Abstract
Objective:
Bacterial loads can be effectively reduced using cavitation-mediated focused ultrasound, or histotripsy. In this study, gram-negative bacteria (Escherichia coli) in suspension were used as model bacteria to evaluate the effectiveness of two regimens of histotripsy treatments: cavitation histotripsy (CH) and boiling histotripsy (BH).
Methods:
10 mL volumes of Escherichia coli (E. coli) were treated at different negative focal pressure amplitudes and over time periods up to 40 minutes. Cavitation activity was characterized with coaxial passive cavitation detection (PCD) and synchronized plane wave B-mode imaging.
Results:
CH treatments showed a threshold behavior that was consistent with PCD metrics of cavitation. Above the threshold, bacterial inactivation followed a monotonically increasing log-linear relationship that indicated an exponential inactivation rate. BH exhibited no threshold, but instead followed a different monotonically increasing inactivation rate. Inactivation rates were larger for BH at or below the CH threshold, and larger for CH substantially above the threshold. CH studies performed at different pulse lengths at the same duty cycle showed similar inactivation rates, suggesting that at any given pressure amplitude the “on time” was the most important variable for inactivating E. coli. The maximum inactivation was produced by CH at the highest pressure amplitudes used, leading to a log reduction of over 4.2 for a 40 min treatment.
Conclusion:
The results of this study suggest that both CH and BH can be used to inactivate E. coli in suspension, with the optimal regimen depending on the attainable peak negative focal pressure at the target.
Keywords: Histotripsy, cavitation histotripsy, boiling histotripsy, Shock-scattering histotripsy, focused ultrasound, bactericide, inactivation
Introduction
Acoustic fields have been known to inactivate bacteria for over 90 years [1]. A substantial literature on the topic has developed, largely in response to the food industry's desire to reduce microbial contamination of food without the requirement for quality-altering high temperature treatments (see reviews: [2-6]). Erriu et. al., reviewed medical applications of low- and high-frequency acoustic treatments [7]. More recently, the literature has expanded for both thermal and non-thermal ultrasound treatments of biofilms [8-10], as well as bacteria in suspension [11, 12]. Our interest is in improving the treatment of infected abscesses where bacteria can be found in the viscous liquid contents of the abscess, not necessarily as biofilms.
Non-thermal inactivation of bacteria by intense pulsed ultrasound - termed histotripsy - results from inertial cavitation via direct mechanical effects related to the shear forces associated with bubble collapse [13]. The shear forces can arise from volumetric bubble oscillations, non-spherical oscillations, and/or jetting. Earlier results obtained with Escherichia coli (E. coli) in suspension are consistent with this general conclusion. Transmission electron microscopy (TEM) images of treated E. coli bacteria showed disruption of the cytoplasm, cellular debris, and evidence of mechanical shearing of the membrane [11].
Histotripsy includes multiple regimens, two of which are shock scattering cavitation histotripsy [14] (CH) and boiling histotripsy [15] (BH). CH and BH regimens include a range of pulsing parameters that can be generalized as follows: 5-50 cycle pulses with a pulse repetition frequency (PRF) of about 200-2000 Hz for CH, and 1,000-10,000 cycle pulses at a PRF of 1-10 Hz and lower focal pressure levels for BH. For either regimen, the duty cycles are typically 1% or less. Each CH pulse generates a smaller cavitation cloud than a BH pulse, but it is repeated at a higher PRF. Due to the difference in pulse duration, solid tissue displacement and/or liquid streaming induced by acoustic radiation force is more pronounced in BH treatments. The combination of the above effects results in BH lesions in tissue being larger than CH ones within the same treatment time. In this work we selected typical parameters from both CH and BH regimens. Exposures were performed with varying peak negative pressure amplitude and treatment time using the same fixed-focus, stationary 1 MHz transducer. In one case, we evaluated treatments using 2% duty cycle for comparative purposes.
Our previous in vitro work was exclusively performed with CH. In contrast, a pilot in vivo study was performed with both CH and BH parameters on abscesses [16, 17] generated in a large animal (porcine) model [18, 19]. This in vivo study applying CH or BH to the abscesses suggested that BH bactericidal effectiveness was much more variable and less effective on average than CH [17, 20]. However, treatment duration and volumes were not controlled. Because CH and BH inactivate bacteria via mechanical pathways (cavitation-induced shear stresses vs. heating), this confounding result suggested a need for a more controlled environment to quantify similarities and differences between CH and BH for inactivating E. coli.
All studies used the same transducer and custom sample vial system. Instead of moving the transducer’s focus through the sample volume, we relied on cavitation-induced streaming to circulate material through the focus. One advantage of histotripsy is that streaming induced at the focus brings new material into the focus via the acoustic radiation force and Kelvin impulse [21-23]. Cavitation enhances streaming due to the interaction between the sound wave and bubbles. Bubble oscillations themselves create microstreaming patterns that are also known to disrupt cells due to high transient shear forces [24]. Thus, the treatment volume is larger than the acoustic pressure focal volume. This was previously demonstrated in an in vitro study that used a fixed-focus system to treat a 10 mL sample by drawing bacteria into a comparatively small acoustic pressure focal volume [11].
The manuscript is organized as follows: Variable peak negative pressure studies were performed to determine the threshold for E. coli inactivation. Cavitation noise was quantified and used to explain the observed threshold behavior. Treatment time studies were then performed at various pressure amplitudes above the threshold. For CH, different pulse parameters (pressure amplitude, pulse length, duty cycle) were compared. With BH, different pressure amplitudes were compared.
