Section 6—Mechanical Bioeffects in the Presence of Gas-Carrier Ultrasound Contrast Agents

Abstract

This review addresses the issue of mechanical ultrasound-induced bioeffects in the presence of gas carrier contrast agents (GCAs). Here, the term “contrast agent” refers to those agents that provide ultrasound contrast by being composed of microbubbles, encapsulated or not, containing one or more gases. Provided in this section are summaries on how contrast agents work, some of their current uses, and the potential for bio-effects associated with their presence in an ultrasonic field.

6.1 Introduction

Enhanced sonographic contrast is desirable in many imaging applications, e.g., detection of tumors and myocardial ischemia. In recent years, a number of gas carrier contrast agents (GCAs) have been developed or are in development. Some are now in clinical use in the United States or Europe and have demonstrated their utility and limitations. An extensive primary literature exists on the physical and clinical properties of GCAs, as well as several reviews (e.g., Ophir and Parker, 1989; Schlief et al, 1993; Goldberg, 1993a; Balen et al, 1994; Burns, 1994a; Goldberg et al, 1994; Winkelmann et al, 1994). However, the minutiae of clinical experience with GCAs are not dealt with here.

GCAs have been available for bioeffect research for only a short time. Consequently, the literature pertaining to mechanical bioeffects in the presence of GCAs is sparse, and thus our knowledge of such phenomena is limited. However, this topic is the subject of current research in several laboratories. Most of the mechanical bioeffects arising from insonation of cells or tissues with GCAs present can be attributed to the occurrence of inertial cavitation.

Subsection 6.2 provides an overview of GCAs in use or under development and discusses briefly some clinical GCA applications. These discussions are not exhaustive but offer an overview of GCAs: what they are, their general properties, and how they have been applied. In subsection 6.3, some clinical applications of GCAs are reviewed. Subsection 6.4 summarizes the literature pertaining to mechanical ultrasound bioeffects in the presence of GCAs.

6.2 Overview of GCAS in Use or Under Development

6.2.1 General Properties of Contrast Agents

Most GCAs comprise micrometer-diameter gas bodies, variously referred to as microbubbles or microspheres, that may or may not be surrounded by a shell of material that stabilizes the gas against diffusion. These are intended for intravascular or intraluminal injection.

A major problem in the development of GCAs has been obtaining a satisfactory size distribution. For imaging applications following intravenous injection of the agent, the microbubbles must be smaller than ~8 μm in diameter to pass through capillaries, thus allowing passage through the pulmonary circulation. The GCA must be tolerated by patients, should be stable with respect to rapid dissolution in the blood, and should be tolerant of hydrostatic pressures produced by the heart. A variety of engineering solutions to these problems has been attempted.

Microbubbles composed of atmospheric gases that are not stabilized in some way dissolve rapidly in aqueous media because of the pressure exerted on the gas by surface tension at the microbubble surface and outward diffusion of the gas (Porter and Xie, 1995a). Stability against dissolution can be provided by a shell of material around the gas or by using relatively insoluble gases. Various stabilizing shell materials have been used, including lipids (Unger et al, 1992; Simon et al, 1992b, 1993; D’Arrigo and Imae, 1992; Barbarese et al, 1995; Schneider et al, 1995), albumin (Bleeker et al, 1990b), dextrose-albumin (Porter et al, 1995a, 1995b, 1995c, 1995d), sugars (Schurmann and Schlief, 1994; Cennamo et al, 1994), gelatin (Prat et al, 1993), and polymers (Schneider et al, 1992). Various gases also have been used, ranging from relatively soluble gases, e.g., air (Bleeker et al, 1990b), nitrogen (Unger et al, 1992), and carbon dioxide (Kudo et al, 1994; Veltri et al, 1994), to gases with relatively low solubility or slow diffusion rates, such as helium (Porter and Xie, 1995a), sulfur hexafluoride (Porter and Xie, 1995a; Schneider et al, 1995), perfluoropropane (Porter et al, 1995d), and dodecafluoropentane (Quay, 1994).

6.2.2 Acoustic Properties of Contrast Agents

The ultrasonic beam generated by a transducer or array is fairly complicated (Shung and Zipparo, 1996; Zagzebski, 1996). However, for the sake of simplicity, here a plane wave is assumed. As an ultrasonic plane wave penetrates a distribution of scatterers, in the present case microbubbles, assuming that the nonlinear effects caused by these bubbles can be neglected, the sound velocity in the medium, c, is given by

$$\text{c}=\sqrt{\frac{1}{\rho \text{G}}}$$ (6-1)

where ρ and G are the effective density and compressibility of the medium, respectively. In general, if the volume concentration V of the bubbles is small, the mixture relationships, which are represented by

$$\rho ={\rho }_b\text{V}+{\rho }_m(1-\text{V})$$ (6-2)

$$\text{G}={\text{G}}_b\text{V}+{\text{G}}_m(1-\text{V})$$ (6-3)

where subscripts b and m stand for bubble and surrounding medium, are valid (Ophir and Parker, 1989; Mobley et al, 1998). Since sound velocity in a gas is lower than in fluids, the sound velocity in a liquid containing a GCA is lower than that of the liquid alone. Figure 6-1 shows that the sound velocity in Albunex decreases as the Albunex concentration increases at 7.5 MHz (Bleeker, 1990a). Recent studies (Wu et al, 1995; Mobley et al, 1998) indicate that the sound velocity in a distribution of microbubbles is a function of frequency and of bubble concentration, and may be affected by the nonlinear resonance behavior of the bubbles at frequencies near the resonance frequencies of the bubbles.

Figure 6-1.

Sound velocity in an Albunex suspension as a function of concentration. The solid circles and the line represent the data points and the least squares regressions line, respectively.

As the wave penetrates the medium, the pressure amplitude, P_z, and intensity, I_z, decrease according to the following equations (Shung et al, 1992; Zagzebski, 1996):

$${\text{I}}_z={\text{I}}_0{e}^{-2\beta \text{z}}$$ (6-4)

$${\text{P}}_z={\text{P}}_0{e}^{-\beta \text{z}}$$ (6-5)

where z = distance traveled, I₀ = intensity at z = 0, P₀ = pressure at z = 0, and β= pressure attenuation coefficient in np/cm (1 np = 8.686 dB) of the medium.

For a distribution of low bubble concentration, the attenuation coefficient is given by

$$\beta =\frac{\text{n}{\sigma }_{\text{e}}}{\text{2}}$$ (6-6)

where n = the number of bubbles per unit volume, and σe = the extinction cross section (Ishimaru, 1978; de Jong et al, 1992). The extinction cross section, σe, with dimensions of cm², is frequency dependent and represents the total energy loss for an ultrasonic wave resulting from the presence of one scatterer in a unit volume; e.g., per cm³. It is the sum of the absorption cross section, σa, and the scattering cross section, σs (Ishimaru, 1978; de Jong et al, 1992)

$$\sigma \text{e}=\sigma \text{a}+\sigma \text{s}$$ (6-7)

Both σa and σs have the dimension of cm² and represent the energy absorbed and scattered by one scatterer in a unit volume, respectively. The question that often arises is the relative contribution of each of these terms to the total attenuation. The attenuation, if estimated accurately, can sometimes give information about the properties of the scatterer but, in general, attenuation is not desirable in diagnostic applications, especially its absorption component. To achieve optimal ultrasonic contrast, the absorption of the contrast agent should be made as low as possible to avoid losing energy and shadowing biological structures, while the scattering is maximized.

The scattering properties of biological tissues and blood have been investigated intensively because these signals are used to form ultrasonic images (Shung and Thieme, 1993). Several investigators have also studied the acoustic properties of liquids containing gas bubbles (Devin, 1959; Medwin, 1977; Anderson and Hampton, 1980). The total scattering cross section of a single bubble can be expressed as follows (Medwin, 1977):

$${\sigma }_s=\frac{4\pi {\text{R}}^2}{{\left(\frac{{\text{f}}_{\text{r}}^{\text{2}}}{{\text{f}}^{\text{2}}}-1\right)}^2+{\delta }^2}$$ (6-8)

where R = bubble radius, f_r = bubble resonance frequency, f = frequency of the incident ultrasonic wave, and δ = total damping constant, caused by the surrounding liquid medium and consisting of terms due to re-radiation, thermal conductivity, and shear viscosity (Devin, 1959).

The resonance frequency of a bubble can be determined assuming the following: the wavelength of the ultrasonic wave is much larger than the bubble diameter, the radial displacement of the bubble is small relative to its radius; and the surrounding fluid is incompressible. The first two conditions are fulfilled by the majority of GCAs and the third by water and blood, which are virtually incompressible.

The simplest equation of resonance frequency for a bubble in water is

$${\text{f}}_{\text{r}}=\frac{1}{2\pi \text{R}}\sqrt{\frac{3\gamma {\text{P}}_0}{{\rho }_{\text{m}}}}$$ (6-9)

which can be modified to

$${\text{f}}_r=\frac{1}{2\pi }\sqrt{\frac{{\text{S}}_{\text{a}}}{\text{m}}}$$ (6-10)

where γ = ratio of specific heats of the gas (~1.4), P0 = ambient hydrostatic pressure (= 1.103 × 10⁶ dyne/cm), ρm = density of the surrounding fluid (=1.03 g/cm³), S_a = adiabatic stiffness (=12πγP₀R), and m = effective mass of the system (=4πR³ρm). Here an adiabatic equation of state is assumed. However, for bubbles of small radii, surface tension becomes a significant additional restoring force and must be considered. The oscillation in this case is closer to an isothermal process. When these points are considered the equation for resonance frequency changes. P₀ in Equation 6.10 is replaced by the average interior pressure including surface tension, ξP₀, and γ is replaced by the effective ratio of specific heats in the presence of thermal conductivity, γb (Medwin, 1977; de Jong et al, 1992; de Jong, 1996)

$${\text{f}}_{\text{r}}=\frac{1}{2\pi \text{R}}\sqrt{\frac{3\gamma \text{b}\zeta {\text{P}}_0}{{\rho }_{\text{m}}}}$$ (6-11)

where

$$\text{b}={(1+{\text{B}}^2)}^{-1}{\left[1+\frac{3(\gamma -1)}{\text{X}}\left(\frac{sinh\hspace{0.17em}\text{X}-sin\hspace{0.17em}\text{X}}{cosh\hspace{0.17em}\text{X}-cos\hspace{0.17em}\text{X}}\right)\right]}^{-1}$$ (6-12)

$$\text{B}=3(\gamma -1)\left[\frac{\text{X(\hspace{0.17em}}sinh\hspace{0.17em}\text{X}+sin\hspace{0.17em}\text{X)\hspace{0.17em}}-\text{2(cosh X}-\text{cos\hspace{0.17em}X})}{{\text{X}}^2\hspace{0.17em}(cosh\hspace{0.17em}\text{X}-\text{cos}\hspace{0.17em}\text{X)}+\text{3(}\gamma -\text{1)X(sinh X}-\text{sin X)}}\right]$$ (6-13)

$$\zeta =1+\frac{2\sigma }{{\text{P}}_0\text{R}}\left(1-\frac{1}{3\gamma \text{b}}\right)$$ (6-14)

$$\text{X}=\text{R }{\left(\frac{\text{2}\omega {\rho }_{\text{a}}{\text{C}}_{\text{pg}}}{{\text{K}}_{\text{g}}}\right)}^{\frac{1}{2}}$$ (6-15)

In these equations, Kg= thermal conductivity of gas (=5.6 × 10⁻⁵ cal/cm-s-°C); ρa = density of free gas at sea level (=1.29 × 10⁻³ g/cm³); σ = surface tension (=75 dyne/cm); C_pg = specific heat at constant pressure for air (=0.24 cal/g), λ = ratio of specific heats and ω = angular frequency.

The parameters b and ζ are functions of the bubble radius. Equation 6.11 is plotted in Figure 6-2, which shows the bubble resonance frequency as a function of the bubble size. Note here that smaller bubbles yield higher resonance frequencies. Figure 6-3 shows the scattering cross section of a bubble of 1.7 μm radius at an ambient pressure of 1 atm calculated using the numerical values of parameters given above. Measured data are in reasonable agreement with the calculated curve (de Jong et al, 1992; Chang et al, 1993).

Figure 6-2.

Computed resonance frequency versus the radius of a bubble.

Figure 6-3.

Computed scattering cross section of a bubble of 1.7 μm radius.

