Recent Advances in Molecular, Multimodal and Theranostic Ultrasound Imaging

Abstract

Ultrasound (US) imaging is an exquisite tool for the non-invasive and real-time diagnosis of many different diseases. In this context, US contrast agents can improve lesion delineation, characterization and therapy response evaluation. US contrast agents are usually micrometer-sized gas bubbles, stabilized with soft or hard shells. By conjugating antibodies to the microbubble (MB) surface, and by incorporating diagnostic agents, drugs or nucleic acids into or onto the MB shell, molecular, multimodal and theranostic MB can be generated. We here summarize recent advances in molecular, multimodal and theranostic US imaging, and introduce concepts how such advanced MB can be generated, applied and imaged. Examples are given for their use to image and treat oncological, cardiovascular and neurological diseases. Furthermore, we discuss for which therapeutic entities incorporation into (or conjugation to) MB is meaningful, and how US-mediated MB destruction can increase their extravasation, penetration, internalization and efficacy.

1. Introduction

1.1 Current indications for using ultrasound imaging

Due to its non-invasive nature, low cost, broad diagnostic applicability and easy handling, ultrasound (US) imaging is the second-most used imaging modality in clinical practice after conventional x-ray radiography [1]. It is used by medical doctors from various different disciplines, including radiologists, gynecologists, cardiologists, gastroenterologists, surgeons and many more as an initial screening tool, as well as for fast-look follow-up examinations. Its ability to visualize blood flow, blood velocity and blood vessels by Power and Color Doppler further recommends US imaging for vascular diagnosis, e.g. for measuring the degree of stenosis in carotid arteries [2], and for looking at the perfusion of tumors [3] and organs after transplantation [4].

Besides these diagnostic applications, High-Intensity Focused US (HIFU) has been attracting ever more attention as a valuable therapeutic option to destroy ureteric stones [5], and to ablate benign uterus myomas and other benign and malignant tumors [6]. In this context, the acoustic energy focused to one defined spot is moved over the pathological tissue. Due to absorption of the acoustic energy and the resulting local temperature increase, the pathological tissue is destroyed. Recently, the first commercial HIFU-systems that can be used inside clinical MR scanners have been introduced which enable highly personalized and well-controlled tissue ablation by getting anatomical information about the pathology and the local temperature rise from MR imaging.

However, the diagnostic and therapeutic potential of US imaging has not yet been fully explored and translated to clinic. In this regard, US contrast agents, which are gas-filled microbubbles (MB) stabilized by a shell made of lipids, proteins or polymers can enormously improve US imaging. In particular, the use of MB significantly expands the diagnostic potential of US for characterizing pathologies based on functional and molecular vascular characteristics. Furthermore, the use of MB-based contrast agents in US imaging offers possibilities for image-guided (theranostic) interventions. In the present manuscript, recent developments in this emerging and interdisciplinary field are summarized and discussed.

1.2 Impact of contrast-enhanced US imaging on routine clinical practice

US contrast agents in combination with contrast agent-specific US imaging techniques are increasingly accepted in routine clinical practice for diagnostic imaging of several organs and pathologies. Particular interest is given to examinations of the liver, because of the significant improvement over conventional US in both, the detection and characterization of focal liver lesions. Recent studies even show that the diagnostic performance of contrast-enhanced ultrasound (CEUS) can reach that of contrast-enhanced computed tomography (CT) and magnetic resonance imaging (MRI) (figure 1) [7-9]. The high diagnostic accuracy of CEUS in liver imaging is based on two characteristics:

1. the detection and early enhancement of a malignant liver lesion during the arterial phase

2. the rapid wash-out of the contrast agent in malignant liver lesions.

A further benefit of US contrast agents in the clinical routine is their good safety profile, which enables the administration of contrast agents to patients who have contra-indications for contrast-enhanced CT or MRI (e.g. patients with severe renal dysfunction). As a consequence, focal liver diseases have evolved into the single most important application of CEUS. The recommendations for CEUS for liver imaging are summarized in the guidelines for good clinical practice of the EFSUMB [10].

Figure 1.

Examples for the use of CEUS in clinical liver imaging in comparison with MRI. A+B: In B-mode US and in T₂-weighted MRI, a benign liver tumor (fibronodular hyperplasia, FNH) can be delineated. C: Twenty minutes after the administration of the hepatocyte-specific MR contrast agent Gd-EOB-DTPA, the lesion and the surrounding tissue provide comparable contrast enhancement, which is typical for a FNH. D-F: In contrast-specific US mode, one can depict the rapid centrifugal (“spoke-wheel”) enhancement of the lesions. In the late phase (G; i.e. 180 s after injection), the lesion provides no wash-out pattern. Knowledge on the microbubble (MB) kinetics in the late phase enables the exclusion a malignant tumor. The enhancement pattern in the arterial phase further supports the diagnosis of a FNH.

A second major clinical application is contrast echocardiography, where MB are used for left ventricular opacification and endocardial border delineation. The superior anatomical delineation of the cardiac boarders leads to specific clinical scenarios in which US contrast agents could / should be used, including the assessment of left ventricular systolic function, elevation of the left ventricular apex, mechanical complications of myocardial infarction and the characterization of intracardiac masses. The consensus statement on the use of ultrasound contrast agents was published in 2008 by the American Society of Echocardiography [11] and a summary of the clinical impact of the guidelines was provided two years afterwards [12].

