An overview of the influence of therapeutic ultrasound exposures on the vasculature: High intensity ultrasound and microbubble-mediated bioeffects

Abstract

It is well established that the interaction of ultrasound with soft tissues can induce a wide range of bioeffects. One of the most important and complex of these interactions in the context of therapeutic ultrasound is with the vasculature. Potential vascular effects range from enhancing microvascular permeability to inducing vascular damage and vessel occlusion. While aspects of these effects are broadly understood, the development of improved approaches to exploit these effects and gain a more detailed mechanistic understanding is ongoing and largely anchored in preclinical research. Here a general overview of this established yet rapidly evolving topic is provided, with a particular emphasis on effects arising from high-intensity focused ultrasound and microbubble-mediated exposures.

• Blood vessels
• drug delivery
• HIFU
• microbubbles
• microvessels
• therapeutic ultrasound
• ultrasound

Introduction

It has long been recognised that ultrasound exposures have the potential to cause vascular bioeffects. In diagnostic ultrasound imaging systems, exposure levels are limited in order to avoid inducing these or other bioeffects [1,2]. In a therapeutic ultrasound context, vascular effects can occur and are of importance from both a safety perspective – to avoid potentially undesired or off-target effects [3] – or they can be a primary, intended therapeutic effect [4–6]. A spectrum of vascular effects can be elicited, ranging from enhancements for microvascular permeability [7,8], to haemorrhage and vascular occlusion [9,10]. These are highly dependent on the specific exposure conditions that are employed (e.g. frequency, intensity, or exposure duration) but ultimately they originate from initial thermal, mechanical or chemical insults, which give rise to a cascade of downstream events. Thermal effects arise from temperature elevations that are caused by the absorption of ultrasound energy by tissue. Mechanical effects such as acoustic streaming [11] and radiation force-induced tissue displacement occur progressively with increasing ultrasound intensity [12,13]. The occurrence of cavitation is a threshold-dependent phenomenon that can cause vascular bioeffects through mechanical, chemical or thermal processes. Cavitation can be initiated within tissue [14] and blood vessels [15] using sufficiently high peak rarefactional pressures, and the resulting gas and vapour filled cavities can grow and oscillate violently, causing damage to both parenchymal tissue and the vasculature. With the introduction of microbubble contrast agents into the systemic circulation, the pressure threshold for vascular bioeffects is reduced.

Effects such as thermal-based vascular damage have been investigated over the course of decades, whereas others, particularly those involving the cavitation of microbubbles, are more recent. While elements of these effects are understood in a general sense, the detailed mechanistic underpinnings of many these processes remain to be elucidated. In order to fully harness their therapeutic potential, a great deal of work remains to be undertaken and much of this must necessarily be conducted at a preclinical level through mechanistic investigation and the development and refinement of methodologies. This paper provides a brief overview of the vascular effects of ultrasound with an emphasis on preclinical studies relating to therapeutic ultrasound. This is a large and rapidly evolving field that encompasses a wide spectrum of exposure regimes and applications. The review is divided first into a high intensity focused ultrasound (HIFU) section followed by one relating to microbubble-mediated effects.

High intensity focused ultrasound effects

HIFU employs highly focused ultrasound beams to concentrate acoustic energy into small focal volumes with the objective of inducing tissue bioeffects [16]. The frequencies employed in extracorporeal ultrasound applications typically range from approximately 0.3–3 MHz, with sub-MHz systems being used primarily for transcranial applications. For thermal applications, the basis of their operation relies upon the absorption of a portion of the incident wave energy by tissue through loss mechanisms, which gives rise to temperature elevations. The degree of thermal elevation is proportional to the absorptive properties of the target tissue, which increases with frequency, and the time-averaged intensity. HIFU exposures can result in rapid temperature rises (10 s of °C) within soft tissue which can induce coagulative necrosis within seconds [17]. Factors such as perfusion and flow in large vessels will act to reduce levels of temperature increase through the convective removal of heat.

