FUSF Sponsors Landmark MRgFUS Brain Workshop


The Focused Ultrasound Surgery Foundation and Department of Neurosurgery at the University of Virginia hosted a Brain Workshop for Magnetic Resonance guided Focused Ultrasound Surgery (MRgFUS) March 23-25, 2009 at a retreat in Chantilly, VA. The primary objective of the workshop was to develop a comprehensive R&D roadmap that would fast-track the use of this technology to treat a variety of brain disorders including benign and malignant tumors, Parkinson’s disease and essential tremor, stroke, and epilepsy.

Participation was by invitation only, and representatives from academia, industry, NIH, and the Focused Ultrasound Surgery Foundation were in attendance. Participants included 45 leading neurosurgeons, neurologists, neuroradiologists, neuroscientists, biomedical engineers, physicists, product-development managers and medical device company executives from around the world.

Pre-work in the form of online surveys, review articles, and brain disease summaries was used by the participants to help identify and prioritize clinical and technical issues for discussion.

The group consensus was to prioritize:

  • movement disorders (Parkinson’s Disease, Essential Tremors, and Dystonia)
  • meningiomas
  • metastatic brain tumors
  • stroke
  • epilepsy

The results of the workshop will be communicated in a White Paper and include the outline of a comprehensive R&D roadmap, the current best treatment for these conditions, an actionable short-term plan for technology development, preclinical studies, and pilot clinical trials including the identification and prioritization of clinical indications to be addressed, technologies required to treat these indications, and sites where the research will be performed.

The workshop also enabled the creation of a collaborative research environment and structure that will enable rapid execution of this plan and accelerate the development and adoption of focused ultrasound surgery for treating a variety of disorders of the brain.

Primary Objectives of the Brain Workshop
  1. Develop an outline of a preliminary comprehensive Research & Development (R&D) plan leading to reimbursable treatments and a quantum improvement in the management of a wide spectrum of intracranial pathologies (including benign and malignant tumors, Parkinson’s disease and essential tremor, stroke, and epilepsy).
  2. Develop an actionable plan leading to pilot clinical studies that could be completed in 1-2 years for several indications that will provide proof of concept and validate the field.
  3. Create enthusiasm and establish momentum that will serve as a foundation for rapid growth.
  4. Establish guidelines for allocating foundation resources. We want to invest around $10M into brain research over the next three years.
The Program

Participation consisted of 64 thought leaders from industry, academia, the NIH and the Focused Ultrasound Surgery Foundation. Participants included 18 neurosurgeons, 5 neurologists, 5 neuroradiologists, 14 biomedical engineers and Radiology PhDs, and 22 other scientists, product-development managers and medical device

The first session was comprised of state-of-the-art presentations from experts in the field:

  1. Thalamotomy for neuropathic painDaniel Jeanmonod, Ernst Martin, (Zürich)
  2. Brain tumorsFerenc Jolesz (Brigham & Women’s)
  3. Intracerebral / Intraventricular hematomaSagi Harnof (Sheba)
  4. Acute Ischemic StrokeThilo Hoelscher (UCSD)
  5. Blood brain barrier openingNathan McDannold (Brigham & Women’s)
  6. Neuromodulation Alexander Bystritsky (UCLA)
  7. Jamie Tyler (Arizona State)
  8. Seung-Schik Yoo (Brigham & Women’s)

The second set of presentations comprised of technology reviews and updates from experts and industry:

  1. Fundamentals of Brain MRgFUS – Ferenc Jolesz
  2. Supersonic Imagine – Claude Cohen-Bacrie
  3. Philips Healthcare – Ari Partanen
  4. InSightec – Eyal Zadicario

There were also several reports from discussion groups on specific subjects. These were followed by facilitated discussions using a framework based on the online survey results. An outline research roadmap for the priority indications was defined by the group.

Specific breakout sessions planned on the first day (for simulations and neuromodulation) were followed the next two days with additional specific workgroups that evolved during the active meeting discussions (in movement disorders, neuromodulation follow up, animal models and a technical workgroup assessing “ringing” and “standing waves” within the skull). A forum for informal discussions was provided each evening in the form of receptions and dinners.

A facilitator with prior experience running such workshops was hired to expedite the process and keep sessions on time. Specialist expert panels, including the primary discussants, were assembled for each section to provide summary discussions and conclusions.

