In vitro thrombolytic efficacy of single- and five-cycle histotripsy pulses and rt-PA

Abstract

While primarily known as an ablative modality, histotripsy can increase the efficacy of lytic therapy in a retracted venous clot model. Bubble cloud oscillations are the primary mechanism of action for histotripsy, and the type of cavitation activity is dependent on the pulse duration. A retracted human venous clot model was perfused with and without the thrombolytic recombinant tissue-plasminogen activator (rt-PA). The clot was exposed to pulses of single- or five-cycle duration and peak negative pressures of 0 to 30 MPa. Bubble activity within the clot was monitored via passive cavitation imaging. The combination of histotripsy and rt-PA was more efficacious than rt-PA alone for single- and five-cycle pulses with peak negative pressures of 25 and 20 MPa, respectively. For both excitation schemes, the detected acoustic emissions correlated with the degree of thrombolytic efficacy. These results indicate that rt-PA and single or multi-cycle histotripsy pulses enhance thrombolytic therapy.

Introduction

Deep vein thrombosis (DVT) affects an estimated 45 to 177 per 100,000 people annually in the USA (Shaikhouni et al. 2018), and creates up to $10 billion in annual associated healthcare costs (Benjamin et al. 2017; Grosse 2012). For iliofemoral DVT, catheter-directed thrombolytics are the front-line approach for vessel recanalization (Jaff et al. 2011; Kearon et al. 2008; Meissner et al. 2012). The thrombolytics are administered over the course of several days, are associated with hemorrhagic side effects (Mewissen et al. 1999) and have limited efficacy for organized, chronic thrombi (Hirsh and Hoak 1996). Thus, there is a need for an adjuvant therapy to decrease the time to vessel recanalization and increase long term efficacy.

One promising approach is mechanical ablation. Histotripsy is a form of therapeutic ultrasound that liquefies tissue through the mechanical action of bubble clouds (Khokhlova et al. 2015; Kieran et al. 2007). Bubble expansion macerates tissue while inducing vigorous fluid mixing (Maxwell et al. 2014). These physical mechanisms have been exploited to enhance the thrombolytic efficacy in an in vitro clot model nonresponsive to lytic therapy (Bader et al. 2016). The insonation conditions necessary to enhance the thrombolytic efficacy should be weighed against the potential for off-target effects. Depending on the duration and peak negative pressure of the pulse, different mechanisms are responsible for histotripsy bubble cloud generation (Bader et al. 2019; Khokhlova et al. 2015). Shock-scattering histotripsy employs pulses that are approximately 20 cycles or fewer in duration with highly asymmetric focal pressure waveforms due to the combined effects of nonlinear propagation and diffraction (Rosnitskiy et al. 2017). A bubble cloud forms when the shock wave scatters from a microbubble in the focal region (Maxwell et al. 2011b). Intrinsic threshold histotripsy, also referred to as microtripsy, employs pulses with a single large tensile phase in excess of approximately 25 MPa that activates nanoscale nuclei intrinsic to the target tissue (Maxwell et al. 2013; Vlaisavljevich et al. 2015).

Both intrinsic threshold and shock-scattering pulses have been explored in vitro and in vivo for clot ablation (Sukovich et al. 2017; Zhang et al. 2017). Shock-scattering pulses have been exploited to enhance the thrombolytic efficacy in an in vitro clot model that is otherwise nonresponsive to lytic therapy (Bader et al. 2016). Off-target hemorrhage to the venous adventitia, and fat and muscle surrounding the femoral vein, have been noted in vivo for these types of pulses (Maxwell et al. 2011a). The precise nature of intrinsic threshold pulses allows generation of bubble activity exclusively with the lumen of pre-clinical models (Zhang et al. 2015b; Zhang et al. 2015a; Zhang et al. 2017). However, the adjunctive nature of intrinsic threshold bubble activity for lytic therapy has yet to be established.

The goal of this study was to compare the thrombolytic efficacy of rt-PA in combination with shock-scattering histotripsy or intrinsic threshold histotripsy using a venous human whole-blood clot. Bubble activity was mapped spatially with passive cavitation imaging (PCI) and correlated with the degree of clot lysis. Particulates generated by the treatment were sized using a flow cytometer. Histological analysis was performed to assess structural changes in the clots following exposure to histotripsy and rt-PA. Single- or five-cycle pulses combined with rt-PA had significantly improved thrombolytic efficacy compared to rt-PA alone for peak negative pressures greater than 25 and 20 MPa, respectively. Acoustic emissions from the bubble cloud, a surrogate for the histotripsy mechanical activity, correlated with the thrombolytic efficacy for both insonation schemes.