Several types of mechanical effects can be induced with HIFU (Figure 1). Ultrasound waves impart momentum to the propagating medium. In blood- or fluid-filled cavities, this causes acoustic streaming flow in the direction of the ultrasound propagation [11]. In tissue, this ‘radiation force’ induces soft tissue displacements which then tend towards recovery to its equilibrium position during off times [12,13]. As the magnitude of radiation force is proportional to intensity, these effects are concentrated within the focal zone of the transducer. At sufficiently high rarefactional pressures, acoustic cavitation can be initiated, which involves the formation, growth and oscillation of gas- or vapour-filled bodies [18]. Cavitation is preferentially initiated at nucleation sites such as interfaces or microscopic gas bodies that are present within the body [15]. While the nature of cavitation is complex and multifaceted [18], in the context of therapeutic ultrasound it is useful to highlight two primary classes of cavitation: stable and inertial. Stable cavitation involves sustained acoustically stimulated oscillations of a bubble about its equilibrium state, wherein its dynamics are largely governed by the gas compressibility [18,19]. These oscillations may be linear or non-linear, the latter being associated with harmonic and potentially sub- and ultra-harmonic emissions [19]. Inertial cavitation involves a large and unstable bubble expansion, followed by a rapid collapse that is dominated by the inertia of the inrushing surrounding liquid. Inertial cavitation is a violent process that is accompanied by high local temperatures, the formation of free radicals and broadband acoustic emissions. The violent nature of inertial cavitation can cause direct mechanical damage to blood vessels [3] and tissue and, under certain conditions, can result in macroscopic temperature elevations and the release of free radicals [18,20]. The likelihood of inertial cavitation inception increases with decreasing ultrasound frequency [15,18]. Cavitation can be initiated with conventional HIFU systems (e.g. those designed primarily for ablative purposes), where it is most effectively accomplished with pulsed rather than continuous wave transmission. Specialised techniques, such as histotripsy [21] and shockwave lithotripsy [22], have been developed to exploit cavitation and involve extremely high pressures (typically > ∼10 MPa) in combination with particular pulse sequences that can result in tissue destruction (please see later section).

Figure 1. Schematic of HIFU-induced effects. (A) Thermal elevation is proportional to tissue absorption and time average intensity, occurring preferentially within the transducer focal region. This can ablate vessels. (B) Flow within fluid (e.g. blood) is induced by acoustic streaming, which occurs as a result of momentum imparted by the ultrasound wave to the propagating medium. (C) In tissue, radiation force induces displacements in the direction of ultrasound propagation – a phenomenon that can displace blood vessel walls. (D) At sufficiently high peak negative pressures, cavitation can be induced – leading to a spectrum of vascular effects such as haemorrhage or the formation of thrombus.

Both thermal and mechanical effects are responsible for HIFU induced bioeffects in the vasculature. These effects include permeability enhancement [7], haemorrhage [3], vascular spasm [3,23,24], vascular occlusion [9,10,24] and haemostasis [4]. We begin with an examination of microvascular effects, followed by those occurring in large vessels using HIFU, and follow this with a section on histotripsy and lithotripsy exposures.

Microvascular effects

The microvasculature is comprised of an interconnected network of arterioles, capillaries and venules and is the primary site of nutrient and waste exchange between blood and tissue. Microvessels are typically defined as being below 200–300 microns in diameter and have flow velocities in the order of tens of mm/s and below. It has long been recognised that HIFU is capable of inducing microvascular damage, which includes capillary rupture, perivascular haemorrhage and thermal damage [7,25].

Perivascular haemorrhage and microvascular rupture are caused by cavitation. Cavitation can also increase vascular permeability and lead to tissue oedema and an inflammatory response [7,26]. As a means of controlling vascular permeability, however, the violent and threshold dependent nature of HIFU-induced cavitation presents control challenges as it is frequently accompanied by the occurrence of haemorrhage [3]. This motivates the use of microbubble-based approaches for permeability enhancement, as there can be a more predictable pressure range over which permeability is enhanced in the absence of vascular damage [27], described in the section on microbubble-mediated effects below. In general, haemorrhage has been reported to be more likely to occur with increases in pulse duration and exposure time, and with decreasing frequency [26].

Mild, sub-ablative hyperthermia (<45 °C) is also capable of enhancing microvascular permeability which can facilitate the delivery of therapeutic agents to tissues, such as tumours [28]. Here, the use of temperature feedback is important to maintain appropriate temperature ranges [29], which has been established with MRI-guided HIFU systems [30].