Being a closed meeting, the workshop facilitated collaboration and discussion. It was an iterative process, wherein discussion proceeded towards a general consensus among the experts. The mix of clinicians and engineers provided a basis for improving the technology to support the projected clinical indications.

Extensive pre-work in the form of 3 online surveys, review articles, and brain disease summaries were distributed to the participants to help identify and prioritize clinical and technical issues for discussion. After reviewing the key papers participants completed an online survey, which had two sections. Firstly, to apply a weighting score to a list of criteria used to decide which indications should be given priority. These criteria included an assessment of unmet need, and expected overall impact. The group consensus was to prioritize movement disorders, meningiomas, metastatic brain tumors, stroke and epilepsy. Secondly, after ranking the priority indications, participants applied weighting scores to potential technical research areas for each indication.

Metrics of Success for the Brain Workshop

Dr. Neal Kassell outlined the Metrics of Success during his opening remarks:

  1. Will we have answered the burning questions? Undoubtedly we will develop new questions.
  2. Develop an outline of a comprehensive R&D roadmap.
  3. Develop an actionable plan leading to pilot clinical studies for several indications that can be completed in 12-24 months.
  4. Established a structure and environment for collaboration that will allow rapid execution of this plan.

He also outlined the diverse agendas envisioned for the groups represented:

  1. Industry – creates value through productivity and quality. Better treatments for large numbers of patients. No more noble way to make money than to save lives and relieve suffering.
  2. Academic community – searching for glory derived from discoveries.
  3. Foundation – Mission is to develop new applications for and accelerate the global adoption of FUS.

Collaboration and cooperation in the digital era are critical to this effort. We want to amalgamate investigators and facilities in a boundary-less model. There is no room to duplicate efforts when there is so much work to be done.

The workshop also enabled the creation of a collaborative research environment and structure that will enable rapid execution of this plan through the synthesis of investigative efforts in academia and industry and linkage to funding sources. This will accelerate the development and adoption of focused ultrasound surgery for treating a variety of disorders of the brain.

In addition to fulfilling the specific ives related to focused ultrasound surgery and the brain, the workshop established a contemporary model for accelerating the development of large-scale therapeutic modalities.

Priority Clinical Indications

During the pre-work phase participants were asked to prioritize the potential clinical indications from a list of 17 possible brain disorders where focused ultrasound has potential. After being provided with background information on each indication, the participants were asked to score the relative importance of each indication using the online survey method. This process took into account a range of factors so that indications were not selected solely on the basis of prevalence.

  • Metastatic Tumors
  • Cavernomas (cerebral cavernous malformations)
  • Gliomas
  • Pituitary macroadenomas
  • Pituitary micro-adenomas
  • Acoustic Neuromas (Vestibular Schwannomas)
  • Meningiomas
  • Chordomas
  • Intracerebral and Intraventricular hemorrhage
  • Ischemic Stroke
  • Arteriovenous Malformations (AVMs)
  • Epilepsy – Mesial Temporal Lobe Sclerosis
  • Movement Disorders
  • Trigeminal Neuralgia (Tic Douloureux)
  • Neuropathic Pain
  • Alzheimer’s Disease
  • Psychiatric Disorders

Movement Disorders – Workshop Discussion

Issues related to Pilot Studies
In considering Parkinson’s Disease (PD), Essential Tremor (ET) and dystonia, essential tremor is best suited to initial evaluation using MR guided focused ultrasound surgery (MRgFUS). Deep brain stimulation (DBS) is the best current therapy given its efficacy similar to that of lesioning, but without the same risk of permanent complication. However, stimulators do have associated mechanical malfunction or infection risks that are more significant. The progressive nature of Parkinson’s disease also makes essential tremor a more favorable indication for the initial assessment of this technology. Dystonia patients are not as favorable because they often take a longer time to show symptomatic improvement.

Deep brain stimulation works best in Parkinson’s patients who are dopamine responsive. Ideally, a procedure would be very effective even when the patients are resistant to medical therapy. But the reversibility of DBS is very appealing if the risk of neurologic complication is high from a misplaced lesion.

The main limitations of DBS for movement disorders are cost in both money and time, and the necessity of a neurosurgical operation that entails some risk of hemorrhage, infection, mechanical failure, neurologic damage, etc. One tremendous advantage of FUS is the non-invasive nature of the technique. Any form of lesioning, including FUS, gains significant advantage through targeting techniques that assure a safe, effective lesion location.

