Leveraging the power of ultrasound for therapeutic design and optimization

Abstract

Contrast agent-enhanced ultrasound can facilitate personalized therapeutic strategies by providing the technology to measure local blood flow rate, to selectively image receptors on the vascular endothelium, and to enhance localized drug delivery. Ultrasound contrast agents are micron-diameter encapsulated bubbles that circulate within the vascular compartment and can be selectively imaged with ultrasound. Microbubble transport-based estimates of local blood flow can quantify changes resulting from anti-angiogenic therapies and facilitate differentiation of angiogenic mechanisms. Microbubbles that are conjugated with targeting ligands attach to endothelial surface receptors that are upregulated in disease, providing high signal-to-noise ratio images of pathological vasculature. In addition to imaging applications, microbubbles can be used to enhance localized gene and drug delivery, either by changing membrane and vascular permeability or by carrying and locally releasing cargo. Our goal in this review is to provide an overview of the use of contrast-enhanced ultrasound methodologies in the design and evaluation of therapeutic strategies with emphases on quantitative blood flow mapping, molecular imaging, and enhanced drug delivery.

1. Introduction

Our goal in this review is to provide an overview of contrast-agent enhanced ultrasound methodologies in the design and application of therapeutic systems. Ultrasound contrast agents are micron-diameter encapsulated bubbles that circulate in the vasculature in a manner similar to erythrocytes; these microbubbles can be selectively imaged with ultrasound. The restriction of intact microbubbles to the intravascular space simplifies the analysis of imaging studies, facilitating quantitation of circulation and accumulation. In this review, we first focus on microbubble imaging, as applied in blood flow estimation and molecular imaging. Second, we detail recent progress in the use of microbubbles to enhance local drug delivery.

The typical microbubble has a core of perfluorocarbon gas and a shell containing a matrix-forming phospholipid and an emulsifying lipid[1]. The shell imparts a barrier between the gas and water for stability and reducing thrombogenic effects in vivo. The emulsifying lipid usually contains a polyethelene glycol (PEG) group that is required for efficient production of lipid-coated particles and provides a steric brush that minimizes shell interactions with blood components. Since particle-based imaging and drug delivery systems may be recognized by opsonins and cell-surface receptors and thus removed by the reticuloendothelial system (RES), the presence of a PEG group enhances circulation. The effective emulsifier concentration is limited; exceeding a concentration of 20 mol% 2-kDa PEG can reduce stability [2]whereas reducing the emulsifier concentration or shortening the PEG length can result in coalescence.

The micron-diameter gas core of microbubbles expands and contracts in response to the ultrasound wave, typically expanding slowly and contracting rapidly[3]. As a result of the nonlinear response to the transmitted wave, microbubble-generated sound waves contain frequency components that are multiples and sub-multiples of the driving frequency. Depending on the diameter and on the shell components, a resonance frequency can be defined for each microbubble construct. When driven by ultrasound at a frequency near resonance, microbubble oscillation is efficient and periodic [4]. Further, when driven by a high intensity ultrasound wave, the wall of a microbubble contracts at a velocity of hundreds to thousands of meters per second, and this rapid collapse generates very strong echoes [5].

With rapid collapse due to application of a high ultrasound pressure (~250 kPa or greater), microbubble contrast agents will fragment; the shell can then reform around the bubble fragments to serve as a barrier to passive diffusion, or the remaining gas fragments can diffuse. Fragmentation is often leveraged when microbubbles are applied to quantify blood flow and such techniques are widely used to assess response to therapies, as described in Section 2.

Small molecules, peptides or antibodies are covalently attached to the distal end of PEG for targeting purposes[1]. Such molecularly-targeted agents show great promise to characterize diseases in which vascular receptors are upregulated, pointing the way to effective, individualized therapies; their application is summarized in Section 3. With microbubble collapse near a vessel wall, liquid jets can impinge on the endothelium and disrupt cell membrane integrity. While the generation of liquid jets has been shown to result in holes within cell membranes at high ultrasound intensities, mechanisms responsible for enhanced transport with lower ultrasound intensities are still under investigation[6–8]. Rapid microbubble collapse and associated biological effects facilitate enhanced drug and gene transport and the applications of such techniques are described in Section 4.