In neurology and intensive care medicine, contrast-enhanced transcranial Doppler ultrasound has been established as a reliable tool to evaluate the cerebral circulation, e.g. to outline vessel stenosis and occlusion as well as ultimately, to diagnose brain death [13, 14]. Besides liver, cardiac and brain imaging, indications for CEUS have expanded to applications in the kidney [15, 16], in vesico-ureteric reflux [17, 18], in the pancreas [18-21], in trauma patients [22] and in cerebral circulation, as well as in oncological studies [23, 24]. In this context, an emerging clinical field might represent the assessment of novel targeted drugs such as anti-angiogenic therapies. Here, contrast enhanced ultrasound enables the early indentification of responders to an antiangiogenic treatment for gastrointestinal stromal tumors, renal cell carcinoma, and hepatocellular carcinoma [25, 26]. Despite this promising data derived in clinical trials, however, a broad application into the daily clinical routine has not yet been established.

In this context, the continuous revision of the existing “Guidelines for Good Clinical Practice” in consensus meetings of the US societies and the continuous medical training of “CEUS-examiners” in specialized CEUS courses are - besides advancements in US machines and contrast agents - the most important preconditions for maintaining and increasing the clinical success of CEUS.

2. Molecular US imaging

An important precondition for in vivo molecular imaging is the use of contrast agents which can be detected with high sensitivity and specificity. MB applied for US imaging fulfill these demands. In principle, even a single MB can be detected and there are imaging techniques that detect MB selectively (see 2.2). Due to their size, which typically ranges from 1-5 μm, MB do not extravasate from the vasculature. This is both, an advantage and a disadvantage. On the one hand, no unspecific accumulation in the interstitial space is observed, leading to a low unspecific background signal when performing molecular US imaging. On the other hand, since there is no extravasation, only intravascular targets can be addressed, which significantly narrows the diagnostic options. Nevertheless, there are many intravascular targets suitable to characterize angiogenesis and inflammation, which can significantly help to better diagnose diseases and monitor therapy responses. The following sub-sections elaborate on preferentially used targets for molecular US imaging, on the design of MB for molecular imaging purposes and on MB-specific US imaging techniques.

2.1 Targets and Contrast agents

2.1.1 Targets

Common targets for molecular US imaging are surface receptor molecules expressed on the luminal side of activated endothelium, either in response to inflammatory or to angiogenic stimuli. Inflammation is accompanied by the recruitment and transmigration of leukocytes through the endothelium to the site of inflammation. This process includes the successive interaction of adhesion molecules on activated endothelial cells with leukocytes. While endothelial- and platelet-specific E- and P-selectins promote the initial attachment and rolling of circulating leukocytes, the adhesion molecules ICAM-1 (intercellular adhesion molecule 1) and VCAM-1 (vascular cell adhesion molecule 1) mediate the firmer adhesion of leukocytes to the endothelium, in order to allow for transmigration. The up-regulation of these adhesion molecules is induced by inflammatory cytokines such as TNF-α (tumor necrosis factor-α) and interleukins (e.g. IL-1). Furthermore, since they are expressed on the luminal side of blood vessels, these adhesion molecules represent prominent targets for molecular US imaging of inflammation-associated processes (as described in section 2.3 “Applications”).

Besides inflammation, also angiogenesis is an important process involving changes in the expression pattern and the behavior of endothelial cells. Angiogenesis is triggered in hypoxic tumor areas via the over-expression of angiogenic factors like vascular endothelial growth factor (VEGF) or fibroblast growth factor (FGF), and it is a crucial pre-requisite for tumor growth beyond a size of a few cubic millimeters [27, 28]. The most prominent angiogenesis-related targets in molecular US imaging are VEGFR2 receptors and α_vβ₃ integrins. VEGFR2-mediated signaling stimulates endothelial cell proliferation and angiogenesis, and enhances the permeability of blood vessels, thus playing a crucial role in developmental and pathological angiogenesis, e.g. in tumor angiogenesis [29]. VEGF and VEGF-receptors are known to be over-expressed in malignant tumors and their enhanced expression generally correlates with poor clinical outcome [30]. Due to its strong up-regulation during tumor angiogenesis and its almost absence on quiescent endothelial cells, VEGFR2 represents a highly attractive target for molecular US imaging of tumor angiogenesis.

α_vβ₃ integrins belong to the integrin family of heterodimeric surface glycoproteins, mediating cell adhesion to components of the extracellular matrix. In addition to their adhesive functions, integrins are involved in various signaling pathways and thus influence the proliferation and survival of cells, including endothelial cells. Like VEGFR2, the expression of α_vβ₃ integrins is elevated on activated endothelium during angiogenesis, while it is only weakly expressed by quiescent endothelial cells [31, 32]. α_vβ₃ integrins recognize different extracellular matrix components (e.g. vitronectin, fibronectin, fibrinogen) via their RGD-binding-site and thus mediate the adhesion of endothelial cells to the extracellular matrix. Blockade of α_vβ₃-mediated cell–matrix interactions between endothelial cells and the extracellular matrix induces apoptosis of endothelial cells [32]. In addition, direct interactions between activated α_vβ₃ integrins and VEGFR2, as well as intracellular signaling cross-talk, have been reported to crucially regulate the VEGF-induced angiogenic response in endothelial cells [33].

Another surface glycoprotein used for molecular US imaging is endoglin, a co-receptor of the transforming growth factor (TGF) -beta receptor. Endoglin is upregulated during inflammation and angiogenesis, and plays an important role in vascular remodelling, homeostasis and angiogenesis [34]. Endoglin expression is considered as a marker for cancer aggressiveness, as a negative correlation was found between the amount of endoglin-expressing blood vessels and the overall survival and metastasis in patients suffering from different types of solid tumors [34].

Apart from the well established inflammation and angiogenesis related markers discussed above, the feasibility of several other targets for CEUS have been tested preclinically. Amongst others, specific approaches include the targeting of prostate specific membrane antigen (PSMA) and thymocyte differentiation antigen-1 for pancreatic cancer imaging [35, 36] and glycoprotein IIb/IIIa for imaging of inflammatory thrombosis [37].

