochemical sensing. Biomicrofluidics 2011, 5, 013415. [CrossRef][PubMed] 43. Windmiller, J.R.; Zhou, N.; Chuang, M.C.; Valdés-Ramírez, G.; Santhosh, P.; Miller, P.R.; Narayan, R.; Wang, J. Microneedle array-based carbon paste amperometric sensors and biosensors. Analyst 2011, 136, 1846–1851. [CrossRef][PubMed] 44. Dumett, M.A.; Rosen, I.G.; Sabat, J.; Shaman, A.; Tempelman, L.; Wang, C.; Swift, R.M. Deconvolving an estimate of breath measured blood alcohol concentration from biosensor collected transdermal ethanol data. Appl. Math. Comput. 2008, 196, 724–743. [CrossRef][PubMed] 45. Dougherty, D.M.; Charles, N.E.; Acheson, A.; John, S.; Furr, R.M.; Hill-Kapturczak, N. Comparing the detection of transdermal and breath alcohol concentrations during periods of alcohol consumption ranging from moderate drinking to binge drinking. Exp. Clin. Psychopharmacol. 2012, 20, 373–381. [CrossRef] [PubMed] 46. Gurvich, E.M.; Kenna, G.A.; Leggio, L. Use of novel technology-based techniques to improve alcohol-related outcomes in clinical trials. Alcohol. Alcohol. 2013, 48, 712–719. [CrossRef][PubMed] 47. Leffingwell, T.R.; Cooney, N.J.; Murphy, J.G.; Luczak, S.; Rosen, G.; Dougherty, D.M.; Barnett, N.P. Continuous objective monitoring of alcohol use: Twenty-first century measurement using transdermal sensors. Alcohol. Clin. Exp. Res. 2013, 37, 16–22. [CrossRef][PubMed] 48. Luczak, S.E.; Rosen, I.G.; Weiss, J. Determining blood and/or breath alcohol concentration from transdermal alcohol data. In Proceedings of the American Control Conference (ACC), Washington, DC, USA, 17–19 June 2013; pp. 473–478. [CrossRef] 49. Barnett, N.P.; Meade, E.B.; Glynn, T.R. Predictors of detection of alcohol use episodes using a transdermal alcohol sensor. Exp. Clin. Psychopharmacol. 2014, 22, 86–96. [CrossRef][PubMed] 50. Lee, S.; Nayak, V.; Dodds, J.; Pishko, M.; Smith, N.B. Glucose measurements with sensors and ultrasound. Ultrasound Med. Biol. 2005, 31, 971–977. [CrossRef][PubMed] 51. Lee, S.; Smith, N.B.; Markley, D.; Snyder, B.; Uzgur, A.E.; Pishko, M.; Newnham, R.E. Composite transducer arrays for the treatment of diabetes. Int. J. Appl. Ceram. Technol. 2005, 2, 308–316. [CrossRef] 52. Laing, S.; Jamieson, L.E.; Faulds, K.; Graham, D. Surface-enhanced Raman spectroscopy for in vivo biosensing. Nat. Rev. Chem. 2017, 1, 1–19. [CrossRef] 53. Gadre, A.P.; Nijdam, A.J.; Garra, J.A.; Monica, A.H.; Cheng, M.C.; Luo, C.; Srivastava, Y.N.; Schneider, T.W.; Long, T.J.; White, R.C.; et al. Fabrication of a fluid encapsulated dermal patch using multilayered SU-8. Sens. Actuators A Phys. 2004, 114, 478–485. [CrossRef] 54. Shoji, E. Fabrication of a diaphragm micropump system utilizing the ionomer-based polymer actuator. Sens. Actuators B Chem. 2016, 237, 660–665. [CrossRef] 55. Li, D.; Ji, Y.; Liang, W.; Zhang, X.; Yu, H.; Xu, K. A portable instrument for continuous glucose monitoring by the integration of microfluidic chip and micro-glucose sensor. In Proceedings of the Microfluidics, BioMEMS, and Medical Microsystems XI, San Francisco, CA, USA, 9 March 2013; International Society for Optics and Photonics; Volume 8615, p. 861519. [CrossRef] 56. Yu, H.X.; Li, D.C.; Ji, Y.J.; Zhang, X.L.; Xu, K.X. An Interstitial Fluid Transdermal Extraction Chip with Vacuum Generator and Volume Sensor for Continuous Glucose Monitoring. Key Eng. Mater. 