The effect of ultrasound pulse length on sonoreperfusion therapy

Abstract

In recent years, long and short pulse ultrasound (US)-targeted microbubble cavitation (UTMC) has been shown to increase perfusion in healthy and ischemic skeletal muscle, in pre-clinical animal models of microvascular obstruction, and in the myocardium of patients presenting with acute myocardial infarction. There is evidence that the observed microvascular vasodilation is driven by the nitric oxide pathway and purinergic signaling, but the time course of the response and the dependency on US pulse length are not well elucidated. Since our prior data supported that long pulse US confers enhanced sonoreperfusion efficacy versus short pulses, in this study, we sought to compare long (5000 cycles) and short pulse (500 × 10 cycles) US at a pressure of 1.5 MPa with an equivalent total number of acoustical cycles hence constant acoustic energy and at the same frequency (1 MHz), in a rodent hindlimb model, with and without microvascular obstruction (MVO). By quantifying perfusion using burst replenishment contrast enhanced US imaging, we found that: (1) long and short pulses result in different vasodilation kinetics in an intact hindlimb model. The long pulse causes an initial spasmic reduction in flow that spontaneously resolved at 4-min, followed by sustained higher flow rates (~2-folds) compared to baseline, starting 10 min after therapy (p<0.05). The short pulse caused a short-lived ~2-folds increase in flow rate that peaked at 4-min (p<0.05), but without the initial spasm; (2) the sustained increased response with the long pulse is not simply reactive hyperemia; and (3) both pulses are effective in reperfusion of MVO in our hindlimb model by restoring blood volume, but only the long pulse caused an increase in flow rate after treatment 2, compared to MVO (p<0.05). Histological analysis of hindlimb muscle post UTMC with either pulse configuration indicates no evidence of tissue damage or hemorrhage. Our findings demonstrate that the microbubble oscillation induces vasodilation and therapeutic efficacy for the treatment of MVO can be tuned by varying pulse length; relative to short pulse US, longer pulses drive greater microbubble cavitation and more rapid microvascular flow rate restoration after MVO, warranting further optimization of the pulse length for sonoreperfusion therapy.

Introduction

Microvascular obstruction (MVO) is a complication of percutaneous coronary intervention that occurs when the myocardial microvascular network remains hypo-perfused despite recanalization of the epicardial artery. This multifactorial complication involves vasospasm, edema, inflammation, thrombosis, and distal coronary embolization, and is highly prevalent (up to 80% of cases)(Costantini et al. 2004; Wu et al. 1998). It is also an independent predictor of worsened short and long term clinical outcome (Gibson 2003; Ito 2014; Khumri et al. 2006). Utilization of microbubble (MB) targeted ultrasound (US) cavitation (UTMC) therapy for sonothrombolysis of acute myocardial infarction (AMI) is a promising therapy undergoing active investigation. Using an in vivo model of embolic microvascular obstruction (MVO), which often accompanies AMI, our group was the first to demonstrate that this same strategy can also be used to restore perfusion (‘sonoreperfusion, SRP’). In the setting of MVO, we showed that long pulse US confers enhanced SRP efficacy versus short pulse ultrasound, which derives from our prior data in: 1) in vitro models of MVO (Black et al. 2016; Chen et al. 2014; Goyal et al. 2017; Leeman et al. 2012; Roos et al. 2016b); 2) in vivo hind limb studies of perfusion kinetics and nitric oxide (NO) generation (Yu et al. 2017; Yu et al. 2020); and 3) rodent hind limb models of MVO (Istvanic et al. 2020; Pacella et al. 2015; Yu et al. 2017). Specifically, in our in vitro model of MVO, we demonstrated the mechanical superiority of long versus short pulse ultrasound (Leeman et al. 2012). In our perfusion kinetics study (Yu et al. 2017; Yu et al. 2020), we showed that long pulse ultrasound of 5000 cycles resulted in a sustained and marked rise in basal microvascular perfusion and NO output. Finally, in our hindlimb model of MVO, we compared long pulse ultrasound with short pulse ultrasound, using clinically available ultrasound systems and showed the long pulse US results in more rapid reperfusion of MVO compared to short pulse ultrasound (Istvanic et al. 2020).