2.1.2 US contrast agents

US contrast agents are gas-filled MB with diameters between 1 and 5 μm. Due to their size, MB have optimal acoustic responses in the MHz range used for US imaging. MB also possess the capability of going through small blood capillaries in the body, but stay strictly intravascular. In order to increase the circulation time and reduce the risk side effect that can arise from coalesced gas bubbles, MB are stabilized by a shell usually made of lipids [38], proteins [39], polymers [40, 41] or a mixture of these [42, 43]. Apart from the shell, the type of gas in the MB plays an important role in the stability and consequently the circulation time of the constructs. In this context, low solubility gases (such as perfluorochemicals) have been shown to substantially increase the stability and circulation times of MB in vivo [44]. In addition to its stabilizing function, the shell also determines the extent to which the MB can oscillate during insonication [45]. In this regard, one can distinguish between soft- and hard-shelled MB, with the former being more flexible and thus more suitable for harmonic (non-linear) imaging, while the latter are more suitable for destructive US imaging procedures, such as Power Doppler US.

US molecular imaging requires the use of target-specific MB that can selectively bind to intravascular targets. This is achieved by attaching specific ligands, mostly antibodies or peptides, to the MB surfaces, which enable MB binding to the respective molecular markers. Such ligand-decorated MB can be produced either by incorporating the specific ligands during MB synthesis, or by attaching the ligands to preformed MB. Details on the production of such target-specific MB are beyond the scope of this manuscript, and have been extensively reviewed by us and other groups [46-48].

2.2 Measurement techniques

In general, the echogenicity of MB is strong enough for their detection in fundamental B-mode imaging. For a better differentiation of MB from tissue echoes, i.e. for a high contrast-to-tissue-ratio, non-linear imaging is applied. This technique was originally developed for non-contrast-enhanced tissue US imaging, resulting in a greater lateral resolution and decreased acoustic noise. Thus, images are clearer, with higher contrast and they show more details [49]. In CEUS, the basis for harmonic or non-linear imaging, is the non-linear oscillation of MB at higher wave amplitudes, i.e. with a low to mid-high mechanical index (MI) [49]. Thereby, the waves backscattered by a MB also consist of higher frequency components compared to the center frequency (f_c) and are called harmonics (e.g. 2f_c, 3f_c). There are several techniques to detect the harmonic frequency components of MB. Among these, Pulse/Phase Inversion (PI), Amplitude Modulation (AM, also: Power Modulation) and Contrast Pulse Sequencing (CPS) are the most common ones. In PI, two pulses are emitted, with the second pulse 180 degrees phase-shifted to the first one [50]. For non-linear scatterers like MB, the summation of the response from the two pulses gives a sum signal, whereas for linear scatterers, like tissue, the responses will be zero (or very close to zero, when summed). In AM, as suggested by the name, the amplitude is modulated rather than the phase, resulting in a similar cancellation of the linear responses [51]. This is also valid during CPS imaging, where three pulses are emitted with the first and third pulse half of the inverse of the second pulse [52]. These techniques rely on the steady oscillation of MB (stable cavitation) and can therefore be classified as non-destructive imaging techniques.

For molecular US imaging, in most cases, the detection is performed in only one slice (2D) and the late-phase enhancement after the clearance of freely circulating MB from the blood, approximately 5-10 min after the i.v. injection of MB, is detected. Sometimes, some freely circulating MB might be left even after 10 min. Thus, a more accurate method to detect bound MB is the destruction-reperfusion technique, which was originally proposed by Wei et al. for the quantification of tissue perfusion [53], and which is nowadays routinely applied for molecular US imaging [54, 55]. Here, 5-10 min after i.v. injection, a set of images of the region of interest is recorded, followed by a high MI pulse, which leads to the disintegration of MB. Immediately afterwards, a second set of images is recorded to detect potentially freely circulating MB. Subtraction of the mean signal intensity (SI) of the second set of images from the mean SI of the first set of images produces the SI of the bound MB.

A further destructive technique is based on imaging the destruction events of stationary, target-bound MB, where the disintegration is induced by Doppler pulses [56]. The high amplitude of the Doppler pulse leads to the destruction of MB, causing an emission of a strong broad-band signal. This signal, referred to as stimulated acoustic emission (SAE), is misinterpreted as movement (loss of correlation LOC) and thus registered as a pseudo Doppler shift signal, which can be quantified. There is a linear correlation of the Doppler signal with the concentration of MB [56]. However, above a certain concentration, the signal is saturated and quantification is not possible anymore. Reinhardt et al. developed a method for the quantification of higher MB concentrations, called sensitive particle acoustic quantification (SPAQ) [57]. This is a 3D technique in which the transducer is being moved stepwise in predefined increments of 50-150 μm over the imaged region. In the small, non-overlapping insonified regions only few MB are destroyed and can thus be quantified (Figure 2A). This technique has already been successfully applied in several studies [55, 58-60]. A major advantage of this technique is the quantification of bound MB throughout a complete volume of interest, e.g. a whole tumor, encompassing possible heterogeneities of molecular marker expression within the whole volume of interest (Figure 2B). Nevertheless, due to the difference in acoustic signals from MB with different sizes, weak acoustic signals from small (< 1 micron) MB could be suppressed thereby leading to mistakes in the quantification. A combination of molecular US imaging with volumetric scanners, for instance a volumetric breast scanner, is thinkable and would provide more operator-independent, accurate and reproducible information as compared to conventional 2D US [55].

Figure 2.

