- Email: [email protected]
Hemolysis in vivo from exposure to pulsed ultrasound
Copyright
PI1 SO301-5629( 96)00203-7
ELSEVIER
@Original
Ultrasound in Med. & Biol., Vol. 23. No. 2, PP. 307-313, 1997 0 1997 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/97 $17.00 + .OO
Contribution
HEMOLYSIS D. DALECKI,*’
IN VZVO FROM
EXPOSURE
TO PULSED
ULTRASOUND
C. H. RAEMAN,+ S. Z. CHILD,*+ C. Cox,+* C. W. FRANCIS,‘~ R. S. MELTZER+~’ and E. L. CARSTENSEN*+
Departmentsof *Electrical Engineering,‘Biostatistics, The “Cardiologyand “HematologyUnits of the Departmentof Medicine, and +TheRochesterCenterfor BiomedicalUltrasound,University of Rochester,Rochester,NY (Received 9 April 1996; in jnal fom 1 October 1996)
Abstract-Ultrasonicaily induced hemolysisin viva was studied when a commercial ultrasound contrast agent, Albunexo, was present in the blood. Murine hearts were exposed for 5 min at either 1.15 or 2.35 MHz with a pulse length of 10 w and pulse repetition frequency of 100 Hz. During the exposure period, four bolusesof Albunexo were injected into a tail vein for a total of - 0.1 mL of Albunexo. Following exposure,blood wascollected by heart puncture and centrifuged, and the plasmawasanalyzed for hemoglobin concentration. With Albunexo present in the blood, the threshold for hemolysisat 1.15 MHz was 3.0 2 0.8 MPa (mean 2 SD) peak positive pressure( - 1.9 MPa negative pressure,- 180W cn-’ pulseaverage intensity). For the highest exposure levels (10 MPa peak positive pressureat the surface of the animal), the mean value for hemolysiswas - 4% at 1.15 MHz and 0.46% at 2.35 MHz, i.e., the threshold at 2.35 MHz is >lO MPa peak positive pressure.In contrast, hemolysisin control mice receiving saIine iqjections at 10 MPa or sham-exposed(0 MPa) mice receiving Albunexo was - 0.4%. 0 1997 World Federation for Ultrasound in Medicine & Biology. Key Words: Ultrasound, Hemolysis,Contrast agent, Bioeffect.
to differentiate between these two alternatives. A partial answer to this question has been given in the studies of Brayman et al. ( 1995, 1996), who found statistically significant increases in free hemoglobin in whole human blood containing the contrast agent, Albunexe’ (Molecular Biosystems, San Diego, CA, USA, and Mallinckrodt Medical, Inc., Hazelwood, MO, USA) after exposure in vitro to pulses as short as 5 ~CLS (duty cycle of 0.01) and at acoustic pressures as small as
INTRODUCTION Attempts have been made to detect ultrasound-induced
hemolysis in viva. Even when hemolysis was detected after exposure to continuous-wave ultrasound, the evidence suggested that the phenomenon was primarily a thermal effect that occurred as blood perfused organs were heated by ultrasound (Williams et al. 1986; Wong and Watmough 1983). But, thus far, there have been no reports of hemolysis in viva with pulses as short as those used in diagnostic ultrasound. Under exposure conditions that avoid excessive heating, there is a high probability that hemolysis is indicative of acoustic cavitation (Carstensen et al. 1993). Failure to observe hemolysis in viva may indicate either that it is difficult to excite bubbles in the environment provided by whole blood or that nuclei appropriate for acoustic cavitation may be very rare in blood in viva. The recent availability of stabilized microbubbles for use as ultrasonic contrast agents makes it possible
0.5 MPa (Brayman et al. 1996). Without added bub-
bles, no significant hemolysis was seen even with much higher pressures and longer pulse lengths (Brayman et al. 1995). We report here qualitatively similar findings of ultrasound-induced hemolysis in mice. EXPERIMENTAL
METHODS
The acoustic sources were 1.15- and 2.35-MHz focused transducers fabricated from 1.5- and l-inch (38.1 and 25.4 mm) diameter piezoceramic elements, respectively, cemented to planoconcave aluminum lenses. The transducer was mounted on the bottom of a plastic exposure tank filled with deionized, degassed
Address correspondence to: Dr. Diane Dale&i, Department of Electrical Engineering, University of Rochester, Rochester, NY 14627. 307
Ultrasound in Medicine and Biology
308
water at room temperature (- 21°C). The acoustic field was measured daily with a bilaminar, PVDF membrane hydrophone (Marconi Research Centre, Chelmsford, UK) calibrated by comparison to a sphere radiometer (Dunn et al. 1977). The focal length of the 1.15MHz source was 6.5 cm, and the -6 dB beam width at the focus was 3.5 mm. The focal length of the 2.35MHz source was 4.7 cm, and the -6 dB beam width at the focus was 3.0 mm. For perspective, the murine heart is approximately 6 mm X 9 mm. Fields were calibrated for peak positive pressures of 2.5, 5, 7.5 and 10 MPa. In addition, spatial peak, pulse average intensities were obtained by integrating the square of the measured pressures over the duration of the pulse using a digital oscilloscope (LeCroy Model 9400, Chestnut Ridge, NY, USA) and dividing the result by the acoustic impedance of water. The results are reported in terms of the peak positive pressure. However, Table 1 presents the corresponding negative pressures and spatial peak, pulse average intensities for each exposure level. Note that these are field parameters in water at the surface of the animal. Mice weighing approximately 35 g were anesthetized with ketamine (200 mg/kg) and Rompun (10 mg/kg). All animal protocols were approved by the University of Rochester Committee on Animal Resources. The ventral chest surface was shaved and depilated ( NeetTM , Reckitt & Colman Inc., Wayne, NJ, USA) to minimize trapped air on the skin surface. Preliminary autopsies of several mice determined the mouse heart was located approximately 1 cm cephalad to the caudal tip of the sternum and 2-3 mm to the animal’s left side. This location was marked on the skin surface with an ink dot to aid in positioning the heart in the acoustic focus. The animal’s circulatory system was accessed through a tail vein using a modified, 25-ga Butterfly@ needle (Abbott Laboratories, Chicago, IL, USA) as shown in Fig. 1. The Butterfly@ apparatus was sealed with an injection cap to allow for multiple injections during ultrasound exposure. A 1-mL syringe with a 26-ga needle was filled with 0.2 mL of undiluted Albunex@ (or saline). For each injecTable 1. Exposure parameters. f (MHz) 1.15 1.15 1.15 1.15 2.35
P+ (MPd
2.5 5.0 7.5 10.0 10.0
P-
NW
1.8 2.7 3.2 3.6 4.4
Is,
(W
cm-‘)
120 3.50 590 790 1025
f = frequency; p+ = peak positive pressure; p- = maximum negative pressure; IsPPA= spatial peak pulse average intensity.