One might consider that Parkinson’s disease patients exist along a spectrum from tremor-dominant to akinesia-dominant. Akinesia may be linked to the limbic system, which is not as well understood as the motor system associated with tremor. Parkinson’s disease involves the supplementary motor area and the entire paralimbic frontal system. Stress can play a major role in the symptoms of the disease.

Patient age can play a role in treatment decision-making. Reversibility may be more important in younger patients, who will be dealing with the disease for a longer time. In more elderly or infirm patients, DBS is much less attractive. The elderly are not as likely to live long enough to benefit from gene insertion or cellular implantation, techniques that might be options in a decade or so. The simpler lesioning procedure, without all of the time and effort spent in programming a DBS system, might be preferable for them.

The nucleus intermedius (Vim) has been a preferred target for thalamotomy. The best general coverage of symptoms is through the regulatory pallido-thalamic pathways. The subthalamic nucleus of Luys (STN) does not cover the relevant pathways as completely. Stimulation of STN is limited by dyskinesias. Lesioning of pallidal output can be utilized bilaterally, if necessary. The relevant discussion is where to focus treatment, not whether to use lesioning or stimulation.

Placebo effects can play a significant role in pain patients and in those with Parkinson’s disease.

Relevant Models
The MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) neurotoxin model performed in vivo in pigs is a good animal model for Parkinson’s Disease, being less politically charged than the non-human primate model. This might be useful for studying more advanced techniques, such as blood brain barrier (BBB) disruption for drug or gene fragment delivery, or arrest of apoptosis.

Some dogs have spontaneous essential tremor. Various animal models have been tried at the different sites using the Exablate 4000 in the 660 kHz configuration, including dogs, monkeys and pigs. MR guidance in animal models has been at the Brigham & Women’s Hospital and in Paris.

Neuromodulation to Optimize Targeting
The use of focused ultrasound surgery for neuromodulation, specifically for functional mapping and ideal target localization during the FUS procedure before making the permanent lesion, is an exciting prospect. There is much research to be done before this might become routine. If it yields much higher assurance of a lesion being in exactly the right site for maximum effect and minimal complications, it may well enable the optimal treatment to be FUS lesioning, even in the subthalamic nucleus (STN, a current favorite for DBS placement) guided by FUS neuromodulation.

In addition, FUS may offer the prospect of thermal release of heat shock proteins, manipulation of apoptosis, or transient blood-brain barrier (BBB) opening for drug or gene fragment delivery as a more refined treatment than thermal lesioning alone.

Astrocytes are ten times more sensitive to non-thermal FUS effects than neurons, and mediate spreading depression. FUS can be used to stimulate an action potential (AP). One might even stimulate an AP with FUS, observe some suppression, and then use for lesioning. There is no refractory period.

Acoustic beam imaging might be used to assess treatment effect, by measuring tissue motion with ultrasound but without using thermal effects.

Feasibility study in Essential Tremor
The proposed study would involve unilateral MRgFUS treatment. This will be a non-inferiority study against DBS in 20 patients over 12 months. Sites would include the University of Virginia and the Kindershospital in Zurich, Switzerland, as well as possible contributions from the University of Toronto, the Brigham & Women’s Hospital in Boston, the University of California in San Diego, and Sheba Medical Center in Israel.

The feasibility study will measure tremor reduction and complications as end points in patients who are surgical candidates with uni- or bi-lateral tremor. The worst side will be the one treated in bilateral cases. Evaluation of various pulse stimulations as an intraoperative physiologic localization tool would not be incorporated in this trial, but await evaluation later. Patients should stop anticoagulants for two days before MRgFUS, until there is no evidence of increased risk of hemorrhage.

It is important to determine the level of energy and temperature dosing that will allow for reversibility. This necessitates a slow, progressive temperature increase. If the patient develops hemiparesis, sonication is discontinued until neurologic function normalizes. If the temperature reaches 45 deg C without any neurological effect, the surgeon decides whether to adjust the target location or discontinue the procedure. Using the 650 kHz ExAblate 4000 system (Insightec, Inc, Haifa, Israel), the surgeon will employ a progressive rise in temperature technique to assess neurological effect in the target region. Motoring in the arms and legs is easily accomplished without moving the patient out of the bore of the magnet, which is critical in maintaining sub mm accuracy. The patient may be moved out of the bore to assess facial motor function if necessary, but this will entail significant time in re-imaging to verify accurate repositioning of the target.