Methods
Culture and Preparation
Escherichia coli FDA strain Seattle 1946 was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and maintained as streak cultures on 3% tryptic soy broth medium and 2% agar. Details of the methods to prepare and use the bacteria have been described in detail earlier [12], except as noted. Bacterial growth was monitored by periodically measuring the optical density at 600 nm (OD600). The culture was used once the bacteria reached the stationary phase, as established from previous growth characteristics studies.
Histotripsy Exposure Apparatus and Treatment Protocols
A custom experimental apparatus (Fig. 1) was used for these experiments. Vials containing the bacterial suspensions were immersed in a cylindrical degassed water bath (initially at ~20 °C) with an attached 1 MHz spherically focused transducer (F# = 1, aperture 85 mm). The transducer was driven by a high power driving electronics system (not shown, but detailed elsewhere: [25]) and had a circular opening in the middle that harbored a coaxially aligned 64-element phased-array ultrasound imaging probe (ATL P4-2 probe, Phillips, Bothell, WA, USA. The imaging probe was attached to a Verasonics V1 ultrasound system (Kirkland, WA, USA), and the position of the histotripsy transducer focus within the B-mode ultrasound image was pre-registered with the system. The cylindrical plastic exposure vials (48 mm tall; 19 mm inner diameter; 3 mm wall thickness) accommodated 10 mL samples. The bottom of the vials was fitted with a cling-film acoustically transparent ‘window’ held in place by an O-ring which nested in an annulus machined into the outer walls of the vials. After filling with the desired sample, the vial was sealed with a friction-fitted cap containing two magnets that clamped against two magnets on the apparatus. This allowed reproducible alignment of replicate samples within the acoustic field. The bottom of the cap was machined into a cone with its point immersed in the sample to reduce acoustic reflections. When filled, a small air gap remained along the edge of the conical cap. The position of the vial corresponded to having the focus 15 mm from the bottom membrane, approximately at the center of the vial’s 10-mL volume. Immediately prior to a new study the water was degassed to 20% O2 concentration or less. Over the treatment time, the water temperature and O2 concentration were measured periodically by an O2 meter (Oxi 3310, Xylem Analytics, Washington DC, USA). Additionally, the proximal wall of the transparent acoustic window was wiped with a gauze sponge to remove pre-existing bubbles that could act as cavitation nuclei during treatments. During treatments, B-mode imaging and Passive Cavitation Detection (described below) were used to look for pre-focal cavitation signals from the transparent acoustic window. None was observed.
Figure 1.
Custom experimental apparatus. A 1 MHz transducer with integrated inline P4/2 imaging probe was attached to the bottom of a degassed water bath. The apparatus was designed so that the transducer’s focus was at the center of a sample vial containing 10 ml of E. coli bacteria (~ 1 x 109 cells/ml). A thin acoustically transparent membrane was placed at the bottom of the vial for improved acoustic energy transmission. The vial was aligned to the transducer using two pairs of magnets to allow for fast reproducible alignment of replicate samples.
Calibration of the acoustic focal pressure levels at full range of system driving voltages was performed in degassed water with a fiber-optic probe hydrophone (FOPH 2000; RP Acoustics, Leutenbach, Germany; 100 μm active diameter, 100 MHz bandwidth). The sample holder was removed for characterization. The resulting peak positive and peak negative pressures and representative focal waveforms are shown in Fig. 2. Both CH and BH regimens use the acoustic output corresponding to the formation of shocks at the focus; peak positive (P+) and peak negative (P−) focal pressures corresponding to fully developed shock formation (i.e. where P+ is equal to shock amplitude [26]) were P+=76 MPa and P−=12 MPa (red line in Fig. 2a). Peak focal pressures corresponding to the maximum output level used in this work were P+=103 MPa and P=−=23 MPa. Most studies were performed at P− ranging from 12 to 23 MPa.
Figure 2.
Transducer characterization. (a) Peak positive and negative amplitudes vs. input voltage. Most studies were performed between peak negative pressure amplitudes 12 MPa to 23 MPa. The red line indicates the output level corresponding to fully developed shock formation. (b) Representative waveforms corresponding to the output levels below (black line), and above (blue and red lines) shock formation.
Two different CH exposures (5 or 10 cycles at 2.0 kHz or 1 kHz PRF, respectively) and one BH exposure (10,000 cycles at 1 Hz PRF) were used in most studies. Except where noted, a duty cycle of 1% was used for all exposures. In the first series of experiments, the focal pressure threshold for inactivation of E. coli was determined for each regimen. Each sample was treated for ten minutes at different voltages corresponding to peak negative focal pressure ranging within 1-23 MPa. Once the amplitude dependence was established, the influence of treatment time within 5-40 min at a fixed pressure level was investigated.
Viability Assessment
Samples withdrawn from the treated cell suspensions were subjected to serial dilutions with EPA dilution water (2 mM MgCl2, 0.6 mM KH2PO4, pH 7.1) as a standard diluent for E.coli using aliquots no smaller than 25 μL, as described previously [12]. Compact Dry EC100 assay plates (Hardy Diagnostics, Santa Maria, CA, USA) were used for colony counting, on which E. coli specifically produces a blue colony while non-E. coli coliforms produce red colonies. Colonies were manually counted after incubation at 37°C for ~18 h, at which time the colonies were visible macroscopically. Control samples were taken periodically (every 30-40 minutes) throughout the study.