Materials and Methods

Venous whole-blood clot model

Human whole-blood clots were manufactured following Mercado-Shekhar et al. (2018a). After local internal review board (IRB) review approval and written, informed consent, venous human blood was drawn from a pool of 15 volunteer patients undergoing invasive catheterization procedures at the University of Chicago Medicine cardiac catheterization laboratory. Volunteer patients on medications that would alter the clotting cascade of the drawn blood were screened out of the study a priori. A list of volunteer demographics is displayed in Table 1. Blood aliquots of 2 mL were transferred into borosilicate glass Pasteur pipettes (Fisher Scientific, Pittsburg, PA, USA). The pipettes were incubated in a water bath at 37 °C for 3 hrs., and then stored at 4 °C for a minimum of 3 days to allow for full clot retraction (Shaw et al. 2008). The clots were used within 2 weeks after the retraction period. A consistent clot stiffness and response to rt-PA has been noted over this period for in vitro studies (Holland et al. 2008; Mercado-Shekhar et al. 2018; Hendley et al. 2019).

Table 1.

Demographics of blood draw patient volunteers

Age	Years		Race	Count		Sex	Count
Average	55		African American	8		Female	9
Range	35–73		White	7		Male	6

Preparation of recombinant tissue-type plasminogen activator and human fresh-frozen plasma

Recombinant tissue plasminogen activator (rt-PA) was obtained from the manufacturer (Activase, Genetech, San Francisco, CA, USA) as lyophilized powder. Each vial was mixed with sterile water to a concentration of 1 mg/mL, aliquoted into 0.5 mL centrifuge tubes, and stored at −80 °C. The enzymatic activity of the rt-PA is stable over a period of at least 7 years utilizing this protocol (Shaw et al. 2009). Human fresh-frozen plasma was procured from a blood bank (Lifesource Blood Center, Chicago, IL), aliquoted, and stored at −80 °C. For each data set, 30 mL of the plasma were thawed and allowed to equilibrate to atmospheric pressure for at least two hours at 37 °C.

Histotripsy insonation

Histotripsy pulses were generated by an eight-segment, spherically-focused 1-MHz transducer with a 10 cm aperture and 7.5 cm focal length. Co-axial image guidance of the histotripsy pulse was achieved with an L11–4v imaging array (Verasonics, Inc., Kirkland, WA, USA) placed through a concentric opening on the histotripsy transducer. All elements of the transducer were excited in parallel by a custom designed and built class D amplifier and matching network (Hall and Cain 2006; Maxwell et al. 2017). The nonlinear acoustic field and pressure waveforms at the focus of the transducer were measured with a fiber optic hydrophone (HFO-690, Onda Corp., Sunnyvale, CA, USA) for nominally single and five-cycle excitation pulses (Figure 1).

Figure 1:

Measured waveforms for (a) nominally single-cycle and (b) five-cycle excitation pulses at the focus of the transducer. The peak negative pressures were approximately 18 and 21 MPa, respectively. The geometric focus of the transducer is at 75 mm, or 50 μs. The associated normalized spectra of the (c) single-cycle and (d) five-cycles excitations. The spectra for both pulse durations were multiplied by a Hamming window to reduce side lobes. The peak positive pressure (PPP) and peak negative pressure (PNP) at a given voltage applied to the class-D amplifier for (e) single cycle, and (f) five cycle excitation pulses, respectively. The legend in (e) is the same for (f). The arrowhead in (f) notes the approximate inflection point of the peak positive pressure curve, indicative of the onset of shock formation.

Experimental procedure

An in vitro setup was adapted from Radhakrishnan et al. (2013) for histotripsy exposure of clots (Figure 2). The flow channel consisted of a 6.35 mm inner diameter and 0.79 mm wall thickness latex tubing (McMaster-Carr, Elmhurtst, IL, USA) attached to a syringe pump in withdrawal mode (EW-74900–20, Cole-Parmer, Vernon Hills, IL, USA) and a reservoir of plasma. The reservoir temperature was maintained by submerging it in a 36 × 36 × 30 cm acrylic tank of degassed (20% dissolved oxygen), deionized water, heated to 37.3 ± 0.5 °C. The water temperature was maintained by a custom-built temperature controller circuit (ITC-308, Inkbird, Pengji Industrial Zone, Luohu District, Shenzhen, China) and submerged heating elements (HT 300 Titanium, Won Brothers, Fredericksburg, VA, USA).

Figure 2:

Experimental setup. A) shows the flow phantom setup. The flow was from right to left in this diagram. The lateral dimension for the imaging array is into the page. B) is the timing diagram for histotripsy insonation and data acquisition. An a priori scanline B-Mode image was taken at each location. Each location is then treated with 2000 pulses at a 40 Hz pulse repetition frequency. Every tenth pulse passive cavitation imaging (PCI) waveforms are acquired for offline analysis. Due to time of flight propagation of the histotripsy pulse to the focal zone (Histotripsy Focal Insonation in diagram), the acquisition of cavitation emissions was delayed 64 μs after the trigger of the pulse.