In regions of ablated tissue, microvessels undergo widespread thermal coagulation. Due to their narrow lumens and thin wall structures this process appears to readily result in vessel collapse and thereby the cessation of flow. Both cavitation and thermal effects can result in microvascular occlusion. A number of studies support that vessels below 0.2 mm can be occluded by HIFU exposures, which can be accompanied by oedema and an apparent inflammatory response [25,31].

Large vessel effects

As vessel diameters increase, their walls become more robust and blood flow velocities are higher. Vessels above several hundred μm in diameter become thermally significant as a result of the convective removal of heat by blood flow, which makes it more difficult to elevate their wall temperatures [32]. Larger vessels therefore typically require higher exposure levels (intensity and/or duration) to induce thermal damage.

Considerable efforts have been directed at attempting to occlude blood flow in larger vessels for therapeutic purposes. Potential applications include treating vascular malformations, haemorrhage control or to deprive tumours of a blood supply. Transient vessel spasm has been observed in a number of studies [3,23,24], which can result in a temporary reduction or cessation of blood flow. The specific mechanism of spasm has not been fully elucidated but it has been associated with the occurrence of cavitation [3]. The sustained occlusion of intact vessels has also been reported, beginning with the work of Fallon et al. [9] in rabbit auricular vessels and subsequently in a range of other models such as rabbit femoral arteries [10,24,33–38]. The dominant effect giving rise to occlusion is thermal coagulation of the vessel wall, which results in tissue shrinkage and therefore constriction of the lumen. A frequent challenge with such treatments is to obtain sufficiently high temperatures in the presence of flow-mediated cooling. It has been shown that inducing spasm in vessels prior to occlusion reduces blood flow and therefore enables more rapid heating [10]. Radiation pressure-induced wall motion and acoustic streaming within the blood may also assist in this process. A challenge with increasing exposure levels is that cavitation can occur, which is also capable of inducing vessel rupture and haemorrhage [10,35,39]. The interplay between these factors is complex and depends upon frequency and exposure duration. A comprehensive listing of exposure conditions that have been employed for the purposes of vascular occlusion can be found in Shaw et al. [5].

In addition to thermal coagulation of vessel walls, thrombogenesis induced by ultrasound can also play a role in creating vascular occlusions. Indeed, the presence of thrombus has been observed in a number of previous studies of vascular occlusion [9,33,36,37]. Ultrasound-mediated thrombosis is thought to occur as a result of several contributing factors. One is haemostasis, whereby possible vessel constrictions, acoustic streaming and radiation forces on the vascular wall act to reduce or halt blood flow, which in turn aids the thrombogenic cascade. Prothrombotic factors can also be upregulated through several paths. One is the release of prothrombotic factors by endothelial cells that have undergone thermal or mechanical damage [40]. If cavitation is present it has been shown that, presumably due to the associated microstreaming, endothelial cells can be detached from the vessel wall and thereby expose the basement membrane [41,42], which will enable the formation and adherence of thrombus. Finally, it has been shown that platelets can be activated by cavitation [43,44]. It has been suggested that a combination of thrombus formation and thermal occlusion may be advantageous to promote the sustained occlusion of large vessels [5,24].

HIFU has also been shown to be capable of stopping bleeding from lacerated vessels and tissue [4,45–47] This approach has considerable potential in a number of application areas, such as abdominal bleeding cessation. The predominant mechanism that has been exploited is thermal coagulation. It has also been reported that acoustic streaming facilitates the cessation process by countering the flow of blood ejecting from the incision, causing it to remain within the vessel [4]. The addition of fibrinogen or pro-inflammatory agents may further enhance the formation of occlusions [48].

Histotripsy and shock wave lithotripsy

While cavitation can be induced by conventional HIFU systems and methods, particularly operating in pulsed mode, specific techniques have also been developed to exploit cavitation by employing very high rarefactional pressures and specific pulsing sequences. Histotripsy is a cavitation-based technique involving very high peak rarefactional pressure (>∼10 MPa) short pulse length (μs long), low duty cycle ultrasound. It exploits the use of sustained cavitation clouds to fractionate soft tissue into a liquefied acellular homogenate [49,50]. With this approach, microvessels within tissue are fractionated, while large vessels such as arteries are typically spared. This approach may have advantages over thermal approaches for treating tissue immediately adjacent to vessels, in that convective cooling does not undermine its performance. Tissue surrounding the emulsified region may exhibit signs of microvascular haemorrhage. A recent study indicates that larger vessel walls may remain intact due to their larger mechanical strength [51].

