Sonothrombolysis with magnetically targeted microbubbles

Microbubble-enhanced sonothrombolysis is a promising approach to increase the safety 23 and efficacy of current pharmacological treatments for ischemic stroke. Maintaining 24 therapeutic concentrations of microbubbles and drugs at the clot site however poses a 25 challenge. The objective of this study was to investigate the efficacy of magnetic 26 microbubble targeting upon clot lysis rates in vitro . Retracted whole porcine blood clots 27 were placed in a flow phantom of a partially occluded middle cerebral artery. The clots were 28 treated with a combination of tissue plasminogen activator (0.75µg/mL), magnetic 29 microbubbles (~10 7 microbubbles/mL), and ultrasound (0.5MHz, 630kPa peak rarefactional 30 pressure, 0.2Hz pulse repetition frequency, 2% duty cycle). Magnetic targeting was achieved 31 using a single permanent magnet element (0.08-0.38T and 12-140T/m in the region of the 32 clot). The change in clot diameter was measured optically over the course of the 33 experiment. Magnetic targeting produced a three-fold average increase in lysis rates and 34 linear correlation was observed between lysis rate and total energy of acoustic emissions. 35


Introduction 40
Despite recent therapeutic advances, ischemic stroke remains a leading cause of disability 41 and mortality worldwide (Naghavi, et al. 2015, Feigin, et al. 2017). At present, intravenous 42 tissue-plasminogen activator (tPA) is the only thrombolytic drug recommended by the 43 British National Institute of Health and Care Excellence (NICE) (NICE 2008) and the U.S. Food 44 and Drug Administration (FDA) (FDA 1996). It carries however the risk of potentially fatal 45 side effects including intracranial haemorrhage. As a result, intravenous tPA administration 46 is subject to rigid exclusion criteria (American College of Emergency Physicians and 47 American Academy of Neurology 2013). Among the most limiting of these is the therapeutic 48 time frame for intravenous administration of the drug (Barber, et al. 2001). In a study 49 conducted with the California Acute Stroke Pilot Registry, only 4.3% of all ischaemic stroke 50 patients arrived in the Emergency Department within the recommended 3-hour window 51 after stroke onset and were thus eligible for tPA. Of equal concern is the fact that even if all 52 patients had arrived within the required time frame, only 28.6% of them would have been 53 eligible for thrombolytic treatment (California Acute Stroke Pilot Registry Investigators 54

2005). 55
Against this background, there is a clear need for an adjuvant therapy that improves the 56 safety and efficacy of thrombolytic treatment. Sonothrombolysis, i.e. the use of ultrasound 57 (US) for enhanced thrombolysis is a minimally or non-invasive physical approach that can be 58 used to enhance the action of tPA, thereby reducing the required dose, or facilitating clot 59 breakdown in the absence of a drug. Multiple in vitro and in vivo studies have shown that 60 intravenously administered microbubbles can significantly improve the rate of clot 61 breakdown, with or without concomitant use of a thrombolytic drug (Lu, et al. 2016, 62 inner diameter and 0.18 mm wall thickness, Heatshrink-Online Ltd, Huntingdon, UK). The 108 internal diameter of the tube was selected to correspond to the typical dimensions of the 109 middle cerebral artery (Serrador, et al. 2000), one of the cerebral vessels most commonly 110 affected by ischaemic stroke (Demchuk, et al. 2001). A solution of dilute plasma (1.25% v/v 111 fresh frozen plasma in PBS) was circulated once through the chamber at a rate of 3 mL/min 112 by a peristaltic pump (Minipuls 2, Gilson Inc., Middleton, WI, USA). The flow rate was 113 selected to obtain a mean blood velocity past the clot at the lower end of the range 114 observed in a partially occluded middle cerebral artery (Ogata, et al. 2004). Lipid-shelled 115 magnetic microbubbles were infused continuously, along with tPA, into the diluted plasma 116 at a rate of 140 µL/min using a syringe pump (World Precision Instruments Ltd, Hitchin, UK). 117 The final concentrations of tPA and microbubbles in the diluted plasma were 0.75 µg/mL 118 and 8.1 x10 6 ± 1.2 x 10 5 microbubbles/mL respectively. Four experimental groups were 119 defined: exposure to tPA only, US + tPA, US + tPA + microbubbles (no external magnet), and 120 water to the concentration of 1 mg/mL and stored in 400 µL aliquots at -80°C. Under these 127 storage conditions, the enzyme is stable for up to a year (Alkatheri 2013, Jaffe, et al. 1989. 128 The phospholipid 1,2 distearoyl-sn-glycero-3-phosphocholine (DSPC) was purchased from 129 Avanti Polar Lipids Inc., Alabaster, USA. The ferrofluid (10 nm spherical magnetite 130 nanoparticles in isoparaffin, 10% volume fraction) was obtained from Liquids Research Ltd, 131 Bangor, UK. 132 The protocol for magnetic microbubble preparation was based on the method first 133 described by (Stride, et al. 2009). The microbubbles were prepared by sonication using a shaken vigorously for 45 s. Microbubble concentration and size were assessed as previously 142 described by (Sennoga, et al. 2010). A 10 µl sample was placed on a microscope slide and 25 143 images were taken with a x40 objective. The images were then analysed using purpose 144 written software in MATLAB (The MathWorks Inc., Natick, MA, USA) to count particles of 145 diameters ranging from 0.2 µm (~ 2 pixels) to 15 µm. 146