3D molecular ultrasound imaging with SPAQ, showing the principle of SPAQ imaging (A). Stepwise movement of the US-transducer leads to MB destruction in the overlapping slice. Smaller step sizes yield less saturated and therefore better quantifiable images. (B) Shows a 3D-reconstructed SPAQ image of VEGFR2-targeted MB binding in an experimental breast cancer xenograft. White arrows outline the tumor. Yellow arrows show artifacts due to breathing of the mouse. The MB-destruction events are displayed as red dots (see e.g. blue arrowheads). The higher VEGFR2 expression at the angiogenic tumor margin compared with the tumor center is clearly demonstrated.

Molecular US imaging is constantly improving, for example with the application of acoustic radiation forces (ARF) to increase MB binding. Acoustically manipulating MB was already introduced by Fowlkes et al. in the early 1990s, suggesting it as means for improved image contrast or for localized drug delivery [61]. Zhao et al. later applied ARF and demonstrated a >25-fold increase in the binding of α_vβ₃ integrin targeted MB to human umbilical vein endothelial cells (HUVEC) in vitro [62]. Subsequently, Rychakk and colleagues and later Frinking and colleagues demonstrated enhanced binding of targeted MB in vivo upon ARF application in experimental models of inflammation and cancer. As compared to normal vessels, they reported a 20-fold increase in binding of P-selectin specific MB to an inflamed femoral artery, as well as an enhanced binding of VEGFR2-targeted MB (BR55) in the vasculature of experimental prostate cancers in rats respectively [63][64]. All of the abovementioned methods have been applied in many preclinical studies and show the potential of applying molecular US imaging for improving diagnosis, disease staging and localized image-guided drug delivery (see below; sections 2.3 and 4.1). Moreover, the first step toward molecular US imaging in clinical settings has already been taken; in a phase 0 clinical trial, BR55 MB has been used to identify regions of VEGFR2 expression in human prostate cancer [65].

2.3 Applications of CEUS for molecular imaging and drug delivery

Molecular US imaging of E-selectin, P-selectin, ICAM-1 or VCAM-1 has been successfully applied for imaging alterations in the endothelium that occur during the process of acute and chronic inflammation. Recently, ischemic myocardium was successfully identified by MB against P- and E- selectin due to their enhanced accumulation [66, 67]. In a mouse model of atherosclerosis, VCAM-1-specific MB specifically bound to the inflamed endothelium at the site of the aortic plaque, and MB attachment correlated very well with disease stage [66]. Similarly, Masseau and colleagues showed that CEUS and VCAM-1-taregted MB can also be applied to monitor vascular inflammation in pigs [68]. VCAM-1- as well as P-selectin-specific MB furthermore showed a high sensitivity for the early detection of the atherogenic phenotype even before obstructive atherosclerotic lesions appeared [69]. Direct thrombus detection was achieved by MB targeting to glycoprotein IIb/IIIa receptors on clotted platelets [70]. Targeting activated glycoprotein IIb/IIIa using a single-chain antibody on the surface of phospholipid MB also enabled to track the reduction of thrombus size during antithrombotic urokinase therapy in carotid arteries of mice, which were exposed to a ferric chloride injury [71]. Furthermore, novel MB functionalized with a recombinant P-selectin glycoprotein ligand-1 (PSGL-1) analogue, which binds to both E- and P-selectin, showed an even stronger binding to the inflamed endothelium after intramuscular injection of endotoxin compared to antibody-coupled or sialyl Lewis X-containing MB [72].

Even more frequently than for imaging inflammation, molecular US imaging is employed for monitoring tumor angiogenesis, including for the assessment of various therapeutic drugs on the vasculature [73-75]. For a broader characterization of angiogenesis, more than one angiogenic marker has been addressed in different studies. Using MB against endoglin, α_vβ₃ integrin and VEGFR2, the varying marker expression during angiogenesis could be longitudinally recorded in breast, ovarian and pancreatic cancer xenografts [76]. MB against the VEGF/VEGFR-complex, VEGFR2 and endoglin were used for monitoring angiogenesis, as well as the effects of anti-angiogenic treatment or chemotherapy in a mouse model of pancreatic carcinoma. In this context, the binding of the targeted MB significantly decreased upon therapy, correlating with the vascularity of the tumors and with the expression of surface markers [77]. Similarly, a significantly lower accumulation of VEGFR2- and α_vβ₃ integrin-targeted MB was recorded in human squamous cell carcinoma xenografts upon inhibition of matrix-metalloproteinases [60]. Immunohistochemical analyses revealed that the general decrease in vascularization was responsible for the lower MB binding, rather than the decline in VEGFR2 or α_vβ₃ integrin expression on the endothelium, strongly suggesting the additional analysis of functional parameters such as the relative blood volume when evaluating molecular marker expression by US [60]. Furthermore, simultaneous imaging of two markers using dual-targeted MB against VEGFR2 and α_vβ₃ integrin, showed a stronger retention in the tumor endothelium of ovarian cancer xenografts than single-targeted MB [78].