The portion of the energy absorbed or converted into heat by a scatterer from an ultrasonic wave is referred as the absorption cross section, and it is related to scattering cross section as follows (Medwin, 1977; de Jong et al, 1992):

$${\sigma }_{\text{a}}={\sigma }_{\text{s}}\left(\frac{\delta }{{\delta }_{\text{r}}}-1\right)$$ (6-16)

where the damping constant δ is the sum of three components (Devin, 1959):

$$\delta ={\delta }_{\text{r}}+{\delta }_{\text{t}}+{\delta }_{\text{v}}$$ (6-17)

and where δ_r = kR = damping constant due to re-radiation, k = wave number, δ_t = B(fr/f)² = damping constant due to thermal conductivity, δ_v = 4η/(ρωR²) = damping constant due to shear viscosity and η is the shear viscosity of the surrounding liquid (= 0.01 g/cm-s).

6.2.2.1 Bubbles with an Elastic Shell

The discussion above applies only to a free bubble. However, most GCAs used today have some form of a shell. Unfortunately, there has been very little study on encapsulated microbubbles. An early work by Fox and Herzfield (1954) discussed the issue that gas bubbles acting as cavitation nuclei are stabilized by an organic monomolecular skin that would act as an elastic shell and as a mechanical barrier to diffusion. The skin provides rigidity and, therefore, would increase the resonance frequency of the microbubbles. Microbubble stabilization by an organic skin is essential for GCAs composed of soluble gases in order to increase their persistence in the blood stream. A theoretical model describing ultrasonic propagation in a distribution of shelled microbubbles was made by de Jong et al (1992), who considered the shell surrounding Albunex microbubbles as layers of elastic solids. Recently, Church (1995) treated the surface layers by including numerical values for such parameters as the modulus of elasticity, density, viscosity and thickness.

de Jong et al (1992) hypothesized that the shell introduced an additional restoring force to the system, which would increase the resonance frequency and decrease the scattering cross section. The contribution of the shell to the bubble stiffness is given by

$${\text{S}}_{\text{shell}}=8\pi \frac{\text{Et}}{1-\text{v}}={8\pi \text{S}}_{\text{p}}$$ (6-18)

where E = shell elasticity, t = wall thickness, v = Poisson ratio, and S_p = Et/(1 −v) = the shell parameter given in dynes/cm. Although the shell elasticity, wall thickness, and Poisson ratio are unknown, the shell parameter can be estimated by matching the measured attenuation coefficient to the theory. The model has shown better agreement for large microbubbles (lower resonance frequencies) than for small ones. To improve the agreement between theory and measurements, de Jong and Hoff (1993) added an additional damping term referred to as the shell friction parameter, S_f, which accounted for the internal friction or viscosity within the shell. The damping coefficient due to the shell friction introduces an additional term to the total damping coefficient,

$${\delta }_{\text{tot}}={\delta }_{\text{r}}+{\delta }_{\text{t}}+{\delta }_{\text{v}}+{\delta }_{\text{f}}$$ (6-19)

where

$${\delta }_{\text{f}}=\frac{{\text{S}}_{\text{f}}}{\text{m}\omega }$$ (6-20)

The modified resonance frequency, taking the shell stiffness into account, is given by

$${\text{f}}_{\text{rs}}=\frac{1}{2\pi }\sqrt{\frac{{\text{S}}_{\text{a}}\text{b}\zeta +{\text{S}}_{\text{shell}}}{\text{m}}}$$ (6-21)

6.2.2.2 Cloud of Bubbles

The dynamics of individual microbubbles is of limited importance when considering the response of GCAs to an acoustic wave. A method for determining the effects of a cloud of microbubbles is needed. When the scatterer concentration is low, the scattered power is proportional to the scatterer concentration (Ishimaru, 1978). For a distribution of microbubbles of different sizes, the mean attenuation or scattering property at a single frequency, <φ>, can be calculated by considering the contribution of each microbubble divided by the total concentration:

$$\langle \phi (\text{f})\rangle =\frac{{\int }_0^{\infty }\phi (\text{R},\text{f})\text{n}(\text{R})\text{dR}}{\text{N}}$$ (6-22)

where R = radius of the scatterer, f = ultrasound frequency, n(R)dR = number of bubbles of radius between R and R + dR per unit volume of scattering medium, φ(R,f) = the attenuation or scattering property of a scatterer with radius R, and N = total concentration of scatterers.

6.2.2.3 Attenuation Coefficient of Albunex

Figure 6-4 shows measured results of the attenuation coefficient of Albunex as a function of the ultrasound frequency (Chang and Shung, 1993). Experimental results obtained using a broadband approach (Bleeker et al, 1990b; de Jong et al, 1992; Chang et al, 1995, 1996) agree well with the theory developed for encapsulated microbubbles discussed above, especially in the vicinity of microbubble resonance. The size distribution of Albunex shown in Figure 6-5 was used to calculate the theoretical curve. The attenuation coefficient peaks at 2.1 and the values of S_p and S_f were assumed to be 5,500 dyne/cm and 0.004 g/s, respectively. The attenuation coefficient of Albunex at different concentrations has also been measured (Marsh et al, 1997).

Figure 6-4.

Experimental and theoretical results of attenuation coefficient of an Albunex suspension as a function of frequency. Mean diameter of Albunex = 3.15 μm.

Figure 6-5.

Size distributions of Albunex and FS069.

6.2.3 Some Specific Contrast Agents

The first GCAs used for research and clinical purposes were handmade materials produced immediately before administration. Many different materials were tried; e.g., agitated saline, x-ray contrast media, or combinations of the two (Tei et al, 1983; Kaul et al, 1984). These procedures trapped air in suspension, but the microbubbles were unstable, often varied dramatically from batch to batch, had wide size distributions, and failed to cross the pulmonary circulation after intravenous injection. More sophisticated methods of making GCAs evolved that led to commercially prepared products (Keller et al, 1986; Feinstein et al, 1989). These GCAs are easier to use than the “handmade” materials, demonstrate less lot-to-lot variability, provide an excellent pharmacological safety profile and, after intravenous injection, consistently produce contrast in the left ventricle and enhance the Doppler signals from the arterial circulation. Many different GCAs are either on the market or are in development as of this writing.

Echovist (SHU 454; Schering AG, Berlin, Germany) was the first commercially available GCA. This material starts as a powder of desiccated galactose microparticles <12 μm in diameter (3.5 μm median diameter; Fritzsch et al, 1988) that are suspended or dissolved in an aqueous galactose solution. Air attached to or trapped within the particles produces microbubbles with a median diameter of 3 μm, 97% of which are <7 μm in diameter. Echovist microbubbles do not cross the pulmonary circulation after intravenous administration and thus are limited to use in right heart or intralumenal studies; e.g., in evaluation of fallopian tube patency (Smith MD et al, 1984; Venezia and Zangara, 1991). A closely related product is Levovist [SHU 508 (A), Schering AG], which starts as desiccated microparticles of galactose and 0.01% palmitic acid (Fritzsch et al, 1990; Schwarz et al, 1994). The surfactant activity of the fatty acid stabilizes the microbubbles and facilitates their passage through the pulmonary circulation (Smith MD et al, 1989). Intravenous injection of Levovist increases the backscatter of the blood pool, opacifies the left ventricle of the heart, and enhances Doppler ultrasound signals from abdominal organs (Schlief et al, 1990; Goldberg et al, 1993b). Levovist is approved for use in Europe.

Albunex (Molecular Biosystems, Inc., San Diego, CA) is the first GCA approved by the Food and Drug Administration (FDA) for use in the United States. Controlled sonication of 5% human serum albumin produces stable, air-filled microbubbles (Barnhart et al, 1990). The preparation possesses a microbubble concentration of 3–5 × 10⁸/mL, with a mean microbubble diameter of 3–5 μm, more than 95% of which are <10 μm in diameter. These possess a denatured albumin shell ~15 nm thick (Christiansen et al, 1994). Intravenously injected Albunex traverses pulmonary capillaries, increases left ventricular contrast, and enhances arterial Doppler signals (Keller et al, 1987; Jakobsen et al, 1996). Left ventricular opacification by Albunex improves endocardial border definition in echocardiograms and thus facilitates evaluation of wall motion and estimation of ejection fraction (Feinstein et al, 1990; Crouse et al, 1993).

A GCA undergoing clinical trials in Europe is BY963 (Bracco-Byk Gulden, Konstanz, Germany). Air bubbles are surrounded by a phospholipid (3-SN-phosphatidyl-D,L-glycero-disteroyl-Na [DSPG-Na]; Solleder et al, 1996). The lyophilized material is rehydrated and then passed back and forth several times through a small chamber to induce microbubble formation. This agent possesses a gas volume of 40 μL/mL and 1.7 × 10⁸ microbubbles/mL. The mean microbubble diameter is 3.8 μm, with 95% of the microbubbles smaller than 8 μm. Intravenous administration of BY963 in humans opacifies the left ventricle and enhances the Doppler signal during transcranial color Doppler imaging (Belz et al, 1994; Kaps et al, 1995).

The commercial materials discussed thus far might be considered to be “first generation” GCAs. All use air as the active agent. However, because of an inherent unsaturation of the air-gases in blood, the air contained within these GCAs readily diffuses out of the microbubbles when injected into the body (Van Liew and Burkard, 1995a).

GCAs are unstable under sonication by diagnostic ultrasound; the problem may be severe at high intensity (Mor-Avi et al, 1994; Porter et al, 1996a; Uhlendorf and Scholle, 1996). This significantly reduces the GCA effectiveness in ultrasonic imaging. A significant fraction of GCA microbubbles may be destroyed by several minutes of exposure to diagnostic ultrasound in vitro, resulting from structural alteration of the microbubble shell, leading to gas loss by diffusion and the formation of microbubble clusters due to the attractive force among nearby microbubbles in an ultrasound field (Wu and Tong, 1998a). With loss of air, the GCAs lose the ability to produce ultrasonic contrast. The next advancement made in GCAs (i.e., the second generation) was the use of gases having lower solubility and diffusivity. These GCAs retain their gas for longer periods, thus increasing the duration of contrast and Doppler enhancement from several seconds, as with first generation agents, to several minutes (Van Liew and Burkard, 1995b).

Molecular Biosystems’ second generation GCA is Optison (FS069), consisting of a proteinaceous shell surrounding a perfluoropropane bubble (Meza et al, 1996). This relatively insoluble gas is inert and is eliminated from the body via normal gas exchange. Optison is produced by controlled sonication of 1% human serum albumin in the presence of perfluoropropane. The microbubble concentration is 6.3–9.0 × 10⁸/mL and the mean microbubble diameter is 2.0–4.5 μm (Dittrich et al, 1995a). Optison opacifies the cardiac ventricular chambers at smaller doses than required for Albunex (0.2 mL versus 15–20 mL, respectively). The duration of contrast produced by Optison dramatically exceeds that of Albunex (>5 min versus 30–45 s, respectively). After intravenous administration, Optison enhances Doppler signals from abdominal and peripheral organs (Dittrich et al, 1994; Brown et al, 1996), enhances two-dimensional ultrasound reflectivity from the heart, demonstrates myocardial perfusion (Dittrich et al, 1995a, 1995b; Meza et al, 1996), and increases the echogenicity of the parenchyma from other organs (Nada et al, 1995; Aronson et al, 1996). At dose volumes much larger than those required to achieve parenchymal enhancement, Optison causes no changes in hemodynamic or blood gas measurements in anesthetized dogs (Dittrich et al, 1994, 1995a; Meza et al, 1996). Phase One human trial results demonstrate an excellent margin of pharmacological safety (Dittrich et al, 1995b). Optison is approved by the FDA for use in the United States.

Another perfluoropropane-based GCA is Aerosomes (MRX115; ImaRx Pharmaceutical, Tuscon, AZ). These microbubbles are stabilized by a phospholipid coating <10 nm thick (Unger, 1995a). The microbubble concentration is ~8 × 10⁸/mL, with a mean microbubble diameter of ~2.5 μm (Unger, 1995b). Intravenous administration of Aerosomes at dose volumes of 0.01–0.05 mL/kg provides myocardial enhancement in subhuman primates (Grauer et al, 1996). Small doses also opacify the left ventricle and enhance Doppler signals from the peripheral vasculature for prolonged periods (Unger, 1995b; Metzger-Rose et al, 1996). In anesthetized monkeys, Aerosomes produce no significant hemodynamic or blood gas changes at dose volumes of 0.05–0.10 mL/kg (Grauer et al, 1995). As of this writing, the agent is undergoing clinical trials to evaluate safety and efficacy in humans.