Magnetic targeting 147
The magnetic field was applied as illustrated in Figure 1, using a single 12.7 mm cubic 148 permanent magnet element (N52 grade NdFeB, NeoTexx, Berlin, Germany). It was placed 149 below the clot at an angle of 45 o , its upper vertex in contact with the flow channel directly 150 below the ultrasound focus. During experiments that did not involve magnetic targeting, the 151 magnet was replaced by a stainless steel block of identical dimensions in order to maintain a 152 comparable sound field. 153 The magnetic field and the magnetic field gradient in the flow chamber were first computed 154 using a numerical model developed by (Barnsley, et al. 2015). The magnetisation value of 155 the magnet, an essential input parameter to the model, was then determined 156 experimentally by measuring the field profile of a single NdFeB element using a three-axis 157 Hall probe connected to a Model 460 3-channel gaussmeter (Lake Shore Cryotronics, 158 Inc.,OH, USA). The field profile was fitted to an analytical expression for the field generated where I(t) is the total greyscale intensity in the ROI at time t, t 0 is baseline time (before bolus 171 injection), and t 1 corresponds to bolus arrival in the ROI. For the ultrasound experiments, two ultrasound transducers were mounted coaxially, as 175 described by (Hockham, et al. 2010). The ultrasound source was a focused, circular, single 176 element, 0.5 MHz transducer (Sonic Concepts, Bothell, WA, USA). The transducer centre 177 frequency was chosen to balance considerations of treatment safety and efficacy. Lower 178 frequency ultrasound is more effective at breaking down blood clots (Nedelmann,et al. 179 2005) (Schafer, et al. 2005) and is less attenuated by tissue and bone, but safety concerns 180 have been raised in several studies conducted with low-to mid-frequency transcranial 181 ultrasound (60 to 300 kHz) (Daffertshofer, et al. 2005) (Nedelmann, et al. 2008). This 182 transducer had a diameter of 64 mm and a focal length of 62.64 mm. The measured -3 dB 183 focal volume dimensions were 4 mm laterally and 37 mm axially. It was driven by a function 184 generator (3325DA, Agilent Technologies Inc., Santa Clara, CA, USA) and connected to a 185 power amplifier (A300, Electronics & Innovation Ltd., Rochester, NY, USA) with a custom-186 made impedance matching network (Sonic Concepts). The clot was exposed for 60 min to 5 187 x 10 4 cycle bursts at a 0.2 Hz pulse repetition frequency, corresponding to a 2% duty cycle. 188 The peak rarefactional pressure at the focus was set to 630 kPa, measured in the absence of 189 the clot and vessel with a calibrated fibre optic hydrophone (Precision Acoustics, 190 Dorchester, UK).

Histological analysis 235
After treatment completion, the clots were immersed in formalin, and fixed in paraffin. 5-236 µm slices were prepared and stained with eosin and haematoxylin. They were analysed in 237 bright field microscopy (Ti Eclipse, Nikon Instruments Inc., NY, USA) using a x20 objective (S 238 Plan Fluor EWLD, Nikon). 239 240