2. Parametric imaging of blood flow

Images of intact microbubbles within a target tissue provide estimates of fractional blood volume and contrast agent arrival time[8]. Fragmenting microbubbles locally within a tissue and imaging the replenishment of the vasculature yields local estimates of blood flow. Typically, a gray scale B-mode image is overlaid with a colored parametric map of blood flow, facilitating an understanding of spatial variations in flow in disease or in response to therapy.

In addition to ultrasound, single photon emission computed tomography (SPECT), positron emission tomography (PET), and magnetic resonance imaging (MRI) have been used to quantify blood volume, flow and vascular permeability[9–11]. In particular, SPECT has been routinely utilized to assess myocardial function and regional perfusion in nuclear cardiology [12] and MRI is widely used to assess flow in cancer and cardiovascular disease[13, 14]. The advantages of ultrasound include the use of an entirely intravascular contrast agent, thus facilitating quantitation, cost, portability and ease of use, and a spatial resolution on the order of hundreds of microns to millimeters.

2.1. Differentiating microbubble from tissue echoes

Traditional B-mode ultrasound creates images of the summed backscatter from both microbubbles and tissue, without discriminating between the two components. Without the use of multiple transmitted pulses, one method to specifically identify microbubble echoes is to use the higher frequency components in microbubble echoes to create images [8]. Alternatively, microbubble-specific imaging techniques can differentiate microbubble and tissue echoes by transmitting pulse sequences that exploit nonlinear changes in microbubble echoes in response to changes in the phase or amplitude of the transmitted pulses. The echoes are then summed algebraically to yield images in which tissue echoes are cancelled and microbubble echoes are summed coherently (Fig. 1) [15–17]. For example, using contrast pulse sequencing (CPS) and the summation of echoes resulting from pulses that were transmitted with phase and amplitude modulation, the contrast agent-to-tissue ratio after summation exceeds 10 times (20 dB) and the image contains little tissue background (Fig. 1b). Using a high frequency scanner (20–50MHz), the spatial resolution of the image improves compared with clinical frequencies (Fig. 1a, 1c); however, some tissue echoes may remain present, although diminished, after pulse summation (Fig. 1c).

Figure 1. Sample images and methodologies used to distinguish between tissue and microbubble echoes from molecularly targeted agents with clinical (1–15 MHz) and high frequency (20–50 MHz) ultrasound in mouse models of cancer (red circle indicates tumor).

In each case tumors with diameters on the order of 5 mm are displayed. Trains of pulses are applied to the transducer with variations in the phase and amplitude of the pulse. Pulse inversion and power modulation vary only the phase or amplitude of the pulses, respectively, while CPS varies both. Algebraic combinations of these echoes cancel contributions from tissue. (a) High frequency (~40 MHz) image obtained in a linear mode (B-mode) in which the green pixels indicate bound agents and the grey indicates surrounding tissue. Microbubbles were detected based on their entry and accumulation. (b) 7 MHz ultrasound image of bound microbubbles using CPS. Here, the entire image represents microbubble echoes as tissue echoes are efficiently suppressed. (c) High frequency image (~40 MHz) obtained using power modulation. Note the high spatial resolution (compared with b), but the skin line at the top of the image results from reduced tissue cancelation at high frequencies.

2.2. Destruction-replenishment strategy to estimate blood flow

Contrast agent-enhanced ultrasound imaging can be used to detect perfusion defects and follow cardiac wall motion using a destruction-replenishment method[18]. With this method, microbubbles are perfused into the myocardium and then locally fragmented by a high pressure ultrasonic pulse. The mean myocardial microbubble velocity is estimated based on the local replenishment of microbubbles, indicated by a change in video intensity (Fig. 2a). The relationship between the acquired video amplitude, the microbubble replenishment rate (β) and the microvascular cross-sectional area (A) are described by an exponential function, derived by fitting the video amplitude curve to determine the parameters A and β (Fig. 2b). Using this strategy, myocardial perfusion can be visualized, evaluated, [18–20] and correlated with parameters obtained using SPECT [21].

Figure 2. Destruction-replenishment methods for blood flow estimation and applications.

After locally fragmenting microbubbles with an ultrasound pulse and imaging their return, the local blood flow rate can be estimated (a). The increasing echo amplitude is modeled as an exponential function (b) and critical parameters, blood velocity (β), blood volume (A) and blood flow (A×β) can be estimated in each pixel. (c–e) The microvascular physiology of a rat glioma tumor (labeled by “T”) was depicted by these three parameters (c, d, and e). (f–g) Parametric imaging of blood flow (A×β) was utilized to evaluate rat tumor vasculature after treatment with anti-angiogenic drugs. Compared to the control tumor (f), blood flow in the treated tumor (g) was reduced. (c–e) and (f–g) reproduced with permission from [22] and [23], respectively.