In most of the abovementioned examples of targeting the angiogenesis marker VEGFR2, the ligands were bound to MB via (strept)avidin-biotin coupling. However, since (strept)avidin-biotin coated MB are not recommended for clinical use due to their potential immunogenicity, MB functionalized by covalently integrated binding epitopes are favoured for clinical applications. The first clinically evaluated molecular MB type is BR55, where a heterodimeric peptide against VEGFR2 has been directly integrated in the phospholipid shell. BR55 showed a strong accumulation in the tumor vasculature of breast [79] and prostate cancer xenografts [80], and could sensitively record the decrease in angiogenic activity during therapy using an anti-VEGF antibody in a colon cancer xenograft model [81]. Furthermore, BR55 showed a high sensitivity for discriminating the angiogenic activity of two differentially aggressive breast carcinomas in mice, as well as for assessing the angiogenic activity in evolving breast cancers of very small sizes (Figure 3) [82]. In addition, molecular US with BR55 was highly sensitive and specific in differentiating benign from malignant breast lesions in a transgenic mouse model of mammary carcinoma, and enabled the detection of ductal carcinomas in situ and invasive breast cancers with high accuracy [83]. An interventional (phase 0) clinical trial has been performed in patients with respect to the potential of BR55 for identifying VEGFR2 positive areas in prostate cancer lesions [65]. There are a number of other US contrast agents that are potentially suited for use in humans. For example, in a more experimental stage, is another lipid (DSPC, Palmitic acid and DSPE-PEG₂₀₀₀)-coated MB, targeted to E- / P-selectin by thiol bonding of PSGL-1 on the MB surface. These MB have been used in mice with inflammatory bowel disease, and for quantifying the level of inflammation as well as for monitoring the response to anti-inflammatory treatments [84].Similarly, we recently developed potentially clinically translatable E-selectin-specific poly(n-butyl cyanoacrylate) (PBCA) MB by covalent (amine) bonding of a short E-selectin-specific peptide with the recognition sequence IELLQAR to the MB surfaces. Significant binding of these MB was shown in vitro on HUVEC stimulated with TNF-α, as well as in vivo using human ovarian carcinoma bearing mice [85].

Figure 3.

BR55 highly sensitively depicts very early breast cancer lesions. Representative SPAQ images of one slice in a 4 mm³ (A) and 34 mm³ (B) MCF-7 tumor (tumor marked by yellow arrowheads; white arrows show representative signals of destructed BR55 MB (red overlay)). Quantitative analysis of 3D SPAQ imaging demonstrates the highest binding of BR55 in 4 mm³ small tumours and a significantly reduced binding in larger tumors (C), whereas the relative blood volume (rBV) is constant (E). US data were confirmed by immunohistochemistry (D, F). *p<0.05; **p<0.01 (adapted with permission from [82])

Although the above mentioned MB formulations with a direct ligand conjugation provide a valuable platform for clinical translation, the actual clinical implementation of such MB formulations still requires a rigid, time consuming and cost intensive clinical trials process.

3. Multimodal US Imaging

3.1 Multimodal US contrast agents

Multimodal US contrast agents are particularly useful in (whole-body) biodistribution and histological validation studies. In this context, they enable the non-invasive and quantitative imaging of the fate of MB and of their shell fragments after systemic application. The ability to image drugs released from MB in vivo and ex vivo, to investigate the coverage of MB surfaces with targeting ligands, to characterize the binding of targeted MB to cells, and to image the opening of biological barriers (such as cellular membranes and the blood brain barrier (BBB)) after US-mediated MB destruction are further application fields of multimodal US contrast agents, which will be elaborated upon in this section.

In principle, multimodal US contrast agents make use of the entrapment, attachment or adsorption of other imaging agents (mostly nanoparticles (NP), radiotracers and small molecules, such as fluorescent dyes for optical imaging) in or on the shell of MB. Because the shell presents the only means of combining MB with other imaging agents, the shell properties (thickness, charge etc) are directly related to the amount of agents that can be entrapped. In this context, polymeric shells (~50 – 500 nm thick) can entrap more imaging agents compared to protein (~15 – 150 nm thick) and lipid shells (~3 nm thick) [42, 86]. So far, the synthesis of multimodal US contrast agents has involved the addition of imaging agents for different modalities during MB synthesis by mechanical agitation [87, 88], as well as by the use of microfluidic devices [89, 90]. Alternatively, imaging agents can also be coupled by chemical bonding or by passive entrapment on preformed MB [91]. To date, US has been combined with several other imaging modalities, as exemplarily discussed below.

3.2 US – Magnetic resonance imaging

Magnetic resonance imaging (MRI) provides whole body images of tissues based on the change in relaxation of the constituent water protons under the influence of an external magnetic field. Apart from its ability to make whole body images, MRI also provides excellent soft-tissue contrast, making a combination of MRI with US highly advantageous. Contrast agents for US and MRI dual-modality imaging feature MB loaded with magnetic species. For example, Liu et al. developed and characterized PBCA-based polymeric MB containing ultrasmall superparamagnetic iron oxide (USPIO) nanoparticles (NP) by a one-pot polymerization reaction, and subsequently showed both in vitro and in vivo that such hybrid MB are suitable contrast agents for both US and MRI (Figure 4 A-C) [87]. Similarly, USPIO-containing MB have been synthesized by layer-by-layer deposition as well as by a double emulsion polymerization process [91, 92]. Interestingly, UPSIO-containing MB were observed to have a stronger non-linear response under US treatment compared to standard MB. This phenomenon could be attributed to the increased resistance to compression of NP-loaded MB compared to their expansion, which leads to non-linear oscillations of the NP-loaded MB [90].

Figure 4.

Multimodal US contrast agents. USPIO-loaded PBCA MB as contrast agents for MRI / US showing TEM images of PBCA MB with increasing USPIO (a–e) concentrations (A), phantom MR imaging showing signal enhancement in MRI, which increases with MB destruction (B), and signal enhancement observed in vivo in both US and MRI upon the i.v. injection of USPIO-loaded MB in tumor bearing mice (C; adapted with permission from [87]). D-F: Rhodamine-b loaded PBCA MB for use in optical and US imaging. Comparison between fluorescent and non-fluorescent MB for the evaluation of targeted-MB binding to cells in vitro (D). Two-photon microscopic validation and evaluation of ICAM-1 targeted MB binding to activated (upper panel) vs non-activated (bottom panel) HUVEC (E). Panel F shows the application of fluorescent MB for validation of in vivo molecular US studies. I-III represents sections of the tumor during molecular imaging, while IV shows the evaluation of bound MB by destruction replenishment analysis. By subsequent fluorescence microscopy of tumor cryosections, the attachment of fluorescent MB (red) to FITC-lectin-stained (green) tumor vasculature (V) could be validated (adapted with permission from [101]). Biodistribution analysis of ¹¹¹In-labelled PBCA-MB over time by gamma counting (G–H) and US (I). Both modalities showed a high amount of MB accumulated in the liver compared to kidney and tumor. Unlike gamma counting, where both signals from MB and shell fragments are registered and quantified over time, upon SPAQ imaging, all MB are destroyed and cannot be monitored longitudinally by US. Radioactively labeled MB therefore provide a more reliable means for quantitatively monitoring the biodistribution of MB and their shell fragment in vivo over time as compared to US (adapted with permission from [107]).