Statistical analysis
Regression lines were fitted to the data sets shown in Figs. 3, 5a, 6, 7, and 8. In the two-group comparison, the homoscedasticity of the groups was first evaluated using Fisher–Snedecor’s test (F-test) . When the groups were homoscedastic (i.e., Figs 3, 6, 7b, 8), p-values were calculated using Student's t-test. When the groups were heteroscedastic (Fig. 7a), the p-value was calculated using Welch's t-test. P-values less than 0.05 were considered significant.
Figure 3.
Mean log reduction (± SD) in bacterial load with increasing peak negative pressure amplitude. (a) The CH threshold for inactivating E. coli depended on pulse parameters. 5 cycle pulses at 2 kHz (1% duty cycle) had a higher threshold than 10 cycle pulses (2% duty cycle). Above the threshold, the bacterial log reduction increased linearly with amplitude (R2=0.93 for both curves). (b) With BH (10,000 cycles at 1 Hz PRF), there was no measured threshold. However, the log reduction in bacteria also increased linearly with amplitude (R2=0.90). The open circles in each figure have the same log reduction over 10 minutes (0.9) at the same pressure amplitude (17.8 MPa).
Figure 5.
Passive cavitation detection metrics obtained from the 5-cycle CH exposures in Fig. 3a: (a) Doppler power associated with spatio-temporal changes in the bubble cloud. The red circles help identify the change in slope, where cavitation occurs each pulse. (b) Average amplitude of broadband noise emissions over the 10-minute exposure. Both metrics exhibit a threshold-like behavior at the focal pressure of 16 MPa that also corresponds to the threshold for inactivating E.coli and indicates the onset of consistent cavitation. The error bars represent standard deviation. Trendlines are used to aid the eye.
Figure 6.
A comparison between CH (5 cycles, 2000 Hz PRF; red circles) and BH (10,000 cycle pulses and 1 Hz PRF; open triangles) bacterial load reduction at a pressure of 17.8 MPa and 1% duty cycle. R2 = 0.99 (CH) and 0.97 (BH). See Table I for converting treatment time to pulse number.
Figure 7.
Mean log reduction (± SD.) in bacterial load over time at different peak negative pressures. Dashed lines represent best fit linear trendlines. (a) CH treatments used 5 cycle bursts at 2,000 Hz PRF. R2 values for 23 MPa and 17.8 MPa = 0.95 and 0.99, respectively. The slope of the trendline for 23 MPa and 17.8 MPa = 0.11 and 0.07 respectively. (b) BH treatments used 10,000 cycle bursts at 1 Hz PRF. R2 values for 17.8 MPa and 14.5 MPa = 0.98 and 0.99, respectively. The slope of the trendlines for 17.8 MPa and 14.5 MPa = 0.075 and 0.054, respectively. See Table II for converting treatment time to pulse number.
Figure 8.
Mean log reduction (± SD.) in bacterial load over time. Dashed lines represent best fit linear trendlines. CH pulse parameters: Black triangles: 10 cycles, 1000 Hz PRF. Red circles: 5 cycles, 2000 Hz PRF. P--= 16 MPa for both. R2 values for 5 cycles and 10 cycles = 0.93 and 0.96, respectively. See Table III for converting treatment time to pulse number.
Ultrasound imaging and passive cavitation detection
Plane wave B-mode imaging at the frequency of 3.5 MHz was used prior to all exposures to confirm the repeatable positioning of the sample vials. During both CH and BH treatments the emission of each histotripsy pulse was followed by the acquisition of one (for CH) or 31 (for BH) plane wave B-mode images every 20 ms. In addition, emission of CH pulses was synchronized with a 200-microsecond passive RF data capture by the ultrasound probe. The data corresponding to backscattered histotripsy pulses and broadband noise emissions from bubble collapses were acquired at 12 MHz sampling rate every 30 seconds (for example, 20 acquisitions for a 10-minute treatment).
The acquisitions were post-processed in two ways to extract metrics of cavitation activity: broadband noise emission quantification and Doppler power processing. Note that the processing was not successful for BH exposures due to high levels of clutter from the 10 ms pulse reflections and reverberations from air-liquid interfaces and rigid boundaries of the experimental assembly.
Broadband noise quantification followed our previously reported method: each acquisition was filtered in the frequency domain by a combination of a band-pass 1000th order Hamming filter within 1.5–4.7 MHz and a second order IIR comb filter with a notch bandwidth of 400 kHz applied at the fundamental frequency of 1 MHz and its harmonics [27, 28]. The filtered signals were then analyzed in the time domain to determine whether a cavitation event occurred. The section of the signal arriving prior to the round-trip time of flight to the transducer focus was considered as background noise; the section corresponding to the arrival time from within the sample vial was considered as the region of interest (ROI). Cavitation was considered present if the peak signal value within the ROI exceeded that of the background noise by a factor √5, corresponding to the Rose criterion which ensures that the signal is distinguishable from the background noise [29].
For each sample cavitation persistence was defined as the percentage of the CH pulses that induced a cavitation event among all recorded pulses for the sample. If a cavitation event was identified, broadband noise amplitude was calculated as the root-mean-square (RMS) value of the filtered passive cavitation detection (PCD) signal within the ROI. The mean and standard deviation of cavitation persistence and broadband noise amplitude integrated over all the exposures was calculated over the replicate samples corresponding to the same CH pressure level.
Doppler processing was used to estimate the magnitude of changes in backscatter over 20 histotripsy pulses. The 20 PCD acquisitions, described above, were treated as a Doppler-PCD ensemble. High pass filtering removed non-changing echoes from the vial enclosure leaving only the temporally changing signals from cavitation events. The remaining power in the filtered Doppler-PCD ensemble estimated the amount of cavitation activity in the treatment volume. This approach is a PCD embodiment of the approach reported by Arnal et al [30].