Volume 23, Number 2, 1997
Fig. 1. Butterflya’ injection apparatus.A 25G modifiedButterfly@needle,fitted with an injection cap, wasinsertedinto the tail vein of a mouse.A 1 mL.syringe with a 26G needle was filled with either 0.2 nL of Albunex@or saline and insertedinto the shankof the Butterfly* needle.Four separate 0.05~mLinjectionsweregiven at l-mm intervalsduring the 5-minexposureduration.Betweeninjections,the syringe with 26G needlewas removed from the injection cap and gently rolled to dispersethe contrastagent.
tion into the animal, the syringe was first rotated to disperse the contrast agent bubbles uniformly in the suspension, and then the needle was inserted through the injection cap to the shank of the Butterfly@ needle (Fig. 1) . Exploratory observations, using a commercial ultrasound scanner operating at 7 MHz, confirmed that Albunexm, administered in this way, reached the heart. Each animal was mounted horizontally, ventral surface down, on a plastic small animal holder with the tail and Butterfly@ injection apparatus taped to the holder. The animal and holder were then attached to a three-axis positioning system (Velmex, East Bloomfield, NY, USA). No part of the support apparatus was exposed directly to ultrasound. The depth of water in the exposure tank was equal to the acoustic focal length. Coordinates from a fixed pointer in the water tank to the acoustic focus were determined daily with the hydrophone. These coordinates were used to position the marked dot on the animal’s chest at the acoustic focus. Each mouse was exposed for 5 min to lo-ps pulses of ultrasound at a repetition frequency of 100 Hz. At 1.15 MHz, exposure levels were either 0 (sham), 2.5, 5, 7.5 or 10 MPa peak positive pressure (Table 1). At 2.35 MHz, animals were exposed only at the highest exposure level, 10 MPa. During the 5min exposure period, four separate, evenly spaced, 0.05~mL boluses of commercial concentration Albunex@ were injected for a total of 0.2 mL. Injections were given at 1, 2, 3 and 4 min following the start of exposure. Pressure on the suspension during injection was minimized to avoid damage to the bubbles. Control animals were exposed at 1.15 MHz at 10 MPa and experienced the same protocol as above, except they were injected with four, 0.05~mL boluses of saline instead of Albunex@. Two independent experiments were conducted to determine the extent of hemolysis
HemoIysis ipr viva from exposure to pulsed ultrasound
in the mouse from exposure to lo-ps pulses of 1.15MHz ultrasound. The first experiment, performed on C3H mice (Jackson Laboratories, Bar Harbor, ME, USA), was intended to be exploratory and provide a basis for choice of exposure levels for a subsequent experiment that would establish the threshold for hemolysis with greater precision. The second experiment, using CD1 mice (Harlan Sprague Dawley, Inc., Indianapolis, IN, USA), became essentially a replicate of the first experiment. The only difference, other than mouse strain, was that more mice were used in the second experiment to reduce uncertainties in the determination of the threshold for hemolysis. After exposure, the animal was killed by cervical dislocation, its chest was opened and blood was collected by heart puncture with a 25ga needle, using a 1-mL syringe flushed with heparin. The heart and lungs were then removed and inspected for gross hemorrhage. A 0.5-mL sample of whole blood was diluted with 1 mL of physiological saline and centrifuged at 10,000 rpm for 10 min. The supernatant was removed and analyzed for hemoglobin concentration by the cyanmethemoglobin method (Drabkin and Austin 1932). Samples were diluted with a commercial reagent (Fisher Diagnostic Cyanmethemoglobin) and their absorbance measured at 540 nm with a calorimeter. Total hemoglobin in whole blood (also diluted as 0.5 mL of whole blood to 1 mL of physiological saline) was determined by cyanmethemoglobin analysis of uncentrifuged samples. At the start of each experiment, the hub of the 25-ga Butterfly@ needle was filled with saline (Fig. 1). Initial injections of Albunex@ were, therefore, diluted by saline. The extent of this dilution and the effective volume of Albunex@ delivered to the mouse from the Butterfly@ injection apparatus was determined in a separate series of experiments. First, we replaced the Albunex@ suspension in the syringe with whole murine blood. (We can analyze hemoglobin concentration more reliably than we can measure the concentration of Albunexe.) Then, the same protocol used for injecting Albunex” into the mouse vein was repeated but, in this case, the tip of the Butterfly@ needle was placed in a known volume of cyanmethemoglobin reagent. A baseline was obtained by injecting the same quantity of blood directly from the syringe into the reagent. Comparison of the amount of blood leaving the Butterfly@’ needle apparatus (Fig. 1) to that ejected from the syringe directly yielded the efficiency of the Butterfly@ injection system. Attenuation of the chest wall was measured at 1.15 MHz to obtain an estimate of the exposure level at the surface of the heart. Animals were anesthetized,
0
D.
DALECKI
309
et al.
shaved and depilated as above, and cervically dislocated. The back was opened, and the heart and lungs were removed. An Imotec Type 80-0.5-4.0 PVDF needle hydrophone was positioned at the acoustic focus. The mouse carcass was positioned in the water tank so that the needle hydrophone was inside the chest cavity at the position normally occupied by the heart. Transmission measurements through the chest wall were made by comparing the field measured by the hydrophone with and without the mouse in place. Attenuation measurements were made at several locations on the chest wall and the values averaged. Statistical analyses compared the thresholds for hemolysis in the two strains and the magnitudes of suprathreshold hemolysis between animals with Albunex@ and those injected with saline. The dependence of the internal pressure of a bubble upon the acoustic pressure is highly nonlinear and is characterized by a very abrupt increase in collapse pressure near the “threshold” for inertial cavitation. In some way, the degree of hemolysis depends upon the violence of the collapse of these bubbles. In the absence of a mechanistic model for the overall relationship between hemolysis and acoustic exposure, it is useful to describe the relationship between percent lysis H and pressure amplitude p by a nonlinear equation with three parameters, a threshold value PT for the effect of the exposure, a constant (mean) level B of response below the threshold and a slope S for the increase in the level of the response above the threshold value.
H=
B i B + S(p - P.,)
ifp 5 PT if p > PT
Through nonlinear regression analysis (McCullagh and Nelder 1989)) values of the three parameters that minimize the sum of the squares of the deviations of observations from the model are chosen. A first regression analysis estimated separate thresholds for each of the two strains. A second analysis, involving a model having a single threshold (but separate background levels and slopes) for both strains, was used to provide a statistical comparison of the separate threshold estimates. To test whether hemolysis in the animals injected with saline was consistent with the model for the Albunex@ data, the data were analyzed with and without the saline observations. The results of the two regression models were then compared statistically, to test the null hypothesis that the saline data were consistent with the regression model for the Albunex@ data. If the null hypothesis is rejected, we conclude that the saline data cannot be described by the same regression.
310
Ultrasound in Medicine and Biology
RESULTS Some lung hemorrhage was observed in all animals exposed to ultrasound. However, the lung damage was not great enough to compromise the survival of the animals over the period of the experiment. The exposure levels used in these studies, even several millimeters off the axis of the sound field, were well above the threshold for lung hemorrhage (Child et al. 1990). Accuracy of the positioning technique was confirmed by the observation on autopsy that the extent of lung hemorrhage on the left and right sides of the heart was equivalent. Two mice showed signs of hemorrhage on the chest wall and on the right ventricle. These mice were exposed at either 5 or 10 MPa and received Albunex@ injections. Earlier studies have shown that without Albunex@, there were no observable effects on heart function from exposures comparable to those used in the present investigation (Dale&i et al. 1993). Pulse lengths in the range 1- 10 ms and acoustic pressures above 2 MPa are required to alter the rhythm of murine (MacRobbie et al. 1997, in press} and amphibian hearts (Dale&i et al. 1993)) but when pulse lengths comparable to those used in Doppler ultrasound ( zz 10 ps) were used, no effects on heart function were detected (Dale&i et al. 1993). At the start of each exposure, the hub of the 25-ga But&lye needle was filled with saline (Fig. 1) . As a result, the first and second boluses of the injections of Albunex* were diluted by saline. Experiments to determine the actual volume of Albunex@ ejected from the injection system are summarized in Table 2. Four 0.05 mL boluses of whole blood were injected sequentially into a known volume of cyanmethemoglobin reagent, removing the syringe from the Butterfly@ needle assembly between each bolus as in the animal experiments. Three trials were performed, and the percentage of blood injected was determined for each of the four sequential injections. Upon averaging the results, it was found that
Table 2. Summaryof volume ejectedfrom the Butterfly@ injection apparatus. Injection no.