While the studies discussed thus far have made use of the thick shell of polymeric MB for loading NP for MRI, some studies suggested lipid-shelled MB for the same purpose [93, 94]. For instance, Fan et al. created US-MRI contrast agents by using superparamagnetic iron oxide (SPIO) NP modified with a four-carbon-atom-long aliphatic terminal end, thus enabling hydrophobic interactions between the SPIO NP and the phospholipids composing the MB shell. They subsequently showed their constructs provide good contrast enhancement in both US and MR imaging. The deposition of magnetic NP and doxorubicin (encapsulated by electrostatic interaction within the MB shell) in brain tissue after focused US treatment was a further feature of their study, highlighting the ability to open up and deliver drugs across the BBB with such a probe [94].

Another concept for dual-modality MB was proposed by Feshitan et al., who produced Gd³⁺-DOTA carrying MB by post-modification of pre-formed lipid-shelled MB [95]. In their study, lipid (90 % DSPE, 10 % DSPE-PEG₂₀₀₀) -shelled MB were synthesized by sonication, followed by a covalent coupling of DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) to the MB shell surface. Subsequently, the complexation of Gd³⁺ ions to the DOTA on the MB surface was performed. A loading of 7.5 × 10⁵ Gd³⁺ ions per μm² of the MB surface could be achieved. Interestingly, MR signal enhancement was only observed after MB destruction. The authors attribute this to the limited access of bulk-water to lipid head-groups containing Gd³⁺ for the intact MB compared to the lipid fragments [95, 96].

3.3 Photoacoustic (PA) – US Imaging

Photoacoustic (PA) imaging is based on the excitation of tissue chromophores using a short-pulsed laser beam, which leads to thermoelastic expansion of the tissue and thus wideband ultrasonic emission. The emitted ultrasonic waves are then captured by an ultrasonic transducer and utilized for image reconstruction. This thereby bestows PA imaging with the molecular sensitivity of optical imaging and the spatial resolution of US imaging [97]. The contrast agents for PA imaging are chromophores such as haemoglobin, indocyanine green (ICG), india ink, gold nanorods, etc. The combination of PA with US imaging is still in its infancy [98-100], but has already been shown to be useful at the preclinical level for staging deep vein thrombosis [99] and sentinel lymph node mapping [100].

3.4 US – Optical Imaging

US-OI contrast agents have been prepared by incorporating OI contrast agents (mostly organic dyes and quantum dots) during MB synthesis or into preformed MB, leading to their encapsulation within the MB shell [101, 102]. Such dye-loaded fluorescent MB have been predominantly used for in vitro studies. In this context, US/OI probes are very useful, e.g. to quantify the surface coverage of MB with targeting ligands [41]. US/OI probes can also be used to study MB binding kinetics to biological targets and to investigate the fate of bubbles after intravenous administration. In this regard, Koczera et al. showed that fluorescence analysis instead of phase contrast microscopy can reduce user-dependency and variability in the quantification of bound target-specific MB to cells in cell culture (Figure 4D). In addition, they demonstrated that fluorescent MB can be used for in vivo and ex vivo validation. In this context, using two-photon laser scanning microscopy, the binding of rhodamine-loaded, ICAM-1 targeted MB to activated HUVEC was reported (Figure 4E). Additionally, by fluorescence microscopy, the attachment of targeted rhodamine-loaded MB to tumor endothelium could be validated in histological sections (Figure 4F) [101]. For theranostic purposes, organic dyes have also been used as model drugs to investigate the loading capacity of such small molecules into the MB shell, as well as their release upon US-mediated MB destruction (see below) [103, 104].

A very innovative application of fluorescently labeled MB was presented by Yuan in 2009 [105]. Here, a fluorophore quencher-labelled MB system was used to measure external pressure. The authors claim that pressure variations as low as 1 mm Hg can be measured. If reliably working in vivo, such systems may become interesting tools in oncology research, e.g. to study the impact of pressure on tumor spread and metastasis.

3.5 US – Nuclear imaging

Radiolabeled MB for PET/SPECT-US imaging are generally used for investigating the biodistribution of MB after i.v. injection [106-108]. In comparison to OI, nuclear imaging techniques offer better quantification, which is particularly true in biodistribution analyses. A problem with both OI and nuclear imaging, however, is that intact MB and their fragments cannot be distinguished. In this context, Palmowski et al. labeled PBCA-MB with ¹¹¹In by incorporating it into DTPA attached to the MB shell. After i.v. injection, they measured the activity in different organs of mice over several time points up to 48 hours post injection (Figure 4G-H). While the activity in the liver and spleen stayed relatively high throughout the experiment, the values in lungs and blood decreased over time. In comparison to the liver and spleen, the authors reported much less signal for tumor, gastrointestinal tract and kidney (Figure 4H). Despite the activity detected in the kidneys, there was no activity in urine, which speaks against kidney clearance of PBCA MB and of their fragments. This was confirmed by the absence of MB signal during US-SPAQ imaging of the kidney 48 h post injection (Figure 4I). The activity measured in the kidney therefore likely resulted from accumulated MB fragments [107].