Imagent US (AFO150; Alliance Pharmaceutical Corp., San Diego, CA) is another GCA that utilizes a perfluorocarbon to increase stability in vivo. It is composed of surfactants, phosphate buffers, NaCl, and a blend of perfluorohexane vapors and nitrogen (Mulvagh et al, 1996). The materials start as a powder and gas mixture that is reconstituted with water (20 mg/mL). This preparation results in microbubbles with a median diameter of 6.0 μm and a concentration of ~5.0 × 10⁸ microbubbles/mL. Intravenous administration (0.5 to 2.0 mL) produces homogeneous left ventricle opacification and enhancement of the myocardial tissue in anesthetized dogs. In rabbits, Imagent (0.007 to 0.2 mL/kg ) dramatically enhances color Doppler signals from the kidney for extended periods; e.g., from ~300 to >1,000 ss, depending on the dose (Taylor et al, 1996). At dose volumes of 0.2 mL/kg, Imagent produces a modest gray scale enhancement in the renal cortex. At doses up to 40 mL in a dog, Imagent caused no changes in hemodynamic parameters (Mulvagh et al, 1996). Clinical evaluations of this agent are under way as of this writing.

Sonovue (BR1; Bracco Research SA, Geneva, Switzerland) is composed of phospholipids (di-stearoylphosphatidylcholine and dipalmitoylphosphatidylglycerol), ethylene glycol 4000 and sulfur hexafluoride gas (Schneider et al, 1995). This lyophilized material is dispersed in saline, resulting in a suspension of 2 × 10⁸ microbubbles/mL and a mean microbubble diameter of 2.5 μm, with >90% smaller than 8 μm. The preparation has a gas volume of 2–10 μL/mL. After intravenous administration, Sonovue opacifies the right and left ventricles of the heart. Doses of Sonovue above 0.025 mL/kg cause no further increase in contrast intensity but prolong the duration of contrast in the ventricles, from 27 ± 23 s to 104 ± 35 s at a dose of 0.2 mL/kg (Rovai et al, 1995). Sonovue also enhances the Doppler signals from the vasculature of abdominal organs (Schneider et al, 1996). As of this writing, Sonovue is undergoing clinical evaluation.

A unique second generation GCA is the perfluoropentane emulsion EchoGen (SONUS Pharmaceuticals, Bothell, WA). EchoGen is formulated as a liquid-liquid aqueous emulsion. Dodecafluoropentane, the active ingredient in EchoGen, undergoes a phase change as it warms from room temperature to body temperature (Quay, 1994). EchoGen emulsion contains 2% dodecafluoropentane in stabilized droplets with a diameter of ~0.3 μm and a concentration of 10¹² droplets/mL (Correas and Quay, 1996); as these warm, the liquid boils (at 28°C) to form microbubbles with a calculated final mean diameter of 2–5 μm. When injected at small doses and without preactivation (see below), EchoGen produces contrast in the ventricular chambers of the heart and enhances Doppler signals from the abdominal organ vasculature (Grayburn et al, 1995; Forsberg et al, 1995). At higher doses, EchoGen enhances the echogenicity of the myocardium and parenchyma of peripheral organs (Sehgal et al, 1995). Activated prior to injection by either sonication or hypobaric manipulation, EchoGen enhances the myocardium and parenchyma of the liver and kidneys at dose volumes one fourth and one tenth those used previously (Cotter et al, 1995; Forsberg et al, 1996). With preactivation and the use of smaller dose volumes, EchoGen causes no hemodynamic or blood gas alterations. As of this writing, EchoGen is under FDA review and is undergoing clinical trial evaluations.

The second generation GCAs discussed above are the ones that are the furthest along in their journey toward clinical utility. Other GCAs are also being developed by several companies. Their development status, physical characteristics, and imaging profiles are less well known.

6.3 Clinical Applications of Contrast Agents

6.3.1 Harmonic and Transient Mode Imaging

6.3.1.1 Harmonic Mode Imaging

Ultrasound imaging is based on the pulse-echo method of the fundamental frequency (f). An ultrasonic transducer transmits repeated ultrasonic pulses (the center frequency is f) to a target, and the echoes received by the transducer have the same center frequency. The brightness of the pixels on a monitor corresponding to the image of the target is proportional to the echo amplitude. GCAs significantly enhance the backscattered echo signals of the fundamental frequency, as the acoustic impedance of the gas-filled encapsulated microbubbles is quite different from that of soft tissue.

It is well known that a bubbly liquid is highly nonlinear. The second order nonlinearity parameter, B/A, characterizes the nonlinearity in the relationship between acoustic pressure and density. The magnitude of the equivalent nonlinearity parameter B/A of a bubbly liquid can be four orders greater than that of most soft tissue (Wu et al, 1995). Direct experimental measurements (Schrope et al, 1992; Wu and Tong, 1998b) indicated that B/A for Albunex and Levovist solutions can reach 3–4 digits; in contrast, the B/A for most soft tissues is only in the single digits. The highly nonlinear property of GCAs has been used to create a new ultrasonic imaging mode; viz., harmonic imaging. Harmonic imaging is realized by transmitting ultrasonic pulses with a center frequency f, but receiving the echoes at nf, where n is an integer greater or equal to 2 (Schrope et al, 1992; Chang et al, 1995, 1996). Since B/A of soft tissue is much smaller than that of GCAs, the contrast of the backscattered signals at the harmonic frequencies is greatly enhanced relative to the contrast obtained at the fundamental frequency. Using this technique, it is possible to detect and measure slow and small volume blood flow. This is not possible by using the current fundamental frequency imaging technique, as the echo from more prevalent tissue at the fundamental frequency dominates the echo from the blood (Burns et al, 1994b). Additionally, since the wavelength and beam width of the nth harmonic frequency are smaller than those of the fundamental by nearly a factor of 1/n (Ward et al, 1997), the axial and lateral spatial resolutions of the image are improved.

6.3.1.2 Transient Mode Imaging

Although newer generation GCAs improve the amount of myocardial contrast produced from an intravenous injection of contrast agent, the doses required to produce this contrast are very large and cause acoustic shadowing of myocardium in the near field of insonation (Carstensen et al, 1992; Villanueva et al, 1992; Porter et al, 1995b). Harmonic imaging significantly improves the signal to noise ratio by taking advantage of the nonlinear reflective characteristics of microbubbles (Villanueva et al, 1993; Schrope and Newhouse, 1993; see also subsection 6.3.1.1. above).

A second method of improving myocardial contrast from an intravenous administration of microbubbles is by reducing their exposure to ultrasound. The peak negative acoustic pressures produced by diagnostic ultrasound transducers, which range from 0.5 to 3 MPa (de Jong et al, 1991; Patton et al, 1994), can destroy air-filled as well as fluorocarbon-filled microbubbles (Wray et al, 1992; Mor-Avi et al, 1994; Vandenberg and Melton, 1994). GCA destruction by ultrasound can involve inertial cavitation (Crum et al, 1992; Everbach et al, 1996b); indeed, the rapidity of microbubble destruction (and decrease in video intensity) is related directly to the magnitude of the peak negative pressure (Vandenberg and Melton, 1994). Less frequent exposure (i.e., lower frame rates) has been shown to reduce microbubble destruction rates (Porter et al, 1997).

Transient mode imaging was first observed in animal studies when the imaging ultrasound was frozen for 30–60 s after an intravenous injection of perfluorocarbon-exposed sonicated dextrose albumin microbubbles. Even when using conventional (i.e., >30 Hz) frame rates, a marked increase in myocardial contrast was observed in the first few frames after turning the freeze button off. Often, this increase in contrast was so transient that the only means to observe it was to search the analogue videotape and view each individual frame obtained after turning the freeze button off. Because of the short duration of contrast obtained when using conventional imaging frame rates, this type of imaging was initially termed “transient response” (or “transient mode”) imaging. An equivalent increase in myocardial contrast can be observed by transmitting ultrasound at only one point triggered to every one (or several) cardiac cycles after the intravenous GCA administration.

Even with a high diagnostic peak negative pressure (e.g., 1.1 MPa), a reduced frame rate (e.g., 1 Hz) destroys significantly fewer microbubbles than a conventional frame rate (Porter et al, 1996b, 1997). Some of the myocardial contrast improvement obtained with transient mode imaging is actually due to the enhancement of cavitation activity produced under these conditions (Porter et al, 1998a). Upon initial exposure to diagnostic ultrasound pressures, perfluorocarbon-exposed microbubbles have exhibited cavitation activity (Everbach et al, 1996b). The magnitude of this activity varies as a function of both the ultrasound pulse repetition frequency and pulse duration. Interestingly, the greatest initial cavitation activity occurs in response to the lowest pulse repetition frequencies, presumably because of less rapid depletion of gas nuclei. Inertial cavitation activity upon initial exposure to diagnostic ultrasound pressures has also been observed with sonicated albumin-coated microbubbles (Crum et al, 1992). The duration of this transient growth and collapse of the latter microbubbles was even shorter than that observed with perfluorocarbon exposed microbubbles. However, transient mode imaging results in sufficient enhancement of myocardial contrast to allow the use of relatively low acoustic pressures. Porter et al (1997) have shown that the threshold for producing an increase in contrast with transient mode imaging is well below commonly used diagnostic ultrasound acoustic pressures. Even at 0.4–0.5 MPa, contrast increases can be observed with triggered imaging. However, this lower peak negative pressure is not nearly as destructive of the microbubbles as is 0.7–0.8 MPa when using frame rates of 10–15 Hz (Porter et al, 1997), and the operator can still see increased myocardial contrast at frame rates that permit simultaneous assessment of wall thickening (Porter et al, 1998b).

The dramatic impact this has on the amount of contrast produced from intravenously injected microbubbles is demonstrated in Figure 6-6. The image on the left is from a patient during a continuous infusion of perfluorocarbon-exposed sonicated dextrose albumin microbubbles during standard, conventional imaging at 30–40 frames/s. The right side of Figure 6-6 is an image from the same patient during the same continuous infusion of microbubbles, but in this case the frame rate had been reduced to one frame in every two cardiac cycles. The large increase in contrast within the myocardium is obvious.

Figure 6-6.

An example of the increase in myocardial contrast observed during continuous infusion of PESDA microbubbles when using a low frame rate (right panel) as opposed to conventional frame rates (left panel); the latter are usually > 30 Hz.

6.3.2 Cardiovascular Applications

The contrast echocardiography effect was described by Gramiak and Shah (1968), who observed a “cloud” of echoes during injection of indocyanin green dye for performance of a dye curve in a catheterization laboratory. Contrast echocardiography has since been used for structure identification, identification of intracardiac and intrapulmonary shunts, valvular regurgitation, Doppler enhancement, and myocardial perfusion imaging (Jayaweera et al, 1994; Grayburn et al, 1995; Porter et al, 1995a, 1995c). Early developments in contrast echocardiography are reviewed elsewhere (Meltzer and Roelandt, 1982a; Meerbaum and Meltzer, 1989).

One of the original uses of contrast echocardiography was for the detection of intracardiac and intra-pulmonary shunts, which remains the most common use for contrast; e.g., quantitation of shunts in atrial septal defect (Okura et al, 1995). For such investigations, agitated saline or dextrose is usually adequate as the GCA, and an intravenous injection is made to search for right-to-left shunting. Contrast echocardiography is particularly important in pediatric cardiology.

Myocardial perfusion imaging by contrast echocardiography was demonstrated in animals in 1982 (Armstrong et al, 1982; Meltzer et al, 1982b) and in humans in 1985 (Santoso et al, 1985). At first, investigators noted the “geographic” extent of contrast distribution in the myocardium and related this to the coronary distribution area. Many subsequent studies of myocardial contrast focused on the “kinetics” of contrast wash-in and wash-out from the myocardium or in the cardiac chambers. These video density curves (or sometimes integrated backscatter curves) are similar to more traditional indicator dilution curves. However, “geographic” studies of contrast distribution within the myocardium are more promising than those from many myocardial “kinetic” contrast studies. Myocardial contrast echocardiography can help delineate areas at risk within coronary territories; i.e., the area likely to infarct related to the occlusion of a coronary artery in experimental animal models. It can be used to delineate the extent of myocardium that has an effective collateral circulation (Sabia et al, 1992a, 1992b); i.e., the area of myocardium opacified by each of two separate injections—one into the right and one into the left coronary artery—is presumably functionally perfused by flow from both coronary arteries. Such myocardium is less likely to infarct and more likely to be viable after an infarction than myocardium not perfused by dual circulation (Galiuto et al, 1994; Camarano et al, 1995). Another important effect of microvascular physiology on gross left ventricular function relates to the “no reflow” phenomenon seen after reperfusion in the setting of acute myocardial infarction (Ito et al, 1992).