Our findings were consistent with studies performed by others in the murine hindlimb (normal and ischemic) and in patients with sickle cell disease (Belcik et al. 2017; Belcik et al. 2015), which reported that ultrasound-induced microbubble oscillation can cause vasodilation and a subsequent increase in local microvascular perfusion. Note that there was no MVO in those studies. Microbubble cavitation was shown to cause purinergic signaling and phosphorylation of endothelial nitric oxide synthase (eNOS). Interestingly, in these studies, microbubble inertial cavitation was induced by a clinical scanner in harmonic power-Doppler mode (Sonos 7500, Philips Ultrasound) at 1.3 MHz, at a pulse repetition rate of 9.3 kHz and a mechanical index (MI) of 1.3. The Power Doppler pulses used in the latter study is relatively short (about 4 μs, much less than that used in our studies (i.e. 5 ms)) in length and are repeated at a fast pulse interval of 1/9.3 kHz or 108 μs. Therefore, these data support that short ultrasound pulses causing inertial cavitation are capable of producing vasodilation in skeletal muscle. Furthermore, it was also recently shown in a clinical trial that diagnostic ultrasound using high mechanical index (HMI) with frequencies ranging from 1.3–1.8 MHz at a MI of 1.1–1.3 using a clinical scanner (iE33, Philips) and pulse lengths ranging from 3–20 μs in ST-elevation myocardial infarction (STEMI) patients before and after percutaneous coronary intervention improved myocardial perfusion score index and left ventricular ejection fraction at 1 month and at 1 year follow-up (Mathias et al. 2019; Mathias et al. 2016). Interestingly, increasing the pulse length from 5 to 40 cycles caused an increase in perfusion flux rate (β) with a low line density (but not at high line density) in an intact mouse hindlimb and substantially increased ATP released at a high line density (Mason et al. 2019), indicating that with clinical scanners, the effect of the pulsing is complex and multifactorial. Altogether, these studies support that UTMC holds therapeutic potential for improving perfusion in MVO, ischemic muscle and sickle cell disease and improves the clinical outcome for STEMI patients in combination with percutaneous coronary intervention. However, in our studies, and in those of other investigators comparing long and shorter pulse ultrasound, the total amount of ultrasound energy delivered was not kept constant, and variations in total ultrasound energy delivered could have important implications for efficacy. Therefore, in the current study, we sought to parcel the contribution of pulse length to SRP efficacy by comparing the reperfusion cause by a long versus short pulse ultrasound regime, with both energy and frequency matching using a single element transducer. Our goal was to define the principles that govern SRP efficacy, and in so doing, we assessed microvascular perfusion, rescue from MVO, and MB cavitation activity following SRP and report our results herein.

Materials and Methods

Ultrasound theranostics: Image guided therapy and monitoring

A 1 MHz single element transducer (A303S, 0.5 inch, Olympus NDT, Waltham, MA) was used for therapeutic US delivery, positioned on top of the hindlimb and was coupled to the shaved skin using acoustic gel. The therapeutic transducer was calibrated in a degassed water tank using a lipstick hydrophone (HGL-0200, Onda Corporation, Sunnyvale, CA). The long therapeutic pulse (LP) comprised US tone bursts 5,000 cycles (5000 μs) in duration delivered every 3 seconds at a peak negative pressure of 1.5 MPa. The rapid short pulse (RSP) consisted of a train of 500 × 10 cycle pulses (500 × 10 μs), each separated by 100 μs, also delivered every 3 s at a peak negative pressure of 1.5 MPa. Thus both LP and RSP therapeutic pulses consisted of 5000 cycles delivered every 3 s with identical time-averaged ultrasound intensity.

The alignment of the therapeutic transducer was verified by visualization of the destruction of microbubbles in the central area of the image by the therapeutic pulse using contrast perfusion imaging. Contrast perfusion imaging was performed with a clinical scanner (15L8 probe, Sequoia 512, Siemens Healthineers, Erlangen, Germany) during continuous infusion (via right internal jugular vein) of Definity™ microbubbles (2 mL/h). The syringe containing the MB was rotated continuously to avoid MB floatation. The imaging probe was positioned horizontally with respect to the hindlimb and in a plane perpendicular to the therapy transducer beam to view the hindlimb in the longitudinal plane, such that the femoral bone was located in the upper right side of the image and the major feeding vessels could be observed, with blood flow traversing from right to left (Figure 1). Imaging was performed in Cadence Pulse Sequencing mode (CPS) at 7 MHz, at an imaging mechanical index (MI) of 0.2 and a framerate of 5 Hz. The imaging probe was used to deliver a burst pulse (5 frames at an MI of 1.9), which cleared the MB from the field of view. It was demonstrated elsewhere that these burst pulses did not cause a vasoactive response (Yu et al. 2017). Perfusion kinetics was then tracked as the MB replenished the vascular network, and the cine loops were stored for offline processing. The dynamic range (60 dB), digital gain (0 dB), scanner compression curve (linear), along with other image processing parameters were fixed. The therapeutic transducer beam profile and other details about the setup geometry can be found in Yu et al. 2017, Figure S2.

Figure 1: Experimental setup –

a) A therapeutic probe was perpendicularly positioned with respect to the imaging plane of a clinical probe operated in contrast mode, which used to guide and monitor therapy. Therapeutic microbubbles were infused via a contralateral femoral access for direct first pass arterial delivery. This port was also used for microclot injection in the MVO model. Imaging microbubbles (Definity™) were infused via a jugular access and were used for burstreplenishment imaging to quantify perfusion. B) Experimental protocol for the intact hindlimb experiment, including therapy and imaging timepoints. C) Experimental protocol for the MVO model, including therapy and imaging timepoints.

Microbubbles

Phospholipid-encapsulated MB containing perfluorocarbon (PFC) gas (Perfluorobutane, Fluoromed, Round Rock, TX) were prepared in house for therapy, and measured using a Coulter counter (Multisizer 3, 50 μm aperture, Beckham Coulter, Brea, CA), as described previously. (Weller et al. 2002) In brief, 1,2-distearoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids Inc., Alabaster, AL), polyoxyethylene(40)stearate (Sigma Aldrich, St Louis, MO) and 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethyleneglycol)-2000] (Avanti Polar Lipids) were mixed in 88:11:1 molar ratios in chloroform, dried under a stream of argon gas and stored under vacuum overnight. The film was rehydrated in 4 mL saline in a glass vial filled with an overhead of PFC. The lipids and PFC were sonicated for 75 s (XL2020, Qsonica LLC, Newtown, CT) and the resulting MB were washed 3 times in saline. The microbubbles had a diameter of 3.5 ± 1.1 μm in diameter (Multisizer 3, Beckman Coulter, Brea, CA), a concentration of 1×10⁹ MB/mL and were stored in sealed glass vials filled with PFC until usage (within 2 weeks of fabrication). These larger MB, delivered arterially directly upstream of the hindlimb was preferred for maximizing the MB cavitation effects.