Trial 1
Trial 2
Trial 3
1
0% 18%
0% 13%
0%
23% 23%
22% 24%
2 3 4
7% 17% 22%
Mean
SD
0%
0%
13% 21% 23%
5% 3%
1%
Four 0.05~mL injections (column 1) of whole blood were delivthrough the Butterfly@ injection apparatus (Fig. 1). In three trials (columns 2-4), the percentages of blood ejected from the ButterSye needle were measured for each of the four sequential injections. The last two columns are the means and standard deviations of the percentages of the three trials.
Volume 23, Number 2, 1997
Table 3. Resultsof exploratory study at 1.15 MHz with C3H mice. Percent hemolysis Exposure 0 MPa 2.5 MPa 5 MPa 7.5 MPa 10 MPa 10 MPalSaline
n
Mean
SE
2 2 3 3 3 3
0.6 0.5 0.9 4.5 4.8 0.4
0.2 0.2 0.2 0.7 0.8 0.06
The first five rows are results of exposures with Albunexe injections. The last row is the result of the control exposure with saline injections. Results are presented as mean percentage of hemolysis and standard error (SE), where n is the number of mice exposed.
no blood left the Butterfly@ needle following the first injection, while 13%, 21% and 23% of the blood was ejected on the second, third and fourth injections, respectively. Of the 0.2 mL of blood initially leaving the syringe, only 57% (0.11 mL) actually was ejected from the Butterfly@ apparatus. During the 5-min exposure period, therefore, each animal received approximately 0.1 mL of commercial concentration Albunex@. Furthermore, the exposure time during which contrast agent was present in the blood of the animal was - 3 min. Total blood volume for a mouse is approximately 2-2.5 mL. This gives an average Albunex@ concentration of up to 5%. Local concentrations could be transiently higher. Losses in the injection system and natural losses in viva would decreasethe actual concentrations in the blood during the experiments. Results at 1.15 MHz for the exploratory study using C3H mice are shown in Table 3, and the followup study with CD1 mice is summarized in Table 4. Percentages of hemolysis for both studies are presented in Fig. 2. At 1.15 MHz, the thresholds (standard devia-
Table 4. Resultsof follow-up study at 1.15 MHz with CD1 mice. Percent hemolysis Exposure 0 MPa 2.5 MPa 5 MPa 7.5 MPa 10 MPa 10 MPalSaline
n
Mean
SE
6 6 6 6 6 6
0.4 0.3 1.2 2.0 3.0 0.3
0.06 0.04 0.2 0.7 0.9 0.07
ered
The first five rows are results of exposures with Albunexe injections. The last row is the result of the control exposure with saline injections. Results are presented as mean percentage of hemolysis and standard error (SE), where n is the number of mice exposed.
Hemolysis in
vivo
from exposure to pulsed ultrasound
6 5
q
C3H (saline)
4 .t? A a 6
3
0 , 0
II 2 Incident
II 4 Peak Positive
II 6 Pressure
II
II
a
IO (MPa)
Fig. 2. Hemolysisin C3H and CD1 strainsof mice. Means and standarderrors are plotted as a function of the peak positive pressureat the surfaceof the animal. The curves are drawn on the assumptionof a commonthresholdbut different slopesand backgroundlevels for the two strains as discussedin the text. Ultrasoundcausedno measurable hemolysisin animalsinjectedwith salineandexposedat I. 15 MHz, nor in animalsinjectedwith Albunexe andexposedto 2.35-MHz ultrasound.
tion) for hemolysis with Albunex@ for the C3H and CD1 mice analyzed separately are 3.8 ( 1.1) MPa and 2.9 ( 1.0) MPa, respectively. An approximate F-test indicated that these threshold values do not differ significantly (p = 0.67). In light of this finding, the results (Fig. 2) are presented under the assumption that the thresholds were the same (but slopes and backgrounds were different). The threshold for hemolysis in viva from this analysis was 3.5 (0.9) MPa peak positive pressure at the surface of the animal at 1.15 MHz. Using the data in Table 1, we find that these values are equivalent to - 2 MPa negative pressure and 200 W cmp2 pulse average intensity. Measurements of chest wall attenuation were made for three chest wall samples for each strain of mice. In contrast with earlier measurements of attenuation near the back of the rib cage (Child et al. 1990)) the attenuation over the location of the heart was not a strong function of position. (Separation of the ribs near the sternum is not as distinct as it is on the lateral and dorsal surfaces of the rib cage.) The variation in attenuation with acoustic pressure and the attenuation difference between strains were not statistically significant. Pooling all of the data for both strains gave an attenuation (standard error) of 1.2 (0.1) dB for the peak positive pressure and 0.6 (0.06) dB for the negative pressure. These values are statistically significantly different (p = 0.001) . Correcting for attenuation
0 D. DALECKI
et al.