Contrast echocardiography is also used to enhance endocardial border definition during stress echocardiography and in technically difficult cases. It can enhance Doppler signals as well as improve two-dimensional (2D) and M-mode imaging. Contrast material can opacify the left ventricle (Sonne et al, 1995) and has been used to enhance weak color Doppler signals, such as those in diastole and in the left ventricular apex. GCAs may prove to be useful for the measurement of ejection fraction.

The ultimate market for contrast echocardiography is still unclear, but new applications and contrast agents designed for these applications will likely continue to evolve; e.g., second and third generation agents that may allow myocardial perfusion imaging after intravenous injections. There is also an exciting possibility that GCAs may enhance therapeutic ultrasound applications involving acoustic cavitation, such as accelerating thrombolysis and perhaps ultrasound angioplasty (Meltzer et al, 1986, 1991; Kornowski et al, 1994; Makin et al, 1995).

6.3.3 Obstetric and Gynecologic Applications

Although GCAs are gaining increased acceptance in abdominal or cardiac applications, their use in obstetrics and gynecology is still limited (Abramowicz, 1997). The reason for this is obvious in obstetrics: GCA manufacturers have not attempted to obtain FDA approval for use of their products in pregnancy. Nonetheless, GCAs offer advantages in placental imaging. In experimental settings, two GCAs, iodipamide ethyl ester and Albunex, improved placental imaging and visualization of maternal and fetal placental blood flow (Panigel et al, 1996; Abramowicz et al, 1996). The GCAs permitted color demonstration of flow in fetal capillaries otherwise not resolved by gray-scale, color, or Doppler ultrasound. This could be used to better delineate placental function, e.g., in cases of restricted fetal growth. In vivo, conventional ultrasound does not differentiate between areas of adequate or reduced perfusion. In the future, GCAs may allow this type of analysis of placental physiopathology.

As of this writing, the only published report on the use of GCA in obstetrics is a case report of twins with unclear chorionicity. Levovist was injected into the circulation of one twin and observed appearing in the second twin, thus confirming monochorionicity (Denbow et al, 1997).

In gynecology, the situation is very different. Adequate visualization of the uterus and ovaries is possible by gray-scale ultrasound, particularly via the transvaginal approach. However, demonstration of the fallopian tubes is possible only in the presence of peritoneal-abdominal fluid, and then only with difficulty. Intrauterine cavity pathologies are also difficult to discern because there is apposition of the uterine cavity walls, except in pregnancy or during menses when blood is present. Furthermore, the uterus is a solid tissue and so are anomalies of the uterine wall (myometrium) or uterine cavity lining (endometrium). Since tissue differentiation by ultrasound is still less than ideal, the addition of a GCA may improve diagnostic capabilities.

Use of a contrast medium in the uterine cavity was first reported by Richman et al (1984). A 32% solution of Dextran 70 was used to distend the uterus under abdominal ultrasound for tubal patency investigation. This medium was used in another study in which 10 anomalies that had been missed by conventional gray-scale ultrasound were detected in 21 “normal” uteri (Van Roessel et al, 1987). A logical approach would be to inject a clear, neutral, non-irritant material. Normal saline is the natural choice, described originally in 1988 in a study of 30 patients with sterility problems, menstrual irregularities, or suspected tumors (Deichert et al, 1988). The procedure has since been extensively studied under different names: contrast ultrasonography or echography (Crequat et al, 1993), hystero-contrast-sonography or Hy-Co-Sy (Deichert et al, 1989; Campbell et al, 1994), hysterosalpingosonography (Bonilla-Musoles et al, 1992), or sonohysterography, echo-hysterosalpingography (Venezia and Zangara, 1991), or hydro-gynecography (Maroulis et al, 1992). The diagnostic accuracy is excellent (Bonilla-Musoles et al, 1992; Parsons and Lense, 1993; Gaucherand et al, 1995; Goldstein, 1996). Evaluation of the postmenopausal uterus with its usually atrophic endometrium is facilitated by introduction of saline (Achiron et al, 1995). Although transvaginal ultrasound is the usual scanning method, abdominal ultrasound (with the addition of saline) has also yielded good results (Cicinelli et al, 1994, 1995). Refinements of these methods include modification of saline by the addition of air bubbles (Allahbadia, 1992) and the use of actual GCAs.

Saline or a contrast medium can be injected into the uterine cavity to study anomalies of the uterus (Fujiwaki et al, 1995). Sonographic examination of the fallopian tubes is facilitated by natural fluid collections in the fallopian tubes and peritoneal cavity (Tufecki et al, 1992; Yarali et al, 1994; Degenhardt et al, 1995). The diagnostic accuracy is comparable to that of hysterosalpingography (Mitri et al, 1991). More sophisticated sonographic technology can also be used for further delineation of the anatomy or pathology; e.g., 2D or color Doppler for detection of fluid flow through the fallopian tubes (Deichert et al, 1989; Peters and Coulam, 1991), as well as three-dimensional (3D) ultrasound.

A recent report described the following additional potential for GCAs in gynecology: After injection of Levovist, blood flow was demonstrated in small vessels of ovarian tumors (Suren et al, 1994). Flow velocity is extremely low in these vessels and, therefore, beyond the limit of resolution of conventional color or Doppler imaging.

The application of ultrasound contrast media in obstetrics and gynecology is a recent phenomenon, but an extensive literature already exists on gynecologic applications of contrast, including detailed descriptions of methodology and an atlas of anatomical and pathological findings (Cullinan et al, 1995; Parsons et al, 1996). It seems that the use of ultrasound contrast media in gynecology (and perhaps in obstetrics) can only expand in the future.

6.3.4 Other Applications

GCAs have been used to enhance sonographic contrast in a variety of other applications. A few recent examples are listed here. In urogynecology, Echovist was instilled into the bladder, leading to improved imaging of the bladder neck. Known pathology present in 39 patients was demonstrated in 38 of these patients when examined using GCA-enhanced ultrasound, but in only 19 of the same patients when similarly examined without GCA (Schaer et al, 1995). In breast ultrasound, Doppler signals were enhanced after injection of GCA. Transit of the GCA was prolonged in malignant lesions as compared with benign masses. Detected vessel number and “tortuosity” indications were also increased. This resulted in 100% sensitivity and specificity in 34 patients (Kedar et al, 1996). GCAs have been used to enhance the detection of brain gliomas (Simon et al, 1992b), cancerous liver growths (Kudo et al, 1992a, 1992b, 1994; Nomura et al, 1993), and ophthalmic (Cennamo et al, 1994) and renal tumors (Forsberg et al, 1995). Contrast has been used to guide peri-cardiocentesis (Chiang and Lin, 1993), to visualize renal perfusion (Porter et al, 1995b), for the detection of venous thrombosis (Coley et al, 1994), for the enhancement of Doppler signals from large and small blood vessels (Goldberg et al, 1993b), and blood flowmetry (Shung and Flenniken, 1995).

GCAs have been used in limited therapeutic applications for the intentional enrichment of tumors with gas nuclei followed by the induction of cavitation and resultant tissue destruction (Prat et al, 1993; Simon et al, 1993), and in accelerating thrombolysis, apparently via a cavitation-related mechanism (see, e.g., Tachibana and Tachibana, 1995; Porter et al, 1996c). These are discussed in subsection 6.4.

6.4 Bioeffects

6.4.1 Introduction

Acoustic cavitation is defined as “any interaction between an ultrasound field and any gaseous inclusion in the medium” (Miller DL and Thomas, 1995b), and they note that by this definition, “cavitation occurs … whenever bubble-based contrast agents are exposed to ultrasound.” The pulsations of air-filled (Schrope et al, 1992; Chang et al, 1995) and perfluorocarbon-enhanced (Chang et al, 1996; Krishna and Newhouse, 1997) GCAs in an ultrasonic field have nonlinear characteristics, including the generation of scattered signals at harmonic frequencies relative to the applied ultrasound field; the second harmonic is exploited in Doppler ultrasound imaging. As will be developed, GCAs or the derivative microbubbles formed upon their ultrasonic modification can also nucleate inertial cavitation, which can increase the signal strength for imaging (Uhlendorf and Hoffmann, 1994), but can also produce biological damage. In many of the studies discussed here, the acoustic pressures used were sufficiently large to ascribe the observed bioeffects to inertial cavitation. In others, the involvement of inertial cavitation is indicated by physical measures of cavitation activity (see, e.g., Miller DL and Bao, 1998a). Bioeffects associated with inertial cavitation can result from the mechanical forces generated by bubble collapse, shock waves generated by bubble rebound, or sonochemical activity.

6.4.2 Effect of Contrast Agents on Cavitation Nucleation and Cavitation Thresholds

Holland and Apfel (1990) explored the effect of Albunex on the threshold for inertial cavitation. Highly filtered water was used as a host fluid and control; filtration removes many of the endogenous gas nuclei. A passive cavitation detector, based on broad-band scattering of acoustic energy from cavitation microbubbles, was used. With 10 μs pulses of ~0.8–2.3 MHz ultrasound, the threshold for inertial cavitation in filtered water ranged from 1.94–2.43 MPa. A small amount of Albunex reduced the threshold for inertial cavitation at 0.757 MHz to 0.52–0.64 MPa; i.e., by a factor of ~1/3. The threshold for inertial cavitation in control preparations containing filtered, 5% human albumin was similar to that of filtered water. Miller and Thomas (1995a) also studied the ability of GCAs to nucleate inertial cavitation. Sonochemical production of H₂O₂ (a product of inertial cavitation) was the endpoint. Filtered saline with or without either Levovist or Albunex were exposed for 5 min to pulsed 2.17 MHz ultrasound, followed by assay for accumulated H₂O₂. At 0.82 MPa, no measurable H₂O₂ was produced in the filtered saline. With either Levovist or Albunex present, detectable H₂O₂ production was associated with a pressure amplitude of ~0.4 MPa. At higher ultrasound frequencies, the threshold for inertial cavitation was greater than at lower frequencies for filtered saline and filtered saline containing either Albunex or Levovist. At 2.95 MHz, the threshold for H₂O₂ production in the fluids containing GCA was about 0.6 MPa, and at 2.17 MHz, the threshold was about 0.4 MPa, as noted above. In whole human blood exposed to 2.5 MHz ultrasound, the pressure threshold for actively detected inertial cavitation activity is on the order of 4–5 MPa (Deng et al, 1996); with GCAs present, the pressure threshold for hemolysis in whole blood is on the order of 0.5 MPa (see below).

These results make it clear that GCAs significantly lower the threshold acoustic pressures required for the inception of inertial cavitation and may increase the extent of such activity relative to non-nucleated or poorly nucleated fluids. These findings have potentially important consequences to the occurrence of mechanical, ultrasound-induced bioeffects. As discussed below, mechanical bioeffects of ultrasound exposure may be potentiated by GCAs. However, a recent microscopic study of Albunex microbubbles adherent to a petri dish and exposed to the 5 MHz output of a Hewlett Packard Sonos 500 medical imager indicates that at maximum output power, the microbubbles slowly diminished in size, but did not appear to undergo inertial collapse (Klibanov et al, 1998).

6.4.3 Potentially Undesirable Bioeffects

6.4.3.1 Simple Model Systems

As of this writing, most of the research on the enhancement of mechanical ultrasound-induced bioeffects by GCAs has been conducted using in vitro systems. Because the results obtained in these experiments often depend critically on the methods used (e.g., the use of stationary versus rotating exposure vessels), the inclusion of some significant experimental detail in this subsection is unavoidable.

The published report of which we are aware concerning a GCA-enhanced bioeffect is that of Williams et al (1991). Dilute suspensions of human erythrocytes in plasma were used to explore the hemolytic effect of ultrasound exposure in the presence of Echovist. Most samples were exposed for 5 min to continuous wave, 0.75 MHz ultrasound in a rotating vessel, although some were exposed in stationary vessels. Echovist was added to experimental samples at concentrations ranging from 5–10 mg/mL. The volume fraction of red cells in suspension was also varied. In very dilute cell suspensions (0.5% volume fraction or hematocrit) exposed with vessel rotation to a spatial average, temporal average intensity (SATA) of 0.5 W/cm², ~33% of the cells lysed in the absence of Echovist. With 8 mg Echovist/mL present, hemolysis increased to ~60%. In other dilute suspensions, there appeared to be a dose:response relationship between Echovist concentration and percent hemolysis. Below ~5 mg/mL, the fraction of cells lysed in the presence of Echovist was comparable to that in the controls, while at Echovist concentrations of ≥8 mg/mL, hemolysis was ~twofold greater. Exposure vessel rotation was found to be an important variable in determining the extent of hemolysis, which was greater with than without rotation, both with and without GCA present. However, the relative enhancement of hemolysis by the GCA was much greater in the stationary than in the rotated vessels.