Specifically, high pass filtering of the Doppler-PCD ensemble (equation 1 of [30]) was performed by a singular value decomposition approach [31]. The 4 lowest frequency projections were removed to eliminate temporally stationary reflections from the vial enclosure. The remaining high frequency projections captured cavitation events occurring over the Doppler-PCD ensemble (equation 2 of [30]). The first lag of the autocorrelation was used to estimate the Doppler power of the remaining high frequency projections, which was then averaged over the treatment volume.
For Doppler power processing the 20 passive acquisitions of RF were treated as a Doppler ensemble. High pass filtering was applied over the 20 acquisitions (i.e. in slow time) within the vial through removal of the lowest frequency projections following singular value decomposition (SVD) [31]. Removal of the four lowest frequency projections eliminated the most spatio-temporally correlated components of the PCD signals, primarily corresponding to the reflections from the vial enclosure. The remaining high frequency projections captured cavitation events occurring over the 20 acquisitions across the array. The first lag autocorrelation was used to estimate the power of those remaining high frequency projections, providing an average power of the spatial and temporal variations of the cavitation events and bubble cloud activity over the time of treatment.
Results
Inactivation threshold
Initial studies were performed to determine the relative rate of bacteria inactivation as a function of focal pressure amplitude. 10 min exposures were performed at different pressures up to P − = 23 MPa. CH exposures (Fig. 3a) resulted in a clear threshold phenomenon: no reduction in bacterial load was observed until P − exceeded a threshold value that itself depended on the pulse parameters. The reduction in bacterial load increased exponentially above the threshold. The slope of the line (rate of inactivation) was higher for the longer pulse (higher duty cycle). For BH exposures (Fig. 3b), no threshold was observed. Instead, the reduction in bacterial load followed an exponential curve as a function of peak negative pressure. Comparing CH (5-cycle pulses) to BH, the open black circles indicate where the two trendlines cross. That is, at that specific pressure (17.8 MPa), the log reduction for both CH and BH was the same (0.9).
Cavitation detection
During CH, a hyperechoic bubble cloud confined to the focal area could be seen in B-mode images (Fig. 4 and supplementary videos V2-4), appearing intermittently from P− = 14.5 MPa and without interruption from P− = 16 MPa (at 1% duty cycle). The appearance of the bubble cloud also correlated with distinct audible cavitation noise at the pitch of 2 kHz corresponding to the PRF. As the output level increased, the bubble cloud shifted pre-focally and slightly elongated.
Figure 4.
Representative B-mode ultrasound images observed during CH (top) and BH (bottom). The hyperechoic CH bubble cloud was generally confined to the focal area; it slightly elongated and shifted pre-focally with the increase of P− from 14.5 to 23 MPa (see supplementary videos V2-4, respectively). Conversely, following each BH pulse hyperechoic bubbles filled the entire sample volume through visible streaming-induced mixing (see supplementary video V1) and persisted until the next pulse arrived. Note the HIFU reverberation artifact in the image immediately following the BH pulse (40msec).
Conversely, during BH the hyperechoic bubbles were seen to appear and circulate within the vial starting from P− as low as 1 MPa. BH bubbles filled the entire sample volume after one to several pulses (Supplementary Video V1), starting from the pre-focal and focal areas and then spreading post-focally and sideways, consistent with a vortex-like streaming pattern induced by high intensity ultrasound (HIFU) radiation forces. A HIFU reverberation artifact was observed in the B-mode frames immediately following each BH pulse and highlights the difficulties encountered with the interpretation of PCD signals for BH exposures. Broadband noise and Power Doppler processing of the PCD data for CH associated with Fig. 3a was performed to evaluate cavitation analyses techniques that could be used in vivo for correlation with bacteria inactivation. As shown in Fig. 5, the threshold between intermittent and continuous cavitation was evident in the output of both passive cavitation detection techniques. In Fig. 5a the Power Doppler analysis showed increasing power levels as the pressure increased. Above 16 MPa, there is a distinct change in slope corresponding to the onset of consistent cavitation. The broadband noise analysis (Fig. 5b) showed a step from low to high at the pressure levels corresponding to the transition between occasional and consistent generation of the bubble cloud.
CH vs BH treatment time studies
At a pressure of 17.8 MPa and a 1% duty cycle, both CH and BH result in similar log reductions (Fig. 3, open circle data points). We hypothesized that this equivalence should extend to other treatment times at this specific pressure level. Thus, comparative studies were performed at 17.8 MPa for up to 40 minutes to test this hypothesis. Both CH and BH have linear fits in the log linear plot (Fig. 6). The difference between the slopes of regressions lines corresponding to CH and BH is statistically insignificant. Note that the abscissa in Fig. 6 is given in terms of total treatment time, which is clinically relevant. For comparative purposes, treatment times can also be described in terms of ‘on-time’ or number of pulses. Table I lists the number of pulses corresponding to the labeled treatment time for Fig. 6.
Table I.