311
through the chest wall, the threshold at the surface of the heart is 3.0 MPa peak positive pressure, 1.9 MPa maximum negative pressure and - 180 W cmB2 spatial peak pulse average intensity. Figure 2 also shows the levels of hemolysis produced in animals of both strains that had been injected with saline. An additional regression indicated that the saline observations were inconsistent with the fitted curves (p < 0.0001). This result did not change substantially when a weighted fit was used where the weights were the standard deviations of the groups at each exposure level. Also shown in Fig. 2 is the data point of 0.46% (0.05) hemolysis for mice injected with Albunex@ and exposed to 2.35 MHz with a peak amplitude of 10 MPa. Since the hemolysis at 2.35 MHz with the maximum output of the transducer used in the test was indistinguishable from sham levels of hemolysis (p = 0.43), we were unable to define a threshold at this frequency. It is noteworthy, however, that the 2.35MHz hemolysis is less than one eighth of the level observed at peak positive pressures three times greater than the l-MHz threshold level, suggesting a strong dependence of the threshold itself upon frequency. DISCUSSION The concentration of Albunex@ in the mouse heart as estimated above is roughly an order of magnitude greater than the maximum used in clinical contrast procedures. A direct comparison of these numbers may not be appropriate, however. We know very little about the survival of Albunex@ bubbles under our experimental conditions or in patients under clinical conditions. Furthermore, because of the slow administration of the agent, the bubbles are probably distributed throughout the blood volume of the animal and, therefore, the concentration of bubbles achieved in the mouse in these experiments may not be significantly greater than that in the bolus of contrast agent that is achieved in clinical procedures through an injection lasting for only a few seconds. At best, we can say that bubble concentrations in these experiments are very roughly equivalent to those used in clinical applications. Clearly, when contrast agents are present, hemolysis can take place in uivo with pulse lengths and pulse repetition frequencies that are used in diagnosis. The association of hemolysis with the presence of exogenous contrast bubbles strongly implicates acoustic cavitation as the physical mechanism for the phenomenon. Failure to obtain hemolysis in the absence of added bubbles suggests that the number of appropriate nuclei in the blood of normal mammals is very small. Note that bubbles large enough to be detectable with ultra-
312
Ultrasound in Medicine and Biology
sonic scanning equipment were created in the blood of dogs without exogenous contrast bubbles (Ivey et al. 1997, in press). However, bursts of ultrasound longer than 1 ms and extremely large acoustic pressures were required, e.g., - 20 MPa for a 12-ms burst at 1.8 MHz. Brayman et al. ( 1996) report a “pseudothreshold” for hemolysis in whole human blood in vitro of approximately 3 MPa peak positive pressure with a lo-ps pulse. This is remarkably similar to our observed value for hemolysis in Vito, considering the great differences in the environment for cavitation in the two studies. The exposure itself was almost identical in the two experimental protocols. Only the duty cycle differed: 0.01 in Brayman et al. (1996) compared with 0.001 here. In view of this similarity of threshold values, an important finding from Brayman et al. ( 1996) should be mentioned here. They used the term “pseudothreshold” because they observed small, but statistically significant, levels of hemolysis below the “pseudothreshold” pressure level. Thus, we have to acknowledge the possibility that a small, undetectable amount of hemolysis may occur even in viuo below the threshold levels reported here. The question is largely academic but interesting. We were unable to determine the actual threshold for hemolysis at 2.35 MHz because of limitations in the output of our 2.35-MHz source. After a 10 MPa peak positive pressure exposure at that frequency, the hemolysis was indistinguishable from sham levels, i.e., less than one eighth of the hemolysis at the corresponding 1. l-MHz exposure. The strong dependence of the threshold for hemolysis upon frequency is similar to the frequency dependence of the threshold for hemorrhage in intestine (Dale&i et al. 1995). In both sites, the biological effects are dependent upon the presence of microbubbles, and the most likely candidate for a physical mechanism at each site is inertial cavitation. Hemolysis at the 4% level in a patient would be clinically significant. But, there is no reason from this study to expect this to result from the current diagnostic use of ultrasound. Although further investigation of the subject is warranted, the results presented in Fig. 2 are reassuring. First, the data show a sharp threshold with a peak positive pressure at the surface of the heart of approximately 3.0 MPa. This corresponds to a pulse average intensity of - 180 W cmV2, a negative pressure extreme of - 1.9 MPa and a mechanical index of 1.8. Few, if any, commercial diagnostic devices have derated output levels corresponding to mechanical indices > 2. Second, the threshold for hemolysis clearly is much greater at frequencies most commonly used clinically. This is evident from
Volume 23, Number 2, 1997
our results, since hemolysis at 2.35 MHz was indistinguishable from sham-exposed animals. In fact, the strong dependence of the threshold upon frequency suggests that the mechanical index, which is proportional to the square root of the frequency, is not a useful predictor of hemolysis. Third, the extent of hemolysis will decrease as the pulse length is reduced from the 10 /LCLS used in our experiments. Fourth, a larger fraction of the animal’s blood was exposed in our experiments than would be the case for a human patient. Finally, the effective exposure time was greater in these experiments than would be the case in the vast majority of clinical procedures. In most transthoracic and transesophageal echocardiographic studies, the Albunex@ bolus would only be exposed to ultrasound in the right ventricle for 5-20 beats and in the left ventricle for lo-40 beats, whereas in our study, the blood in the heart probably contained bubbles for a large fraction of the 5-min treatment period. The combination of recirculating contrast agents with examinations that have long dwell times should be evaluated from the point of view of safety if acoustic pressures exceed threshold levels. CONCLUSIONS These hemolysis studies suggest that inertial cavitation can take place in viva if appropriate nuclei are present. Since the degree of hemolysis in the absence of exogenous microbubbles is indistinguishable from sham-exposed blood, we must conclude that the number of spontaneous nuclei in normal blood in viva is very small. The threshold for hemolysis at 1 MHz in murine blood is approximately 3 MPa peak positive pressure or 1.9 MPa negative pressure. This, together with other characteristics of the phenomenon, indicates that there is little likelihood of cavitation-induced damage in the cardiovascular system in most diagnostic procedures. Acknowledgments-This work was supported in part by USPHS Grants No. DK39796, HL30616 and HL50497. The Albunexe used in this study was manufactured by Molecular Biosystems, Inc., San Diego, CA, USA, and supplied by Mallinckrodt Group, Inc., St. Louis, MO, USA.
REFERENCES Brayman AA, Azadniv M, Makin IRS, et al. Effect of a stabilized microbubble echo contrast agent on hemolysis of human erythrocytes exposed to high-intensity pulsed ultrasound. Echocardiography 1995; 12:13-21. Brayman AA, Azadniv M, Cox C, Miller MW. Hemolysis of Albunexo-supplemented, 40% hematocrit human erythrocytes in vitro by 1 MHz pulsed ultrasound: Acoustic pressure and pulse length dependence. Ultrasound Med Biol 1996;22:927-938.
Hemolysis in vivo from exposure to pulsed ultrasound 0 D. DALECKI et al. Carstensen EL, Kelly P, Church CC, et al. Lysis of erythrocytes by exposure to cw ultrasound. Ultrasound Med Biol 1993; 19:147165. Child SZ, Hartman CL, Schery LA, Carstensen EL. Lung damage from exposure to pulsed ultrasound. Ultrasound Med Biol 1990; 16:817-825. Dale& D, Keller BB, Raeman CH, Carstensen EL. Effects of pulsed ultrasound on the frog heart: I. Thresholds for changes in cardiac rhythm and aortic pressure. Ultrasound Med Biol 1993; 19:385390. Dale&i D, Raeman CH, Child SZ, Carstensen EL. Intestinal hemorrhage from exposure to pulsed ultrasound. Ultrasound Med Biol 1995;21:1067-1072. Drabkin DL, Austin JH. Spectrophotometric studies I. Spectrophotometric constants for common hemoglobin derivatives in human, dog and rabbit blood. J Biol Chem 1932;98:719-733. Dunn F, Averbuch AJ, O’Brien WD Jr. A primary method for the
313
determination of ultrasonic intensity with the elastic sphere radiometer. Acustica 1977;38:58-61. Ivey JA, Gardner EA, Fowlkes JB, Rubin JM, Carson PL. Acoustic generation of intraarterial contrast boluses. Ultrasound Med Biol 1997; (In press). MacRobbie AG, Raeman CH. Child SZ, Dale&i D. Thresholds for premature contractions in mouse hearts exposed to pulsed ultrasound. Ultrasound Med Biol 1997: (In press). McCullagh P, Nelder JA. Generalized linear models.’ 2nd ed. London: Chapman and Hall, 1989. Williams AR, Miller DL, Gross DR. Haemolysis in vivo by therapeutic intensities of ultrasound. Ultrasound Med Biol 1986; 12:501509. Wang YS, Watmough DJ. Haemolysis of red blood cells in vitro and in vivo by ultrasound at 0.75 MHz and at therapeutic intensity levels. In: Millner R. Rosenfeld E. Cobet U. eds. Ultrasound interactions in biology and medicine. New York: Plenum Press, 1983:179-184.