Williams et al (1991) observed that with or without GCA, the fraction of cells lysed by the ultrasound decreased as the hematocrit of red cells in suspension increased. No lysis was detectable in most samples having hematocrits >5%. In canine red cell suspensions prepared with variable hematocrits and Albunex concentrations and exposed to 2.25 MHz ultrasound (1.6 MPa, 1 s continuous wave), hemolysis increased monotonically with increasing hematocrit at constant concentration, and with increasing concentration at constant hematocrit (Miller DL et al, 1997). No hemolysis was observed without added Albunex. The observation that the extent of cell lysis appears to decline to immeasurably low levels with increasing cell concentration has often been noted (Veress and Vincze, 1977; Saad, 1983; Williams, 1983; Ellwart et al, 1988; Brayman et al, 1992; Carstensen et al, 1993; Miller MW et al, 1995), suggesting that high cell concentrations either inhibit or eliminate inertial cavitation activity, and that ultrasonic hemolysis in vivo is unlikely. However, Miller et al (Miller MW et al, 1995) noted that while the percentage of erythrocytes lysed by sonication in the presence of Albunex decreased with increasing cell concentration, the number of cells lysed per sample remained more or less constant. Similar results were obtained in a study of cell lysis produced by stable microbubble pulsations (Miller DL, 1988). A subsequent analysis (Brayman et al, 1996b) indicates that in vitro sonolysis of cells at high cell concentrations is limited by the number of microbubbles available, or by the number of cells a microbubble may encounter before being “inactivated” by cell aggregation around pulsating microbubbles, but that hemolysis might be produced by sonication of dense cell suspensions supplemented with GCAs. Although Williams et al (1991) reported that hemolysis in blood cell suspensions declined to undetectable levels as hematocrit increased, regardless of the presence or absence of a GCA, the results of other studies conducted with GCAs and human erythrocyte suspensions of high hematocrit indicate that readily detectable hemolysis can be produced by sonication of whole blood with GCA present.

Miller et al (Miller MW et al, 1995) studied ultrasonic hemolysis of human erythrocytes of 1%–20% hematocrit using Albunex at concentrations of 0–41 μL/mL. Cell suspensions were exposed or sham-exposed to CW ultrasound (1 MHz with a spatial peak, temporal peak intensity (I_SPTP) of 1–5 W/cm² (SPTP intensity) for 60–120 s. Exposure vessel rotation was also explored as an experimental variable. Most experiments used samples prepared from a stock suspension of red cells that had been anticoagulated with CPDA (citrate-phosphate-dextrose-adenine) and stored under refrigeration, although some experiments were conducted using freshly collected heparinized blood. In 1% hematocrit cell suspensions prepared from stored blood and insonated with vessel rotation, sonication alone (5 W/cm², 60 s) produced 28% hemolysis, versus 1% hemolysis in sham-exposed controls. The addition of Albunex to insonated samples increased the level of hemolysis to ~50%. There was no apparent dose:response relationship between Albunex concentration (1.3–41 μL/mL) and hemolytic yield. There was a dose:response relationship between ultrasound intensity (0–5 W/cm²) and hemolysis. A significant elevation of hemolysis, relative to sham-insonated controls, was produced in samples insonated at intensities of ≥ 1 W/cm², regardless of the presence or absence of Albunex, but significantly greater levels of hemolysis were produced with GCA present.

The hematocrit of stored blood samples affected cell lysis, with or without Albunex present; percent hemolysis decreased generally as hematocrit increased over the range of 1%–20%. When samples were rotated during exposure, significantly greater levels of hemolysis were produced with Albunex than without Albunex at all hematocrits <10%; at 20% hematocrit, the difference was nominal. Similar results were obtained using stationary exposure vessels, although the levels of hemolysis produced in the stationary vessels were less than in rotating vessels. Quite different results were obtained when freshly drawn blood was used. Whereas exposure of 20% hematocrit suspensions of stored blood cells to 5 W/cm² ultrasound with vessel rotation lysed ~10% (without Albunex) to ~20% (with Albunex) of the cells, when fresh blood was used, comparable exposures produced very low levels of hemolysis (~0.5%) that did not differ significantly among cells exposed with or without Albunex or sham-exposed to ultrasound. It seems probable that the disparity of results obtained with stored or fresh blood relates to the level of sample aeration. Sample oxygenation increases cavitation activity in “normal” media (Kondo and Kano, 1987, 1988b; Inoue et al, 1990; Riesz and Kondo, 1992; Brayman et al, 1992; Carstensen et al, 1993), but the samples used in these experiments were not intentionally oxygenated. The enhancement of ultrasonic hemolysis by Albunex was later shown to depend strongly on the oxygenation state of the sample (Azadniv et al, 1995). Interestingly, Williams et al (1991) had noted that the ability of Echovist to nucleate cavitation activity was abolished when the agent was added to degassed plasma. Presumably, these effects arise by dissolution of the microbubbles in undersaturated fluids.

Brayman et al (1995) further explored the issue of ultrasonic hemolysis in the presence of 35 μL/mL Albunex. Human erythrocyte suspensions were prepared in autologous plasma to hematocrits of 1%–40% using refrigerated stock suspensions. Samples were exposed for 60 s to pulsed 1.1 MHz ultrasound (1 ms pulse duration, 20 Hz pulse repetition frequency). The peak acoustic pressures of the pulses were 4.7 MPa (peak positive) and 2.7 MPa (peak negative), corresponding to an I_SPPA of ~420 W/cm². Without Albunex, statistically significant levels of ultrasound-induced hemolysis, relative to sham-exposed controls, were observed only at the lowest tested hematocrit (1%), while significant levels of hemolysis were observed at all tested hematocrits when Albunex was present in exposed samples. With Albunex, percent hemolysis declined as sample hematocrit increased, but the total number of cells lysed per sample remained approximately constant over the 5%–40% range of hematocrits. At 40% hematocrit (equivalent to whole blood), the background level of hemolysis in sham-exposed samples was 2.4 ± 0.4%. The level of hemolysis in samples insonated with or without Albunex were 6.7 ± 1.3 versus 2.8 ± 0.4%, respectively. For these samples, hemolysis in the presence of Albunex was significantly greater than in either the sham-exposed control or the insonated, Albunex-free preparations. Hemolysis in the shams and the Albunex-free, insonated samples were statistically indistinguishable. Thus, there was a twofold increase in hemolysis associated with insonation in the presence of Albunex.

A subsequent study (Brayman et al, 1996b) examined the acoustic pressure and pulse duration dependence of hemolysis produced by ultrasound in the presence of Albunex. Freshly collected, oxygenated human blood containing 3.6 V% Albunex was exposed to pulsed 1 MHz ultrasound with sample rotation. Acoustic pressures ranged from 0 to 7.4 MPa (peak positive), 4.0 MPa (peak negative); I_SPTP ranged from 0 to ~1080 W/cm². The duty factor was 0.01, and pulse durations varied between 5 μs–1,000 μs. Statistically significant levels of background-corrected hemolysis were observed in all insonated samples, including those exposed to the lowest acoustic pressure or intensity at the shortest pulse duration tested (0.5 MPa, 5 μs pulses). At constant pulse duration, there was a dose:response relationship between ultrasound intensity and hemolysis. At constant intensity, hemolysis increased generally with increasing pulse duration, but at relatively high intensities, this trend was not monotonic. The hemolysis data for relatively brief pulses (5–30 μs) were analyzed in relation to the MI values associated with the various exposures. Although statistically significant levels of hemolysis were produced at an MI of 0.5 for each of these pulse durations, the level of hemolysis was generally low (~0.5%–1.0%). Hemolysis increased slowly with increasing MI up to a value of ~2, and then increased sharply. The point at which this sharp inflection appeared to increase as pulse duration decreased. Since most diagnostic pulses are on the order of 1 μs in duration, this observation is encouraging from the perspective of the potential for diagnostic ultrasound exposures in the presence of GCAs to produce little or no hemolysis in vivo.

The frequency dependence of hemolysis produced by pulsed ultrasound exposure in the presence of 3.5 V% Albunex was investigated using erythrocyte suspensions prepared as described above (Brayman et al, 1997b). Peak negative acoustic pressures ranged from 0.0 to ~3.0 MPa, the ultrasound frequencies were 1.0, 2.2, or 3.5 MHz, pulse durations were 20 (see Figure 6-7) or 200 μs, the duty factor was 0.01, and total treatment time was 60 s. Hemolysis increased with increasing acoustic pressure at each frequency and depended weakly on pulse duration. At low peak negative acoustic pressures, the frequency dependence of ultrasonic hemolysis was relatively weak, but it was very strong at high pressures (e.g., at 0.5 MPa, the data appear to follow a P–/f^0.5 relationship, whereas at 3 MPa, the data appear to follow a P–/f³ or P–/f⁴ relationship, where P– is the peak negative acoustic pressure and f is the ultrasound frequency). Because most diagnostic ultrasound equipment emits ultrasound in the 2–7 MHz range and inertial cavitation-related hemolysis in vitro (and in vivo; see below) declines very rapidly with ultrasound frequency at constant acoustic pressure, it therefore appears that large-scale ultrasound-induced hemolysis in vivo is unlikely under diagnostic examination conditions. However, Brayman et al (1997b) estimate that under worst case conditions during an echocardiographic examination in the presence of Albunex, it might be possible to lyse ~1% of the systemic erythrocyte population in the body. Dalecki et al (1997d) have demonstrated with an animal model that ultrasound exposures with GCA present in the vasculature can produce systemic levels of hemolysis on the order of 1% (see subsection 6.4.3.2). Killam et al (1998) detected no hemolysis in vivo due to diagnostic insonation of rabbit hearts following administration of Optison. However, the exposure conditions were such that hemolysis would probably not be expected to occur (see subsection 6.4.3.2).

Figure 6-7.

Hemolysis of whole human blood containing 3.6 V % Albunex and exposed with sample rotation to 20 μs pulses of 1.0, 2.2, or 3.4 MHz ultrasound. The duty factor was 0.01 and the total exposure duration 60 s. (Data from and reprinted by permission of Elsevier Science, from Brayman AA, et al: Hemolysis of 40% hematocrit, Albunex-supplemented human erythrocyte suspensions by intense pulsed ultrasound: Frequency, duty factor, pulse length and sample rotation dependence. Ultrasound Med Biol 23:1237, 1997b.)

Miller and Thomas (Miller DL and Thomas, 1996a) studied the effect of Albunex on hemolysis in whole canine blood of ~50% hematocrit. A lithotripter (peak positive pressure ~24 MPa, peak negative pressure ~5 MPa) and a focused 1.3 MHz ultrasound source were used. The focused ultrasound was applied in continuous or burst mode, with burst durations of 20, 100, or 1,000 μs, and with pressure amplitudes ranging from 1 to ~18 MPa. Albunex was added to some samples at final concentrations of 0.1, 1, or 10 V%. The presence of Albunex in samples exposed to the lithotripter shocks decreased the number of shocks required to observe greater hemolysis than in the shams. Without Albunex, 500 shocks were necessary; with 1% or 10% Albunex, 200 and 100 shocks, respectively, were required. However, while there was a nominal twofold increase in hemolysis associated with the presence of 10% Albunex in the shock-exposed samples, this difference was not statistically significant, leading the authors to conclude that the shock waves were capable of reliably nucleating inertial cavitation without the need for exogenous nuclei. With continuous 1.3 MHz ultrasound, the presence or absence of Albunex markedly affected hemolysis. The threshold acoustic pressure required for significant hemolysis varied with exposure duration and the presence or absence of Albunex. For example, without Albunex, the threshold pressure was > 17.8 MPa when the exposure duration was 1 ms, 17.8 MPa at 100 ms, and 10 MPa with 10 s exposure. With 1% Albunex present, the threshold acoustic pressure was ~10 MPa with 100 and 1,000 ms exposures, and 3.2 MPa with 10 s exposure. Thus, 1% Albunex reduced the apparent threshold for hemolysis by a factor of two or more. With 100 ms exposures, the threshold pressure with 0.1% Albunex was the same as for the Albunex-free control (viz., ~18 MPa); with 1% or 10% Albunex, the threshold pressures were 10 MPa. Thus, the 0.1% Albunex was not effective in reducing the threshold pressure for hemolysis relative to control blood, and 10% Albunex was no more effective in reducing the threshold pressure than was the 1% concentration. Additional experiments were conducted using 1.3 MHz ultrasound applied in burst mode, with a total “on” time of 100 ms and a constant duty factor of 0.01. The threshold acoustic pressures were similar to those noted above for the continuous wave exposures; as before, 0.1% Albunex had no effect on the threshold pressure for hemolysis, while 1% and 10% Albunex were equivalently effective in lowering the threshold to ~10 MPa. At very high acoustic pressures (e.g., ~18 MPa), hemolysis occurred in all insonated samples, regardless of the presence or absence of Albunex, with the exception of those exposed to 20 μs bursts without Albunex, in which there was none. Hemolysis increased with increasing burst length (20, 100, or 1,000 μs), and was greater with Albunex than without at suprathreshold acoustic pressures.