Clot debris analysis 241
Secondary embolism due to circulation of clot debris is a potential risk of sonothrombolysis. 242 In order to assess clot debris size, 15 mL of effluent were collected halfway through 243 treatment and immediately stored at -20°C. The samples were then defrosted for 12 hours 244 at 4°C. This low temperature was maintained so as to inhibit the enzymatic activity of tPA, It must however be noted that the numerical model breaks down in a region within about 1 265 mm away from the magnet. This is due to the approximation of the magnetic medium as a 266 lattice of evenly distributed point dipoles, which is no longer valid in the short range. Precise 267 estimation of the magnetic field in this narrow region is, however, not relevant to the 268 present study, as the region is occupied by the clot and no microbubbles can be present. 2.9 ± 2.8 x 10 -3 mm/min respectively). Lysis rates were substantially increased in the 300 presence of microbubbles (11.2 ± 4.1 x 10 -3 mm/min with US + tPA + microbubbles, non-301 significant); they were further accelerated (x 3.3 on average) with a magnetic field (36.6 ± 302 23.4 x 10 -3 mm/min with US + tPA + microbubbles + magnet, p<0.05). 303 The effect of treatment on clot structure was investigated on H&E stained clot samples 304 ( Figure 6). The surface of clots treated without microbubbles (Figure 6a. and b.) was 305 smooth. On the clot treated with tPA with US, it was not possible to identify with precision 306 the focal region. In the samples treated with US + tPA + microbubbles, the damaged region 307 has a smooth surface and several small areas of erythrocyte depletion (Figure 6 c., direction 308 of US exposure shown by the black arrow). Closer inspection shows that these cavities still 309 contain fibrin. These observations are in agreement with a recent report by Petit et al. 310 (2012). All of the samples treated with US + tPA + microbubbles + magnet were entirely 311 lysed at the transducer focus after 60 minutes of treatment. Figure 6d   were insufficient data to fit more complex curves. 347 348

Clot debris 349
The effluent was collected during treatment and analysed. In all experimental groups, over 350 99.9% of the particles were smaller than 10 m, i.e. of similar size to red blood cells, 351 indicating that the risk of downstream embolisation with this treatment is low. 352 353

Discussion 354
In this study, magnetic targeting was associated with a significant increase in cavitation 355 energy near the clot, confirming previous reports that magnetically targeted microbubbles 356 can locally enhance cavitation (Crake, et al. 2015). As this enhanced cavitation was 357 associated with accelerated clot lysis, this study demonstrates that magnetic targeting has 358 the potential for a doubly positive impact on thrombolysis. Firstly, it may accelerate vessel 359 recanalisation, increasing tissue salvage, and limiting the adverse side-effects associated 360 with prolonged ultrasound exposure. These findings are particularly promising in the 361 context of ischaemic stroke therapy, where early tissue reperfusion is crucial to improve 362 patient outcomes (Rha and Saver 2007). Secondly, stronger cavitation signals may facilitate 363 real-time treatment monitoring and thus enhance patient safety. 364 To the best of the authors' knowledge, this is the first report of magnetically targeted 365 microbubbles as a method to accelerate sonothrombolysis. Targeted sonothrombolysis has 366 been previously investigated in vitro and in vivo using various "biological" microbubble 367 targeting methods (e.g. antibody conjugation) (Alonso, et al. 2009, Chen, et al. 2009, Culp, 368 et al. 2004, Wu, et al. 1998, Xie, et al. 2009). Higher plasma D-dimer levels have been 369 reported in rats treated with pulsed 2-MHz ultrasound and platelet-specific, abciximab-370 coated immunobubbles, compared to the non-specific immunobubble control group. 371 Histological analysis also showed clot disintegration in 4/5 clots in the target group, 372 compared to 1/5 in the controls (Alonso, et al. 2009). (Xie, et al. 2009 loaded echogenic liposomes. When exposed to 120 kHz ultrasound, these particles achieved 380 more effective thrombolysis in vitro than free tPA + US. In comparison to these biologically 381 targeted agents, the main advantage of magnetically targeted microbubbles is the relatively 382 long potential working distance: magnetically targeted agents do not need to be initially in 383 contact with the target to be retained. This property is particularly relevant in the context of 384 thrombosed vessels, which are characterised by reduced blood flow and complex 385 haemodynamic patterns (Strony, et al. 1993, Wootton andKu 1999). 386

387
Other researchers have provided an in vitro proof-of-concept for magnetically targeted 388 enzymatic sonothrombolysis, developing magnetic polymer microparticles that release tPA 389 upon exposure to very low frequency ultrasound , Torno, et al. 2008. 390 The present study uses magnetic microbubbles, which have the added advantage of strongly 391 promoting cavitation. The microbubbles were effectively retained against physiological flow 392 rates using a magnetic field within the 8 -3.8 x10 -2 T range, and a 12-140 T/m magnetic field 393 gradient. Such flux density and flux density gradients are achievable with permanent 394 magnet arrays at depths up to a few tens of mm in the body depending on the flow rate 395 (Barnsley, et al. 2015, Barnsley, et al. 2016. This is sufficient for treating 396 clots in the middle cerebral artery (Gillard, et al. 1986), one of the brain vessels most 397 commonly affected by ischaemic stroke (Demchuk, et al. 2001). 398