PI1 SO301-5629( 96)00203-7
ELSEVIER
@Original
Ultrasound in Med. & Biol., Vol. 23. No. 2, PP. 307-313, 1997 0 1997 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/97 $17.00 + .OO
Contribution
HEMOLYSIS D. DALECKI,*’
IN VZVO FROM
EXPOSURE
TO PULSED
ULTRASOUND
C. H. RAEMAN,+ S. Z. CHILD,*+ C. Cox,+* C. W. FRANCIS,‘~ R. S. MELTZER+~’ and E. L. CARSTENSEN*+
Departmentsof *Electrical Engineering,‘Biostatistics, The “Cardiologyand “HematologyUnits of the Departmentof Medicine, and +TheRochesterCenterfor BiomedicalUltrasound,University of Rochester,Rochester,NY (Received 9 April 1996; in jnal fom 1 October 1996)
Abstract-Ultrasonicaily induced hemolysisin viva was studied when a commercial ultrasound contrast agent, Albunexo, was present in the blood. Murine hearts were exposed for 5 min at either 1.15 or 2.35 MHz with a pulse length of 10 w and pulse repetition frequency of 100 Hz. During the exposure period, four bolusesof Albunexo were injected into a tail vein for a total of - 0.1 mL of Albunexo. Following exposure,blood wascollected by heart puncture and centrifuged, and the plasmawasanalyzed for hemoglobin concentration. With Albunexo present in the blood, the threshold for hemolysisat 1.15 MHz was 3.0 2 0.8 MPa (mean 2 SD) peak positive pressure( - 1.9 MPa negative pressure,- 180W cn-’ pulseaverage intensity). For the highest exposure levels (10 MPa peak positive pressureat the surface of the animal), the mean value for hemolysiswas - 4% at 1.15 MHz and 0.46% at 2.35 MHz, i.e., the threshold at 2.35 MHz is >lO MPa peak positive pressure.In contrast, hemolysisin control mice receiving saIine iqjections at 10 MPa or sham-exposed(0 MPa) mice receiving Albunexo was - 0.4%. 0 1997 World Federation for Ultrasound in Medicine & Biology. Key Words: Ultrasound, Hemolysis,Contrast agent, Bioeffect.
to differentiate between these two alternatives. A partial answer to this question has been given in the studies of Brayman et al. ( 1995, 1996), who found statistically significant increases in free hemoglobin in whole human blood containing the contrast agent, Albunexe’ (Molecular Biosystems, San Diego, CA, USA, and Mallinckrodt Medical, Inc., Hazelwood, MO, USA) after exposure in vitro to pulses as short as 5 ~CLS (duty cycle of 0.01) and at acoustic pressures as small as
INTRODUCTION Attempts have been made to detect ultrasound-induced
hemolysis in viva. Even when hemolysis was detected after exposure to continuous-wave ultrasound, the evidence suggested that the phenomenon was primarily a thermal effect that occurred as blood perfused organs were heated by ultrasound (Williams et al. 1986; Wong and Watmough 1983). But, thus far, there have been no reports of hemolysis in viva with pulses as short as those used in diagnostic ultrasound. Under exposure conditions that avoid excessive heating, there is a high probability that hemolysis is indicative of acoustic cavitation (Carstensen et al. 1993). Failure to observe hemolysis in viva may indicate either that it is difficult to excite bubbles in the environment provided by whole blood or that nuclei appropriate for acoustic cavitation may be very rare in blood in viva. The recent availability of stabilized microbubbles for use as ultrasonic contrast agents makes it possible
0.5 MPa (Brayman et al. 1996). Without added bub-
bles, no significant hemolysis was seen even with much higher pressures and longer pulse lengths (Brayman et al. 1995). We report here qualitatively similar findings of ultrasound-induced hemolysis in mice. EXPERIMENTAL
METHODS
The acoustic sources were 1.15- and 2.35-MHz focused transducers fabricated from 1.5- and l-inch (38.1 and 25.4 mm) diameter piezoceramic elements, respectively, cemented to planoconcave aluminum lenses. The transducer was mounted on the bottom of a plastic exposure tank filled with deionized, degassed
Address correspondence to: Dr. Diane Dale&i, Department of Electrical Engineering, University of Rochester, Rochester, NY 14627. 307
Ultrasound in Medicine and Biology
308
water at room temperature (- 21°C). The acoustic field was measured daily with a bilaminar, PVDF membrane hydrophone (Marconi Research Centre, Chelmsford, UK) calibrated by comparison to a sphere radiometer (Dunn et al. 1977). The focal length of the 1.15MHz source was 6.5 cm, and the -6 dB beam width at the focus was 3.5 mm. The focal length of the 2.35MHz source was 4.7 cm, and the -6 dB beam width at the focus was 3.0 mm. For perspective, the murine heart is approximately 6 mm X 9 mm. Fields were calibrated for peak positive pressures of 2.5, 5, 7.5 and 10 MPa. In addition, spatial peak, pulse average intensities were obtained by integrating the square of the measured pressures over the duration of the pulse using a digital oscilloscope (LeCroy Model 9400, Chestnut Ridge, NY, USA) and dividing the result by the acoustic impedance of water. The results are reported in terms of the peak positive pressure. However, Table 1 presents the corresponding negative pressures and spatial peak, pulse average intensities for each exposure level. Note that these are field parameters in water at the surface of the animal. Mice weighing approximately 35 g were anesthetized with ketamine (200 mg/kg) and Rompun (10 mg/kg). All animal protocols were approved by the University of Rochester Committee on Animal Resources. The ventral chest surface was shaved and depilated ( NeetTM , Reckitt & Colman Inc., Wayne, NJ, USA) to minimize trapped air on the skin surface. Preliminary autopsies of several mice determined the mouse heart was located approximately 1 cm cephalad to the caudal tip of the sternum and 2-3 mm to the animal’s left side. This location was marked on the skin surface with an ink dot to aid in positioning the heart in the acoustic focus. The animal’s circulatory system was accessed through a tail vein using a modified, 25-ga Butterfly@ needle (Abbott Laboratories, Chicago, IL, USA) as shown in Fig. 1. The Butterfly@ apparatus was sealed with an injection cap to allow for multiple injections during ultrasound exposure. A 1-mL syringe with a 26-ga needle was filled with 0.2 mL of undiluted Albunex@ (or saline). For each injecTable 1. Exposure parameters. f (MHz) 1.15 1.15 1.15 1.15 2.35
P+ (MPd
2.5 5.0 7.5 10.0 10.0
P-
NW
1.8 2.7 3.2 3.6 4.4
Is,
(W
cm-‘)
120 3.50 590 790 1025
f = frequency; p+ = peak positive pressure; p- = maximum negative pressure; IsPPA= spatial peak pulse average intensity.