Albumin-stabilized microbubbles are acoustically labile (subsection 6.3), and there is evidence that ultrasonic modification of these GCAs creates “derivative” gas bodies capable of nucleating inertial cavitation. With highly concentrated Albunex, different levels of hemolysis are produced by pulsed or continuous wave, 2.25 MHz ultrasound exposures of constant total “on” time, with greater hemolysis produced by the continuous wave exposures (Miller DL et al, 1997). At 1.6 MHz, continuous wave bursts of 0.001–100 s produced statistically significant levels of hemolysis relative to shams. With pulsed exposures of 1 s total “on” time but variable duty factors, hemolysis declined rapidly as duty factor decreased. These results suggest that (1) ultrasound modifies the GCA, producing free microbubbles that then serve as cavitation nuclei; (2) with continuous wave exposure, the biological effect is expressed quickly; and (3) without continuous exposure to replenish the nuclei via sustained inertial cavitation activity (i.e., with pulsed exposures), the free microbubbles dissolve rapidly, thus producing a smaller biological effect than continuous wave exposures of the same total duration. Albunex microbubbles can nucleate inertial cavitation in vitro after the bulk of the microbubbles has been destroyed by ultrasound (Brayman and Miller, 1997a); because free microbubbles generated by the acoustic destruction of Albunex are expected to dissolve rapidly, this result suggests that a small number of the Albunex microbubbles escaped initial ultrasonic destruction, and that these can nucleate significant cavitation activity. For a dramatic example of the latter point, see Dalecki et al (1997e).

Everbach et al (1997) used a passive, 20 MHz acoustic detector to measure inertial cavitation activity in dilute human erythrocyte suspensions, and demonstrated that sonolysis induced in the presence of 3.5 V% Albunex is correlated with inertial cavitation activity occurring in the sample. Samples were exposed or sham-exposed to 1 MHz ultrasound (peak positive and peak negative pressures of 5.0 and 2.8 MPa, respectively) of various pulse durations for 60 s. Inertial cavitation activity within the samples was monitored throughout the exposure period and hemolysis assessed subsequently. At a 20 Hz pulse repetition frequency, both inertial cavitation activity and hemolysis increased in parallel with increasing pulse duration when samples contained Albunex; these correlations were statistically significant at the p<0.0001 level. Therefore, this study provides direct physical evidence in support of the assumption usually made in GCA bioeffect studies; viz., that the effect arises via an inertial cavitation mechanism.

Perfluorocarbon-based GCAs, by virtue of their greater persistence in an ultrasonic field, appear to have a greater potential for enhancing mechanical bioeffects than do air-based GCAs. Miller and Geis (Miller DL and Geis, 1998b) studied in vitro hemolysis in relation to the presence of different GCAs or filtered buffered saline solution. Two perfluoropentane-based GCAs (FSO69 and MRX-130) and two air-based GCAs (Albunex and Levovist) were used. The different agents were used at comparable volume fractions in the experimental mixtures, but differ somewhat in microbubble concentration in the stock material (Albunex: 0.3–0.9 × 10⁹/mL; FS069: 0.6–0.9 × 10⁹/mL; MRX–130: ~1 × 10⁹/mL; Levovist: unquantified. (See Miller DL and Geis, 1998b, subsection 6.2.3, and Figure 6-5). In 50:50 mixtures of whole canine blood and GCA exposed (stationary) for 1 s to continuous wave, 2.4 MHz ultrasound, the pressure thresholds for hemolysis were similar (0.2–0.4 MPa) for Albunex, FS069 and MRX-130; the threshold for Levovist treatment could not be determined. However, at suprathreshold acoustic pressures, the hemolytic yield was consistently greater for the perfluoropentane-based agents than for Albunex; e.g., at 1.6 MPa, hemolysis was 8, 24, and 48% for Albunex, MRX-130, and FS069 addition treatments, respectively. With pulsed ultrasound exposure (10 μs pulse duration, 0.01 duty factor) of the same total “on” time used in the continuous wave experiments, hemolysis was generally lower than produced by continuous wave exposures at comparable acoustic pressures. The general trends observed in the continuous wave experiments were again observed, and at acoustic pressures within the range of diagnostic ultrasound, levels of hemolysis were considerable. At a pressure amplitude of 3.2 MPa, insonation produced ~60 or ~10% hemolysis with FS069 or MRX-130 present, respectively, as compared with ~2%–5% hemolysis with Albunex or Levovist present. Hemolysis produced in the presence of the perfluorocarbon-based GCAs had a marked dependence of GCA concentration. Pulse duration was not a strong determinant of hemolysis. Most of the difference in the ability of the different GCAs to promote ultrasonic hemolysis appears to be related to the ability of the perfluorocarbon gases to persist in the cell suspension during insonation. A comparison of hemolysis produced with Albunex or FSO69 present and application of either 1, 10, 100, 1,000, or 10,000 pulses (10 μs, 1 ms pulse repetition period) of 2.4 MHz ultrasound indicated that for low numbers of pulses (up to 100), the enhancement of hemolysis produced by these two GCAs were equivalent. However, with 1,000 or 10,000 pulses, the levels of hemolysis produced in the presence of FSO69 were sixfold to eightfold greater than produced with Albunex. This factor is far greater than the expected differences in microbubble concentration in the samples (see above).

Erythrocytes are neither the sole corpuscular element of blood, nor are they the only blood cell type susceptible to ultrasonic damage. At comparable concentrations, human erythrocytes and lymphocytes are lysed in similar numbers by ultrasound exposures, the white cells being somewhat more sensitive than the red cells (Miller MW and Brayman, 1997). Platelets are also susceptible to inertial cavitation-induced membrane damage. Everbach et al (1996a) studied ultrasound-induced platelet lysis and membrane permeability changes (pre-loaded ⁵¹Cr leakage) in relation to inertial cavitation activity, and the effect of Albunex on these endpoints. Cavitation activity was assessed using a passive detector. Platelets were exposed to pulsed 1 MHz ultrasound with either 0 or 3.5 V% Albunex present. With Albunex, lysis increased with increasing pulse duration; without Albunex, lysis remained at control levels. Platelet lysis in samples containing Albunex was highly correlated with measures of inertial cavitation occurring in the sample. The assay for sublytic damage also showed that the number of cells damaged increased with increasing pulse duration, but this endpoint was not well correlated with inertial cavitation activity.

Monolayers of CHO cells grown on Mylar membranes undergo membrane damage, as evidenced by ATP release, when insonated at 3.3 MHz in the presence of 5% Albunex (Miller DL and Bao, 1998a). The effect is dependent on the applied acoustic pressure and exposure mode, with a threshold of ~0.3 MPa for 1 s of continuous exposure and ~0.6 MPa for pulsed exposures (10 μs pulses, 0.01 duty factor, 1 s “on” time). ATP release was highly correlated with the strength of subharmonic signals emitted by the microbubbles, indicating that cavitation was involved. The cell populated surface orientation was an important determinant of the amount of membrane damage observed, with much greater ATP release obtained when the populated surface was at the site of ultrasound entry into the exposure vessel than when at the site of exit. However, the harmonic signals were unaffected by vessel orientation. Qualitatively similar results have been obtained by Brayman (unpublished; work in progress). Data obtained using confluent V79 fibroblasts on Mylar membranes and exposed to pulsed 1.0, 2.2, or 3.5 MHz ultrasound indicate that (1) 3% Albunex reduces the pressure threshold for an erosive effect of insonation on the monolayer relative to that when Albunex is absent, (2) without Albunex, cell populated surfaces at the site of ultrasound exit from the exposure vessel are more susceptible to erosion than are those at the site of entry, but that this differential disappears when Albunex is present, and (3) the effect is frequency and pressure dependent. Because these studies involve diagnostically realistic ultrasound frequencies, acoustic pressures and pulse durations, these results suggest that the use of GCAs during vascular ultrasound imaging may damage the endothelial cells of blood vessels.

6.4.3.2 Laboratory Animals

Since tissues that contain gas bodies naturally are susceptible to damage from exposure to diagnostic levels of ultrasound (see Section 5), and because of the evidence from in vitro studies that GCAs enhance ultrasound-induced bioeffects, it is reasonable to hypothesize that GCAs will increase the likelihood of ultrasound-induced bioeffects in vivo. Dalecki et al (1997d) studied ultrasound-induced hemolysis in vivo when Albunex was present in the blood. Murine hearts were exposed for 5 min to 1.2 or 2.4 MHz pulsed ultrasound (10 μs pulse duration, 100 Hz PRF) at pressure amplitudes ranging from 0–10 MPa. At evenly spaced intervals during the exposure period, four ~25 μL boluses of Albunex were injected into a tail vein. Two studies, using different strains of mice, were performed at 1.2 MHz to determine the threshold for hemolysis. At 2.4 MHz, animals were exposed to only the highest pressure amplitude (10 MPa). Control animals were exposed at 1.2 MHz with a peak amplitude of 10 MPa and were injected with four boluses of saline. Following exposure, the animals were killed, the blood collected and the blood assayed for hemolysis. Figure 6-7 presents the percent hemolysis as a function of exposure amplitude for all experiments in this investigation. Analyses of the data at 1.2 MHz for the C3H and CD1 strains of mice indicated that the threshold values did not differ significantly. The threshold for detectable hemolysis in vivo at 1.2 MHz was 3.5 MPa peak positive pressure, 2 MPa peak negative pressure and ~200 W/cm² SPPA intensity at the surface of the animal. After correcting for the chest wall attenuation, the threshold at the surface of the heart at 1.2 MHz was 3.0 MPa peak positive pressure, 1.9 MPa peak negative pressure, corresponding to an MI of 1.8 and an SPPA intensity ~180 W/cm². Hemolysis in mice injected with Albunex and exposed at 2.4 MHz with the highest exposure amplitude (10 MPa) was only 0.46%, and was comparable to that in sham exposed mice receiving Albunex (~0.4% hemolysis) and in control mice receiving saline injections and exposed at 10 MPa (~0.4% hemolysis). This strong frequency dependence of hemolysis was also observed in whole blood in vitro (Brayman et al, 1997b) and is similar to the frequency dependence of the threshold for ultrasound-induced hemorrhage in the intestine (Dalecki et al, 1995b). These results indicate that when GCAs are present in the blood, ultrasonically induced hemolysis can occur in vivo at diagnostically relevant pulse durations and pulse repetition frequencies. The relatively high threshold for detectable hemolysis at 1.2 MHz combined with the strong dependence of the threshold on frequency suggest there is little likelihood of extensive (i.e., more than ~1% of the systemic red cell pool) ultrasound-induced hemolysis under most current diagnostic imaging procedures. Killam et al (1998) detected no hemolysis in rabbits injected with Optison and exposed in vivo to 5 MHz pulsed ultrasound. A Hewlett Packard Sonos 500 with a 5 MHz phased array transducer was used at maximum output power (MI ~0.5) (A. Killam, personal communication). The pressure amplitude was therefore ~1 MPa). Blood samples were collected from the carotid and femoral arteries at baseline, and at 0, 3, and 6 min after infusion of 0.6 mL/kg Optison. Analysis of blood samples showed no increase in serum free hemoglobin or serum lactate dehydrogenase, both markers for hemolysis. However, at the acoustic frequency and pressure used in this study (i.e., 5 MHz, ~1 MPa), the available in vitro and in vivo data indicate that little or no hemolysis would be likely to arise (see, e.g., Fig. 6-8). These data indicate that under conditions of “conventional” B-mode imaging, the administration of Optison did not produce detectable lysis of red blood cells.

Figure 6-8.

Hemolysis in vivo in mice exposed to pulsed ultrasound. Percent hemolysis is plotted as a function of peak positive pressure at the surface of the animal. Solid squares and circles are data for C3H and CD1 strains of mice, respectively, exposed to 1.2 MHz ultrasound and injected with Albunex. Open squares and circles are data for C3H and CD1 mice, respectively, exposed to 1.2 MHz ultrasound and injected with saline. Open triangle is datum for CD1 mice exposed to 2.4 MHz ultrasound and injected with Albunex. (Figure adapted from and reprinted by permission of Elsevier Science, from Dalecki D, et al: Remnants of Albunex nucleate acoustic cavitation. Ultrasound Med Biol 23:1405, 1997e). Data are presented as the mean percentage of hemolysis; error bars represent the standard error of the mean.