This destruction-replenishment strategy has been applied to clinical applications including cancer [22, 23] (Fig. 2c–e), renal disease [24]and peripheral vascular disease [25]. Tumor blood velocity is typically lower than that of the surrounding tissue; while tumor blood volume can be greater than or less than the surrounding tissue[22]. Microbubble replenishment kinetics can be quantified by using bolus injections as well as continuous infusion, where, following a bolus injection, tumor perfusion based on the A β analysis has been applied to indicate the tumor response to anti-angiogenic drugs[26] (Fig. 2f, g).

Following the local destruction of microbubbles, the time required for replenishment of microbubbles within the region of interest can also be estimated, often termed the “replenishment time” and contrasted with measures of the initial arrival time or washout rate after bolus injection [27]. The advantage of using the replenishment time measure is that this value is insensitive to the effects of gain and attenuation and is easily estimated. The replenishment time measurement has been used to visualize neovascularization in rabbit liver VX2 tumors, resulting in better delineation of tortuous and meandering tumor vessels[28]. Motion-corrected replenishment time estimates have been assigned to a nonlinear color map to easily visualize changes in blood flow in the kidney medulla or cortex with therapy (Fig. 3a, b)[29, 30]. Similarly, the response of rodent tumor vasculature to anti-vascular endothelial growth factor (VEGF) and anti-Activin Like Kinase-1 (ALK-1)therapies has been differentiated by contrasting the effect of these therapies on the replenishment time [31]. Anti-VEGF therapy alters blood flow across a wide range of flow rates and is often interpreted as providing vascular normalization (Fig. 3c, d)[32]. In these studies, the replenishment time derived by ultrasound was correlated with early arrival-time maps acquired with contrast-enhanced MRI [32]. Similarly, ultrasound replenishment time estimates demonstrated that anti-ALK-1 therapy significantly retarded vascular maturation by disrupting the co-localization of endothelial and perivascular cells[31]and thus primarily affected fast replenishment times (Fig. 3e, f).

Figure 3. Applications of destruction-replenishment perfusion imaging.

The time required for contrast agent replenishment was used to quantify flow in the kidney before (a) and after (b) administration of Dopamine, where yellow indicates fast replenishment (~ 1 second) and pink indicates slow replenishment (10 seconds). As a result of dopamine, the medulla image changes from deep blue to lighter blue indicating a decrease in replenishment time (faster flow) in the kidney. Time for contrast replenishment has also been used to assess the response to anti-angiogenic therapy in rodent tumors. (c–d) Response to anti- VEGF therapy in a rat adenocarcinoma demonstrated that VEGF therapy altered vasculature with fast and slow blood flow. (c) Upper is treated, lower is control. (e–f) Response to anti-ALK-1 therapy in a chimeric mouse model demonstrated that ALK-1 therapy interferes with vascular maturation and influenced flow only in mature vessels with fast flow. (e) The upper frame is ALK-1 treated, lower is diluent treated.

2.2.1. Other blood flow estimation methods

The pharmacokinetics of contrast agent arrival can also be used to estimate local blood flow without high amplitude ultrasound pulses that fragment microbubbles. The area under the echo amplitude curve and area under the contrast agent wash-out correlate with tumor response and have been routinely used to evaluate new drugs in phase I, II and III clinical trials[33, 34]. The regeneration of the microvasculature can be detected by ultrasound imaging before the morphological changes are detected by computed tomography (CT). Cerebral perfusion deficits have been evaluated by time-to-peak and peak intensity parametric imaging during acute ischemic stroke [35, 36]. Microbubble-based imaging of blood flow complements extracranial and transcranial color-coded duplex sonography in imaging brain perfusion [35, 36]. Brain hemorrhage and glioblastoma with and without large necrotic areas demonstrated different perfusion patterns and such estimates have been shown to change with therapy [37].