Volume 23, Number 2, 1997
Fig. 1. Butterflya’ injection apparatus.A 25G modifiedButterfly@needle,fitted with an injection cap, wasinsertedinto the tail vein of a mouse.A 1 mL.syringe with a 26G needle was filled with either 0.2 nL of Albunex@or saline and insertedinto the shankof the Butterfly* needle.Four separate 0.05~mLinjectionsweregiven at l-mm intervalsduring the 5-minexposureduration.Betweeninjections,the syringe with 26G needlewas removed from the injection cap and gently rolled to dispersethe contrastagent.
tion into the animal, the syringe was first rotated to disperse the contrast agent bubbles uniformly in the suspension, and then the needle was inserted through the injection cap to the shank of the Butterfly@ needle (Fig. 1) . Exploratory observations, using a commercial ultrasound scanner operating at 7 MHz, confirmed that Albunexm, administered in this way, reached the heart. Each animal was mounted horizontally, ventral surface down, on a plastic small animal holder with the tail and Butterfly@ injection apparatus taped to the holder. The animal and holder were then attached to a three-axis positioning system (Velmex, East Bloomfield, NY, USA). No part of the support apparatus was exposed directly to ultrasound. The depth of water in the exposure tank was equal to the acoustic focal length. Coordinates from a fixed pointer in the water tank to the acoustic focus were determined daily with the hydrophone. These coordinates were used to position the marked dot on the animal’s chest at the acoustic focus. Each mouse was exposed for 5 min to lo-ps pulses of ultrasound at a repetition frequency of 100 Hz. At 1.15 MHz, exposure levels were either 0 (sham), 2.5, 5, 7.5 or 10 MPa peak positive pressure (Table 1). At 2.35 MHz, animals were exposed only at the highest exposure level, 10 MPa. During the 5min exposure period, four separate, evenly spaced, 0.05~mL boluses of commercial concentration Albunex@ were injected for a total of 0.2 mL. Injections were given at 1, 2, 3 and 4 min following the start of exposure. Pressure on the suspension during injection was minimized to avoid damage to the bubbles. Control animals were exposed at 1.15 MHz at 10 MPa and experienced the same protocol as above, except they were injected with four, 0.05~mL boluses of saline instead of Albunex@. Two independent experiments were conducted to determine the extent of hemolysis
HemoIysis ipr viva from exposure to pulsed ultrasound
in the mouse from exposure to lo-ps pulses of 1.15MHz ultrasound. The first experiment, performed on C3H mice (Jackson Laboratories, Bar Harbor, ME, USA), was intended to be exploratory and provide a basis for choice of exposure levels for a subsequent experiment that would establish the threshold for hemolysis with greater precision. The second experiment, using CD1 mice (Harlan Sprague Dawley, Inc., Indianapolis, IN, USA), became essentially a replicate of the first experiment. The only difference, other than mouse strain, was that more mice were used in the second experiment to reduce uncertainties in the determination of the threshold for hemolysis. After exposure, the animal was killed by cervical dislocation, its chest was opened and blood was collected by heart puncture with a 25ga needle, using a 1-mL syringe flushed with heparin. The heart and lungs were then removed and inspected for gross hemorrhage. A 0.5-mL sample of whole blood was diluted with 1 mL of physiological saline and centrifuged at 10,000 rpm for 10 min. The supernatant was removed and analyzed for hemoglobin concentration by the cyanmethemoglobin method (Drabkin and Austin 1932). Samples were diluted with a commercial reagent (Fisher Diagnostic Cyanmethemoglobin) and their absorbance measured at 540 nm with a calorimeter. Total hemoglobin in whole blood (also diluted as 0.5 mL of whole blood to 1 mL of physiological saline) was determined by cyanmethemoglobin analysis of uncentrifuged samples. At the start of each experiment, the hub of the 25-ga Butterfly@ needle was filled with saline (Fig. 1). Initial injections of Albunex@ were, therefore, diluted by saline. The extent of this dilution and the effective volume of Albunex@ delivered to the mouse from the Butterfly@ injection apparatus was determined in a separate series of experiments. First, we replaced the Albunex@ suspension in the syringe with whole murine blood. (We can analyze hemoglobin concentration more reliably than we can measure the concentration of Albunexe.) Then, the same protocol used for injecting Albunex” into the mouse vein was repeated but, in this case, the tip of the Butterfly@ needle was placed in a known volume of cyanmethemoglobin reagent. A baseline was obtained by injecting the same quantity of blood directly from the syringe into the reagent. Comparison of the amount of blood leaving the Butterfly@’ needle apparatus (Fig. 1) to that ejected from the syringe directly yielded the efficiency of the Butterfly@ injection system. Attenuation of the chest wall was measured at 1.15 MHz to obtain an estimate of the exposure level at the surface of the heart. Animals were anesthetized,
0
D.
DALECKI
309
et al.
shaved and depilated as above, and cervically dislocated. The back was opened, and the heart and lungs were removed. An Imotec Type 80-0.5-4.0 PVDF needle hydrophone was positioned at the acoustic focus. The mouse carcass was positioned in the water tank so that the needle hydrophone was inside the chest cavity at the position normally occupied by the heart. Transmission measurements through the chest wall were made by comparing the field measured by the hydrophone with and without the mouse in place. Attenuation measurements were made at several locations on the chest wall and the values averaged. Statistical analyses compared the thresholds for hemolysis in the two strains and the magnitudes of suprathreshold hemolysis between animals with Albunex@ and those injected with saline. The dependence of the internal pressure of a bubble upon the acoustic pressure is highly nonlinear and is characterized by a very abrupt increase in collapse pressure near the “threshold” for inertial cavitation. In some way, the degree of hemolysis depends upon the violence of the collapse of these bubbles. In the absence of a mechanistic model for the overall relationship between hemolysis and acoustic exposure, it is useful to describe the relationship between percent lysis H and pressure amplitude p by a nonlinear equation with three parameters, a threshold value PT for the effect of the exposure, a constant (mean) level B of response below the threshold and a slope S for the increase in the level of the response above the threshold value.
H=
B i B + S(p - P.,)
ifp 5 PT if p > PT
Through nonlinear regression analysis (McCullagh and Nelder 1989)) values of the three parameters that minimize the sum of the squares of the deviations of observations from the model are chosen. A first regression analysis estimated separate thresholds for each of the two strains. A second analysis, involving a model having a single threshold (but separate background levels and slopes) for both strains, was used to provide a statistical comparison of the separate threshold estimates. To test whether hemolysis in the animals injected with saline was consistent with the model for the Albunex@ data, the data were analyzed with and without the saline observations. The results of the two regression models were then compared statistically, to test the null hypothesis that the saline data were consistent with the regression model for the Albunex@ data. If the null hypothesis is rejected, we conclude that the saline data cannot be described by the same regression.
310
Ultrasound in Medicine and Biology
RESULTS Some lung hemorrhage was observed in all animals exposed to ultrasound. However, the lung damage was not great enough to compromise the survival of the animals over the period of the experiment. The exposure levels used in these studies, even several millimeters off the axis of the sound field, were well above the threshold for lung hemorrhage (Child et al. 1990). Accuracy of the positioning technique was confirmed by the observation on autopsy that the extent of lung hemorrhage on the left and right sides of the heart was equivalent. Two mice showed signs of hemorrhage on the chest wall and on the right ventricle. These mice were exposed at either 5 or 10 MPa and received Albunex@ injections. Earlier studies have shown that without Albunex@, there were no observable effects on heart function from exposures comparable to those used in the present investigation (Dale&i et al. 1993). Pulse lengths in the range 1- 10 ms and acoustic pressures above 2 MPa are required to alter the rhythm of murine (MacRobbie et al. 1997, in press} and amphibian hearts (Dale&i et al. 1993)) but when pulse lengths comparable to those used in Doppler ultrasound ( zz 10 ps) were used, no effects on heart function were detected (Dale&i et al. 1993). At the start of each exposure, the hub of the 25-ga But&lye needle was filled with saline (Fig. 1) . As a result, the first and second boluses of the injections of Albunex* were diluted by saline. Experiments to determine the actual volume of Albunex@ ejected from the injection system are summarized in Table 2. Four 0.05 mL boluses of whole blood were injected sequentially into a known volume of cyanmethemoglobin reagent, removing the syringe from the Butterfly@ needle assembly between each bolus as in the animal experiments. Three trials were performed, and the percentage of blood injected was determined for each of the four sequential injections. Upon averaging the results, it was found that
Table 2. Summaryof volume ejectedfrom the Butterfly@ injection apparatus. Injection no.