The clot was cut to 1 cm in length, blotted (Holland et al. 2008; Hitchcock et al. 2011; Sutton et al. 2013; Bader et al. 2016), and weighed with a laboratory balance (P-114, Denver Instrument, Bohemia, NY). The clot was then inserted into the flow channel with tweezers. The flow channel was perfused with plasma alone, or plasma and rt-PA (2.68 μg/mL (Hilleman and Razavi 2008)) at 0.65 mL/min. The flow rate induced by the pump is consistent with a near total occluded iliofemoral vessel (Jensen et al. 2017), though the clot does not occlude the lumen fully. A nylon rod (1.6 mm diameter) in the lumen prevented movement of the clot due to flow (see Figure 2a). The flow channel was submerged in the tank of degassed water.

Alignment of the transducer focus along the length of the clot was conducted similar to Maxwell et al. (2011a). The transducer focus was determined by generating histotripsy pulses in the degassed water outside the lumen. The resultant bubble cloud generated at the focus was identified as a hyperechoic region on a B-mode image acquired with the confocal imaging array. Once the location of the focus was determined, the transducer was then moved and aligned with the proximal end (i.e. towards the plasma reservoir in Figure 2) of the clot. Confinement of the bubble cloud within the lumen was confirmed at 5 mm increments along the length of the clot with short (<1 s) bursts. Once target locations were ascertained, a custom script interpolated an automated path for the transducer to apply histotripsy pulses at 0.5 mm increments along the length of the clot via a three-axis positioning system (Velmex Inc., Bloomfield, NY, USA). The 0.5 mm distance is less than the −6 dB focal width of the histotripsy source (5 mm × 1 mm × 1 mm for the imaging range, elevational, and lateral directions, respectively, in Figure 2a). The clots were 4.2 ± 0.3 mm in diameter, which was larger than the focal zone for the histotripsy source. At each location, 2000 pulses were applied at a 40 Hz rate for a total insonation time of fifty seconds. An additional seven seconds were required to download passive cavitation emissions and B-mode imaging data. The duration of the 1-MHz fundamental frequency pulse was either 1 or 5 μs to nucleate bubble activity primarily through the intrinsic threshold or shock scattering mechanisms, respectively. The peak negative pressure of the pulse was either 0, 15, 20, 25, or 30 MPa. The approximate total treatment duration was 20 min.

Prior to the histotripsy excitation, a scanline B-mode image was acquired to visualize the lumen and clot (Figure 2b). During the insonation, acoustic emissions from the bubble cloud were passively acquired with the L11–4v imaging array and processed offline to form passive cavitation images using a delay, sum, and integrate algorithm (Haworth et al. 2017). Pixel amplitudes were proportional to the energy of the received cavitation emissions. Radiofrequency signals were passively recorded over 64 μs with an appropriate delay to account for the ultrasonic time of flight (Figure 2b). Acoustic emissions were processed between 2 and 10 MHz, the nominal sensitivity range of the imaging array. Due to the data transfer rate limitations, acoustic emissions were acquired once every tenth histotripsy pulse (200 total PCI frames/location). A minimum of 8 clots were used for each treatment arm (171 total).

Following treatment, the residual clot segment was removed from the flow channel, blotted, and weighed in an identical manner as done before treatment. Thrombolytic efficacy was reported as percent mass loss. In a subset of each arm, a 500 μL sample of the perfusate was collected within 5 minutes after treatment and mixed with 0.1 mg/μL of 6-aminocaproic acid (Sigma-Aldrich, St. Louis, MO, USA) to halt further lysis of any clot debris (Petit et al. 2015). The size distribution of clot debris was measured utilizing flow cytometry (ImageStreamX Mark II, MilliporeSigma, Burlington, MA, USA). Based on the particle size, debris was categorized as subcellular (less than 6 μm), cellular (6 μm - 8 μm), and super-cellular (greater than 8 μm) (Turgeon 2004). The flow cytometer can detect particles 4 μm in diameter and larger. A minimum of 3 perfusate samples were collected per treatment arm.

Passive cavitation image analysis

For each location within the clot, passive cavitation images were summed pixel-by-pixel to generate a cumulative passive cavitation image with a pixel amplitude proportional to the total received power. For a given data set, B-mode images were segmented manually to determine the clot area (Figure 3). The passive cavitation images were co-registered with the B-mode images, and the pixel values within the clot were summed to compute the total acoustic power within the clot, P_clot:

$$P_{\mathit{\text{clot}}}=\sum _{i=1}^{21}\sum _x^{\forall x\in \text{ clot}}{P}_i\left(x\right),$$

where P_i(x) is the PCI acoustic power at a pixel location x for an insonation location i. For the equation above, the first summation is over insonation location within the clot (21 locations/clot), and the second summation is over pixel locations within the clot.

Figure 3:

Duplex B-mode/passive cavitation images for (a) single-cycle and (b) five-cycle pulse duration. For both panels, the peak negative pressure was 15 MPa. Note that the color bar has different limits in each subfigure. B-mode images are acquired prior to histotripsy exposure (Figure 2b).