For contrast perfusion imaging, Definity™ microbubbles (Lantheus Medical Imaging, North Billerica, MA) were administered via jugular vein access at a flow rate of 2 mL/h. These MBs have a mean size of 1.1–3.3 μm and concentration of 1.2×10¹⁰ MB/mL. These commercial MB were chosen for imaging for a reproducible quantification of perfusion.

Microthrombi preparation

Fresh porcine blood in citrated buffer was obtained (Lampire Biological Laboratories, Pipersville, PA) and used within 14 days. On the day of the experiment, blood was reconstituted with 25 mM CaCl₂ in type 1 borosilicate glass vials (Supelco Analytical, Bellefonte, PA) and stored at room temperature for 2 h. The clotted blood was then shaken using a vial mixer (Vialmix, Lantheus Medical Imaging) for 10 sec, centrifuged at 1000 g for 5 min. The buffy coat and pellet were collected and successively passed through needles with progressively decreasing diameters (20G, 25G, 27G and 30G). The suspension containing the microthrombi was then filtered through a 200 μm pore mesh and counted (400 μm aperture, Multisizer 3, Beckman Coulter), yielding a large fraction of microthrombi in the 10–30 μm range [(6.3±1.6)×10⁵ clots/mL] and a tail in the 30–200 μm range [(6.5±3.4)×10⁴ clots/mL], similar to that reported previously (Yu et al. 2017).

Animal preparation

All animal studies were approved by the Institutional Animal Care and Use Committee at the University of Pittsburgh. We used our previously developed rodent hindlimb model (Pacella et al. 2015; Yu et al. 2017), to induce microvascular obstruction of the hindlimb gastrocnemius muscle. In rats weighing 270g, we induced anesthesia by inhalation of isoflurane (2.5%) and inserted an arterial cannula (PE-10, Becton Dickinson) via contralateral femoral access and advanced it to the abdominal aorta for the direct injection of microthrombi and therapeutic MB into the left hindlimb (Figure 1) in order to achieve first pass MB delivery. An angiocatheter (24G, ½ inch long) was placed in the right internal jugular vein for infusion of Definity™ microbubbles (Lantheus Medical Imaging, North Billerica, MA) for contrast imaging. Separate groups of animals were used in the following models (see details in Table 1).

Table 1 :

Summary of animal used for each experiment

Figure #	Substudy	Number of rats per group	Total
Figure 2	Cavitation detection	n=1	1
Figure 3	Intact hindimb	RSP n=7 LP n=7	14
Figure 4	Reactive hyperemia	n=3	3
Figure 5	MVO RSP vs Control	RSP n=5 Non treated control n=3	8
Figure 6	MVO RSP vs LP	RSP n=8 LP n=8	16
Figure 8a	RTPCR (VCAM1)	RSP n=8 LP n=8 Control muscle n=8
Figure 8b	Elisa (VCAM1)*	RSP n=7 LP n=7 Control muscle n=4
TOTAL			42 rats

^*For Elisa, the number of samples analysed were chosen to accommodate one 48-well plate, including all control samples, as recommended by the manufacturer. All samples were measured in duplicates. For figures 8a and 8b, tissue samples were taken from the same rats used in MVO RSP vs. LP study (Figure 6).

Intact hindlimb model

For the perfusion kinetics experiments (without MVO), therapeutic ultrasound was applied for 2 minutes and burst replenishment imaging was performed at baseline and at fixed intervals after therapy for 60 min. There was no injection of microthrombi in these animals. Separate animal experiments (n=7 each group) were performed to study both pulse configurations.

Reactive hyperemia experiment

An over the wire 2.0 mm angioplasty balloon (Maverick™, Boston Scientific, Marlborough, MA) was placed in the descending aorta over a guide wire in three rats, distal to the origin of the renal arteries via contralateral femoral access. The balloon was inflated to occlude blood supply to the hindlimb for 5 minutes, to reproduce a similar duration of transient hindlimb microvascular hypoperfusion observed after 2 minutes of the long pulse ultrasound. Balloon placement was performed by an experienced interventional cardiologist (JP) and guided by angiography (ARCADIS Avantic, Siemens, Malvern, PA). Total occlusion of blood flow was confirmed by continuous CEUS imaging and angiography. Imaging microbubbles (Definity™) were infused using a jugular vein access. Upon release, burst replenishment imaging was performed at discrete timepoints over 15 min (see Figure 4). Three rats were studied in this substudy.

Figure 4: Balloon induced reactive hyperemia –

a) The flow was obstructed for 5 min using an angioplasty balloon placed in the descending aorta, guided by fluoroscopy imaging; b) Flow rate kinetics, were measured at different time point following release of the occlusion (n=3, p<0.05, T-test).