After stimulation has been used to determine the correct site, the surgeon will make a 3-4 mm ablative lesion (~75 uL) in the Vim (nucleus intermedius of the thalamus) using a 10 to 20 sec sonication, with a target maximum temperature of 55 to 59 deg C. The surgeon has discretion of the exact target location within Vim, and is able to make decisions about relocalization of the target within that region based on stimulation results. Patients who undergo stimulation and assessment will be included in the study, even if no lesion is actually made.

Outcome measures will include video, accelerometer, and MRI. These will be obtained at 24-48 hours, 1-2 weeks (clinical assessment without MRI), 3 months and 1 year.

The follow up pivotal trial will compare DBS vs. MRgFUS in the same patient, one technique on each side. Which technique is used on the more affected side will be randomized. Endpoint will be a validated rating system for tremor.

Outstanding questions for trial design:

  • What is the threshold for reversible change?
  • Can one stimulate with FUS pulse sequences?
  • What is the durability of the clinical result after a FUS lesion for tremor?

Summary: The Way Forward
The next step is to create the clinical protocol for the study of Essential Tremor, primarily by Dr. Jeff Elias at the University of Virginia and Dr. Daniel Jeanmonod at the University of Zurich. In addition, the Neuromodulation Research Program Planning Group (NM-RPPG), created during the Brain Workshop, will create a research roadmap to plan the specific collaborative research that will enable FDA approval of neurophysiologic techniques to verify target localization during functional neurosurgical procedures.

Meningiomas – Brain Workshop Discussion

Best Current Treatment
Meningiomas are generally well-managed with surgical resection or stereotactic radiosurgery. Skull base meningiomas, especially those with intimate associations with cranial nerves, continue to vex neurosurgeons dealing with them. Microneurosurgery and stereotactic radiosurgery, often combined, have allowed for improved management of challenging skull base meningiomas adjacent to cranial nerves in the last few decades, but there remains a cadre of very difficult cases.

Stereotactic radiotherapy (SRT) offers a gold standard in the treatment of meningiomas that are intimately involved with cranial nerves. SRT differs from SRS in that it utilizes the biological advantage of fractionation, or giving radiation in discrete doses separated by intervals that allow differential recovery of normal tissue more so than target tissue. One caveat of SRT is that, in general, it cannot be readily repeated because any normal tissue one is trying to protect within the target volume neighboring the lesion has been taken to a near-maximum dose during the course of the initial treatment. MRgFUS may have some advantage in that it can be repeated as often as is necessary or practical. There is no cumulative build up of thermal dose, as there is with ionizing radiation. Based on these properties, this modality may have significant potential benefit to this patient population, particularly to those who recur after standard treatment. However, thermal lesioning will not allow any preferential protection of normal tissue within the target zone. More elegant application of non-thermal FUS techniques may be required to safely treat lesions intimately associated with critical neural structures at a microscopic level (eg. nanoparticle drug release, modulation of inflammatory or apoptotic pathways, control of heat-shock proteins or immune response, etc.). Investigation into these effects is still very preliminary.

Thermal ablation with MRgFUS may also prove useful in eradicating or controlling growth of convexity meningiomas, as well as falcine and intraventricular lesions. It resembles stereotactic radiosurgery (SRS) in being noninvasive and exceptionally effective within a defined “kill zone” that has a very sharp gradient at the edge, allowing for the protection of adjacent normal structures. It has advantages over SRS in that it does not involve ionizing radiation, there is no cumulative dose and therefore it is repeatable without limitation. It also allows for real time visualization of energy location, and real time monitoring of energy delivery, through MR thermometry.

Certain features common to meningiomas have a significant influence on FUS treatment, especially the degree of vascularity, proximity to bone, and involvement with cranial nerves (in the case of skull base lesions) and blood vessels (internal carotid, cavernous sinus). In addition to the heat sink effect of vascularity within a tumor, initial studies will need to assess thermal effects on cerebral arteries and veins along the surface of the tumor, and take care to avoid the possibility of venous thrombosis and subsequent infarction. Bone heating adjacent to the sagittal sinus and other sinuses or large cortical veins will also require careful temperature monitoring so as to avoid the risks of sinus or cortical vein thrombosis with resultant venous infarction. Bone heating is also an important consideration in the skull base tumors, as well as in convexity lesions. However, the cranium has a very effective cooling mechanism which prevents the heating of the brain from heat sources outside the skull (like being in the sun). This mechanism assures the constant temperature of the brain despite the higher tissue temperature outside the skull. This mechanism does not exist at the skull base.