Histological Analysis

Following treatment, clots were sectioned to 2 to 3 mm in thickness, embedded in low gelling temperature agarose (Sigma-Aldrich, St. Louis, MO, USA) to preserve the clot orientation relative to the histotripsy source, and fixed in 10% formalin for 24 to 36 hrs. The clot/agarose samples were embedded in paraffin and stained with Masson’s trichrome (860–031, Roche, Basel, Switzerland) and CD61 (M075301–2, Agilent Technologies, Santa Clara, CA, USA). Masson’s trichrome stains nuclei black, cytoplasm, muscle, and erythrocytes red, and collagen blue. The CD61 stain marks platelets brown. The histological results were qualitatively analyzed for structural clot damage due to the histotripsy insonation and thrombolytic exposures. For each treatment arm histological samples from 2 clots were collected, as well as 2 untreated clots (i.e. not introduced into the flow channel).

Statistical Analysis

Statistical analysis was performed using MATLAB (The Mathworks, Natick, MA, USA). A one-way analysis of variance (ANOVA) test was performed in conjunction with a Tukey’s honest significant difference test using an α level of 0.05 to determine differences in the lytic efficacy or P_clot between arms. Pearson correlation coefficients were calculated for mass loss and P_clot.

Results

Thrombolytic Efficacy

The clot mass loss for all treatment arms is reported in Figure 4. The p-values for comparing the differences in thrombolytic efficacy between peak negative pressures at a given treatment group are shown in Table 2. In the absence of histotripsy or rt-PA exposure, the clot mass loss was 11.12 ± 0.91%, indicating some clot lysis from endogenous factors in the plasma or shearing from the perfusate. The addition of rt-PA increased the mass loss to 24.78 ± 6.15%. For the application of histotripsy pulses alone, increased efficacy compared to rt-PA alone was achieved for the five-cycle pulses with peak negative pressure 25 MPa or greater. Either the one- or five-cycle pulses in combination with rt-PA were more efficacious than rt-PA alone at 25 and 20 MPa, respectively. Thrombolytic efficacy was maximized for the rt-PA and five-cycle histotripsy pulse arm for a given peak negative pressure on average but was only significantly greater than all other arms for 20 and 25 MPa peak negative pressure pulses. The largest mass loss observed was for clot exposed to rt-PA and five-cycle duration pulses at 25 MPa peak negative pressure, 65 ± 8%. The mass loss for this arm was a factor of five greater than the plasma alone arm, and a factor of 2.6 greater than the plasma with rt-PA arm. For all arms, the thrombolytic efficacy increased for peak negative pressure between 15 and 25 MPa. Beyond 25 MPa, no additional increase in the thrombolytic efficacy was observed for any arm.

Figure 4:

(a) Mass loss for each experimental arm as a function of the peak negative pressure of the histotripsy pulse. (b) Mass loss grouped by treatment arm. The asterisk (*) denotes a significant difference with respect to rt-PA alone. The triangle (Δ) denotes a significant difference to all other arms at a given peak negative pressure.

Table 2:

p-values comparing differences in mass loss for a given arm as a function of the peak negative pressure (PNP). This table corresponds to the results shown in Figure 4b. Highlighted values indicate a significant difference (p < 0.05). A p-value of 0.00 indicates p < 0.005.

Treatment Arm	PNP (MPa)	15	20	25	30
1 cycle
	15	-	1.00	0.10	0.00
	20		-	0.20	0.00
	25			-	1.00
	30				-
1 cycle, rt-PA
	15	-	0.96	0.00	0.00
	20		-	0.14	0.00
	25			-	1.00
	30				-
5 cycle
	15	-	0.53	0.00	0.00
	20		-	0.02	0.00
	25			-	1.00
	30				-
5 cycle, rt-PA
	15	-	0.98	0.00	0.00
	20		-	0.00	0.01
	25			-	1.00
	30				-

Bubble cloud emissions

Strong bubble activity was noted within the clot for all insonation schemes as monitored with passive cavitation imaging (Figure 3). Broadband spectra were observed for the single-cycle calibration waveforms (Figure 1c). The five-cycle excitation waveforms had harmonics of the 1-MHz fundamental frequency (Figure 1d) Both spectra decreased monotonically with frequency. The spectra of the peak pixel values of the passive cavitation images were dependent on the pulse duration (Figure 5a–d). Single-cycle pulses nucleate broadband emissions, with peak emissions near 2 MHz that decreased monotonically over the receive bandwidth. For the five-cycle pulses, line emissions at harmonics of the 1-MHz fundamental frequency of the histotripsy pulse were present in addition to the broadband emissions.