Number of pulses corresponding to treatment time for Figure 6.
| Figure | Histotripsy | PRF | Treatment | Pulses |
|---|---|---|---|---|
| CH/BH | (Hz) | Time (min) | (number) | |
| Fig. 6 black | BH | 1 | 5 | 300 |
| Fig. 6 black | BH | 1 | 10 | 600 |
| Fig. 6 black | BH | 1 | 20 | 1,200 |
| Fig. 6 black | BH | 1 | 40 | 2,400 |
| Fig. 6 red | CH | 2000 | 5 | 600,000 |
| Fig. 6 red | CH | 2000 | 10 | 1,200,000 |
| Fig. 6 red | CH | 2000 | 20 | 2,400,000 |
| Fig. 6 red | CH | 2000 | 40 | 4,800,000 |
Abbreviations: CH = Cavitation Histotripsy; BH = Boiling Histotripsy; PRF = Pulse repetition frequency
Treatment time comparison at different peak negative pressures
In addition to comparing CH against BH at one pressure level, we also looked at factors affecting each treatment modality individually. Time-course studies were performed for both CH and BH at different pressure amplitudes, for up to 40 minutes (Fig. 7). Bacterial load reductions are plotted in a log-linear format. In Fig. 7a CH treatments were performed at two different peak negative pressure amplitudes: P − = 23 MPa and 17.8 MPa. Treatments at 23 MPa showed an increase in log reduction over 17.8 MPa treatments. A similar study with BH (Fig. 7b) also yielded increased inactivation with higher applied pressures. In all cases, the number of inactivated bacteria increased exponentially over time (corresponding to straight lines in these log-linear plots). All best-fit trendlines have a coefficient of a determination (R2) between 0.95 and 0.99. Note the data points at 10 min in Fig. 7a were in close agreement with those at corresponding pressures in Fig. 3. As with the previous figure, table II lists the number of pulses corresponding to the treatment time in Fig. 7.
Table II:
Number of pulses corresponding to treatment time for Figure 7.
| Figure | Histotripsy | PRF | Treatment | Pulses |
|---|---|---|---|---|
| CH/BH | (Hz) | Time (min) | (number) | |
| Fig.7 a | CH | 2000 | 5 | 600,000 |
| Fig.7 a | CH | 2000 | 10 | 1,200,000 |
| Fig.7 a | CH | 2000 | 20 | 2,400,000 |
| Fig.7 a | CH | 2000 | 40 | 4,800,000 |
| Fig.7 b | BH | 1 | 5 | 300 |
| Fig.7 b | BH | 1 | 10 | 600 |
| Fig.7 b | BH | 1 | 20 | 1,200 |
| Fig.7 b | BH | 1 | 40 | 2,400 |
Abbreviations: CH = Cavitation Histotripsy; BH = Boiling Histotripsy; PRF = Pulse repetition frequency
CH parameter comparisons
Finally, studies were performed to quantify bacterial load reduction for two different sets of CH pulse parameters. Our hypothesis is that bacterial load reduction depends on the total number of pulses (or on-time). Thus, different sets of pulse parameters should produce the same outcome for the same duty cycle. In the following dose response study, CH treatments were performed at 16 MPa peak negative pressure with either 5-cycle pulses at 2,000 Hz PRF, or 10-cycle pulses at 1,000 Hz PRF. The on-time is thus the same for both parameter sets. Figure 8 shows that both sets of parameters led to the same log reduction at each time point. The difference between the slopes of regression lines corresponding to 5 cycle and 10 cycles is statistically insignificant. Table III lists the number of pulses corresponding to the treatment time in Fig. 8.
Table III:
Number of pulses corresponding to treatment time for Figure 8.
| Figure | Histotripsy | PRF | Treatment | Pulses |
|---|---|---|---|---|
| (Hz) | Time (min) | (Number) | ||
| Fig. 8 black | CH | 1000 | 5 | 300,000 |
| Fig. 8 black | CH | 1000 | 10 | 600,000 |
| Fig. 8 black | CH | 1000 | 20 | 1,200,000 |
| Fig. 8 red | CH | 2000 | 5 | 600,000 |
| Fig. 8 red | CH | 2000 | 10 | 1,200,000 |
| Fig. 8 red | CH | 2000 | 20 | 2,400,000 |
Abbreviations: CH = Cavitation Histotripsy; PRF = Pulse repetition frequency
Statistical analyses
A student’s t-test was used in the case of equal variances between two regression models., Results were statistically significant for Fig. 7b (p= 7.1*10−4)and statistically insignificant for Figs. 6 and 8 (p=0.52, p=0.17 respectively). In the case of unequal variances (Fig. 7a), Welch’s t-test was applied and the results were statistically insignificant. This is likely due to low number of replicates for the case of unequal variances.
Discussion
Histotripsy includes multiple regimens, two of which are shock scattering cavitation histotripsy (CH) and boiling histotripsy (BH). CH generates small lesions in rapid succession (~ 1 kHz), whereas BH generates a much larger lesion, but at a much lower rate (~ 1 Hz). The current work was initially motivated by the results of a prior uncontrolled in vivo study suggesting greater variability in bacterial load reduction with boiling histotripsy (BH) treatments than with shock-scattering cavitation histotripsy (CH) treatments [20]. Possible causes of variability in that study included unequal treatment durations and volumes. Here, a more controlled in vitro study determined that both CH and BH treatments reduce bacterial loads effectively for a 10 ml bacteria suspension. The studies evaluated the effects of peak negative pressure (from 14.5 MPa to 23 MPa), treatment time (10, 20 or 40 minutes), and pulse parameters (BH: 10,000 cycles at 1 Hz PRF; CH: 5 or 10 cycles, at 2,000 or 1,000 Hz PRF, respectively). The duty cycles for these parameters were 1% for both BH and CH, except for the threshold data in Fig. 3a that included a 2% duty cycle study. The pressure amplitude studies showed that CH had a threshold for bacteria inactivation that depended on the pulse parameters, while BH did not show any threshold behavior. For both modalities the rate of inactivation increased with pressure amplitude and treatment time.