2013, 562, 571–575. [CrossRef] 57. Cordeiro, C.A.; Sias, A.; Koster, T.; Westerink, B.H.C.; Cremers, T.I.F.H. In vivo “real-time” monitoring of glucose in the brain with an amperometric enzyme-based biosensor based on gold coated tungsten (W-Au) microelectrodes. Sens. Actuators B Chem. 2018, 263, 605–613. [CrossRef] 58. Ballerstadt, R.; Schultz, J.S. A fluorescence affinity hollow fiber sensor for continuous transdermal glucose monitoring. Anal. Chem. 2000, 72, 4185–4192. [CrossRef][PubMed] Sensors 2019, 19, 1028 35 of 36

59. Ballerstadt, R.; Gowda, A.; McNichols, R. Fluorescence resonance energy transfer-based near-infrared fluorescence sensor for glucose monitoring. Diabetes Technol. Ther. 2004, 6, 191–200. [CrossRef][PubMed] 60. Ballerstadt, R.; Polak, A.; Beuhler, A.; Frye, J. In vitro long-term performance study of a near-infrared fluorescence affinity sensor for glucose monitoring. Biosens. Bioelectron. 2004, 19, 905–914. [CrossRef] [PubMed] 61. Ballerstadt, R.; Evans, C.; Gowda, A.; McNichols, R. In vivo performance evaluation of a transdermal near-infrared fluorescence resonance energy transfer affinity sensor for continuous glucose monitoring. Diabetes Technol. Ther. 2006, 8, 296–311. [CrossRef][PubMed] 62. Umezawa, K.; Yoshida, M.; Kamiya, M.; Yamasoba, T.; Urano, Y. Rational design of reversible fluorescent probes for live-cell imaging and quantification of fast glutathione dynamics. Nat. Chem. 2017, 9, 279–286. [CrossRef][PubMed] 63. Meller, K.; Szumski, M.; Buszewski, B. Microfluidic reactors with immobilized enzymes characterization, dividing, perspectives. Sens. Actuators B Chem. 2017, 244, 84–106. [CrossRef] 64. Kermis, H.R.; Rao, G.; Barbari, T.A. Transport properties of pHEMA membranes for optical glucose affinity sensors. J. Membr. Sci. 2003, 212, 75–86. [CrossRef] 65. Vakoc, B.J.; Fukumura, D.; Jain, R.K.; Bouma, B.E. Cancer imaging by optical coherence tomography: Preclinical progress and clinical potential. Nat. Rev. Cancer 2012, 12, 363–368. [CrossRef][PubMed] 66. Khan, F.; Saxl, T.E.; Pickup, J.C. Fluorescence intensity-and lifetime-based glucose sensing using an engineered high-Kd mutant of glucose/galactose-binding protein. Anal. Biochem. 2010, 399, 39–43. [CrossRef] [PubMed] 67. Heo, Y.J.; Shibata, H.; Okitsu, T.; Kawanishi, T.; Takeuchi, S. Long-term in vivo glucose monitoring using fluorescent hydrogel fibers. Proc. Natl. Acad. Sci. USA 2011, 108, 13399–13403. [CrossRef][PubMed] 68. Shibata, H.; Heo, Y.J.; Okitsu, T.; Matsunaga, Y.; Kawanishi, T.; Takeuchi, S. Injectable hydrogel microbeads for fluorescence-based in vivo continuous glucose monitoring. Proc. Natl. Acad. Sci. USA 2010, 107, 17894–17898. [CrossRef][PubMed] 