Extracorporeal shock-wave therapy, characterised by approximately 1-ms long high-pressure pulses followed by longer (∼5 ms) lower amplitude pulses [22] is known for use in lithotripsy, for example in kidney stone fragmentation applications [52]. Its mode of action is primarily through cavitation and acoustic streaming. It has been reported to have the potential to induce haemorrhage in soft tissue [53]. Lithotripsy has also been shown in several studies to be a possible treatment for ischaemic heart disease through the promotion of angiogenesis [54,55] [56]. Small scale clinical trials of this approach are being undertaken [57]. The origin of these effects is unclear at present [58] but is presumed to be associated with non-thermal cavitational effects, possibly upregulating nitric oxide [59] and/or vascular endothelial growth factor (VEGF) production [60].

Microbubble-mediated effects

The introduction of encapsulated microbubbles into the systemic circulation is widely employed for the purposes of enhancing diagnostic ultrasound vascular imaging. Microbubbles are encapsulated with compliant biocompatible shells (e.g. phospholipids) for stabilisation purposes and are largely within the range of 1–6 μm in diameter and thus remain within the vasculature during circulation. By coupling targeting ligands to their shells, microbubbles can be employed as a platform for molecular imaging of targets such as the angiogenic neovasculature, inflammation and thrombus [61]. In a diagnostic imaging context they are intended to be exposed under conditions that do not produce bioeffects [62,63].

Increasingly, microbubbles are the subject of interest for their ability to produce a wide range of therapeutically relevant bioeffects that have the potential to enhance the treatment of a spectrum of disease processes [6,64–67]. Of particular interest is that microbubbles lower the threshold for inducing therapeutically relevant vascular effects and do so in a manner that is potentially more controllable than HIFU-based cavitation. A primary avenue for their therapeutic use is that they can enhance microvascular permeability and thereby be employed to facilitate locally enhanced drug delivery [6,67–69]. They are also capable of eliciting other bioeffects such as inflammation [70,71], thrombogenesis [41], angiogenesis [72,73], and microvascular shutdown [74–79], depending upon the exposure conditions.

Physical behaviour and interactions with blood vessels

When microbubbles are exposed to ultrasound fields they can exhibit a rich variety of dynamic behaviours, ranging from stable spherically symmetrical oscillations to shape oscillations [80], fragmentation, and highly unstable violent behaviour (e.g. inertial collapse). In general, lower pressure amplitudes (up to 10 s of kPa) produce only modest oscillation amplitudes (in the order of a few per cent or less). The specific amplitude will depend on factors such as the relationship between the ultrasound frequency relative to the bubble resonant frequency, which will in turn be affected by the encapsulating shell properties [81]. Oscillation amplitudes tend become more pronounced and non-linear with increasing pressure, eventually being substantial enough to induce therapeutic effects at pressures in the order of 100 kPA [6,82] – more than an order of magnitude lower than the threshold for HIFU-induced cavitation in the absence of microbubbles [83]. As noted earlier for the case of HIFU-induced cavitation, there is a trend towards lower frequencies favouring the development of inertial cavitation. It is also notable that cavitating microbubbles are constrained to be within the vasculature, whereas with HIFU-induced cavitation bubbles can be present within both tissue and the vasculature.

Microbubbles can also translate in the direction of ultrasound propagation under the influence of primary radiation forces and can be drawn together under the influence of secondary radiation forces [84]. They can be disrupted through a variety of processes such as inertial collapse [20], fragmentation [18,85] or the loss of shell material [86,87]. Both during stable oscillations and collapses local thermal elevations can occur and under certain circumstances these can give rise to macroscopic temperature elevations [88]. Also notable is that certain oscillation characteristics (stable or inertial) can be linked to the acoustic signatures that they emit and therefore there is interest in monitoring these emissions as a means of treatment monitoring and control [89,90].