MVO model

MVO was induced by injection of microthrombi into the femoral artery catheter until there was an estimated 75% decrease in peak video-intensity in the hindlimb microvasculature, as monitored by simultaneous diagnostic US contrast perfusion imaging. Stable hypoperfusion was confirmed for 10 minutes prior to proceeding with US therapy. If spontaneous reperfusion occurred (brightness increased above the 25% threshold of peak videointensity), additional microthrombi were administered as needed to maintain 10 min of stable hypoperfusion. After 10 minutes of stable hypoperfusion, therapeutic RSP (500 × 10 cycles) trains, were delivered every 3 seconds during simultaneous infusion of custom lipid MB into the femoral artery catheter via syringe pump at a flow rate of 3 mL/h. Two successive therapeutic UTMC sessions of 10 minutes each were performed. Burst replenishment imaging during continuous intravenous infusion of Definity™ as described above was performed at baseline, after MVO, and after each of the 10 min UTMC therapy sessions (typically 3min after the end of therapy to allow Definity™ MB to stabilize). Heart rate, respiration and O₂ saturation were continuously monitored. At the end of the experiment, animals were euthanized by isoflurane overdose and heart excision. Five animals received RSP (n=5) and 3 animals received no treatment (n=3) (see Figure 5). In a separate study, we performed a direct head to head comparison of LP and RSP, in which pairs of animals were randomly attributed to each group on each experimental session. n=8 rats/group were used for this sub-study (see Figure 6). Histology of hindlimb muscle was performed post-mortem in RSP group to detect tissue damage (see Figure 7). Fixed samples were stained with hematoxylin and eosin, and the sections were examined microscopically. VCAM-1 expression what quantified at the mRNA level by RT-PCR and at the protein level by ELISA (see details below) (see Figure 8).

Figure 5: UTMC efficacy in MVO model with the rapid short pulse -.

Blood volume (A) and flow rate (AxB) calculated using contrast enhanced burst replenishment ultrasound imaging at different stages of sonoreperfusion therapy using UTMC with the rapid short pulse (n=5) compared to control groups no receiving UTMC (n=3) ( *p < 0.05 vs baseline; + p < 0.05 vs MVO). A was only different from no treatment after treatment 2.

Figure 6: RSP vs LP efficacy in MVO model -.

Blood volume (A) and flow rate (AxB) calculated using contrast enhanced burst replenishment ultrasound imaging at different stages of sonoreperfusion therapy using UTMC (n=8/group) (*p < 0.05 vs baseline; + p < 0.05 vs MVO). Flow rate was only different from MVO for RSP after Tx2.

Figure 7: Histology –

Haematoxilin and eosin staining of muscle following microthrombi injection and treatment with UTMC with the rapid short pulse, indicating no overt evidence of micro vascular damage with mostly patent vessels (A); In some sections (B and C), we could find remaining occluding microthrombi in vessels. Scale bar = 500 μm.

Figure 8. VCAM-1 expression following two sessions of sonoreperfusion therapy:

(A) at mRNA level detected by RT-PCR (n=8 for all groups), and (B) at the protein level detected by ELISA, in untreated control muscle (n=4), and following RSP and LP sonoreperfusion therapy (n=7). No effect was detected by ANOVA testing. p-values indicate pairwise comparison using Tukey post-hoc testing

CEUS Perfusion quantification

A large area excluding the feeding vessels (microcirculation only) was selected and the average video intensity in the microcirculation was quantified following the burst, up until the video intensity plateaued (typically < 30 sec). It has been shown that the blood volume (A) and perfusion rate (A×B) could be estimated by fitting a mono-exponential function to the kinetics of video intensity (VI) using (Wei et al. 1998):

$$VI(t)=A\left(1-e^{-Bt}\right)$$ (Eq. 1)

In this expression A is the maximal peak plateau video intensity and A×B is the slope of the video intensity at t=0 and is consistent with perfusion rate. Typical perfusion data and a fitted model have been previously described (Pacella et al. 2015; Yu et al. 2017). Image analysis was performed offline on the cineloops obtained in CPS mode using custom Matlab software.

In vivo cavitation dose measurements

In an intact hindlimb, a 3.5 MHz transducer focused single element transducer (V383-SU-F1.0-PTF, 0.375 inch, Olympus NDT, Waltham, MA) was placed in confocal alignment with the therapeutic 1 MHz transducer. The signal from cavitation detector probe was amplified (10 dB) low-pass filtered (20 MHz) with a pulser/receiver (5073PR, Panametrics-NDT, Olympus) configured in receive mode, and digitized using an oscilloscope (WaveRunner 6051A, LeCroy, Chestnut Ridge, New York, NY) at a sampling frequency of 10 MHz. Therapeutic microbubbles were infused (2 mL/h) and imaged using CEUS imaging. With both therapeutic transducer and the cavitation detector in fixed position, and under CEUS imaging guidance, three RSP and three LP were sent into the muscle and the cavitation activity was recorded. Sufficient time was given to the muscle for complete microbubble replenishment of the tissue between pulses, which was confirmed by ultrasound imaging. Spectral analysis was performed offline using Matlab. Spectrograms were computing using rectangular windows of 128 points with 50% overlapping. The power in the fundamental (0.9–1.1 MHz), 2^nd harmonic (1.9–2.1 MHz) and 3^rd harmonic (2.9–3.1 MHz), subharmonic (0.4–0.6 MHz) and ultraharmonics (1.4–1.6 MHz) frequency bands were calculated and averaged over the 50 ms acquisitions. The broadband power was defined as inertial cavitation. For each frequency band, means and standard deviations over the three acquisitions are reported. To minimize the variability in cavitation dose measurements caused by changes between animal anatomies, measurements were repeated 3 times for each pulse configuration in the same animal.