3. Molecularly-targeted contrast-enhanced ultrasound

In molecularly targeted ultrasound imaging, microbubbles with covalently conjugated ligands are injected and accumulate at sites where vascular targets have been upregulated. Small molecule ligands are typically attached to a microbubble by coupling the ligand to the lipid prior to particle formation [38]. Attachment of antibodies to preformed microbubbles may be accomplished through the use of a common coupling chemistry, such avidin-biotin, maleimide-thiol or carboxylic acid-amine[1]. We focus here on microbubble contrast agents, although liposomal and perfluorocarbon emulsions have also been used for targeted imaging [39]. This section includes common imaging methods applied in molecular ultrasound imaging and applications of targeted microbubbles.

3.1 Imaging methods for molecularly-targeted ultrasound imaging

The goal of targeted ultrasound imaging is to selectively image microbubbles that are bound to a molecular target on the vascular endothelium, while suppressing signals from surrounding tissues and freely circulating microbubbles. These processes are typically segmented into two parts where differentiation of microbubble from tissue echoes occurs first, followed by separation of bound from freely circulating microbubble echoes. Differentiation of microbubbles from tissue was described in Section 2.1.

3.1.1. Differentiating bound from freely-circulating microbubbles

In order to distinguish between bound and freely circulating microbubble echoes, at least two methods have been developed that build upon the previously discussed microbubble-specific imaging techniques (Fig. 4). Selective imaging of bound microbubbles typically combines sequential data acquired over several minutes to differentiate both between microbubble and tissue echoes and bound and freely circulating microbubbles echoes.

Figure 4. Strategies for the differentiation of bound and circulating microbubbles.

In each case, microbubbles are injected and circulate for ~7 minutes. When bound agents are present the image amplitude decays more slowly. (a) Red bar indicates the timing of a high-pressure ultrasound burst to destroy the bound and circulating agents. The pre-burst and post-burst images (yellow bars) are subtracted to yield the targeted agent image. (b) Alternatively, a ten second window of images can be obtained at the time of injection and a later time point (yellow bars), averaging over the ten second interval to suppress the contribution from circulating agents. The late-average image normalized by the early-average image quantifies accumulation.

Both methods begin with an injection of targeted contrast agents, followed by a delay of several minutes to allow bubbles to circulate and bind to their target. Motion correction and inter-frame image averages can reduce artifacts caused by respiration and signals from freely circulating microbubbles [40–42]. Once the total microbubble signal reaches a plateau, the method outlined in Fig. 4a uses a high-intensity acoustic burst to destroy bound microbubbles within a local region. Assuming that circulating agents will replenish the region of interest, bound agents are selectively imaged by subtracting the post-burst images (assumed to contain freely-circulating agents) from the pre-burst images (assumed to contain bound and freely-circulating agents). Variations of this technique have been used for many applications (Table 1, “Pre-burst minus post-burst”); however, the high-intensity acoustic pulses may not be desirable for the initial clinical evaluation of targeted microbubbles [43].

Table 1.

Pre-clinical applications of targeted microbubbles

Clinical Application	Target	Ligand	Tissue Rejection	Bound vs circulating	Ref
Angiogenesis	VEGFR2	Peptide	CPS, clinical freq	Post-burst was negligible	48
	αvβ3	Knottin peptide	B-mode, high freq	Pre-burst minus post-burst	49
	αvβ3	cRGD	Pulse Inversion, clinical freq	None, video intensity only	50
	VEGFR2	scVEGF	CPS, clinical freq	Pre-burst minus post-burst	51
	αvβ3	cRGD	CPS, clinical freq	Pre-burst minus post-burst and Late- average	40
	VEGFR2	antibody vegf	B-mode, high freq	Pre-burst minus post-burst	52
	VEGFR2	anti-vegf and anti-av	B-mode, high freq	Pre-burst minus post-burst	53
	CD105, VEGFR2, VEGF	anti-CD105, anti-VEGFR2, or anti-VEGF-VEGFR complex specific	B-mode, clinical freq	Targeted minus background	44
	αvβ3, endoglin, VEGFR2	mAB-alpha(v), mAB-VEGFR2, mAB-endoglin	B-mode, high freq	Pre-burst minus post-burst	54
	alpha(v) integrins	mAB-alpha(v) and RGD-containing echistatin	Doppler, clinical freq	Pre-burst minus post-burst	55
	alpha(v)int egrins	echistatin	CPS, clinical freq	Pre-burst minus post-burst	56
Atherosclerosis	P-selectin	Sialyl Lewis A	N/A	N/A-in vitro	59
	P-Selectin and VCAM-1	mAb-MVCAM.A, a sialyl Lewisx polymer	N/A	N/A-In vitro	60
	VCAM-1	antibody (VCAM-1)	CPS, clinical freq	Pre-burst minus post-burst	61
	P-selectin	sialyl Lewis X	N/A	N/A-intravital microscopy	62
	P-selectin	antibody	N/A	N/A-In vitro	63
	P-Selectin and VCAM-1	antibody (VCAM-1or P-Selectin)	CPS, clinical freq	Pre-burst minus post-burst	64
Inflammation	Selectins	sialyl Lewis X	CPS, clinical freq	Pre-burst minus post-burst	65
	P-Selectin	antibody	CPS, clinical freq	Pre-burst minus post-burst	66
	Monocyte-derived Cells	phosphatidylserine	Pulse Inversion, clinical freq	Pre-burst minus post-burst	68
	MAdCAM-1	MECA-367antibody	Pulse Inversion, clinical freq	Subtract average background	69
	GPIIb IIIa	peptide CRGDC	B-mode image intensity	N/A	70
	ICAM-1	antibody	T1.3R3.6	Pre-burst minus post-burst	71