Figure 5:

Average emission spectrum of (a) single-cycle pulses of 15 MPa peak negative pressure, (c) five-cycle pulses of 15 MPa peak negative pressure, (b) single-cycle pulses of 25 MPa peak negative pressure, and (d) five-cycle pulses of 25 MPa peak negative pressure. The acoustic emissions were mapped using passive cavitation imaging (PCI). The spectra were taken from the pixel with the maximum amplitude, normalized by the sensitivity of the imaging array and the maximum amplitude.

The acoustic power within the clot, P_clot, is shown in Figure 6a as a function of the peak negative pressure of the histotripsy pulse. The power of emissions increased for peak negative pressures 15 to 25 MPa for the five-cycle arm with and without rt-PA. For single-cycle excitations (with or without rt-PA), no increase in P_clot was observed between 15 and 20 MPa or between 25 and 30 MPa peak negative pressure. For a given arm, no difference was observed in P_clot between the 25 and 30 MPa peak negative pressure exposures except the five-cycle pulse duration without rt-PA. For a given peak negative pressure and pulse duration, there was no difference in P_clot for arms with and without rt-PA. For a given peak negative pressure or dose of rt-PA, emissions were increased for the five-cycle pulses in comparison to the single-cycle pulse except at 30 MPa.

Figure 6:

(A) Acoustic power, P_clot, from the passive cavitation images for each peak negative pressure and treatment arm. (B) Normalization of P_clot by the excitation pulse duration (1 or 5 μs).

Figure 6b shows P_clot normalized with respect to the excitation duration. No difference was observed amongst the arms in the normalized values of P_clot for 15 MPa peak negative pressure histotripsy pulse. For peak negative pressure greater than 20 MPa, the single-cycle pulses were greater than five-cycle pulses, with or without rt-PA.

In order to discern the relationship between the strength of bubble activity and thrombolytic efficacy, the clot mass loss was assessed as a function of P_clot for each arm (Figure 7). Here, P_clot was not normalized by the excitation pulse duration. The Pearson correlation coefficients for each treatment arm was greater than 0.6. A significant correlation was observed between the thrombolytic efficacy and the strength of the bubble activity for each experimental group (p < 0.05). A diminishing increase in thrombolysis was observed for the strongest two measurements of P_clot for all groups, indicating overtreatment of each location at this scan rate or an upper limit to the synergistic bubble activity for rt-PA. The degree of thrombolytic efficacy, assessed by mass loss, was independent of pulse duration for a given P_clot when comparing arms with or without rt-PA (see markers in Figure 7). The marked locations are different pulse durations as well as peak negative pressures.

Figure 7:

Scatter plot comparing the clot mass loss and bubble activity-induced acoustic power, P_clot, for clots treated with (A) histotripsy alone, or (B) histotripsy and rt-PA. The Pearson correlation coefficients for each treatment arm indicated a significant correlation between thrombolytic efficacy and the strength of the bubble cloud activity for each arm (p < 0.05). Instances where the clot mass loss and P_clot coincided for the single and five cycle pulse durations were observed (arrows), indicating that the thrombolytic efficacy was independent of pulse duration.

Particle debris

Figure 8 depicts the number of subcellular, cellular, and supercellular clot debris particles measured in the perfusate post insonation. For each arm, on average, 90% of debris was subcellular in size. Cellular and supercellular debris particulate counts were a factor of 10 and 100 times smaller than the subcellular counts, respectively. For all arms, the largest particle observed was 38 μm (single-cycle pulse with rt-PA at 30 MPa peak negative pressure). For a given set of insonation conditions or particle categorization, no differences were observed in the total number of particles between any of the arms.

Figure 8:

Flow cytometry particle size measurements. The distribution is split into three size ranges with respect to the diameter of a red blood cell: (a) subcellular (between 4 and 6 μm), (b) cellular (between 6 and 8 μm), and (c) supercellular (between 8 and 38 μm).

Histological analysis

Representative stained clot sections of untreated clots (i.e. not introduced into the flow phantom) are shown in Figure 9. Clots exposed to histotripsy pulses of 0 or 25 MPa peak negative pressure with and without rt-PA are shown in Figure 10. For samples not exposed to histotripsy or rt-PA (Figures 9 and 10a.1), red blood cells and macrophages were apparent throughout the clot section as visualized on the Masson’s trichrome stain. A thin fibrin layer was concentrated at the edge of the clot (Figure 9b), with an intermittent interior fibrin network. Platelets, stained brown on the CD61 stain, were grouped in discrete clusters periodically throughout the cross section of the clot, with a high-density rim along the edge of the clot (Figure 9c and d). Gross damage was apparent on Masson’s trichrome for samples insonified in the presence and absence of rt-PA. Larger regions of clot disintegration were observed for the five-cycle pulses, with damage both within the clot as well as the posterior surface (i.e. the interface between the clot and the lumen that is the distal most point on the clot from the histotripsy source). The shape of the anterior clot surface was deformed for clots exposed to rt-PA and five-cycle pulses (Figure 10f.1). The platelet grouping was occasionally diffused from its tight grouping when exposed to histotripsy compared to control samples (Figure 10c–f).