Trial 1
Trial 2
Trial 3
1
0% 18%
0% 13%
0%
23% 23%
22% 24%
2 3 4
7% 17% 22%
Mean
SD
0%
0%
13% 21% 23%
5% 3%
1%
Four 0.05~mL injections (column 1) of whole blood were delivthrough the Butterfly@ injection apparatus (Fig. 1). In three trials (columns 2-4), the percentages of blood ejected from the ButterSye needle were measured for each of the four sequential injections. The last two columns are the means and standard deviations of the percentages of the three trials.
Volume 23, Number 2, 1997
Table 3. Resultsof exploratory study at 1.15 MHz with C3H mice. Percent hemolysis Exposure 0 MPa 2.5 MPa 5 MPa 7.5 MPa 10 MPa 10 MPalSaline
n
Mean
SE
2 2 3 3 3 3
0.6 0.5 0.9 4.5 4.8 0.4
0.2 0.2 0.2 0.7 0.8 0.06
The first five rows are results of exposures with Albunexe injections. The last row is the result of the control exposure with saline injections. Results are presented as mean percentage of hemolysis and standard error (SE), where n is the number of mice exposed.
no blood left the Butterfly@ needle following the first injection, while 13%, 21% and 23% of the blood was ejected on the second, third and fourth injections, respectively. Of the 0.2 mL of blood initially leaving the syringe, only 57% (0.11 mL) actually was ejected from the Butterfly@ apparatus. During the 5-min exposure period, therefore, each animal received approximately 0.1 mL of commercial concentration Albunex@. Furthermore, the exposure time during which contrast agent was present in the blood of the animal was - 3 min. Total blood volume for a mouse is approximately 2-2.5 mL. This gives an average Albunex@ concentration of up to 5%. Local concentrations could be transiently higher. Losses in the injection system and natural losses in viva would decreasethe actual concentrations in the blood during the experiments. Results at 1.15 MHz for the exploratory study using C3H mice are shown in Table 3, and the followup study with CD1 mice is summarized in Table 4. Percentages of hemolysis for both studies are presented in Fig. 2. At 1.15 MHz, the thresholds (standard devia-
Table 4. Resultsof follow-up study at 1.15 MHz with CD1 mice. Percent hemolysis Exposure 0 MPa 2.5 MPa 5 MPa 7.5 MPa 10 MPa 10 MPalSaline
n
Mean
SE
6 6 6 6 6 6
0.4 0.3 1.2 2.0 3.0 0.3
0.06 0.04 0.2 0.7 0.9 0.07
ered
The first five rows are results of exposures with Albunexe injections. The last row is the result of the control exposure with saline injections. Results are presented as mean percentage of hemolysis and standard error (SE), where n is the number of mice exposed.
Hemolysis in
vivo
from exposure to pulsed ultrasound
6 5
q
C3H (saline)
4 .t? A a 6
3
0 , 0
II 2 Incident
II 4 Peak Positive
II 6 Pressure
II
II
a
IO (MPa)
Fig. 2. Hemolysisin C3H and CD1 strainsof mice. Means and standarderrors are plotted as a function of the peak positive pressureat the surfaceof the animal. The curves are drawn on the assumptionof a commonthresholdbut different slopesand backgroundlevels for the two strains as discussedin the text. Ultrasoundcausedno measurable hemolysisin animalsinjectedwith salineandexposedat I. 15 MHz, nor in animalsinjectedwith Albunexe andexposedto 2.35-MHz ultrasound.
tion) for hemolysis with Albunex@ for the C3H and CD1 mice analyzed separately are 3.8 ( 1.1) MPa and 2.9 ( 1.0) MPa, respectively. An approximate F-test indicated that these threshold values do not differ significantly (p = 0.67). In light of this finding, the results (Fig. 2) are presented under the assumption that the thresholds were the same (but slopes and backgrounds were different). The threshold for hemolysis in viva from this analysis was 3.5 (0.9) MPa peak positive pressure at the surface of the animal at 1.15 MHz. Using the data in Table 1, we find that these values are equivalent to - 2 MPa negative pressure and 200 W cmp2 pulse average intensity. Measurements of chest wall attenuation were made for three chest wall samples for each strain of mice. In contrast with earlier measurements of attenuation near the back of the rib cage (Child et al. 1990)) the attenuation over the location of the heart was not a strong function of position. (Separation of the ribs near the sternum is not as distinct as it is on the lateral and dorsal surfaces of the rib cage.) The variation in attenuation with acoustic pressure and the attenuation difference between strains were not statistically significant. Pooling all of the data for both strains gave an attenuation (standard error) of 1.2 (0.1) dB for the peak positive pressure and 0.6 (0.06) dB for the negative pressure. These values are statistically significantly different (p = 0.001) . Correcting for attenuation
0 D. DALECKI
et al.