69. Rhaman, M.M.; Hasan, M.H.; Alamgir, A.; Xu, L.; Powell, D.R.; Wong, B.M.; Tandon, R.; Hossain, M.A. Highly selective and sensitive macrocycle-based dinuclear foldamer for fluorometric and colorimetric sensing of citrate in water. Sci. Rep. 2018, 8, 286. [CrossRef][PubMed] 70. Badugu, R.; Lakowicz, J.R.; Geddes, C.D. Enhanced fluorescence cyanide detection at physiologically lethal levels: Reduced ICT-based signal transduction. J. Am. Chem. Soc. 2005, 127, 3635–3641. [CrossRef][PubMed] 71. Al-Agel, F.A.; Mahmoud, W.E. “Turn-off-on” fluorescence probe based functionalized InP quantum wires for detection of cyanide ions. Sens. Actuators B Chem. 2013, 183, 441–445. [CrossRef] 72. Long, R.; McShane, M.J. Three-dimensional, multiwavelength Monte Carlo simulations of dermally implantable luminescent sensors. J. Biomed. Opt. 2010, 15, 027011. [CrossRef][PubMed] 73. Aslan, K.; Lakowicz, J.R.; Geddes, C.D. Nanogold plasmon resonance-based glucose sensing. 2. Wavelength-ratiometric resonance light scattering. Anal. Chem. 2005, 77, 2007–2014. [CrossRef][PubMed] 74. Uskoković-Marković,S.; Jelikić-Stankov, M.; Holclajtner-Antunović,I.; Đurđević,P. Raman spectroscopy as a new biochemical diagnostic tool. J. Med. Biochem. 2013, 32, 96–103. [CrossRef] 75. Shi, L.; Zheng, C.; Shen, Y.; Chen, Z.; Silveira, E.S.; Zhang, L.; Wei, M.; Liu, C.; de Sena-Tomas, C.; Targoff, K.; et al. Optical Imaging of Metabolic Dynamics in Animals. Nat. Commun. 2018, 9, 2995. [CrossRef][PubMed] 76. Enejder, A.M.; Scecina, T.G.; Oh, J.; Hunter, M.; Shih, W.; Sasic, S.; Horowitz, G.L.; Feld, M.S. Raman spectroscopy for noninvasive glucose measurements. J. Biomed. Opt. 2005, 10, 031114. [CrossRef][PubMed] 77. Shao, J.; Lin, M.; Li, Y.; Li, X.; Liu, J.; Liang, J.; Yao, H. In vivo blood glucose quantification using Raman spectroscopy. PLoS ONE 2012, 7, e48127. [CrossRef][PubMed] 78. Lyandres, O.; Yuen, J.M.; Shah, N.C.; VanDuyne, R.P.; Walsh, J.T., Jr.; Glucksberg, M.R. Progress toward an in vivo surface-enhanced Raman spectroscopy glucose sensor. Diabetes Technol. Ther. 2008, 10, 257–265. [CrossRef][PubMed] 79. Schatz, G.C.; Van Duyne, R.P.; Chalmers, J.M.; Griffiths, P.R.Electromagnetic Mechanism of Surface-Enhanced Spectroscopy. In Handbook of Vibrational Spectroscopy; Wiley & Sons: New York, NY, USA, 2002; pp. 759–774. [CrossRef] 80. Ju, J.; Liu, W.; Perlaki, C.M.; Chen, K.; Feng, C.; Liu, Q. Sustained and Cost Effective Silver Substrate for Surface Enhanced Raman Spectroscopy Based Biosensing. Sci. Rep. 2017, 7, 6917. [CrossRef][PubMed] Sensors 2019, 19, 1028 36 of 36

81. Yuen, J.M.; Shah, N.C.; Walsh, J.T., Jr.; Glucksberg, M.R.; Van Duyne, R.P. Transcutaneous glucose sensing by surface-enhanced spatially offset Raman spectroscopy in a rat model. Anal. Chem. 2010, 82, 8382–8385. [CrossRef][PubMed] 82. Matousek, P.; Clark, I.P.; Draper, E.R.C.; Morris, M.D.; Goodship, A.E.; Everall, N.; Towrie, M.; Finney, W.F.; Parker, A.W. Subsurface probing in diffusely scattering media using spatially offset Raman spectroscopy. Appl. Spectrosc. 