The dynamic behaviour of microbubbles is influenced by their surroundings, such as vessel walls, formed blood elements (e.g. erythrocytes) and endothelial cells through a variety of mechanisms. When microbubbles are situated within microvessels, for example, they experience higher levels of damping and have a shifted resonant frequency [91–93]. When oscillating bubbles are in contact with or in close proximity to cells, it has been shown that cell membrane and cytoskeletal deformations can occur [94]. Oscillating bubbles also induce microstreaming in the surrounding fluid [95] and when situated near boundaries [96,97] or contained within microvessels [98–102], which can create complex flow patterns and the presence of oscillatory shear stresses [91,103]. Additionally, experimental and theoretical evidence indicates that when bubbles are situated within microvessels, they can induce complex deformation of vessel walls including over-expansion (with accompanying circumferential stresses) (Figure 2A), invaginations, and direct hydrodynamic jets towards the walls [104,105] (Figure 2B). Inertial collapse in the vicinity of the wall can result in localised temperature elevations, shock waves and the release of reactive oxygen species (Figure 2C). These effects are highly dependent upon the pressure amplitude, and their size relative to the resonant frequency.

Figure 2. Schematic of several ultrasound-stimulated microbubble behaviours of relevance to inducing vascular bioeffects. (A) Microbubble adjacent to a vessel wall (left) undergoes expansion that deforms the wall outwards and creates fluid flow patterns that give rise to shear stress at the lumen surface endothelial cells (middle). During rarefaction, the fluid flow patterns change and there can potentially be an invagination of the wall. (B) Hydrodynamic jets have been reported that are directed towards the vessel wall. (C) Bubbles can undergo inertial collapse, a violent process that can result in extremely high local temperatures, shock waves, and the release of reactive oxygen species.

Collectively, these behaviours are capable of inducing a wide spectrum of bioeffects in the vasculature – from transient permeabilisation to sustained damage in both microvessels and larger vessels. There is not necessarily a distinct division between these effects as they may be a part of a continuous spectrum of events, but the discussion below is divided into the categories of microvascular permeabilisation, microvascular damage, followed by sections on large vessel effects and thrombolysis.

Microvascular permeabilisation

In the late 1990s it was observed that ultrasound-stimulated microbubbles were capable of increasing microvascular leakiness [8,106]. The recognition that this bioeffect could potentially be exploited as a means of locally promoting the transport of therapeutic agents from the vascular to the extravascular compartment stimulated intense interest in research ranging from basic physical and biological mechanistic investigations to an assessment of its potential application across a variety of tissue types and organs, disease processes and therapeutic agents [6,64–66]. As this is a broad topic that cannot be covered in detail in an overview paper, the reader is directed to a number of recent review papers which focus on specific application areas (e.g. brain, cancer, cardiovascular disease [45–49]).

Preclinical evidence of effects

Preclinical investigations have been conducted using a number of in vivo and ex vivo tissue preparations such as rabbit ears [41,107], rat mesentery [105], chicken chorioallantoic membrane [108] and cranial preparations [109] in combination with microscopy to assess effects in individual microvessels as a function of exposure parameters. These are supplemented by tissue-specific studies that have employed MR and optical imaging approaches and/or histological examination of excised tissues to assess extravasation in terms of mechanisms, degree and kinetics [110,111]. In more applied studies, the delivery of a wide spectrum of therapeutic agents has been examined in some cases along with the resulting therapeutic effects in the context of preclinical models of disease. A review of these is beyond the scope of the current paper, but there are a range of comprehensive recent reviews related to specific application topics [45–49].

There is now a wealth of evidence that transient microvascular permeability can be enhanced with exposure conditions that do not produce any apparent signs of vascular damage or haemorrhage [27,108]. The large majority of studies have been conducted at frequencies between 0.300 and 2 MHz, employing pulsed ultrasound (0.1–10 ms) at relatively low duty cycles – conditions that typically would not be associated with macroscopic thermal elevations. Permeability enhancements can be stimulated to be transient (in the order of minutes to hours) [42,66] and without the apparent appearance of other damage such as haemorrhage. The speed of onset, duration, and the degree of permeability (e.g. as a function of molecule weight) appears to be dependent upon both the particular exposure scheme employed (pressure, pulse pulsing scheme) and tissue type [108,112]. Due to the strong dependence of bubble oscillations on pressure, the majority of studies tend to report exposure amplitudes in terms of pressure as opposed to thermal studies which report time average intensity. In blood–brain barrier disruption (BBBD) studies for example, it has been found that the BBBD threshold is related to mechanical index (pressure over the square root of frequency). At 1 MHz for example, the range for non-haemorrhagic BBBD has been reported to be from approximately 0.3–0.4 MPa [27]. Studies conducted in other tissue types can be higher, some in the range of 1–2 MPa at 1 MHz [106,113]. Notably, these pressure levels are substantially lower than those employed to initiate cavitation in the absence of microbubbles, where required rarefactional pressures are in the range of 10 MPa or above, near 1 MHz [50,114]. Further, the permeabilisation effect has been linked to characteristics of the detected microbubble emissions. For example, in BBBD experiments, harmonic emissions (associated with therapeutic effects) along with wideband signals (associated with undesired inertial cavitation effects) can be used for the purposes of control [27]. The nature of the control process may depend on the desired effect as well as tissue type – for example the delivery of anticancer agents into tissue has been proposed using broadband emissions [115].