An alternative method that does not require a high amplitude destructive pulse to discriminate bound and circulating microbubbles is summarized in Fig. 4b. Here, a 10-second average is acquired at the peak concentration and after 7 minutes and the difference of the later and earlier averaged images is calculated [40]. The inter-frame averaging suppresses echoes from freely circulating microbubbles and with such a strategy the addition of a destructive pulse did not improve the differentiation of bound and circulating agents [40, 44]. Sophisticated processing methods, such as estimation of the spatial frequencies within the targeted image, can also be employed [40].

3.2 Applications

Pre-clinical studies with targeted microbubbles focus on noninvasive imaging of vascular cell-surface receptors with a goal of diagnosing disease or quantifying therapeutic efficacy [45, 46]. The following sections discuss recent pre-clinical studies applying microbubbles targeted to receptors that are upregulated in inflammation and angiogenesis, in applications including cancer, cardiovascular disease and atherosclerosis.

3.2.1. Molecular imaging of angiogenesis

With the advent of drugs targeted to angiogenic receptors on cancerous endothelium, noninvasive assessment of molecular activity is desirable to individualize therapy [47]. The vascular endothelium of tumors undergoing angiogenesis expresses VEGF receptors and integrins which have been successfully targeted with microbubbles conjugated with antibodies, small molecules and peptides (see Table 1, refs [40, 44, 48–56]).

With an array of ligands now available, multi-receptor quantification can be performed to longitudinally profile tumor angiogenesis and monitor therapeutic efficacy. Microbubbles have been targeted to the angiogenesis integrins α_vβ₃ and vascular endothelial growth factor receptor 2 (VEGFR2)using monoclonal antibodies or small molecular ligands (Fig. 3a–e) [40, 48–58]. Within the same imaging study, multiple microbubbles can be sequentially administered and the density of multiple receptors assayed within minutes. Receptors that are upregulated in angiogenesis (α_vβ₃, endoglin, and VEGFR2) have been repeatedly imaged with targeted microbubbles in murine models of breast, ovarian, and pancreatic cancer to profile the longitudinal expression of these receptors (Fig. 5f). Based on ultrasound studies of early stage breast and ovarian cancers, endoglin expression was found to be upregulated before α_vβ₃ integrin and VEGFR2 expression; alternatively, in early stage pancreatic cancer, the difference in expression of these receptors was not significant [54]. Microbubbles targeted to endoglin and VEGFR2 have been applied in pre-clinical studies to follow changes with gemcitabine chemotherapy of pancreatic tumors [44]. Histology and immunohistochemistry suggested that decreases in endoglin and VEGFR2 signals measured by targeted ultrasound were due to decreased vascular density and decreased expression of endoglin and VEGFR2 [44].

Figure 5. Examples of targeted agents from multiple ligand-receptor pairs obtained with CPS.