311
through the chest wall, the threshold at the surface of the heart is 3.0 MPa peak positive pressure, 1.9 MPa maximum negative pressure and - 180 W cmB2 spatial peak pulse average intensity. Figure 2 also shows the levels of hemolysis produced in animals of both strains that had been injected with saline. An additional regression indicated that the saline observations were inconsistent with the fitted curves (p < 0.0001). This result did not change substantially when a weighted fit was used where the weights were the standard deviations of the groups at each exposure level. Also shown in Fig. 2 is the data point of 0.46% (0.05) hemolysis for mice injected with Albunex@ and exposed to 2.35 MHz with a peak amplitude of 10 MPa. Since the hemolysis at 2.35 MHz with the maximum output of the transducer used in the test was indistinguishable from sham levels of hemolysis (p = 0.43), we were unable to define a threshold at this frequency. It is noteworthy, however, that the 2.35MHz hemolysis is less than one eighth of the level observed at peak positive pressures three times greater than the l-MHz threshold level, suggesting a strong dependence of the threshold itself upon frequency. DISCUSSION The concentration of Albunex@ in the mouse heart as estimated above is roughly an order of magnitude greater than the maximum used in clinical contrast procedures. A direct comparison of these numbers may not be appropriate, however. We know very little about the survival of Albunex@ bubbles under our experimental conditions or in patients under clinical conditions. Furthermore, because of the slow administration of the agent, the bubbles are probably distributed throughout the blood volume of the animal and, therefore, the concentration of bubbles achieved in the mouse in these experiments may not be significantly greater than that in the bolus of contrast agent that is achieved in clinical procedures through an injection lasting for only a few seconds. At best, we can say that bubble concentrations in these experiments are very roughly equivalent to those used in clinical applications. Clearly, when contrast agents are present, hemolysis can take place in uivo with pulse lengths and pulse repetition frequencies that are used in diagnosis. The association of hemolysis with the presence of exogenous contrast bubbles strongly implicates acoustic cavitation as the physical mechanism for the phenomenon. Failure to obtain hemolysis in the absence of added bubbles suggests that the number of appropriate nuclei in the blood of normal mammals is very small. Note that bubbles large enough to be detectable with ultra-
312
Ultrasound in Medicine and Biology
sonic scanning equipment were created in the blood of dogs without exogenous contrast bubbles (Ivey et al. 1997, in press). However, bursts of ultrasound longer than 1 ms and extremely large acoustic pressures were required, e.g., - 20 MPa for a 12-ms burst at 1.8 MHz. Brayman et al. ( 1996) report a “pseudothreshold” for hemolysis in whole human blood in vitro of approximately 3 MPa peak positive pressure with a lo-ps pulse. This is remarkably similar to our observed value for hemolysis in Vito, considering the great differences in the environment for cavitation in the two studies. The exposure itself was almost identical in the two experimental protocols. Only the duty cycle differed: 0.01 in Brayman et al. (1996) compared with 0.001 here. In view of this similarity of threshold values, an important finding from Brayman et al. ( 1996) should be mentioned here. They used the term “pseudothreshold” because they observed small, but statistically significant, levels of hemolysis below the “pseudothreshold” pressure level. Thus, we have to acknowledge the possibility that a small, undetectable amount of hemolysis may occur even in viuo below the threshold levels reported here. The question is largely academic but interesting. We were unable to determine the actual threshold for hemolysis at 2.35 MHz because of limitations in the output of our 2.35-MHz source. After a 10 MPa peak positive pressure exposure at that frequency, the hemolysis was indistinguishable from sham levels, i.e., less than one eighth of the hemolysis at the corresponding 1. l-MHz exposure. The strong dependence of the threshold for hemolysis upon frequency is similar to the frequency dependence of the threshold for hemorrhage in intestine (Dale&i et al. 1995). In both sites, the biological effects are dependent upon the presence of microbubbles, and the most likely candidate for a physical mechanism at each site is inertial cavitation. Hemolysis at the 4% level in a patient would be clinically significant. But, there is no reason from this study to expect this to result from the current diagnostic use of ultrasound. Although further investigation of the subject is warranted, the results presented in Fig. 2 are reassuring. First, the data show a sharp threshold with a peak positive pressure at the surface of the heart of approximately 3.0 MPa. This corresponds to a pulse average intensity of - 180 W cmV2, a negative pressure extreme of - 1.9 MPa and a mechanical index of 1.8. Few, if any, commercial diagnostic devices have derated output levels corresponding to mechanical indices > 2. Second, the threshold for hemolysis clearly is much greater at frequencies most commonly used clinically. This is evident from
Volume 23, Number 2, 1997
our results, since hemolysis at 2.35 MHz was indistinguishable from sham-exposed animals. In fact, the strong dependence of the threshold upon frequency suggests that the mechanical index, which is proportional to the square root of the frequency, is not a useful predictor of hemolysis. Third, the extent of hemolysis will decrease as the pulse length is reduced from the 10 /LCLS used in our experiments. Fourth, a larger fraction of the animal’s blood was exposed in our experiments than would be the case for a human patient. Finally, the effective exposure time was greater in these experiments than would be the case in the vast majority of clinical procedures. In most transthoracic and transesophageal echocardiographic studies, the Albunex@ bolus would only be exposed to ultrasound in the right ventricle for 5-20 beats and in the left ventricle for lo-40 beats, whereas in our study, the blood in the heart probably contained bubbles for a large fraction of the 5-min treatment period. The combination of recirculating contrast agents with examinations that have long dwell times should be evaluated from the point of view of safety if acoustic pressures exceed threshold levels. CONCLUSIONS These hemolysis studies suggest that inertial cavitation can take place in viva if appropriate nuclei are present. Since the degree of hemolysis in the absence of exogenous microbubbles is indistinguishable from sham-exposed blood, we must conclude that the number of spontaneous nuclei in normal blood in viva is very small. The threshold for hemolysis at 1 MHz in murine blood is approximately 3 MPa peak positive pressure or 1.9 MPa negative pressure. This, together with other characteristics of the phenomenon, indicates that there is little likelihood of cavitation-induced damage in the cardiovascular system in most diagnostic procedures. Acknowledgments-This work was supported in part by USPHS Grants No. DK39796, HL30616 and HL50497. The Albunexe used in this study was manufactured by Molecular Biosystems, Inc., San Diego, CA, USA, and supplied by Mallinckrodt Group, Inc., St. Louis, MO, USA.
REFERENCES Brayman AA, Azadniv M, Makin IRS, et al. Effect of a stabilized microbubble echo contrast agent on hemolysis of human erythrocytes exposed to high-intensity pulsed ultrasound. Echocardiography 1995; 12:13-21. Brayman AA, Azadniv M, Cox C, Miller MW. Hemolysis of Albunexo-supplemented, 40% hematocrit human erythrocytes in vitro by 1 MHz pulsed ultrasound: Acoustic pressure and pulse length dependence. Ultrasound Med Biol 1996;22:927-938.
Hemolysis in vivo from exposure to pulsed ultrasound 0 D. DALECKI et al. Carstensen EL, Kelly P, Church CC, et al. Lysis of erythrocytes by exposure to cw ultrasound. Ultrasound Med Biol 1993; 19:147165. Child SZ, Hartman CL, Schery LA, Carstensen EL. Lung damage from exposure to pulsed ultrasound. Ultrasound Med Biol 1990; 16:817-825. Dale& D, Keller BB, Raeman CH, Carstensen EL. Effects of pulsed ultrasound on the frog heart: I. Thresholds for changes in cardiac rhythm and aortic pressure. Ultrasound Med Biol 1993; 19:385390. Dale&i D, Raeman CH, Child SZ, Carstensen EL. Intestinal hemorrhage from exposure to pulsed ultrasound. Ultrasound Med Biol 1995;21:1067-1072. Drabkin DL, Austin JH. Spectrophotometric studies I. Spectrophotometric constants for common hemoglobin derivatives in human, dog and rabbit blood. J Biol Chem 1932;98:719-733. Dunn F, Averbuch AJ, O’Brien WD Jr. A primary method for the
313
determination of ultrasonic intensity with the elastic sphere radiometer. Acustica 1977;38:58-61. Ivey JA, Gardner EA, Fowlkes JB, Rubin JM, Carson PL. Acoustic generation of intraarterial contrast boluses. Ultrasound Med Biol 1997; (In press). MacRobbie AG, Raeman CH. Child SZ, Dale&i D. Thresholds for premature contractions in mouse hearts exposed to pulsed ultrasound. Ultrasound Med Biol 1997: (In press). McCullagh P, Nelder JA. Generalized linear models.’ 2nd ed. London: Chapman and Hall, 1989. Williams AR, Miller DL, Gross DR. Haemolysis in vivo by therapeutic intensities of ultrasound. Ultrasound Med Biol 1986; 12:501509. Wang YS, Watmough DJ. Haemolysis of red blood cells in vitro and in vivo by ultrasound at 0.75 MHz and at therapeutic intensity levels. In: Millner R. Rosenfeld E. Cobet U. eds. Ultrasound interactions in biology and medicine. New York: Plenum Press, 1983:179-184.