Figure 9:

Histological sample of an untreated clot (i.e. no exposure to plasma, rt-PA, or histotripsy). (a) Masson’s Trichrome stain of clot section, (b) magnified Masson’s Trichrome stain of clot section. The fibrin network is faintly visible on the edge of the clot and marked with a black arrowhead. (c) CD61 stain of clot section. Platelets are stained brown. Blue coloring is a counterstain. (d) CD61 stain of a magnified clot section. All slides were scanned at 20× magnification.

Figure 10:

Masson’s trichrome and CD61 stains of clot samples subject to the flow channel. The numerical suffix 1 denotes Masson’s trichrome stains, and 2 denotes CD61 stains. (a) Plasma exposure alone. (b) Plasma and rt-PA exposure. (c) Exposure to histotripsy alone: single-cycle pulses, 25 MPa peak negative pressure. (d) Histotripsy and rt-PA: single-cycle pulses, 25 MPa peak negative pressure. (e) Exposure to histotripsy alone: five-cycle pulses, 25 MPa peak negative pressure. (f) Histotripsy and rt-PA: five-cycle pulses, 25 MPa peak negative pressure. For clots in the second and third rows (c-f), the histotripsy pulse propagated from top to bottom in the image. The orientation in sample d was lost. The mass losses were 5, 24, 21, 55, 29 and 62% for samples a through f, respectively. Black arrowheads denote areas of ablation. The red arrowhead in (f.1) marks deformation of the anterior surface of the clot. The scale bar in panel a.1 applies to all panels. All slides were scanned at 20× magnification. The −6 dB width of the histotripsy focal zone ran 5 mm from top to bottom in the image, and 1 mm along the left to right axis of the image.

Discussion

Thrombolytic efficacy

The objective of this study was to compare the influence of pulse duration on the efficacy of histotripsy in combination with a lytic therapy. Different types of bubble activity were initiated depending on the duration of the histotripsy pulse (Bader et al. 2019; Khokhlova et al. 2015). In this study, five-cycle pulses were assumed to primarily generate bubble clouds via a shock-scattering mechanism (Maxwell et al. 2011b), whereas single-cycle pulses predominately activated nuclei intrinsic within the clot via tension only.

For histotripsy without rt-PA, only five-cycle pulses at peak negative pressures of 25 MPa and 30 MPa yielded a thrombolytic efficacy greater than rt-PA alone (Figure 4). Histotripsy in combination with rt-PA was more efficacious than rt-PA alone at peak negative pressures of 20 MPa or greater for five-cycle excitations, respectively. In a previous study, histotripsy pulses of five-cycle duration and 14 MPa peak negative pressure were sufficient to enhance rt-PA lytic efficacy compared to rt-PA alone (Bader et al. 2016). There may be several reasons why a larger peak negative pressure was necessary in these studies to observe lytic enhancement. In the previous study, strong thrombolytic enhancement was only observed following shock wave formation of the excitation pulse. Notably, the transducer geometry in each study was different. The transducer utilized in this study had an f-number of 0.75, whereas the f-number for the source employed in the previous study was 0.9 (Bader et al. 2016). The increased focal gain comes at a cost of a decreased length of focal zone, necessitating an increased peak negative pressure that coincides with shock wave formation (see Figure 1f, (Rosnitskiy et al. 2017)). The coincidence of shock wave formation with enhanced lytic activity is consistent with a previous observation of shock-enhanced drug delivery (Khokhlova et al. 2018). Together, these data indicate that shock-scattering enhancement of lytics requires knowledge of the onset of shockwave formation. This aspect of transducer geometry should be considered when developing a source based on the target pathology. Another potential reason for the difference is that the previous study utilized porcine clots, which are less susceptible to rt-PA lysis (Flight et al. 2006; Holland et al. 2008; Huang et al. 2017; Mercado-Shekhar et al. 2018b), whereas human whole-blood clots and plasma were employed in this study. Finally, clots were subjected to flow which may alter the clot lytic profile compared to the static model used in the previous study (Bajd et al. 2010).

Single-cycle excitations generate bubble activity primarily via activating intrinsic nuclei. Over a wide range of water-based media, the intrinsic threshold for cavitation is approximately 25 MPa (Bader et al. 2019; Maxwell et al. 2013; Vlaisavljevich et al. 2015). Thrombolytic enhancement relative to rt-PA alone was not observed here for the application of single-cycle pulses to the clot alone (Figure 4). Single-cycle excitations with rt-PA only yielded an increased thrombolytic efficacy for peak negative pressure 25 MPa and greater. The results here indicate that single-cycle pulses can also be utilized to enhance rt-PA efficacy provided the tension is sufficient to overcome the intrinsic threshold. The extent of bubble activity can be largely controlled based on the transducer geometry (Vlaisavljevich et al. 2017), making intrinsic threshold histotripsy a good candidate for confining bubble clouds within the finite extent of a deep vein lumen.