(a) CPS images of the circulating agent in a Met-1 syngeneic mouse tumor (arrow) immediately after injection and (b) the corresponding B-mode, anatomical image of the tumor region (arrow). Note abundant vasculature in the CPS image. (c–d) cRGD-conjugated [40] and scVEGF-conjugated microbubbles [51] accumulate within the Met-1 tumor to a greater extent than non-targeted microbubbles (e). (f) Mean targeted, contrast-enhanced US imaging signal intensities from longitudinal experiments in subcutaneous ovarian cancer xenografts in mice obtained by using integrin (red), endoglin (teal), and VEGFR2-targeted microbubbles (MB) (green). Absolute and relative US imaging signals were different among different tumor types and different MB types. Error bars = standard deviations.(f) reproduced with permission from [54].

3.2.2. Molecular Imaging of inflammation

Pre-clinical studies indicate that ultrasound molecular imaging can detect inflammation associated with atherosclerosis and track the progression of disease (Table 1[59–64]) [45]. The accumulation of microbubbles targeted to markers for inflammation, vascular cell adhesion molecule-1 (VCAM-1) and P-selectin, was greater in mice deficient for the LDL-receptor and the Apobec-1 editing peptide, as compared to control mice, at ages of 10, 20, and 40 weeks [64]. Evidence of atherosclerotic phenotype was detected prior to the formation of plaques by imaging microbubbles targeted to VCAM-1 and P-selectin [64].

Molecular imaging with ultrasound has been used to image inflammation that occurs after myocardial infarction (Table 1, refs [65, 66])[67]. Microbubbles conjugated to Sialyl Lewis^x can target endothelial P-selectin and increase signal in post-ischemic versus normal myocardium in rats with a surgically induced ischemia-reperfusion injury. These studies concluded that molecular ultrasound may provide a method for increasing diagnostic accuracy of cardiac ischemia [65, 66]. Ultrasound molecular imaging has also been demonstrated to noninvasively image cardiac inflammation [68], Crohn’s disease [69], thrombosis [70] and inflammation due to transplant rejection (Table 1) [71].

4. Microbubble- and ultrasound-enhanced drug delivery

Microbubbles have been used to enhance the delivery of genes, proteins and chemotherapeutics in pre-clinical studies [72–74]. Systemic co-injection of a drug-carrying particle and a microbubble, followed by local insonation, has been shown to increase local vascular permeability and allow for the extravasation of the agent [75, 76]. Alternatively, directly incorporating a drug or gene within the shell of the activatable bubble has been shown to locally deliver the drug after insonation [77–79]. Drug-solubilized oils may also be included in the shell to enhance loading [80, 81]. Paclitaxel has been solubilized within an oil coating of the lipid layer and gas core, facilitating the transfer of drugs from the microbubble shell to the endothelium during insonation[82]. Protein and polymer-shelled microbubbles have also been functionalized to carry targeting ligands and genetic payloads [83, 84]. Also, preformed liposomes or other nanoparticles can be conjugated to the membrane, thus enhancing the payload [2, 82, 85]. For example, we find that thousands of small unilamellar liposomes can be attached to a single microbubble without significantly decreasing microbubble oscillation[82]. Hybrid microbubble-liposome vehicles can easily be deflected to the vessel wall by a train of low-pressure ultrasound pulses [2] and the cargo released from the liposomes.

The advantages of microbubbles as drug or gene carriers include enhanced transport across cell membranes and vessel walls resulting from oscillation of the microbubble within tens of microns from the vessel surface (Fig. 6). Hypothesized mechanisms for this enhanced transport include fluid microstreaming as the microbubble oscillates (Fig. 6a), liquid jets directed toward or away from the endothelium (Fig. 6b), transcytosis of particles (Fig. 6c) and the creation of trans-endothelial gaps (Fig. 6d)[7, 8, 86]. The ability to use ultrasound to image, deflect and fragment microbubbles provides local control over when and where a drug is released [86]. The spatial localization of enhanced delivery effects is obtained primarily through positioning of the ultrasound focus. Disadvantages of microbubbles as carriers include the relatively low payload and short circulation half-life, which is on the order of minutes for lipid-shelled vehicles and is extended with polymer-shelled vehicles.

Figure 6. Hypothesized mechanisms for enhanced drug delivery with the insonation of microbubble contrast agents.

(a) Fluid microstreaming surrounding an oscillating bubble can enhance transport. (b) Liquid jets create small pores in cell membranes. (c) Ultrasound microbubble oscillation enhances trans-cytosis and (d) increases the width of gaps between endothelial cells.