Alternatively, Tartis et al. performed biodistribution studies with soft-shell MB. In their study, they incorporated the ¹⁸F-labelled lipid (¹⁸F fluorodipalmitin) to the shell of lipid (DPPC, DPPA, DPPE-PEG₅₀₀₀) MB during synthesis. Subsequently, in line with the experiments by Palmowski et al, they demonstrated in male Fischer-344 rats that most of the lipids accumulated in the liver and spleen compared to other organs. Furthermore, they showed that local deposition of ¹⁸F-labeled lipids in the kidney is observed upon US-mediated MB destruction [106].

4. Therapeutic and Theranostic US

4.1 Therapeutic US

Besides for diagnostic purposes, US can also be used for therapeutic and theranostic purposes. Therapeutic US interventions generally refer to the use of the thermal effects of HIFU. MR-guided HIFU ablation, for instance, is currently used in the clinic for the treatment of deep-seated tumors [5, 6, 109]. In this procedure, the unwanted tissue is destroyed by heating it to around 60 °C using HIFU. This is nowadays performed under the guidance of MRI which provides high-resolution anatomical images for the delineation of target tissue, as well as real-time temperature mapping, ensuring ablation of only the target region [110, 111]. Voogt et al. recently used this technique for ablation of uterine fibroids in 33 patients [109].

Besides for thermal therapy, the effects of US can also be used to trigger release of therapeutic substances from nanocarriers at the target site. In principle, US-induced mild hyperthermia (40-45 °C) is known to increase tissue perfusion, thus e.g. enhancing the deposition of systemically administered therapeutics in the heated tissue. Heating by US can furthermore also be used for target-specific controlled release of therapeutic substances from e.g. temperature-sensitive liposomes [112-114].

Alternatively, also non-thermal effects of US have been used for therapeutic purposes. For example, US-induced MB cavitation has been shown to facilitate thrombolysis. In this regard, representative studies have been performed by Molina et al., who demonstrated in patients with middle cerebral artery (MCA) occlusion that the combination of tissue plasminogen activator (tPA), MB and US treatment was significantly more effective than tPA plus US treatment or tPA alone for thrombolysis [115].

Similarly, therapeutic agents could be released from nanocarriers by non-thermal effects of US. In this context, Wang et al. reported that the hydrolysis of THPMA side chains in PEG-b-THPMA diblock copolymer micelles destabilized the micelles, and enabled the triggered release of nile red [116]. In line with this, also for PLA-b-PEG micelles, Zhang et al pointed out that HIFU-induced degradation of the copolymer was responsible for micelle disruption [117]. Recently, Deckers et al. reported radiation force-induced acoustic streaming to be one of the main mechanism behind the release of hydrophobic and lipophilic substances from liposomes and the release of nile red from polymeric micelles in vitro [118, 119]. In vivo, pulsed HIFU was shown to reversibly enhance the extravasation and the interstitial transport of large (~100 nm) nanospheres in murine muscle [120, 121]. Based on their investigations and previous studies, Hancock et al pointed out acoustic radiation forces as the underlining mechanism behind the enhanced extravasation of nanospheres observed [121].

4.2 Theranostic US

Theranostic refers to the combination of disease diagnosis (in its broadest sense) and therapy [122-124]. Theranostic agents can provide valuable information of drug delivery, drug release and drug efficacy [107]. The ability to perform functional and molecular US, and the possibility to create MB carrying therapeutic substances, therefore make CEUS an attractive platform for theranostic applications [125, 126]. Theranostic US contrast agents are generally synthesized by incorporating therapeutic agents during MB synthesis into the shell of MB [102, 127], by the attachment of drug carriers (liposomes, nucleic acid-containing nanoparticles, etc.) to the MB shell surface [128, 129] or by partial incorporation of drug containing nano-emulsions into the gas core of the MB [130]. Upon injection, drug-carrying MB can be tracked by US imaging and - upon reaching the target site - destroyed to release their contents using high mechanical index US pulses.

Apart from the ability to track and destroy MB non-invasively by US, the use of MB for theranostics is fueled by the fact that the interaction of US with MB leads to stable (sustained) and inertial (destruction) MB cavitation, which can temporally increase vascular and cellular membrane permeability (sonoporation), thereby potentially increasing the extravasation and/or internalization of co-administered or MB-entrapped drugs. As depicted in Figure 5, the temporal opening of endothelial and cellular linings can be the result of:

1. Acoustic micro-streaming

2. Stable cavitation; expanding MB pushing the endothelial and/or cellular lining apart

3. Stable cavitation; contracting MB causing invaginations in the lining

4. Inertial cavitation; MB destruction-related shock waves permeabilizing the lining.

The duration for which the permeabilized endothelial and/or cellular membranes remain open is a subject of ongoing research [121, 131-133], but extensive pre-clinical studies have made use of this phenomenon to deliver drugs and genes to diseased tissues in oncology, across the BBB, and in thrombolysis.

Figure 5.