There is another big difference between convexity and skull base meningiomas when they are treated with FUS. In order to be able to focus on convexity meningiomas lower frequency is required while skull base tumor treatment is possible with higher frequency phased array transducers. In both cases the hemispheric phased array is necessary.

Animal studies in a porcine model have shown no histological damage of the optic nerve or chiasm after 52 deg C thermal exposure with FUS. The sonication target volume is surrounded by a 2-3 mm zone of steep temperature gradient, outside of which there is no significant temperature elevation. In these experiments, however, no functional ing was performed.

Studies in peripheral nerves suggest that increased myelin offers protection against thermal lesioning. Nerves with a significant proportion of unmyelinated c-fibers are at higher risk for damage from heating. FUS animal studies indicate that thermal tissue ablation can be accomplished safely within 2-3 mm of neural tissue. Tolerance of neurovascular bundles adjacent to the prostate has been investigated in a canine model of prostate ablation with MRgFUS. Despite these findings the issue of the nerves thermal sensitivity is not well known. Specifically there is not enough information about the thermal sensitivity and temperature tolerance of the optic nerve. Because the size of the nerve is also a significant factor, animal models may not sufficient to resolve this issue.

Meningiomas occur naturally in dogs and cats, which might offer a convenient animal model if necessary. Matching the design of the technology to the animal in a way that would give meaningful results that are applicable in humans is quite difficult, however. This is due to the size and shape of the human skull, which is much larger than that of the experimental animals. One might perform limited experiments to directly assess the thermal effect of FUS on bone and overlying scalp, meningioma, arteries, veins and venous sinuses. In general, the animal would require a craniotomy for treatment, although a piece of human calvarium might be placed in the beam path to simulate the process of treating a patient transcranially with an FUS brain unit.

Treatment of the prostate with FUS in a dog model also leads to edema within the target region. Thermal lesioning with laser also is often associated with edema, which can be partially mitigated with pre-treatment with dexamethasone.

The Exablate 4000 (Insightec, Inc., Haifa, Israel) has been evaluated in rabbit and dog models at the Brigham & Women’s Hospital, which models will also be assessed in Toronto. The common model uses closed skull with a 2 cm distance from overlying bone to target. Future work will assess the cavitation threshold at varying frequencies.

Issues related to Pilot Studies
Meningiomas are quite similar to uterine fibroids in having different fundamental tissue types that may require different FUS energies to achieve the same temperature. The determinant of tissue ablation is the peak temperature and the time it is maintained. The amount of energy required to achieve that temperature can vary with different types of tumor, depending on the vascularity and other characteristics, even within meningiomas as a group. In general, a longer sonication is required to achieve the ablative temperature in certain tissue types.

Meningiomas may have responses to treatment similar to those seen in breast fibroadenomas. FUS in those lesions causes tumoral edema for 24-48 hours. Initially the tumor hardens, but it later softens.

Initial FUS protocols for brain cases stipulate that at least 2 cm distance be maintained between the target and the inner table of the skull, in order to minimize the amount of bone heating that might occur. In the case of convexity meningiomas, this bone heating can even be used to advantage in helping to create a thermal lesion within the tumor. Care must still be taken to avoid excessive bone heating, and any damage to the overlying skin. The scalp is very well equipped for cooling given its exquisite blood flow. In addition, the ExAblate brain unit has a very efficient liquid cooling system for the scalp.

Protocol option 1: FUS treatment prior to surgery
One option for a treatment protocol mimics a study requested by the FDA in the evaluation of MRgFUS for the treatment of uterine fibroids. MRgFUS treatment would be performed on the central core of the meningioma (subdosal, ie. full ablative dose but to only a fraction of the tumor volume). As above, patients would stop anticoagulants for two days before MRgFUS, until there is no evidence of increased risk of hemorrhage. The entire meningioma would then be removed via a standard craniotomy. This type of study could be used to assess the safety of the technique and gain significant information about the histopathology of MRgFUS in this particular tumor type.