399
The results of this study suggest that sonothrombolysis can enable a reduction in the 400 quantity of drug required. The drug alone did not produce significant lysis, which is 401 attributable to the relatively short time frame of the experiment (60 min), to the low The results also demonstrate the feasibility of non-invasive treatment monitoring using 414 passive cavitation detection, with good correlation between lysis rates and acoustic 415 emissions. This contrasts with a previous study, in which whole porcine blood clots were 416 exposed to tPA, pulsed ultrasound and lipid-shelled microbubbles (Datta, et al. 2008). The The finding that sonothrombolysis can be significantly enhanced in an inertial cavitation 428 regime warrants further discussion both in terms of potential mechanism and in terms of 429 safety. Mechanistically, microstreaming has been routinely associated with non-inertial 430 cavitation, primarily because of the need for persistent cavitation to occur in order for a 431 stable microstreaming field to be established. However, several recent studies in the 432 context of cancer drug delivery have indicated that suitably nucleated, sustained inertial 433 cavitation in fact results in higher streaming velocities than non-inertial cavitation in vitro 434 (Rifai, et al. 2010) and enables significantly enhanced drug delivery, penetration and 435 distribution in vivo (Arvanitis, et al. 2011). These recent findings suggest that strategies 436 which enable sustained inertial cavitation to occur, such as the magnetic retention approach 437 presented here, could offer significant enhancements in terms of drug transport and 438

penetration. 439
In terms of safety, inertial cavitation has traditionally been deemed undesirable for 440 diagnostic and therapeutic applications, in great part due to reports of a higher incidence of 441 unwanted bioeffects (Schafer, et al. 2005). Over the past decade, studies specifically aimed 442 at opening the blood brain barrier have further indicated that inertial cavitation can cause 443 undesirable erythrocyte extravasation and cell death in murine brain models (McDannold, 444 et al. 2006). However, a recent clinical study of opening the blood-brain-barrier in humans 445 (Carpentier, et al. 2016)  The study has a number of limitations that warrant comment. The model mimics a partial 455 vessel occlusion, which does not perfectly represent most in vivo pathological thrombi, 456 although partial occlusions may also lead to acute clinical symptoms (Chesebro, et al. 1987, 457 Xie, et al. 2009). A single vessel orientation (horizontal) was investigated in the present 458 study; future work may focus on the effect of different orientation angles, in order to mimic 459 a broader range of clinical and physiological conditions. Also, blood from a single animal 460 was used in the results presented here. The range of hematocrits in the porcine blood donors 461 used for the current work was consistently found to be in the range 25 -30 % and clots were 462 assigned to individual groups blindly in order to mitigate the possibility of bias. Nevertheless, 463 in future studies, samples from multiple animals may be pooled in order to limit the effects of 464 pig-to-pig variability. To further increase clinical relevance, the composition of the plasma 465 solution in the flow loop may also be improved. A recent study by Huang et al. (2017) 466 demonstrated that using human plasminogen in the liquid surrounding a porcine clot produces 467 sonothrombolysis behaviours that resemble more closely human clots in human plasma. 468 In addition, many published studies use travelling waves when studying sonothrombolysis 469 (Datta, et al. 2006, Petit, et al. 2012, Pfaffenberger, et al. 2005, Xie, et al. 2009), while the 470 ultrasound field in this study included a partial standing wave. This is inherent to the 471 geometry of the setup. It is not expected to affect microbubble oscillations, as the size of a 472 microbubble (~1-10 µm) is much smaller than half the wavelength at 0.5 MHz (~1.5mm), but 473 it may have affected microbubble translation. In clinical applications, standing waves are 474 generally not desirable for patient safety (Baron, et al. 2009, Daffertshofer, et al. 2005, 475 although standing waves may also form in vivo in transcranial applications (Baron, et al. 476 2009, Bouchoux, et al. 2014). Furthermore, the microbubble formulation used was not 477 optimal for human use due to its composition. Magnetic microbubble formulations with 478 improved biocompatibility have recently been developed and characterized, and will be 479 used in future work (Owen, et al. 2012). Finally, the physical properties of the flow medium 480 did not faithfully replicate blood's properties, which may impact on microbubble flow 481 behaviour and acoustic response. Future studies will be conducted using whole blood, which 482 may require changes to the optical imaging technique employed to quantify clot lysis 483 (because of the reduced image quality). 484 485

Conclusions 486
This study demonstrates in vitro that magnetic targeting can significantly accelerate 487 microbubble-enhanced enzymatic sonothrombolysis at 0.5 MHz, with the potential to 488 reduce administered drug doses and to achieve faster vessel recanalisation. A more than 3 489 fold increase in the lysis rate was observed when using magnetic targeting, compared with 490 microbubbles in the absence of magnetic force. The total energy of acoustic emissions was 491 strongly correlated with lytic rates, providing a possible method for real-time treatment 492 monitoring. These results indicate that magnetic targeting has the potential to enhance 493 treatment efficacy and patient safety in the management of occlusive conditions using 494 microbubble mediated sonothrombolysis.