Five cycle excitations with or without rt-PA were generally more efficacious than single cycle pulses. This may be in part due to the increased degree of bubble activity for five-cycle pulses compared to single-cycle pulses, as indicated in Figure 6. The difference in total insonation time between the two pulsing schemes may also have been a contributing factor (total ultrasound exposure duration per clot of 210 ms and 42 ms for five and single-cycle pulses, respectively). Indeed, PCI noted a stronger degree of bubble activity for five cycle pulses in comparison to single cycle pulses for a given peak negative pressure (Figure 6). For all treatment arms, no increase was observed in the thrombolytic efficacy as the peak negative pressure increased from 25 to 30 MPa. Previous studies have noted the maximum size of histotripsy-nucleated bubbles do not increase as the peak negative pressure of the pulse exceeds the intrinsic threshold (Vlaisavljevich et al. 2015). The degree of bubble-induced strain, a primary mechanism for histotripsy liquefaction, is directly proportional the maximum bubble size (Bader 2018). The results here indicate a limiting return in the increase of thrombolytic efficacy as the peak negative pressure of the pulse exceeds 25 MPa. The lack of increase may be attributed to overtreatment of the clot at each location, indicating the need for a feedback loop to assess treatment progress.

Perfusate particle size

For all treatment arms, the particles in the perfusate were submillimeter in diameter, with a maximum particle size of 38 μm. These results were consistent with previous observations of histotripsy clot ablation (Maxwell et al. 2009; Zhang et al. 2016). A limited variability was noted amongst the treatment groups in terms of the number of particles. This may be due to a limitation of utilizing flow cytometry for particle sizing. Previous studies have noted the majority of clot debris is subcellular (Bader et al. 2016), less than 4 μm in diameter. Here, the lower sensitivity limit of our method restricted analysis to particulate larger than 4 μm in diameter. The lack of large particulates is promising in terms of clinical translation. Mechanical thrombectomy procedures have generated debris up to a millimeter in size without causing severe distal embolism (Müller-Hülsbeck et al. 2001; Uflacker et al. 1996; Yasui et al. 1993), indicting the debris generated via histotripsy would be tolerated in a clinical procedure.

Bubble emissions

The difference in the thrombolytic efficacy between the single- and five-cycle pulses may be reflective of the differences in bubble cloud activation between the pulsing schemes. Indeed, there is a strong variation in the spectra of the received emissions indicative of different bubble dynamics, as noted in Figure 5. Single-cycle pulses exhibited a broadband spectrum that decreased monotonically with frequency, with no prominent features. These broadband signals originate either from inertial cavitation emissions or fundamental scatter of the pulses from the clot and latex lumen. Further work is needed to determine the contributions of the received emissions from bubble activity and fundamental scattering. Broadband emissions were accompanied by harmonics of the fundamental frequency for the five-cycle pulses. The harmonics might be attributed to fundamental scattering of the incident shockwave from expanded bubbles within the focal zone (Maxwell et al. 2011b) or from the tube or clot. Together, these data indicate a diverse degree of bubble activity between the two modes of histotripsy.

The relationship between P_clot and the peak negative pressure was dependent on the duration of the histotripsy pulse. For single-cycle pulses with and without rt-PA, sporadic cavitation emissions were observed for peak negative pressures less than 25 MPa. There was a step-wise increase of P_clot for single-cycle pulses, with emissions at similar levels for peak negative pressures between 15 and 20 MPa, and between 25 and 30 MPa. The dividing line between strong and weak emissions at 25 MPa is consistent with exceeding the intrinsic cavitation threshold for water-based media (Maxwell et al. 2013; Vlaisavljevich et al. 2015; Vlaisavljevich et al. 2016). For five-cycle pulses both with and without rt-PA, bubble activity increased for peak negative pressures between 15 and 25 MPa. The intensity of bubble activity nucleated by five-cycle pulses increased between 25 and 30 MPa without rt-PA, but not with rt-PA. Previous observations have noted a lack of increase in the bubble size for peak negative pressures greater than the intrinsic threshold (Vlaisavljevich et al. 2015), which may explain the lack of consistent increased strong bubble activity between 25 and 30 MPa observed here.

When normalized by the pulse duration, bubble activity was stronger for the single-cycle pulses in comparison to the five-cycle pulses for peak negative pressures greater than 15 MPa (Figure 6b). These results may indicate the most intense bubble activity may occur within the first few cycles for multi-cycle pulses (Wang et al. 2010). Newly formed bubbles within the cloud may shield distal portions of the focus, mitigating sustained intense activity throughout the duration of the excitation (Maxwell et al. 2011b).