The inactivation threshold observed for CH provides insight into the mechanism of histotripsy treatment. The CH threshold behavior was consistent with previous observations [12]. Fig. 3a shows that the threshold is reduced with an increase in the duty cycle. The observed threshold level coincided with the pulse parameters that produced cavitation for each pulse. Below the threshold, cavitation was inconsistent, and thus the non-cavitating pulses did not contribute to bacteria inactivation. In addition, non-cavitating pulses do not contribute to sample mixing. Thus, below the continuous cavitation threshold, cavitation is intermittent and bacterial load reduction is more variable. Based on these results, a possible explanation for the absence of a threshold with BH treatments is that cavitation was generated by every pulse at all pressure amplitudes investigated.
CH-induced cavitation activity was quantified through PCD signal analysis with two processing methods to help identify a cavitation metric for bacterial inactivation in future in vivo studies. Due to reverberations of the long BH pulse from the walls of the small sample vial, the analysis was only successful for CH pulses. Specifically, the average amplitude of the broadband noise emissions was measured at a range of pressure levels corresponding to varying levels of inactivation for 10-minute exposures. Broadband noise is commonly used as a measure of the intensity of inertial bubble collapses that are responsible for the observed bioeffect – inactivation in the present case [27]. In addition, the average of the spatio-temporal variation of cavitation events within the cavitation cloud was quantified by treating the PCD acquisitions as a Doppler ensemble. At lower treatment pressures (<16 MPa) cavitation was intermittent: broadband noise emissions were not detectable with every CH pulse and the detectable emissions had low amplitude. The PCD Doppler power signal was dominated by cavitation events that were variable over the treatment time, with singular cavitation events progressing to a cavitation cloud with increased treatment pressure. At 16 MPa, the cavitation persistence reached 100% in a threshold-like manner: broadband emissions were detected following each pulse and their amplitude abruptly increased. Similarly, the power Doppler signal increased, and at pressure levels over 16 MPa it was dominated by a continual mix of spatially consistent temporally-varying cavitation events. This pressure threshold of consistent and violent cavitation corresponded to the threshold in bacterial inactivation.
At pressure levels above the threshold both broadband emission amplitude and Doppler power nearly plateaued (Fig. 5). In contrast, the inactivation rate increased tenfold within the 16-23 MPa range (Fig. 3a, black data points). We speculate that the plateau is due to shielding of the bubble cloud by its proximal front, which is shifted towards the transducer with the increase in pressure levels (Fig. 4). This shielding may prevent the emissions and scattering from the bubbles in the central and distal parts of the cloud to reach the PCD transducer, thus reducing both the broadband emissions amplitude and the Doppler power.
The treatment-time studies (Figs 6 to 8) highlight the noteworthy observation that both CH (above the threshold) and BH treatments resulted in an exponential bacterial load reduction over time, as evidenced by the linear fits to the log-linear plots. The inactivation rates, indicated by the slopes of the trendlines, were dependent on pressure amplitude for both CH and BH: higher pressure amplitudes led to higher inactivation rates (Fig. 7).
In most cases CH and BH did not exhibit equal effectiveness in bacterial inactivation. Below P− = 17.8 MPa, BH exhibited greater bacterial inactivation rates. This was partially due to CH having a threshold below which significant inactivation did not occur, whereas BH did not exhibit such a threshold. At 17.8 MPa, CH and BH performed equally well at all time points examined (see e.g., open circles in Fig. 3 and Fig. 6). Above 17.8 MPa, CH outperformed BH in terms of bacterial inactivation. Figure 3 demonstrates that over a ten-minute period, the maximum inactivation rate for CH was a 1.5 log reduction in bacterial count, while over the same duration, the maximum BH treatment resulted in a 1 log reduction. Thus, in practice, the preferred treatment method may depend on the peak negative focal pressure attainable at the specific location of the targeted abscess within attenuative tissue and the available acoustic window.
These findings suggest that both CH and BH treated the entire 10 mL sample volume, as evidenced by the log-linear trends at all time points (bacteria inactivation is obeying a first-order exponential kinetic process; partial treatment should not lead to linear trendlines). This is somewhat counter-intuitive because BH created a larger cavitation cloud that filled a greater portion of the sample volume, thus resulting in a larger "treatment zone" each pulse. The cost for covering a larger volume was at a much reduced PRF (1 Hz), whereas CH treatments covered a smaller volume, but at a much greater PRF (~ 1 kHz). In both regimens, circulation induced by streaming through the focal region ensured treatment of the entire 10 mL volume.
While levels of disinfection up to 4.2 log reduction were reported, this occurred for a treatment time of 40 min. Faster treatments may be possible with higher average power (increased PRF) or peak power (increased input voltage). For the current studies, higher peak powers were impractical due to hardware limitations. Higher average powers would increase the duty cycle above 1%, which would enhance heating. Also, Tables I-III provide actual pulse numbers used in these studies and can be compared to other studies that use pulse number as a histotripsy parameter.
Finally, note that the gas concentration and sample temperature were not controlled. However, changes in gas concentration and temperature of the water bath surrounding the sample vial were monitored throughout. Over the course of a 40-minute study the gas concentration was observed to gradually increase by approximately 20% and the temperature increased from 20°C to 40°C. A study was performed to determine whether there was a difference in inactivation rate between the initial and final gas concentrations, and between the initial and final temperatures. There were no statistically significant differences in bacterial load reduction in either case (data not shown). Nor were there changes in the rate of inactivation. We conclude that changes in gas concentration or temperature of the coupling water did not affect the experiment.