Sub-micron particles that can be activated by ultrasound have also been described, where the smaller diameter of these particles facilitates extravasation from leaky tumor vasculature and targeting to vascular receptors. Liquid perfluorocarbon-containing nanoparticles have been applied for ultrasound and MRI[87, 88], and more recently for therapeutic delivery [89]. Polymer-coated perfluorocarbon nanoparticles can produce an effective cancer therapeutic when combined with ultrasound [90]. With a stabilizing coating, liquid perfluorocarbons with low phase transition temperatures can be stably incorporated within vehicles.

4.1 Parameters for drug delivery

Critical parameters for quantifying the effects of microbubble oscillation include ultrasound settings, the microbubble diameter and concentration and the vessel diameter[8]. Markers for ultrasound-induced vascular changes that have been used as indications that drug delivery can be enhanced are hemorrhage (extravasation of red blood cells), trans-BBB transport and the extravasation of small imaging probes. The incorporation of targeting ligands is likely to further enhance such effects.

The transmission center frequency (f_o) and the peak rarefactional pressure (PRP) are key parameters in determining the likelihood of biological effects from ultrasound. The mechanical index (MI=PRP/√f_o), displayed on all clinical commercial scanners, does not account for the presence of exogenous microbubble agents[91]. Yet, the threshold for cavitation is much lower (MI=0.4) for petechial hemorrhage in gas-containing tissues (eg. lungs) as compared to tissues that do not contain gas bodies[92]. The addition of exogenous agents similarly reduces threshold for bioeffects, and several recent papers have proposed the ratio of acoustic pressure and center frequency (PRP/f_o) as an index that is predictive of the biological effects of microbubble contrast agents [43, 93], where the likelihood of a direct biological effect increases with an increase in this ratio. When ultrasound was applied to the rat kidney with a value of PRP/f_o above 0.5 in the presence of exogenous agents, hemorrhage was detected [43]. The extravasation of fluorescein isothiocyanate (FITC)-labeled dextran from the vasculature increased with PRP/f_o following insonation of microbubbles in the chorioallantoic membrane model [86].

The pulse duration and time between ultrasound pulses are also important factors in the efficiency of drug delivery. Microbubbles repeatedly fragment and re-combine during a long ultrasound pulse and the remaining gas can diffuse over long intervals between pulses. For example, the pressure threshold for blood brain barrier (BBB) disruption decreased from 0.7 to 0.4 MPa when pulse length increased from 0.1 to 10 msec [94]. Similarly, the pressure threshold for disruption of a gel boundary decreased from 2.5 MPa to 1.2 MPa with the pulse duration increasing from 10 μsec to 10 msec [95]. However, when microbubbles are destroyed within the vasculature, sufficient time must be allowed between pulses for the microbubbles to re-enter the vasculature, and therefore a minimum time between pulses must be maintained. In explorations of the effect of the time between pulses, renal hemorrhage decreased with 33 ms between frames, as compared with a 1 second between frames (for a transmitted center frequency of 1.8 MHz, MI of 1.6) [96]. Ten seconds of continuous insonation yielded a similar number of petechial hemorrhages as a 100 sec continuous insonation and hemorrhage increased when the time between ultrasound pulses increased [97].

Increased microbubble diameter has also been associated with an increased propensity for the microbubbles to interact with the vessel wall [7, 98–100]. Mechanisms for this increase include the extended lifetime of larger microbubbles. In ex vivo rat cecum, bubbles larger than 4 μm were observed to oscillate for 176±139 pulses on average, compared to 31±14 pulses for bubbles smaller than 4 μm (with a transmitted center frequency of 1 MHz, PRP of 800 kPa and 10 cycle pulses)[7]. With this extended lifetime, the large microbubbles can be deflected to the wall by ultrasound radiation force[101, 102]. In addition, assuming the same distance to the wall, the likelihood of a liquid jet increases with a larger microbubble.

Both biological effects and the delivery of a drug from a microbubble increase with increasing dose, particularly since the number of bubbles near boundaries increases with increased concentration. For a dosage above 1 mL/kg, petechial hemorrhages were reported to increase in proportion to Optison dose in mouse intestine and abdominal muscle [97]. In an ex vivo rat cecum, the presence of a high concentration of microbubbles (>20 microbubbles per (200-μm)² field) resulted in microbubble fusion, with the larger, fused microbubbles interacting with the vessel wall [103].