Rupture of microvessels in exteriorized rat spino-trapezius muscle following injection of modified Optison microbubbles and exposure of the muscle to single sweeps of 2.3 MHz diagnostic ultrasound has been reported (Skyba et al, 1998). The numbers of ruptured capillaries and killed nucleated cells scaled with increasing MI value over the range of 0.4–1.0 as indicated by the output display. However, because the muscle was exteriorized and bathed in Ringer’s solution (which has a much lower attenuation coefficient than assumed for tissue), the tissues were exposed to greater acoustic pressures than indicated by the on-screen MI values. Capillaries smaller than ~7 μm in diameter were affected. The mean number of ruptured vessels per g fresh tissue weight increased monotonically with increasing MI value, from ~320 at an MI of ~0.4 to ~2,400 at MI ~1.0. Microscopic observation during the exposure allowed direct visualization of ultrasonic microbubble destruction in vivo.

Ultrasound-induced lung hemorrhage under diagnostically relevant exposure conditions is well documented (Section 4) and depends on the presence of gas in the tissue (Hartman et al, 1990). If cavitation in the lung vasculature were the mechanism for lung hemorrhage, then the addition of GCAs to the blood should increase the extent of ultrasound-induced lung hemorrhage. This was tested by Raeman et al (1997). Murine lungs were exposed to pulsed ultrasound (1.2 MHz, 10 μs pulse duration, 100 Hz PRF, 2 MPa peak positive pressure, 5 min). During exposure, four 25 μL boluses of Albunex were injected into a tail vein. The mean area of ultrasound-induced lung hemorrhage in Albunex-injected mice was not statistically different from that in saline-injected or uninjected mice. For several echocardiographic procedures, using current diagnostic imaging systems, the acoustic pressure amplitude at the surface of the lung may approach the threshold for lung hemorrhage (Carstensen et al, 1992). This study indicates that the presence of Albunex in the lung vasculature does not increase the extent of ultrasound-induced lung hemorrhage.

Miller and Geis (1998b) investigated the effects of the presence of GCAs in the vasculature on ultrasound-induced intestinal hemorrhage in mice. For exposure to continuous wave ultrasound at 1 MHz (0.75 MPa peak positive pressure), the extent of hyperemia, petechiae, and hemorrhages increased in comparison with control animals injected with a blank agent. For pulsed ultrasound exposures at 1 MHz (10 μs pulse duration, 0.01 duty cycle), Albunex injection greatly increased the number of ultrasound-induced petechiae; e.g., for exposures at 2.8 MPa, there was a thirtyfold increase in petechia in the mice with Albunex compared with Albunex-free control animals. The number of petechiae produced increased linearly with increasing Albunex dosage.

Tissue damage resulting from lithotripsy is often attributed to inertial cavitation. Prat et al (1991b) investigated whether the presence of GCAs increases the tissue damage produced by exposure to lithotripter fields at amplitudes typically used clinically to treat kidney stones. Three treatment groups of rabbits were used: Group 1 received 500 lithotripter shocks focused on the right hepatic lobe, Group 2 received an infusion of microbubbles (Plasmion; Roger Bellon, Neuillysur-Seine, France), and Group 3 received 500 shocks and microbubbles during the exposure period. Animals in Group 1 exhibited a few, small subcapsular and intraparenchymal hematomas, while animals in Group 2 showed only moderate liver congestion. Animals exposed to lithotripter fields in the presence of microbubbles (Group 3) showed numerous subcapsular and intraparenchymal hematomas of significantly larger size than those observed in Group 1 animals. Hence, infusion of microbubbles during lithotripsy greatly increased hepatic tissue damage. Tissue damage is enhanced by the intravascular presence of GCAs during lithotripsy treatment even for relatively low pressure amplitude exposures (Dalecki et al, 1997e). In uninjected mice exposed abdominally to 200 piezoelectric lithotripter pulses (2 MPa peak pressure amplitude), there was only minimal damage to the intestine and lung tissue. In similarly exposed animals who received four 25 μL injections of Albunex during exposure there were extensive hemorrhages in the muscle, fat, mesentery, intestine, kidneys, stomach, bladder, and seminal vesicles.

Although Albunex is effective as a diagnostic contrast agent for only a few minutes, Albunex continues to supply cavitation nuclei in vivo for long periods of time after injection (Dalecki et al, 1997e). Mice were given a single injection of Albunex prior to exposure to 200 lithotripter pulses at times ranging from 5 min to 24 hr following the injection. For many tissues, such as intestine, mesentery and bladder, Albunex continued to enhance hemorrhage from exposure to lithotripter fields at amplitudes of 2 MPa for up to 4 hr after injection. For exposures at clinical amplitudes, Albunex continued to enhance hemorrhage in the bladder, muscle and seminal vesicles when exposures were performed up to 6 hr after injection. These studies suggest that great caution should be practiced in the use of GCAs before or during lithotripsy procedures.

6.4.4 Potentially Useful Bioeffects

6.4.4.1 Tumor Disruption and Enhancement of Chemotherapy

A number of papers have been published in which ultrasound was reported to enhance the effectiveness of certain anticancer drugs, apparently via nonthermal mechanisms (see, e.g., Yumita et al, 1989, 1990; Saad and Hahn, 1989; Loverock et al, 1990; Umemura et al, 1990; Harrison et al, 1991; Kessel et al, 1994). However, most of the published reports dealing with this topic have used “normal” media; i.e., media not supplemented with exogenous gas bodies, and will, therefore, not be considered here.

As therapeutic tools for tumor treatment, GCAs have been employed thus far for three general uses: (1) as a means of disrupting tissues via direct mechanical effects resulting from inertial cavitation; (2) as an adjunct to ultrasound:drug interactions in which inertial cavitation enhances anticancer drug effectiveness, apparently via a sonochemical effect, and (3) as a site-specific drug carrier that releases a drug in the neighborhood of the target tissue when the microbubbles are destroyed by insonation of the tissue. The use of GCAs or similar preparations to deliberately nucleate inertial cavitation for these purposes is a relatively recent development.

6.4.4.1.1 Mechanical Tumor Disruption by Microbubble-Enhanced Inertial Cavitation

Lipid-coated microbubbles accumulate in tumor tissue to a much greater extent than in surrounding tissues (Simon et al, 1992b, 1993), thus presenting the possibility of therapeutic intervention by the induction of inertial cavitation in microbubble-enriched tumor tissues (Simon et al, 1993). Simon et al (1993) tested this idea using Walker-256 tumors implanted subcutaneously in rats. There was one control group (no ultrasound, no microbubbles), and two ultrasound-exposed groups, one of which received lipid-coated microbubbles via tail vein injection prior to insonation. Tumors were located using 7.5 MHz ultrasound scans. In the ultrasound-exposed groups, this was followed by exposure to continuous wave, 4.5 MHz ultrasound (I_SATA = 250 mW/cm²) for 8 min. Insonation did not demonstrably affect tissue temperature. Animals were sacrificed at 0–72 hr post-insonation and scored blindly for hemorrhage and tumor necrosis. A consistent elevation of tumor necrosis was associated with microbubble and ultrasound treatment relative to either the controls or the ultrasound-only treatment group. The authors, while not entirely ruling out possible tissue heating effects, interpret their results in terms of inertial cavitation activity. That insonation of microbubble-enriched tumor tissue may enhance tumor necrosis is exciting, but caution is warranted because of the potential for metastasis (see subsection 6.4.4.1.4).

6.4.4.1.2 Increased Tumoricidal Drug Activity with Microbubble-Enhanced Inertial Cavitation

Prat et al (1993) explored the combined effects of shock wave–induced, microbubble-enhanced cavitation and an anticancer drug in vivo. Tumors were implanted in the abdomens of rats, and an electrohydraulic lithotripter (250–500 shocks, peak positive pressure ~60 MPa) used to induce inertial cavitation while the abdomens were being infused with air-filled, gelatin-encapsulated microbubbles with or without co-treatment with the drug fluorouracil. At a relatively high dose, the drug alone cured 100% of the animals, but produced signs of liver compromise. Low dosages of the drug cured none of the animals when used alone. Treatment with shock waves and microbubbles without the drug cured about half of the animals. However, treatment of the rats with shock waves and microbubbles, in combination with a low dose of fluorouracil, was as effective as the high dose of the drug alone.

Jeffers et al (1995) also explored ultrasound-induced inertial collapse of stabilized microbubbles, combined with drug therapy, as a potential tool for tumor destruction. The goal was to enhance the membrane-damaging effect of the polar solvent and tumoricidal drug N,N-dimethylforamide by ultrasound in a localized tissue volume, such that systemic drug doses might be reduced while still producing tumoricidal activity. The related drugs monomethylforamide and dimethylsulfoxide were also tested. In vitro suspensions of HL-60 human promyelocytic leukemia cells were used; membrane damage was assessed by released lactate dehydrogenase activity. The GCAs used in this study were “homemade” albumin stabilized microbubbles having a median diameter of 3.2 μm. Controls (i.e., no bubbles) were prepared by adding an equivalent volume of 5% albumin solution. Cells were exposed or sham-exposed to 1 MHz ultrasound (continuous wave, 15 s) at 0, 0.6, 1.0, 1.9 or 2.5 W/cm² SPTA intensity, which did not result in bulk heating of the suspensions. Subharmonic emissions from the sample were monitored simultaneously as a measure of cavitation activity within the samples. Without microbubbles, neither ultrasound + albumin nor ultrasound + albumin + N,N-dimethylforamide increased levels of membrane damage at any of the tested intensities. In contrast, membrane damage produced by the ultrasound + microbubble treatment increased with increasing ultrasound intensity, reaching ~35% at the highest exposure level. Membrane damage produced by the ultrasound + N,N-dimethylforamide + microbubble treatment also increased with ultrasound intensity, but was significantly greater than that produced by the ultrasound + microbubbles treatment at all nonzero ultrasound intensities. Qualitatively similar results were obtained with the other drugs tested. Cell membrane damage in the ultrasound + N,N-dimethylforamide + microbubble and the ultrasound + microbubble treatments was correlated strongly with the subharmonic emission amplitude measured during sample insonation, indicating that cavitation was involved in the effect. At comparable “levels” of cavitation activity, membrane damage was greater in samples containing N,N-dimethylforamide than in those lacking it. Separate assessments for sonochemical effects failed to detect any significant effect on membrane damage, suggesting that any effective sonochemicals produced by the insonation were short-lived. Other assessments indicated that the presence of N,N-dimethylforamide did not increase cavitation activity relative to N,N-dimethylforamide-free controls, nor did N,N-dimethylforamide sensitize the cells to damage by shear forces applied to the cells by non-acoustic means. When taken together, the observations that (1) there was no bulk heating of the insonated suspensions, (2) enhancement of N,N-dimethylforamide cytotoxicity was proportional to subharmonic emission, and (3) the N,N-dimethylforamide did not affect the occurrence of subharmonic emission or the susceptibility of the cells to rupture by mechanical shear forces, suggest that cavitation is responsible for the effect.

The study of Yumita et al (1997) did not involve the use of exogenous microbubbles, but it is relevant to this discussion because the insonation method used (viz., superposition of a 1 MHz ultrasound field and its 500 kHz subharmonic) was designed specifically to promote the occurrence of cavitation. Colon 26 carcinoma tumors were transplanted into the kidneys of mice. A gallium-porphyrin compound (ATX-70) was administered intravenously. Neither ATX-70 treatment alone nor sonication alone produced tumor necrosis; however, combined ultrasound + ATX-70 treatment produced extensive tumor necrosis. The ultrasound SPTP intensities (~30 W/cm² at both 500 kHz and 1 MHz) produced significant tissue heating (3–8°C), but because ultrasound treatment alone was ineffective, the effect of combined ultrasound + ATX-70 treatment was not ascribed to thermal mechanisms. The authors speculate that active oxygen species are generated by sonodynamically activated (i.e., inertial cavitation-activated) porphyrin, and that these oxygen species then destroy the tissue. Miyoshi et al (1997) further explored this issue, using suspension cultured HL-525 human leukemia cells with ATX-70 treatment. Cavitation activity was generated in the cell suspensions with a 50 kHz commercial sonicator. Sonication alone lysed ~20% of the cells, while ATX-70 alone had no effect. Ultrasound + ATX-70 treatment produced a large increase in cytotoxicity. Accumulation of the ATX-70 in the cells was demonstrated, but only extracellular ATX-70 was effective as a sonosensitizer; cells with incorporated ATX-70 sonicated in ATX-70 free medium were no more sensitive to killing than control cells. Sonication of ATX-70 solutions followed by incubation of cells in the fluid was ineffective, suggesting the involvement of short-lived sono-chemical products. However, the presence of free radicals could not be demonstrated by electron paramagnetic resonance (EPR) spin trapping, nor was the cytotoxic effect of simultaneous ultrasound + ATX-70 treatment overcome by inclusion of spin traps (which can remove radical intermediates) in the suspension fluid. Thus, the mechanism by which the enhanced ATX-70 activity arises remains unclear.