Gene expression profiling by RT-PCR

Total RNA was isolated from the (~50 mg) left hindlimb muscle by TRIzol^Ⓡ reagent (Life Technologies) as per the manufacturer’s protocol. The purity and concentration of RNA were measured by spectraMax QuickDrop Micro-Volume spectrophotometer (Molecular Devices). Reverse transcriptase reaction was performed by iScript cDNA Synthesis Kit (BIO-RAD) from 1 μg of RNA. Primer sequences for vascular cell adhesion molecule-1 (VCAM-1; forward: TTTGCAAGAAAAGCCAACATGAAAG) and reverse TCTCCAACAGTTCAGACGTTA GC) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH; forward: AAACCCATCACC ATCTTCCA and reverse: GTGGTTCACACCCATCACAA) were purchased from IDT, USA. Real time-PCR was performed on CFX Connect Real-Time System, BIO-RAD the SsoAdvanced Universal SYBR^Ⓡ Green Supermix (BIO-RAD, USA). The data obtained were normalized to GAPDH expression as a reference gene.

Quantitative Sandwich ELISA for the Determination of VCAM-1

The level of VCAM-1 in the left hindlimb muscle was determined by a commercially available ELISA kit (Catalog #. MBS027532, MyBioSource, USA). Briefly, ~40 mg tissue homogenate in PBS was centrifuged at 5000 rpm for 15 min at 4°C followed by the collection of the supernatant which was used for VCAM-1 quantification, as per the manufacturer’s protocol. The optical density (OD) at 450 nm was measured using a spectrophotometer (DTX 880 Multimode Detector, Beckman Coulter, USA). Finally, the OD was normalized by protein concentration measured by the Bradford assay. All measurements were assessed in duplicates.

Statistics

All data were expressed as mean ± standard error on the mean. Statistical testing was performed using 2-way ANOVA to compare the effects of treatment time and treatment group. Multiple comparisons analysis to discern differences between groups was performed using Fisher’s LSD post-hoc test. Direct pairwise comparisons were performed using T-tests, as indicated. All tests were performed using Prism 9.2.0 (Graphpad software, LaJolla, CA). A p-value < 0.05 was considered statistically significant.

Results

Cavitation dose quantification

First, we quantified the cavitational activity of the two pulses in vivo. In Figure 2A and 2B, typical cavitation signals and spectrograms are represented for each pulse, over the entire 50 ms and zoomed in the first 3 ms. As expected, the long pulse, comprised of 5000 consecutive cycles, produced stronger activity over the first 5 ms while the 500 × 10 cycles pulse produced a weaker, periodic (every 100 μs) and prolonged activity over 50 ms. To capture and compare the entire cavitation activity for both pulses, we also computed the overall activity averaged over the entire 50 ms (Figures 2C and D). In Figure 2C, the time averaged power spectra indicate the presence of peaks in the harmonic (2^nd and 3^rd), subharmonic (0.5 MHz) and ultraharmonic (1.5MHz) frequency bands. The broadband signal also appeared to increase. Quantitatively (Figure 2D), while the fundamental power was similar for both pulses, the power in all the all non-linear frequency bands, representative of nonlinear scattering and MB oscillations, were significantly higher with the 5000 cycle pulses compared to the RSP (p<0.05), supporting that LP produced a stronger MB activity compared to RSP ultrasound even after time averaging. (see Table 2)

Figure 2: Passive cavitation detection in vivo –

Panels A) and B): representative 50 ms and zoomed in initial 3 ms time traces for the long pulse (A) and RSP (B) and corresponding spectrograms; C) Power spectra averaged over the entire 50 ms; d) Spectral power in respective frequency bands, averaged over the entire 50 ms for both pulses (n=3). H0.5=Subharmonic; F=fundamental; H1.5=ultraharmonics; H2=second harmonic; H3= third harmonic; IC=Inertial Cavitation; * indicates statistical significance (p<0.05, Fisher’s LSD multiple testing).

Table 2:

Cavitation power in respective frequency bands averaged over the 50 ms pulse length.

	Long Pulse [A.U.]	Rapid Short Pulse [A.U.]	p-value
H0.5	164.2 ± 14.7	128.3 ± 10.2	0.01
F	2117.0 ± 427.6	1286.0 ± 151.9	0.06
H1.5	316.9 ± 30.6	138.1 ± 5.4	0.01
H2	646.2 ± 168.9	231.7 ± 24.6	0.05
H3	173.8 ± 22.9	70.1 ± 3.2	0.01
IC	142.6 ± 19.6	68.0 ± 2.7	0.02

H0.5=Subharmonic; F=fundamental; H1.5=ultraharmonics; H2=second harmonic; H3= third harmonic; IC=Inertial Cavitation; p-value based on Fisher’s LSD multiple testing

Contrast perfusion imaging following 2-min UTMC

Next, we applied both LP and RSP to microbubbles, each in separate animal experiments, for two minutes of therapy and compared the effects in an intact hindlimb (Figure 3). This time point was chosen because we observed a marked change in perfusion using live contrast imaging after 2 min of therapy. Initially, the baseline flow rates were similar for RSP and LP (respectively 4.1 ± 0.9 dB/s and 4.5 ± 0.7 dB/s). With RSP, the flow rate (A×B) immediately increased post therapy and peaked (11.2 ± 2.9 dB/s) at 3 min, and decreased to 6.7 ± 1.1 dB/s at 30 min and 4.8 ± 0.9 dB/s at 60 min. With LP, the flow rate initially dropped to 1.2 ± 0.3 dB/s post therapy but then started to rise. It reached 6.7 ± 2.5 dB/s at 3 min, and was higher than baseline at 10, 30 and 60 min (respectively 10.3 ± 2.1 dB/s, 10.0 ± 0.8 dB/s and 8.0 ± 0.9 dB/s. Significant differences between LP and RSP were seen at 0 min (Post), 30 min and 60 min (all p<0.05), as indicated in Figure 3.