Conclusions
Cavitation Histotripsy (CH) and Boiling Histotripsy (BH) reduced bacterial loads in suspension in a consistent manner. In both cases, inactivation increased exponentially with pressure amplitude or treatment time. CH studies showed an inactivation threshold phenomenon as a function of focal pressure amplitude related to the onset of consistent cavitation. In contrast, cavitation activity and inactivation both increased gradually with pressure for BH studies. At or below the CH threshold BH generated greater inactivation compared to CH; conversely, at pressure amplitudes substantially exceeding the threshold CH generated a higher amount of inactivation. In particular, at the highest pressure amplitude of 23 MPa and longest treatment time of 40 min, CH generated the largest bactericidal effect of 4.2 log compared to 2.8 log for BH at the same conditions. Thus, in conditions allowing for the use of high in situ pressure levels – superficial abscesses, relatively small volumes and favorable heat dissipation dynamics - CH would be the preferred treatment modality. For deeper abscesses, where excessively high pressure levels may not be advisable due to the potential for heat buildup in the intervening tissues or in the abscess itself, the use of BH may be preferable. Future studies will compare these regimens for larger treatment volumes.
Supplementary Material
Acknowledgements
This work was supported in part by NIH grants R01AR080120, R01EB023910, R01GM122859, and R01EB031788.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of Competing Interests
The authors declare no competing interests.
Data availability statement
The data sets generated and analyzed for the present study are available from the corresponding author on reasonable request.
References
- 1.Harvey EN and Loomis AL, The Destruction of Luminous Bacteria by High Frequency Sound Waves. J Bacteriol, 1929. 17(5): p. 373–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chemat F, Zill-e-Huma, and Khan MK, Applications of ultrasound in food technology: Processing, preservation and extraction. Ultrasonics Sonochemistry, 2011. 18(4): p. 813–835. [DOI] [PubMed] [Google Scholar]
- 3.Cullen PJ, Tiwari BK, and Valdramidis VP, Novel thermal and non-thermal technologies for fluid foods. 1st ed. Food science and techonology international series. 2012, London ; Waltham, MA: Academic. xvi, 526 p. [Google Scholar]
- 4.Earnshaw RG, Appleyard J, and Hurst RM, Understanding physical inactivation processes: Combined preservation opportunities using heat, ultrasound and pressure. International Journal of Food Microbiology, 1995. 28(2): p. 197–219. [DOI] [PubMed] [Google Scholar]
- 5.Knorr D, Zenker M, Heinz V, and Lee DU, Applications and ultrasonics in food potential of processing. Trends in Food Science & Technology, 2004. 15(5): p. 261–266. [Google Scholar]
- 6.Piyasena P, Mohareb E, and McKellar RC, Inactivation of microbes using ultrasound: a review. International Journal of Food Microbiology, 2003. 87(3): p. 207–216. [DOI] [PubMed] [Google Scholar]
- 7.Erriu M, Blus C, Szmukler-Moncler S, Buogo S, Levi R, Barbato G, Madonnaripa D, Denotti G, Piras V, and Orru G, Microbial biofilm modulation by ultrasound: Current concepts and controversies. Ultrasonics Sonochemistry, 2014. 21(1): p. 15–22. [DOI] [PubMed] [Google Scholar]
- 8.Wardlow R, Sahoo K, Dugat D, Malayer J, and Ranjan A, High Intensity Focused Ultrasound (Hifu) Heating Improves Perfusion and Antimicrobial Efficacy in Mouse staphylococcus Abscess. Ultrasound in Medicine and Biology, 2018. 44(4): p. 909–914. [DOI] [PubMed] [Google Scholar]
- 9.Bigelow TA, Northagen T, Hill TM, and Sailer FC, Ultrasound histotripsy and the destruction of Escherichia coli biofilms. Conf Proc IEEE Eng Med Biol Soc, 2008. 2008: p. 4467–70. [DOI] [PubMed] [Google Scholar]
- 10.Bigelow TA, Thomas CL, Wu H, and Itani KMF, Histotripsy Treatment of S. Aureus Biofilms on Surgical Mesh Samples Under Varying Pulse Durations. IEEE Trans Ultrason Ferroelectr Freq Control, 2017. 64(10): p. 1420–1428. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Brayman AA, MacConaghy BE, Wang YN, Chan KT, Monsky WL, Chernikov VP, Buravkov SV, Khokhlova VA, and Matula TJ, Inactivation of Planktonic Escherichia coli by Focused 1-MHz Ultrasound Pulses with Shocks: Efficacy and Kinetics Upon Volume Scale-Up. Ultrasound Med Biol, 2018. 44(9): p. 1996–2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Brayman AA, MacConaghy BE, Wang YN, Chan KT, Monsky WL, McClenny AJ, and Matula TJ, Inactivation of Planktonic Escherichia coli by Focused 2-MHz Ultrasound. Ultrasound Med Biol, 2017. 43(7): p. 1476–1485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gao S, Lewis GD, Ashokkumar M, and Hemar Y, Inactivation of microorganisms by low-frequency high-power ultrasound: 2. A simple model for the inactivation mechanism Ultrasonics sonochemistry, 2014. 21(1): p. 454–60. [DOI] [PubMed] [Google Scholar]