A primary consideration to mitigate collateral damage is containment of the bubble cloud within the lumen. Strong pre-focal cavitation for the five-cycle pulses necessitated translating the focal zone towards the bottom of the clot. Pulses here were applied at a 40 Hz pulse repetition frequency, which may have contributed to the prevalence of pre-focal cavitation (Zhang et al. 2015a). Bubble deleting pulses interspersed with ablative histotripsy pulses could aid in the reduction of off-target effects (Shi et al. 2018). In contrast, bubble activity was largely contained within the clot for single-cycle pulses.

A significant correlation was observed between P_clot and the thrombolytic efficacy (Figure 7), particularly for peak negative pressures less than 25 MPa. The strength of emissions was not increased between 25 and 30 MPa peak negative pressure, consistent with the lack of increased thrombolytic efficacy observed over this range for each treatment arm. The lack of increased bubble activity beyond 25 MPa indicates this to be an upper insonation limit for thrombotripsy to contribute appreciably to thrombolytic efficacy or overtreatment based on the insonation schemes employed in this study. For a given P_clot, the thrombolytic efficacy appeared to be independent of the pulse duration for arms with or without rt-PA (see markers in Figure 7). Notably, different insonation pressure amplitudes were needed to achieve similar values of P_clot for single-cycle versus five-cycle pulses. This result indicates that the strength of the bubble activity is important in terms of gauging clot liquefaction, not the mechanism of bubble generation (which may be better determined based on the shape of the power spectrum).

Limitations

There are several aspects of this in vitro study that limit the generalizability of these findings. The delay, sum, and integrate beamforming method used to process the passive cavitation images has poor range resolution due to the diffraction-limited point spread function (Haworth et al. 2012; Haworth et al. 2017), which may alter the estimated degree of bubble activity. The flow channel was submerged in a water bath, and clots were targeted under B-mode guidance. Attenuation due to tissue layers in vivo may cause degradation of the image quality and thus targeting. The transducer geometry used in this study would not contain cavitation within the lumen of the human iliofemoral vasculature. This necessitates the development of specialized sources based on the desired bubble activity for the treatment of DVT.

A partial occlusion model was employed in this study, negating the effects of total vascular occlusion (Liebeskind et al. 2011). Total occlusions would inhibit clot thrombolytic exposure when administered systemically, though positioning the catheter directly into the thrombus is the preferred clinical method for the treatment of deep vein thrombosis (Saha et al. 2016). The addition of a histotripsy-induced flow channel may further increase lytic exposure to the clot (Badj et al. 2010). The influence on systemic toxicity or side effects of this treatment could not be ascertained in this in vitro model, thus future in vivo studies are needed to compare adverse side effects against those in existing therapies. The low flow rate considered was fixed, which neglects the contribution of increased flow as the clot lyses (Bajd et al. 2010). Retracted human whole-blood clots were employed in this study, whereas several thrombus phenotypes have been identified in vascular occlusions (Barros D’Sa and Chant 2005; Czaplicki et al. 2017; Liebeskind et al. 2011). To assess thrombolytic efficacy via mass loss, clot samples were blotted to remove excess fluid consistent with previous in vitro studies (Holland et al. 2008; Hitchcock et al. 2011; Sutton et al. 2013; Bader et al. 2016). Blotting would not be possible when treating the clot in vivo, and future studies will consider alternative means to assess treatment efficacy that do not include removal of the excess clot fluid. Histological samples were assessed at a fixed point relative to the length of the clot, preventing analysis on spatial uniformity. Residual clot was observed for all experimental arms due to the insonation protocol employed in this study, and the size of the histotripsy focal zone relative to the clot diameter. The presence of residual clot facilitates comparison amongst treatment arms. Future studies will investigate thrombotripsy schemes for total removal of the thrombus.

Conclusions

Combining histotripsy with a lytic therapy may provide improved thrombus dissolution. The type of bubble activity initiated by histotripsy is dependent on the pulse duration. In this study, the efficacy of thrombotripsy, histotripsy in combination with rt-PA, was explored for bubble activity initiated by intrinsic threshold and shock-scattering histotripsy pulses. When combined with rt-PA, increased thrombolytic efficacy was observed above the peak negative pressures associated with the onset of shockwave formation for shock-scattering pulses (five-cycle duration) and above the intrinsic threshold for single-cycle pulses. Without rt-PA, only shocked pulses had a larger thrombolytic efficacy compared to the lytic alone (peak negative pressures greater than 25 MPa). Mapping of acoustic emissions during the insonation was a good predictor of thrombolytic efficacy for both pulse durations. Further, no increase was observed in the emission level or thrombolytic efficacy for peak negative pressures greater than 25 MPa. Clot spallation particulates were primarily subcellular in size, and no particles larger than 38 μm were detected. Overall, these results indicate that the efficacy of bubble activity initiated via either the shockwave scattering or intrinsic threshold histotripsy mechanism increase the efficacy of clot dissolution when used with rt-PA relative to just histotripsy alone.