Figure 3: Effect of UTMC pulse length on the flow rate -.

Flow rate kinetics following 2 min of UTMC in an intact hindlimb. (BL) Baseline, Post (immediately after UTMC). * indicates statistical significance (p<0.05, T-test).

Balloon reactive hyperemia

Reactive hyperemia occurs after a transient cessation of blood flow, and is accompanied by a marked, transient increase in perfusion. Since we observed a temporary microvascular spasm that spontaneously resolved at 4-min post treatment with the long pulse, we wanted to determine whether simple reactive hyperemia could explain the increased flow observed starting 15 min post-treatment with LP. We therefore performed a balloon occlusion of the infrarenal aorta and quantified flow before and after occlusion using burst replenishment imaging kinetics (Figure 4). The baseline flow rate was 1.2 ± 1.0 dB/s. Upon release of the balloon, the flow rate immediately increased and peaked at 11.5 ± 0.6 dB/s, significantly higher than baseline (p<0.05) at 2-min post release. In comparison to the enhanced flow induced by LP UTMC therapy, this peak flow rate was very brief, and returned to baseline values starting 6-min post release.

Efficacy of RSP in the MVO model

Finally, we assessed whether rapid short pulse ultrasound achieved successful reperfusion in our hindlimb model of MVO and compared this to the efficacy of our previous long pulse US therapy. Similarly to our previous studies (Pacella et al. 2015; Yu et al. 2017), we performed two 10 minute sessions of therapeutic ultrasound and quantified blood volume (A) and flow rate (AxB) using burst replenishment imaging at baseline (BL), 10 min post microthrombi injection (MVO), to confirm stable occlusion, and following each therapy session (Tx1 and Tx2) in a group receiving RSP therapy and a negative control group receiving no US (Figure 5). At baseline, blood volume (A) were >20 dB in both groups (respectively 21.3 ± 4.4 dB and 22.3± 4.6 dB for RSP and for negative control (no US). Blood volume was reduced to < 1dB following injection of microthrombi to create MVO in both groups. Following Tx1, blood volume increased to 7.2 ± 5.0 dB in the RSP group and 4.8 ± 4.9 dB in the control group and was significantly increased compared to MVO in the RSP group (p<0.05). Following Tx2, blood volume increased to 16.4 ± 5.0 dB in the RSP group but only to 6.0 ± 5.3 dB in the control group. This difference was significantly different between the two groups (p<0.05) and for the RSP, blood volume was similar to baseline value, indicating reperfusion. The flow rates (A×B), were similar at baseline, respectively 4.4 ± 2.4 dB/s and 5.1 ± 1.8 dB/s for RSP and control groups. Flow rates significantly decreased after microthrombi injection and never recovered to baseline values (p<0.05). Following sacrifice, we did not observe any gross evidence of hemorrhage on visual inspection of the tissue after therapy with RSP. Additionally, we found no evidence of microvessel trauma or hemorrhage in the UTMC treated area, as illustrated in Figure 7, which looked normal and with patent vessels. In some sections, we could identify some remaining occluding microthrombi in the vasculature. These results were similar to our previous observations with the LP, which can be found in (Pacella et al. 2015). Finally, we measured VCAM-1 after LP, RSP and untreated controls, as a marker of vascular inflammation and endothelial damage at the mRNA and protein levels. We did not find evidence of an increase in VCAM-1 caused by RSP or LP (Figure 8).

Comparison of RSP vs LP efficacy in MVO model

Finally, we directly compared RSP and LP in our hindlimb model of MVO (Figure 6). Both pulses were similarly efficacious at restoring blood volume. Blood volumes were similar for RSP and LP at baseline (respectively 21.0 ± 0.7 dB and 19.8 ± 0.7 dB), MVO (respectively 0.3 ± 0.2 dB and 0.1 ± 0.1 dB), Tx1 (respectively 10.2 ± 3.4 dB and 8.5 ± 2.2 dB) and Tx2 (respectively 14.1 ± 3.7 dB and 16.3 ± 3.1 dB) at which time the blood volumes were restored to BL values. The flow rate for RSP and LP were also similar at BL (respectively 4.3 ± 0.3 dB/s and 4.6 ± 0.6 dB/s) and MVO (respectively 0.04 ± 0.02 dB/s and 0.02 ± 0.01 dB/s), and after Tx1 (respectively 1.5 ± 0.8 dB/s and 1.2 ± 0.6 dB/s). Interestingly, after Tx2, flow rate was significantly higher than MVO for LP (4.3 ± 2.2 dB/s, p<0.05), but not for RSP (2.3 ± 0.8 dB/s).