- 14.Maxwell AD, Wang TY, Cain CA, Fowlkes JB, Sapozhnikov OA, Bailey MR, and Xu Z, Cavitation clouds created by shock scattering from bubbles during histotripsy. J Acoust Soc Am, 2011. 130(4): p. 1888–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Khokhlova TD, Canney MS, Khokhlova VA, Sapozhnikov OA, Crum LA, and Bailey MR, Controlled tissue emulsification produced by high intensity focused ultrasound shock waves and millisecond boiling. J Acoust Soc Am, 2011. 130(5): p. 3498–510. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bruce M, Leotta DF, Wang Y-N, Khokhlova T, Kucewicz J, Maxwell AD, Chan K, Monsky WL, and Matula TJ. Ultrasound Imaging of Abscesses Before and During Histotripsy Treatment. in 2020 IEEE International Ultrasonics Symposium (IUS). 2020. Las Vegas, NV. [Google Scholar]
- 17.Matula TJ, Wang YN, Khokhlova T, Leotta DF, Kucewicz J, Brayman AA, Bruce M, Maxwell AD, MacConaghy BE, Thomas G, Chernikov VP, Buravkov SV, Khokhlova VA, Richmond K, Chan K, and Monsky W, Treating Porcine Abscesses with Histotripsy: A Pilot Study. Ultrasound Med Biol, 2021. 47(3): p. 603–619. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Leotta DF, Bruce M, Wang YN, Kucewicz J, Khokhlova T, Chan K, Monsky W, and Matula TJ, Sonographic Features of Abscess Maturation in a Porcine Model. Ultrasound Med Biol, 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wang Y-N, Brayman AA, Chan KT, Richmond K, Monsky WL, and Matula TJ, A porcine bimicrobial abscess model for testing interventional treatments. bioRxiv, 2020: p. 2020.03.05.979088. [Google Scholar]
- 20.Matula TJ, Wang Y-N, Khokhlova T, Leotta DF, Kucewicz J, Brayman AA, Bruce M, Maxwell AD, MacConaghy BE, Thomas GT, Richmond K, Chan K, and Monsky WL. Histotripsy Treatment of Abscesses. in 2020 IEEE International Ultrasonics Symposium (IUS). 2020. Las Vegas. [Google Scholar]
- 21.Blake JR, Leppinen DM, and Wang Q, Cavitation and bubble dynamics: the Kelvin impulse and its applications. Interface Focus, 2015. 5(5): p. 20150017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Maxwell AD, Cain CA, Duryea AP, Yuan L, Gurm HS, and Xu Z, Noninvasive thrombolysis using pulsed ultrasound cavitation therapy - histotripsy. Ultrasound Med Biol, 2009. 35(12): p. 1982–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Song M, Thomas GPL, Khokhlova VA, Sapozhnikov OA, Bailey MR, Maxwell AD, Yuldashev PV, and Khokhlova TD, Quantitative Assessment of Boiling Histotripsy Progression Based on Color Doppler Measurements. IEEE Trans Ultrason Ferroelectr Freq Control, 2022. 69(12): p. 3255–3269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wu J and Nyborg WL, Ultrasound, cavitation bubbles and their interaction with cells. Advanced Drug Delivery Reviews, 2008. 60(10): p. 1103–1116. [DOI] [PubMed] [Google Scholar]
- 25.Maxwell AD, Yuldashev PV, Kreider W, Khokhlova TD, Schade GR, Hall TL, Sapozhnikov OA, Bailey MR, and Khokhlova VA, A Prototype Therapy System for Transcutaneous Application of Boiling Histotripsy. IEEE Trans Ultrason Ferroelectr Freq Control, 2017. 64(10): p. 1542–1557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rosnitskiy PB, Yuldashev PV, Sapozhnikov OA, Maxwell AD, Kreider W, Bailey MR, and Khokhlova VA, Design of HIFU Transducers for Generating Specified Nonlinear Ultrasound Fields. IEEE Trans Ultrason Ferroelectr Freq Control, 2017. 64(2): p. 374–390. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Li T, Chen H, Khokhlova TD, Wang Y-N, Kreider W, He X, and Hwang JH, Passive cavitation detection during pulsed HIFU exposures of ex vivo and in vivo mouse pancreatic tumors Ultrasound in Medicine & Biology 2014. 40(7): p. 1523–1534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Khokhlova T, Rosnitskiy P, Hunter C, Maxwell A, Kreider W, Ter Haar G, Costa M, Sapozhnikov O, and Khokhlova V, Dependence of inertial cavitation induced by high intensity focused ultrasound on transducer F-number and nonlinear waveform distortion. J Acoust Soc Am, 2018. 144(3): p. 1160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Rose A, Vision: human and electronic. Optical physics and engineering. 1973, New York,: Plenum Press. xv, 197 p. [Google Scholar]
- 30.Arnal B, Baranger J, Demene C, Tanter M, and Pernot M, In vivo real-time cavitation imaging in moving organs. Phys Med Biol, 2017. 62(3): p. 843–857. [DOI] [PubMed] [Google Scholar]
- 31.Demene C, Deffieux T, Pernot M, Osmanski BF, Biran V, Gennisson JL, Sieu LA, Bergel A, Franqui S, Correas JM, Cohen I, Baud O, and Tanter M, Spatiotemporal Clutter Filtering of Ultrafast Ultrasound Data Highly Increases Doppler and fUltrasound Sensitivity. IEEE Trans Med Imaging, 2015. 34(11): p. 2271–85. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data sets generated and analyzed for the present study are available from the corresponding author on reasonable request.