4.2 Drug delivery applications

Applications of microbubble-enhanced drug delivery are expanding rapidly and include therapeutic delivery to solid tumors[104–107], delivery across the BBB[108]and enhanced transfection efficiency[104, 109, 110] (Table 2). Perhaps the fastest developing application has been the use of ultrasound-microbubble techniques to treat cancer. Recently-published cancer drug delivery strategies include the local delivery of doxorubicin[105], cis-diamminedichloroplatinum (II) (CDDP) [107]and paclitaxel[106]. For example, [105] reported the delivery of ~9 % of the injected dose of doxorubicin per cubic centimeter to tumors in a rat model with significant therapeutic improvement with the addition of ultrasound. Other reported strategies include the delivery of herpes simplex virus thymidine kinase (TK) and ganciclovir[104], and the delivery of exogenous antigens into the cytosol of dendritic cells[111].

Table 2.

Examples of recent papers demonstrating the applications in drug delivery

Application	Therapeutic cargo or goal	Reference
Cancer	Chemotherapeutics, gene delivery, immunotherapy	[104–107, 111]
Renal	Modify the size selectivity of the filtration capacity of the kidney	[108]
Brain	Delivery across BBB	[94, 112]
Diabetes/pancreas	Transfect pancreatic β-cells	[116]
Cardiovascular disease	Enhance gene transfection and recruit progenitor cells after myocardial infarction	[110, 117]
Peripheral vascular disease/thrombus	Enhance gene transfection	[113, 114]
Regenerative medicine	Nonviral delivery method of siRNA into MSC	[109]
Surgical guidance	Noninvasively deposit markers under image guidance to help surgeons identify targets for resection	[115]

Applications of ultrasound to enhance drug pharmacokinetics and distribution include the treatment of kidney disease [108]and the delivery of an anti-A beta antibody to the brain [112]. Enhancing delivery across tight endothelial junctions, such as those present within the blood-brain barrier, is one of the most important applications of ultrasound-enhanced delivery, since it represents one of few feasible approaches[94, 112]. Applications of gene therapy with microbubble-enhanced transfection include the treatment of diabetes, cardiovascular disease, peripheral vascular disease and regenerative medicine[104, 109, 110, 113, 114]. Finally, microbubble based delivery has been proposed as method to deposit surgical tattoos and thus enhance the efficacy of surgical resection[115].

5.0 Summary

There are many exciting opportunities to leverage ultrasound contrast agents and molecular imaging in the development of new therapeutic strategies. Those emphasized here include the use of ligand-conjugated microbubbles to detect vascular receptors and thus individualize therapy, the use of ultrasound flow mapping to characterize the anti-vascular effect of various therapies and the use of microbubbles to enhance drug and gene delivery.

At this time, a targeted microbubble agent from Bracco (BR55) is reported to be approaching clinical evaluation and strategies for imaging targeted agents are available on commercial ultrasound scanners[118]. Preclinically-formulated targeted microbubbles are widely available and the imaging of such agents to monitor therapy is widespread. Methods to quantify accumulation are advancing, facilitating longitudinal studies. As described here, such agents can be insonified at a low acoustic pressure and high ultrasound frequency and under such conditions the safety profile is optimized.

The use of ultrasound with microbubbles to enhance drug delivery is now frequently reported in pre-clinical research; however, planning for clinical studies is still underway. Optimization of ultrasound-enhanced drug or gene delivery vehicles is challenging due to wide range of possible ultrasound parameters and the range of potential materials, targeting ligands, and cargos. The use of cross-modality validation methods has increased with nuclear medicine, MRI and optical imaging probes used to image the delivery vehicle or the model drug [119–122]. PET provides real-time and highly sensitive full-body pharmacokinetics [113, 122, 123]. MRI can be used to visualize the biodistribution of drugs or vehicles [120] and the activation of delivery vehicles [119]. Optical imaging can validate the accumulation of targeted microbubbles or drugs [40, 76]. Validation tools are important for the optimization of and the translation of microbubble-based techniques to the clinic.

The optimization of methods to quantify and characterize drug accumulation is particularly critical and timely. In particular, high pressure and low frequency insonation of microbubbles within the vasculature has been shown to alter local blood flow and angiogenesis. Thus, drug accumulation within the altered vasculature may result but may not fully treat the surrounding tumor. Thus, measurement of accumulation within and outside the vasculature during and after insonation will be an important component of system optimization.