Prat et al (1994) explored the combined effect of 5-fluorouracil and inertial cavitation on cancer cell killing and other endpoints. They had previously demonstrated (Prat et al, 1993) that cancer cells in vitro and in vivo can be perturbed by exposure to lithotripter shock waves with microbubbles present. Among their findings was that exposure to shock waves + microbubbles resulted in a greater immediate cell kill and an increase in delayed mortality relative to cells exposed to shock waves alone. Prat et al (1994) facilitated inertial cavitation occurrence by the use of a lithotripter (peak positive pressure ~60 MPa) and by supplementing the cell preparations with gelatin-stabilized, air-filled microbubbles (10–100 μm diameter). HT-29 human colon cancer cells were exposed to shock waves + microbubbles followed by a 12 hr incubation period of the surviving cells. Various concentrations of 5-fluorouracil were then added and the cells allowed to incubate in the presence of the drug for up to 72 hr. The drug was then removed and clonogenic growth assessed. Other shock + microbubble treated cells were subcultured after shock treatment for 24 or 48 hr, followed by assessment of the rate of [³H]deoxuridine incorporation into DNA. In the absence of 5-fluorouracil, cells that survived the initial exposure to shock + microbubble treatment had decreased clonogenic growth, as did cells exposed to 5-fluorouracil only. The effect of shock + microbubble pretreatment decreased the dose of 5-fluorouracil needed to kill 50% of the cells and reduced the rate of deoxuridine incorporation relative to untreated control cells. 5-Fluorouracil treatment alone suppressed deoxuridine incorporation after 24 or 48 hr of subculture, but shock + microbubble pretreatment reduced by fourfold to tenfold the concentration of 5-fluorouracil needed to inhibit deoxuridine incorporation by 50% relative to cells treated with 5-fluorouracil only. Thus, the shock + microbubble treatment sensitized the cells to the toxic effects of the drug despite the passage of many hours between shock treatment and drug application. The mechanism by which this arises is not yet certain.

6.4.4.1.3 Microbubbles as a Local Drug Delivery System

Porter et al (1996a) report the results of an interesting “proof-of-concept” study in which microbubbles were used as a vehicle to carry drugs to a target tissue, with release of the drug in situ produced by insonation of the microbubbles and target tissue, presumably via microbubble destruction. A synthetic antisense oligodeoxyribonucleotide, phosphorothiolate oligonucleotide, was found to bind to the albumin coating of perfluorocarbon-exposed sonicated dextrose albumin (PESDA) microbubbles. To test the ability of PESDA microbubbles to deliver and release the drug in a target tissue, microbubbles bearing bound phosphorothiolate oligonucleotide were injected intravenously into three dogs. The left kidney of each was insonated using a diagnostic ultrasound transducer (peak negative pressure = 1.1 MPa, 30 Hz frame rate, 2.0–2.5 MHz frequency) after the injection; the PESDA microbubbles were visualized simultaneously as sonographic contrast. The right kidney received no ultrasound. The dogs were killed 4 hr later, and the kidneys sampled and assayed for phosphorothiolate oligonucleotide. In two of the three dogs, there was a substantial increase in accumulated phosphorothiolate oligonucleotide in the insonated kidney (ninefold and threefold greater than in the matched control kidneys), while in the third dog, there was no difference. The mean relative increase was ~fourfold; however, possibly because of the small number of subjects, this difference was not statistically significant. Nonetheless, the result indicates that the general technique is worthy of further exploration.

6.4.4.1.4 Inertial Cavitation-Related Metastasis

The foregoing discussion indicates that the intentional induction of cavitation in vivo may have therapeutic value in the treatment of tumors. In this regard, an issue of considerable concern is whether the treatment may increase the potential for metastasis. Only a few studies have addressed this issue, and while most of these have been negative (see, e.g., Geldhof et al, 1989; Oosterhof et al, 1990; Hoshi et al, 1991), a recent positive report by Oosterhof et al (1996) is noteworthy. Although this study was conducted with high pressure lithotripter shocks, it is somewhat unsettling because these effects presumably arise via inertial cavitation and, as discussed, inertial cavitation activity sufficient to disrupt the microcirculation may arise in tissues during ultrasonic imaging with GCA present (see subsection 6.4.3.2).

6.4.4.2 Ultrasound-Enhanced Thrombolysis with GCAs

Ultrasound can enhance the dissolution of thrombi (Hong et al, 1990; Ariani et al, 1991; Francis et al, 1992; Lauer et al, 1992; Tachibana, 1992; Blinc et al, 1993; Harpaz et al, 1993; Luo et al, 1993; Sehgal et al, 1993; Olsson et al, 1994). The bulk of the evidence indicates that this occurs via nonthermal mechanisms, presumably by increasing the permeance of fibrolytic agents into the thrombi. Tachibana and Tachibana (1995) postulated that inertial cavitation contributes to the process, and that increasing inertial cavitation activity by nucleating thrombus-containing fluids with GCAs would further increase ultrasound-enhanced thrombolysis. Artificial thrombi formed from whole blood were placed in ultrasound exposure vessels into which test fluids could be circulated (urokinase or albumin solutions, Albunex or various combinations). These were exposed or sham-exposed for 3 min (170 kHz, 2 s bursts, duty factor = 0.33, SPTP intensity = 0.5 W/cm²) followed by an incubation period. Fibrinolysis was assessed by the loss of thrombus dry weight relative to untreated control thrombi. After 120 min of posttreatment incubation, neither albumin, microbubbles, ultrasound + albumin, nor ultrasound + microbubbles significantly enhanced thrombolysis in the absence of urokinase (~1%–3% fibrinolysis). Without ultrasound treatment, urokinase increased fibrinolysis to ~50% in the presence of either albumin or microbubbles. With ultrasound treatment, fibrinolysis with urokinase + albumin was significantly enhanced (from 51% to 58%; p ≪0.005) and enhanced further with urokinase + microbubbles (from 48% to 72%; p ≪0.005). Tachibana and Tachibana (1995) attributed the increased level of thrombolysis associated with simultaneous treatment with urokinase, ultrasound, and Albunex to the occurrence of inertial cavitation in the fluid surrounding the thrombus, and speculated that the mechanism involves cavitation-induced microstreaming that increases urokinase penetration into the thrombus.

Porter et al (1996c) compared the efficacy of PESDA and room air exposed sonicated dextrose albumin (RASDA) microbubbles in ultrasound-enhanced thrombolysis. Artificial thrombi formed from whole blood were insonated or sham-insonated (20 kHz, 0.85 MPa peak negative pressure, 2 min) in a variety of test solutions. Sonication in saline solution increased thrombolysis relative to simple incubation in saline solution (from 7% to 23%). Saline + ultrasound treatment was about as effective as urokinase treatment alone (23% versus 17%, respectively). Ultrasound + urokinase treatment produced 48% thrombolysis. PESDA alone had no effect on thrombolysis, and PESDA + urokinase was no more effective than urokinase alone. However, treatment of the thrombi with ultrasound + PESDA, in the absence of urokinase, produced a large increase in thrombolysis [c.f. ultrasound + Albunex treatment by Tachibana and Tachibana (1995), described above, and results obtained with RASDA, described below]. The ultrasound + urokinase + PESDA treatment was the most effective, producing 60% thrombolysis. RASDA microbubbles were less effective than PESDA microbubbles. Ultrasound + RASDA treatment, without urokinase, was no more effective than ultrasound treatment of thrombi in saline solution; however, when urokinase was present, the level of thrombolysis produced with RASDA microbubbles present was about the same as that produced by treatment with ultrasound + PESDA without urokinase. Porter et al (1996c) attribute the microbubble enhancement of ultrasound-accelerated thrombolysis to cavitation. They suggest that the difference in the responses to PESDA and RASDA may relate to the number of cavitation nuclei present, because of a fivefold greater microbubble concentration in the PESDA preparations, and because the PESDA microbubbles are more stable and persist longer.

Nishioka et al (1997) conducted a study of the thrombolytic effectiveness of ultrasound with or without perfluorocarbon-based GCAs in the absence of thrombolytic drugs. Dodecafluoropentane emulsion or saline solution was tested for thrombolytic effectiveness with or without ultrasound. Human thrombi in vitro and rabbit ileofemoral arterial thrombi in vivo were the model systems. The former were exposed/sham-exposed for 3 min to 25 kHz ultrasound at an intensity of 2.9 W/cm². The latter were insonated transcutaneously (20 kHz, in situ intensity ~1.5 W/cm²). In vitro clot dissolution was significantly enhanced by ultrasound treatment alone (no ultrasound + saline: 10%; ultrasound + saline: 72%). The GCA without ultrasound had no effect on thrombolysis, while GCA + ultrasound produced 98% thrombolysis. The maximum diameter of clot debris particles formed by insonation with or without dodecafluoropentane was ~4 μm in both cases. The in vivo study produced somewhat different results. Ultrasound treatment alone produced reperfusion in none of six rabbits. Reperfusion was observed in 1 of 11 rabbits (9%) treated with dodecafluoropentane treatment alone, but in 14 of 17 rabbits (82%) treated with ultrasound + dodecafluoropentane. Thermal effects may have played a role, because histological examination of insonated tissues revealed evidence of coagulative necrosis and focal hemorrhage.

A potential problem with the therapeutic use of GCAs to improve ultrasound-enhanced thrombolysis in vivo is that often the thrombus occludes the blood vessel to the extent that blood flow is minimal and, therefore, the GCA may have limited access to the thrombus. Lanza et al (1996) described a method that may help to overcome this problem. Their interest was in site-specific image enhancement rather than production of a therapeutic bioeffect such as thrombolysis, but their approach may hold promise for cavitation-related thrombolysis and tumor necrosis. Their approach was to use ligand-specific binding and accumulation of a nongaseous, lipid-encapsulated perfluorocarbon emulsion (perfluorodichlorooctane). The lipid shell was biotinylated by incorporating biotinylated phosphatidlyethanolamine. The basic idea is as follows: the blood clot (or other target) is treated with a biotinylated monoclonal antibody specific to the target of interest (e.g., a fibrin domain in the case of thrombi). The biotinylated monoclonal antibody binds to the surface of the target. A biotin-binding molecule, avidin, is then administered and conjugates and cross-links the bound biotinylated monoclonal antibody on the target surface. Finally, the biotinylated, lipid-coated perfluorocarbon emulsion is administered. These microparticles bind, via unoccupied biotin binding sites on the immobilized avidin molecules, to the surface of the target. Lanza et al (1996) demonstrated site-specific accumulation of their preparation in vivo using a canine femoral artery thrombus model; image enhancement was marked. If it is possible to similarly biotinylate the lipid shell of a stable gaseous contrast agent, it may be possible to accumulate large numbers of microbubbles directly on the surface of the target structure. Because of their immediate proximity to the target, ultrasound induced inertial cavitation might then be quite effective in disrupting the target. However, because of the potential for this treatment to facilitate metastasis, this possible approach to tumor therapy must necessarily be approached with caution.

6.4.4.3 Enhancement of Sonoporation by GCAs

Enhancement of sonoporation (i.e., transient permeabilization) of mammalian cells by GCAs has been reported. Bao et al (1997) reported that plasmid transfection of CHO cells in vitro exposed to 2.25 MHz ultrasound occurred at a very low level without the inclusion of 10% Albunex and without exposure vessel rotation during insonation; with rotation and 10% Albunex present, the threshold pressure for transfection was ~ 0.3 MPa. Kim et al (1996) report the results of extensive in vitro experiments on ultrasonically induced transfection of mammalian cells in the absence of GCAs. However, the report also describes preliminary efforts to ultrasonically induce transfection in vivo by injection of Albunex and plasmid suspensions into rat knee joints prior to insonation or sham insonation. There was evidence of three successful transfections in four exposed knees, and none in the sham-exposed knees.