Discussion

In this study, we compared two energy-matched ultrasound pulse lengths (at 1 MHz and 1.5 MPa) each delivering the same total number of cycles: LP was 5000 cycles given every 3 s (total duration of the pulse = 5 ms), while the RSP was composed of a train of 500 × 10 cycle pulses, separated by 100 μs (total duration of the pulse = 50 ms) also delivered every 3 s.

Cavitation activity

As expected, our in vivo cavitation data supported that the microbubble cavitation activity was greater with LP, as reflected by the greater power in all the non-linear frequency bands and including the broadband signal: indeed, above the cavitation threshold and after destabilization of the microbubble shell, a long pulse can sustain cavitation activity of the free gas (Chen et al. 2016) and generate daughter bubbles (Tu et al. 2006), which themselves provide additional cavitation nuclei. In contrast, due to the 100 μs idle intervals of the RSP, less cavitation activity ensues.

UTMC induced vasodilation kinetics

In an intact hindlimb, this increased cavitation activity with the long pulse was associated with two major observations: (1) temporary microvascular hypoperfusion (presumed transient vasospasm) in the muscle that spontaneously resolved 4-min post therapy; (2) a sustained marked increase in perfusion observed starting at 10 min up to 60 min post therapy with LP, which was significantly than RSP at 30 and 60min (p<0.05, Figure 3B). Conversely, the RSP was not associated with microvascular spasm and also resulted in an increase in flow rate, which was higher than baseline at 4 min (p<0.05) but decreased to baseline values subsequently. Since the observed temporary microvascular spasm resulted in hindlimb muscle hypoperfusion, we investigated whether simple post occlusion hyperemia could explain the sustained hyperemic response with LP: our balloon occlusion experiment caused a typical ischemia reperfusion response, which peaked at 2-min post-release and quickly returned to baseline 6-min post release. Thus, based on the different kinetics in flow rate changes observed with LP, we conclude that the mechanism for hyperemia triggered by MB oscillation is distinct from the nitric oxide surge observed following transient occlusion. This is consistent with prior observations that the eNOS activation and purinergic signaling following MB cavitation could last for hours (Belcik et al. 2017; Mason et al. 2019; Yu et al. 2017).

Effect of pulse length on sonoreperfusion efficacy

Finally, we investigated the therapeutic efficacy of the RSP strategy in our established rat hindlimb model of MVO (Pacella et al. 2015; Yu et al. 2017). In fact, we found that the RSP configuration had therapeutic efficacy in our MVO model. Blood volume was restored to baseline values after the second therapy session, indicating that microbubbles regained access to areas that had been occluded by the microthrombi injection. However, the flow rates remained slower than baseline values and not different from MVO. Please note that with this dynamic MVO model, we have always observed some reperfusion with time in the no-US group, which highlights the importance of including a no-treatment group in this study (Figure 5). We then directly compared RSP and LP in our MVO model, alternating between the two pulses randomly on each pair of rats. In Figure 6, we can see that both pulses were equivalent for restoring blood volume to baseline values after two 10 min therapy sessions, and the flow rate increased above the MVO level after treatment 2 in Figure 6 for the LP group, consistently with Figure 5 and with our historical experiments with long pulse SRP [see reference (Pacella et al. 2015) Figure 5 and reference (Yu et al. 2017) Figure 8]. This data supports equivalent efficacy for RSP and LP in terms of blood volume, but a slightly better efficacy of LP at restoring the flow rate after MVO. Importantly, please note that the data in figures 6 and 7 were obtained 3 min after stoppage of SRP for Tx1 and Tx2, to allow the level of Definity™ MB to stabilize in the animal. Consistent with figure 3, there was no sustained spams (>3min) with either LP or RSP pulses. Taken altogether, our data suggests that UTMC with a longer pulse, compared to a rapid train of short pulses, with a constant total number of pulses, caused more microbubble activity, increased the vasodilation response and provided more rapid therapeutic efficacy for treating MVO. However, we did observe a temporary spasm when using the long pulse, that spontaneously resolved ~4 minutes post-therapy. This point will be discussed in detail.

Advocating for long pulse therapy

The therapeutic use of long ultrasound pulses (>2 ms in duration) with microbubbles has been reported in many areas of research, including sonothrombolysis (Bader et al. 2015; Leeman et al. 2012), blood-brain barrier opening (Lipsman et al. 2018; Sun et al. 2017; Wu et al. 2018), and drug delivery (Escoffre et al. 2011; Lentacker et al. 2010; Yu et al. 2016) to name a few. With longer pulses, the mechanical effects of microbubbles (cavitation) is sustained, radiation force can push the microbubbles towards the vessels (Dayton et al. 1997) and microbubbles can cluster and form microbubble clouds (Chen et al. 2016), which all contribute to increased mechanical and biological effects. However, increasing the pulse length has some drawbacks: there is an increased risk of standing waves, hemorrhage (Price et al. 1998) or even sterile inflammation (Kovacs et al. 2018; McMahon et al. 2017). In this study, standing waves were minimized by placing an acoustic absorber under the animal and we found no evidence of damage by histology after UTMC with the rapid short pulse (Figure 7), similarly to our previous reports with the long pulse (Pacella et al. 2015). Other potential bioeffects (such as sterile inflammation) remain largely unknown.