Mechanistic perspective

An understanding of the specific mechanisms by which microvascular permeabilisation occurs is at present incomplete. It has been demonstrated to occur in arterioles, venules and capillaries [108–110]. The primary candidate mechanisms that have been proposed are endothelial cell pore formation, trans-endothelial fenestrae, transmural vesicle transport, and widening of the gap junctions between endothelial cells [8,42,110,116]. These points will now be discussed.

One of the earliest hypotheses put forth was that microvascular permeability was due to pore formation in the membrane of endothelial cells. This was based largely on previous work conducted that demonstrated that cell membrane pore formation could occur when cells and oscillating microbubbles were in close proximity. The ‘sonoporation’ of cells with microbubbles was originally reported by Bao [117], and there is now a wealth of papers examining this effect as a function of cell type, including endothelial cells [82,94,118]. While this mechanism can explain how materials can be transported into endothelial cells, the short duration of these pores (milliseconds to seconds [6,119]) does not appear to be compatible with the extended microvascular permeability durations that have been observed (in the order of minutes to hours), which suggests that this is not a primary mechanism for enhancing microvascular permeability. There is now substantial evidence that transcellular vesicle transport is a primary pathway for enhancing extravasation. This is in part based on histological observations which show increased vesicle concentrations present within the cytoplasm of endothelial cells and within microvessel walls [110]. It is also supported by further in vivo [120] and in vitro [82] studies, where it has been shown that this effect is related to caveolin-mediated transcytosis [118]. A third mechanism of extravasation is through the widening of endothelial cell–cell gap junctions after microbubble exposures. This is supported by histological analyses derived from in vivo studies [110], and is also consistent with recent experiments conducted with in vitro layers of endothelial cells [6].

The relative importance of each of these paths may depend upon the exposure conditions used and the tissue types (e.g. brain or tumour or muscle). The details of these processes and their link to interactions between oscillating bubbles have yet to be fully elucidated. The latter two paths have been shown to be accompanied by changes in reactive oxygen species (ROS) levels, cytoskeletal configurations and mechano-sensitive ion channels (Ca2+) [82,119]. Collectively this suggests a prominent role of the activation of mechano-sensitive signalling cascades by the various mechanical actions of vibrating microbubbles (direct deformation, shear stresses), a topic that is the subject of ongoing research.

Microvascular damage and flow cessation

There is now therefore a considerable body of evidence indicating that transient microvascular permeabilisation can be achieved to permit the delivery of therapeutic agents, without any apparent sustained vessel damage. However, it is also the case that under sufficiently strong exposure conditions a plethora of other effects can also occur. Arguably one of the most significant of these from a therapeutic perspective is their capacity to induce a shutdown of microvascular blood flow and thereby create ischaemic lesions. This has been shown most extensively in brain and tumour tissue.

Preclinical evidence of effects

Brain

A number of studies that have subjected brain tissue to ultrasound microbubble treatments have reported the formation of necrotic non-thermal lesions [75,121–123]. Exposure levels were employed that primarily resulted in ischaemic necrosis, consistent with the occurrence of microvascular shutdown. Histological and MRI imaging indicated the presence of oedema, and micro-haemorrhages associated with the occurrence of microvascular damage along with infiltration of inflammatory cells were observed. Long-term (3–4 weeks) timescale shrinkage of targeted tissue was observed without apparent damage to adjacent tissue structures [75]. The exposure conditions employed were either estimated to be associated with negligible macroscopic temperature elevations [75], or measured thermal elevations that were at sub-ablative levels (e.g. ∼10–12 °C in Huang et al. [121]). It has been proposed that the use of such exposure schemes, well below powers required for thermal ablation, may have particular potential for treating brain structures adjacent to bone, such as at the skull base.