Mechanisms of sonoporation. Acoustic micro-streaming under stable cavitation (1). (2) MB compression leading to invagination and membrane opening. (3) MB expansion leading to membrane extension (push-force) and opening. (4) MB destruction releasing acoustic shock-waves and jet-streams that permeabilize the membranes. (Adapted with permission from [134])

4.2.1 Drug delivery

Image-guided drug delivery with MB can be performed either by the co-administration of both drugs and MB, or by the injection of drug-loaded MB. The latter has the advantage of reduced systemic drug exposure, and therefore less damage to healthy tissues. A further advantage is in the delivery of nucleic acids, which will otherwise be rapidly degraded upon systemic injection. However, in comparison to the systemic injection of high amounts of therapeutic drugs or their targeted delivery using nanomedicine formulations, the absolute accumulation of drugs delivered to the pathological site using MB likely is relatively low, due to the very short circulation half-life time of MB, to the fact that MB cannot extravasate and to the limited amount of therapeutic agents that can be loaded into the shell of MB. Nevertheless, such a targeting strategy presents a good avenue for the delivery of highly potent compounds, which otherwise cannot be applied due to the risk of extensive damage to healthy tissue.

Drug-loaded MB are generally produced by incorporating the drugs during MB synthesis (1-step), or by post-loading the drugs into or onto pre-formed MB (2-step) (Figure 6A). In this context, Wheatley et al. showed that doxorubicin and paclitaxel can be efficiently loaded into PLGA-based MB by both 1- and 2-step synthesis and demonstrated US-mediated release of the loaded drugs after MB destruction in vitro [103, 127]. Similarly, Fokong et al. developed PBCA-based MB loaded with hydrophilic (Rhodamine-b) and hydrophobic (coumarine-6) model drugs (Figure 6 C-D), and demonstrated efficient US-mediated release of these agents in vitro and in vivo in tumor-bearing mice (Figure 6E-F) [102]. Alternatively, Kooiman et al. encapsulated a lipophilic model drug (sudan black) into polymeric MB by using a hexadecane oil (as the drug carrier) in the air core of the bubbles [130]. An additional and very elegant MB design for drug delivery purposes utilized the attachment of drug-carrying liposomes to pre-formed MB (Figure 6B), which has the advantage of having much higher drug contents per microbubble as compared to the other MB-based materials [128, 129]. Yet another theranostic approach to drug delivery using MB was presented by Rapoport and colleagues, who prepared doxorubicin-loaded perfluoropentane-based nanobubbles. Upon injection, these nanobubbles passively accumulated in tumors via EPR, coalesced into MB at physiologic temperatures, and could then be imaged and destroyed by US, releasing encapsulated doxorubicin and resulting in effective tumor growth inhibition [135].

Figure 6.

Drug delivery to tumors and across the BBB using (model)-drug loaded MB. A-B: The entrapment of drug molecules in the MB shell can either be achieved during MB synthesis or upon post-loading. C-D: By fluorescence microscopy, the encapsulation of in the MB shell was validated. E-F: Upon US-mediated destruction of VEGFR2-targeted model drug-loaded MB in tumors, the accumulation of rhodamin-b and coumarin-6 in and around tumor blood vessels for animals treated with MB plus US (panels II and III; vs. without US, panel I) exemplified effective model drug delivery upon US guidance and triggering (adapted with permission from [102]). G: US- plus MB-mediated drug delivery across the BBB was studied using lipid-shelled MB. The extravasation of evans blue dye was used as verification of BBB-opening induced by DOX-SPIO-MB with US in normal brain tissue and C6 tumors (delineated by yellow line) H-J: H & E stained images of tumor-bearing brain after applying DOX-SPIO-MB and US validated the absence of brain hemorrhage. Region of interest for further analysis were selected from the tumor (T1, T2), the tumor-tissue boundary (B1, B2), and normal brain tissue (N1, N2). Panel I demonstrates increases in doxorubicin accumulation in brain tissue in the DOX-SPIO-MB + US group compared to the DOX group. K: SPIO accumulation in US-treated brain tissues also increased upon the combined application of MB plus US (adapted with permission from [94]).

Apart from the use of drug-loaded MB for target-specific delivery to tumors, the advantage of using MB for facilitating site-specific drug delivery and drug therapy has also been utilized for the delivery of therapeutic entities into the CNS. To the end, the combination of MB and destructive US pulses has been reported to enhance drug delivery across the BBB [136-139]. In a recent representative study, Fan et al. demonstrated that using lipid-shelled (DSPC, DSPE-PEG₂₀₀₀, DSPG) MB loaded with doxorubicin and SPIO in combination with focused US enhanced targeted drug delivery in rats bearing brain gliomas [94]. They observed a significant opening of the BBB in normal brain tissue for animals that received MB and focused US treatment by looking at the extravasation of evans blue dye (Figure 6G). Using H & E staining, they confirmed the absence of brain hemorrhage in the tumor and in normal brain tissue (Figures 6H and 6J). Furthermore, significant doxorubicin and SPIO deposition was observed for the MB plus focused US treated groups compared to control groups (Figures 6I and 6K). While this study provides proof-of-principle for BBB permeabilization and drug delivery into the CNS using focused US and MB, it also might also present a novel avenue for image-guided drug delivery across the BBB by deposited SPIO particles.

5. Conclusion

In summary, with the introduction of MB as US contrast agents, important diagnostic and therapeutic options have emerged for this extensively used, real-time and low-cost imaging modality. Besides a detailed characterization of tissue (and tumor) microvascularisation, US imaging also allows the assessment of specific molecular alterations, in particular at the vascular level. This is expected to improve the accuracy in pathology characterization and likely also enables more efficient treatment monitoring. The first molecularly targeted MB have now entered clinical trials and several others are ready to be tested. Additionally, many promising preclinical studies have demonstrated the capability of MB to increase vascular and cellular permeability, and to thereby facilitate the transport of drugs and genes. Furthermore, several therapeutic US-based interventions, such as MR-guided HIFU, are currently entering the clinic, and a number of theranostic means for image-guided and/or US-triggered drug delivery are currently being evaluated. It therefore seems reasonable to assume that MB, in combinations with diagnostic, therapeutic and theranostic US, will gain ever more importance in the years to come, both at the preclinical level, and in patients.