Interestingly however, our results support that in the context of MVO, increasing the pulse length increases the vasodilation response and the therapeutic efficacy of sonoreperfusion therapy. Thus, this study justifies continuing exploring long pulses in the context of sonoreperfusion therapy. In fact, another group investigated longer pulses for therapy in the context of peripheral artery disease, and found that increasing the pulse length from 5 to 40 cycles resulted in a higher purinergic signaling level and an increased perfusion in a low line density setting. (Mason et al. 2019)

Vasospasm

The group of Otto Kamp (Roos et al. 2016a) observed epicardial coronary artery vasospasm in a small number of patients following microbubble and ultrasound therapy in the setting of acute myocardial infarction during a clinical trial, prompting early termination. Of note, the total number of patients in the study was 6 and only 3 were observed to have spasm. In contrast, Mathias et al (Mathias et al, 2016) reported on 20 patients treated with high mechanical index US pulses ranging from <5 μs (15 patients) to 20 μs (5 patients) and reported no episodes of coronary spasm. The duration of the observed spasm in (Roos et al. 2016a) remains unknown but there have been no reported poor clinical outcomes related to their clinical observation, indicating that this transient spasm likely spontaneously resolved. Additionally, as stated by the authors, “we cannot rule out that the vasoconstriction in our human patients was catheter induced or caused by STEMI itself”, indicating that the origin of the spasm is not fully elucidated. Nevertheless, it was shown by the same group in a pig model that the long pulse (20 μs) could induce an average 6% vasoconstriction (range 2–13%) downstream, only with concomitant thrombus and severe angioplasty induced vascular injury. In all other tested conditions, UTMC caused vasodilation, consistently with a potentially increased efficacy of longer pulses. Interestingly, in our model, we also observed a transient microvascular spasm, which was followed by a sustained marked increase in perfusion in the long pulse group, which in our model outweighs the effect of transient microvascular hypoperfusion. Our group has also found that co-administration of several pharmacologic agents during long pulse US eliminates this transient event as well (Yu et al. 2020). It remains to be seen how our observations with long pulses, which were obtained in the microcirculation of a healthy muscle, can improve MVO in STEMI patients, which in our opinion warrants further studies with the long pulse, ideally in a more realistic model of STEMI+MVO.

Safety

Our histology and safety data substantiate that in our model, the technique is not deleterious to tissue with either pulsing regime. This is consistent with finding of others in clinical trials, in which the treatment using high MI was well tolerated (Mathias et al. 2019; Mathias et al. 2016). However, given the high doses of ultrasound in this study, especially with longer pulses, other biologic changes, that were not detected, cannot be ruled out. While it is reassuring that major safety concerns have not arisen in the studies thus far, further study is warranted, as with all new technologies, to fully document the safety profile of this technique.

Limitations

There are several limitations to this study. First, the mechanism explaining a sustained vasodilation for >60 min remains to be elucidated. Purinergic signaling, which has been shown to be sustained for up to 24 h and which is implicated in the NO pathway may be involved (Belcik et al. 2017). Whether sterile inflammation, which has been observed in the brain following BBB opening is implicated remains unknown (Kovacs et al. 2017). Moreover, whether this is related to sonoporation and calcium influx is also unknown. In fact, our group has demonstrated pore formation and resealing (Helfield et al. 2016) during a time scale which aligns with observed sustained increase in perfusion in this study. Secondly, it is unknown whether the type of MBs affects SRP efficacy. We have opted to continue using our custom formulation as our therapeutic MBs for historical continuity to our previous reports (see introduction). These in-house lipid MBs are larger than Definity™. Other groups have reported changes in perfusion using Definity™ as the therapeutic MB in skeletal muscle in mice (Belcik et al. 2017; Michon et al. 2022), in mouse tumors (Michon et al. 2022), and in patients (Belcik et al. 2017; Mathias et al. 2019; Mathias et al. 2016). In all cases, vasodilation was observed, which is consistent with our observations. Nevertheless, in the presence of microthrombi in our MVO model, it remains unknown whether MB size affects SRP. For passive cavitation detection, the sampling frequency used was too low due to hardware limitations. However, there was very little energy in the received radio frequency signal beyond the Nyquist limit due the limited bandwidth of the receiving transducer. Therefore the conclusion is unlikely to be affected. Finally, it should be noted that this study does not fully recapitulate clinical MVO in a diseased myocardium but reflects the acute response of a muscular vasculature to UTMC, with and without microthrombi occlusion. Hence the findings reported here should not be directly extrapolated to clinical MVO, which is beyond the scope of this study.

Conclusion

In this study, we compare long and rapid short therapeutic ultrasound pulses with the same total number of pulses in a rat hindlimb model of microvascular obstruction. Our results demonstrate that the use of longer pulses causes an increase in MB cavitation activity, which results in a sustained increase in vasodilation and an increase in therapeutic efficacy. This recent body of work has opened a new paradigm on the use of therapeutic applications of microbubbles and ultrasound, shifting the focus in vascular applications from breaking up large thrombi to restoring microvascular perfusion, with promising long term prognostics (Mathias et al. 2019; Mathias et al. 2016). It is noteworthy to remark that both effects are not mutually exclusive, as the well know mechanical effects of microbubble oscillation and cavitation could also assist microvascular perfusion by breaking up microthrombi. While initial clinical trials in the context of STEMI are very encouraging using short high MI pulses, preclinical studies seem to indicate that longer pulses could increase therapeutic efficacy and support further optimization of the ultrasound pulsing schemes. However, these improvements and the safety of longer pulses still need to be validated in more realistic models of STEMI+MVO and/or in patients.