Tumours

An increasing number of studies have examined flow inhibition in preclinical tumour models. The effect was first reported with continuous wave physiotherapy transducers, which resulted in a vascular shutdown within tumours, under conditions that also produced mild temperature elevations [74,124]. These effects were subsequently shown to be feasible with pulsed exposures (typically at 1 MHz) under conditions that did not produce temperature elevations [77,125] but were associated with inertial cavitation (e.g. 1.6 MPa peak negative pressure [87]). The vascular shutdown has been shown to occur within tens of seconds [78,125,126]. An example of this is shown in Figure 3. Following vascular shutdown, endothelial cell apoptosis has been observed on a 24-h post-treatment time scale [127], along with tissue apoptosis and necrosis [78,128]. These treatments can produce growth delays in tumours [77–79,128–130], highlighting the potential of this approach as a tumour treatment method. Eventually, however, flow has been shown to recover, and this is accompanied by a resumption of tumour growth [78,128]. This ‘vascular rebound’ effect is qualitatively similar to the action of small molecule vascular disrupting agents, whose effects are most pronounced when coupled with taxane chemotherapy, antiangiogenic therapy (to block the vascular rebound) or radiotherapy [131]. Indeed, when these vascular damaging treatments are coupled with such drug approaches [78,128] or radiotherapy [79], profoundly enhanced anti-tumour effects can be achieved.

Figure 3. Figure illustrating vascular shutdown in a subcutaneous mouse tumour by ultrasound-stimulated microbubbles. (A) Example ultrasound contrast images at peak enhancement following microbubble injection prior to treatment (left) and after therapy exposures (right). These images illustrate qualitatively the reduction of perfusion resulting from the treatment, which preferentially affects the central regions of the tumour. Images are 1.5 cm lateral dimension. (B) An example of time–intensity contrast curves (central and peripheral regions of interest) following the bolus injection of contrast for a tumour during treatment. After an initial a rise to a peak as microbubbles enter the tumour, there is a gradual decay over several minutes which is associated with a systemic reduction in microbubble concentration in the bloodstream. Therapy pulses of 1 MHz (1.6 MPa peak negative pressure) are sent every 20 s, resulting in the destruction of agent within the tumour. In the peripheral region, substantial reperfusion occurs following each burst; however, in the central region there is a reduced level of reperfusion with successive burst, consistent with a reduction in blood flow in response to the treatment. (C) Example cavitation signals recorded from the microbubbles within the tumour during exposure to therapy pulses, expressed as a function of frequency in a decibel scale. The ‘baseline’ signal (solid) is taken prior to microbubble injection and arises from scattering of the incident US by tissue. Here a clear 1 MHz component (transmit frequency) along with a 2 MHz signal associated with (non-linear) propagation of the US pulse. Signals outside these frequencies are associated primarily with noise. For the microbubble signals (dashed) there are also pronounced peaks at 0.5 (subharmonic) and 1.5 MHz (ultraharmonic). The remaining substantial energy present across a wide range of frequencies is a hallmark of inertial cavitation, indicating the violent oscillations of microbubbles during the therapy pulses. Figure from Todorova et al. [128].

Mechanistic perspective

The section on microvascular permeabilization above highlighted evidence of modulations in microvascular permeability that appear to be associated to a significant extent with transient alterations in endothelial cell function and connectivity. While the mechanisms responsible for the sustained microvascular functional shutdown shown by studies reported in the present section remain to be fully established, it has been strongly implicated that endothelial cell damage may play an important role. Endothelial cell damage and denudation has been reported in microvessels as well as larger vessels [41,132] [42] as indicated by electron microscopy. The degree of these effects has been linked to inertial cavitation dose, an indicator that they are a result of violent bubble oscillations [41]. Microvascular haemorrhage and red blood cell extravasation associated with violent microbubble oscillations has been widely reported in the literature [8,133–135] and suggests a compromise in the integrity of microvessel walls. In addition to such acute endothelial cell damage it has also been reported that microbubbles can initiate endothelial cell apoptosis on a longer timescale, through the ceramide pathway [127]. As with permeability alterations, it appears likely that these effects may be substantially a consequence of mechanical stresses such as fluid shear stress, hydrodynamic jets, over-stretch injury and possibly vessel wall invaginations. It is also possible that temperature changes, when present, may also contribute to endothelial cell injury [121].