2005, 59, 393–400. [CrossRef][PubMed] 83. Ma, K.; Yuen, J.M.; Shah, N.C.; Walsh, J.T., Jr.; Glucksberg, M.R.; Van Duyne, R.P. In vivo, transcutaneous glucose sensing surface-enhanced spatially offset Raman spectroscopy: Multiple rats, improved hypoglycemic accuracy, low incident power, and continuous monitoring for greater than 17 days. Anal. Chem. 2011, 83, 9146–9152. [CrossRef][PubMed] 84. Gao, Z.F.; Sann, E.E.; Lou, X.; Liu, R.; Dai, J.; Zuo, X.; Xia, F.; Jiang, L. Naked-eye point-of-care testing platform based on a pH-responsive superwetting surface: Toward the non-invasive detection of glucose. NPG Asia Mater. 2018, 11, 117. [CrossRef] 85. Oliver, N.S.; Toumazou, C.; Cass, A.E.G.; Johnston, D.G. Glucose sensors: A review of current and emerging technology. Diabetic Med. 2009, 26, 197–210. [CrossRef][PubMed] 86. Brauker, J. Continuous glucose sensing: Future technology developments. Diabetes Technol. Ther. 2009, 11, S25–S36. [CrossRef][PubMed] 87. Aye, T.; Block, J.; Buckingham, B. Toward closing the loop: An update on insulin pumps and continuous glucose monitoring systems. Endocrinol. Metab. Clin. N. Am. 2010, 39, 609–624. [CrossRef][PubMed] 88. Weigl, W.; Milej, D.; Gerega, A.; Toczylowska, B.; Sawosz, P.; Kacprzak, M.; Janusek, D.; Wojtkiewicz, S.; Maniewski, R.; Liebert, A. Confirmation of brain death using optical methods based on tracking of an optical contrast agent: Assessment of diagnostic feasibility. Sci. Rep. 2018, 8, 7332. [CrossRef][PubMed] 89. Lieber, C.A.; Majumder, S.K.; Ellis, D.L.; Billheimer, D.D.; Mahadevan-Jansen, A. In vivo nonmelanoma skin cancer diagnosis using Raman microspectroscopy. Laser Surg. Med. 2008, 40, 461–467. [CrossRef][PubMed] 90. Lui, H.; Zhao, J.; McLean, D.I.; Zeng, H. Real-time Raman spectroscopy for in vivo skin cancer diagnosis. Cancer Res. 2012, 72, 2491–2500. [CrossRef][PubMed] 91. Oh, S.G.; Li, X.; Lee, H.W.; Singh, T.D.; Lee, S.B.; Ji, H.D.; Yoon, G.; Cho, S.J.; Lee, I.K.; Jeong, S.Y.; et al. Non-invasive visualization of mast cell recruitment and its effects in lung cancer by optical reporter gene imaging and glucose metabolism monitoring. Biomaterials 2017, 112, 192–203. [CrossRef][PubMed] 92. Tsalikian, E.; Fox, L.; Weinzimer, S.; Buckingham, B.; White, N.H.; Beck, R.; Kollman, C.; Xing, D.; Ruedy, K. Feasibility of prolonged continuous glucose monitoring in toddlers with type 1 diabetes. Pediatr. Diabetes 2012, 13, 301–307. [CrossRef][PubMed] 93. Ichimori, S.; Nishida, K.; Shimoda, S.; Sekigami, T.; Matsuo, Y.; Ichinose, K.; Shichiri, M.; Sakakida, M.; Araki, E. Development of a highly responsive needle-type glucose sensor using polyimide for a wearable artificial endocrine pancreas. J. Artif. Organs 2006, 9, 105–113. [CrossRef][PubMed] 94. Wu, W.; Shen, J.; Li, Y.; Zhu, H.; Banerjee, P.; Zhou, S. Specific glucose-to-SPR signal transduction at physiological pH by molecularly imprinted responsive hybrid microgels. Biomaterials 2012, 33, 7115–7125. [CrossRef][PubMed]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).