In vitro platelet activation has been observed to occur following ultrasound stimulated microbubble exposures [43,44,136,137]. Recent work in a tumour model [126] suggests that platelet activation and aggregation may be a major factor in the acute microvascular response that ultimately progresses to shutdown. This would be consistent with the rapid timescale of the initial shutdown, which can be in the order of 10 s of seconds (e.g. Figure 3). Damage to endothelial cells coupled with the exposure of the vascular basement membrane to blood when denudation occurs [41,108,132] are implicated in the formation of thrombus which can potentially act to occlude blood vessels and/or reduce blood flow [65,66,72]. This may be exacerbated by microvessel spasms that have been observed with microscopy [109]. Finally, the observations of infiltration of exposed tissue with inflammatory cells [75], coupled with separate evidence that inflammation can be up-regulated by microbubbles suggests that this process may also be a contributing factor. Therefore, microbubble-mediated vascular shutdown appears to be complex and potentially involves multiple processes that appear to unfold in a range of timescales.

Angiogenesis and arteriogenesis

It has been demonstrated in muscle tissue that ultrasound-stimulated microbubbles can induce angiogenesis and arteriogenesis [73,138,139]. Somewhat paradoxically, the exposure schemes that resulted in these effects were associated with an initial rupture of capillaries. This process has been shown to be driven at least in part by an initial inflammatory response [140,141] following the mechanical insult by microbubbles. The inflammation then promotes the recruitment of bone marrow derived cells [73] which subsequently produce cytokines that contribute to the stimulation of angiogenesis. While this has been achieved with microbubbles alone, it can also be coupled with the delivery of proangiogenic factors that may enhance these effects, a treatment that has potential for the revascularisation of ischaemic myocardium [142].

Large vessel effects

In the section on high intensity focused ultrasound effects above, the use of HIFU to initiate cavitation to cause damage within larger vessels (non-microvessels) was discussed. It has been shown that when microbubbles are used in conjunction with HIFU, damage to the vessel walls and in particular to endothelial cells can be promoted. The resulting endothelial cell damage, coupled with the exposure of the basement membrane and possible activation of platelets in the presence of microbubbles undergoing inertial cavitation has been shown to promote thrombogenesis [41]. The use of microbubbles may therefore be employed to aid in the formation of vascular occlusions or to aid in bleeding cessation procedures [24]. These effects can be exacerbated with the addition of fibrinogen or inflammatory agents [48]. It has been observed in the context of HIFU vessel occluding experiments that these effects may couple with thermal effects to yield more robust and sustained occlusions [36].

Sonothrombolysis

For completeness, it is also relevant to note that ultrasound can be employed to degrade thrombotic vessel occlusions with the objective of restoring blood flow [143]. This is a topic of considerable interest in the context of acute ischaemic stroke, deep venous thrombosis and coronary artery thrombosis or myocardial microthrombi. This approach has been under investigation for decades [143,144], with and without lytic enzymes; and with [145–148] and without [149] microbubbles; and using cavitational HIFU or histotripsy approaches [150–152]. Aside from its therapeutic potential, it must also be considered that vascular occluding therapies that exploit to some extent the formation of thrombus, may also have to avoid resolving the thrombi during treatment.

Conclusion

Under the appropriate exposure conditions, ultrasound is therefore capable of inducing a wide range of vascular bioeffects. While in a diagnostic imaging context exposures are constrained in order to avoid such effects, in a therapeutic context they are one of the central effects that occur, either as a by-product of treatments (e.g. tissue ablation) or as a primary intended effect. Techniques such as microvascular permeabilisation and vascular occlusion have considerable potential across a wide range of application areas. In order to fully harness this potential, a great deal of work remains to be undertaken and much of this must necessarily be conducted at a preclinical level in terms of mechanistic investigation and the development and refinement of methodologies.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.