Protocol option 2: FUS treatment in patients unsuitable for surgery
Another suggested protocol would recruit patients who are not candidates for surgical resection (due to age, concomitant illness, coagulopathy, etc). This study might include patients who have failed stereotactic radiosurgery, or as an alternative to radiosurgery in patients who cannot undergo surgery. Malignant meningiomas, well known for failing radiation, would not be good candidates for these initial feasibility and safety studies. In addition, prior radiation would be a contraindication to meaningful histopathological assessment. Since meningiomas can be quite stable in size over an extended time, tumor growth or progressive symptoms must be shown prior to a decision to treat the patient.

Tumor growth or progressive symptoms must be shown prior to a decision to treat the patient. Meningiomas can be quite stable over time, without obvious growth.

Given the relative success of surgery and/or radiation (either stereotactic radiosurgery [SRS] or stereotactic radiotherapy [SRT]) in managing benign meningiomas, it is important to define clear endpoints in a well-defined cohort. It may be that the phase 1 route followed for uterine fibroids and breast fibroadenomas may well apply to meningiomas: treat with FUS to establish safety, then surgically remove the tumor. A parallel study could be performed in patients who are not surgical candidates.

Currently, the following institutions are interested in pursuing the phase 1 safety study: the Brigham & Women’s Hospital in Boston, the University of Virginia, and the University of Toronto.

Brain Metastases – Workshop Discussion

Discussion: Best Current Treatment
The best current therapy for non-disseminated metastases is surgical resection in the absence of definitive diagnosis, and generally involves stereotactic radiosurgery in patients with a known diagnosis. A significant number of metastases fail to respond to radiosurgery, or are associated with significant peritumoral edema and symptom worsening months later that might represent a mix of tumor recurrence and radionecrosis of the tumor.

Issues related to Pilot Studies
A significant number of metastases would be excluded if they were within 2 cm of the inner table of the skull, due to the effort to minimize skull heating. The initial safety study would exclude posterior fossa metastases, due to the risk of edema and sudden neurological demise due to acute hydrocephalus. It would also exclude melanoma and renal cell carcinoma due to hemorrhage risk.

Protocol option 1: FUS Feasibility in Radiosurgery Failures
The feasibility protocol would recruit 20 patients over a year who had failed stereotactic radiosurgery and had four or fewer lesions. Only one lesion would be treated, generally the most symptomatic one. The treated lesion must be at least 2 cm from the inner table of the skull, with a maximum diameter of 4 cm. Patients with a worsening radiographic picture due to either tumor re-growth or radio-necrosis due to SRS or both would be included. Study sites might include the Brigham & Women’s Hospital in Boston, the University of Virginia, the University of Toronto and Sheba Medical Center.

The feasibility study protocol would involve treatment using the lower frequency (230 kHz) brain unit, which allows for slightly larger sonication volumes and target location well away from the midline. As above, patients would stop anticoagulants for two days before MRgFUS, until there is no evidence of increased risk of hemorrhage. Tumors known to have a high propensity towards hemorrhage would also be excluded (renal cell carcinoma, melanoma), as would those in eloquent regions or with significant mass effect from peritumoral edema.

Future work will involve extending the accessible treatment area closer to the skull. In addition, simulations will investigate parameters leading to cavitation in relation to the cranial vault and the skull base.

Protocol option 2: FUS Feasibility in WBXRT Failures
In Toronto, Loch MacDonald reported they are not as aggressive in treating patients with brain metastases using stereotactic radiosurgery as in the USA. This represents an opportunity to obtain de novo data in patients who have not had focused treatment. The protocol might involve MRgFUS in a head-to-head comparison, or as an adjuvant to whole brain radiation therapy (WBXRT). Brain metastases are a common disease, so a single site could generate enough data, especially for a safety study. Dr. MacDonald estimates easily recruiting 10 patients over 6 months. The protocol would involve treating patients with four or fewer brain metastases who have failed WBXRT, with treatment volume < 2.0 cm diameter, lesions in non-eloquent locations. The requirement for the lesion to be greater than 2 cm from the inner table will be a problem because most lesions occur in that range. Consensus of the group: to exclude melanomas and renal cell carcinoma due to very high hemorrhage risk. They would also exclude posterior fossa lesions until safety has been demonstrated (ie., no overwhelming edema post treatment).

Acute Ischemic Stroke – Workshop Discussion

Discussion: Best Current Treatment
The best current therapy is tissue Plasminogen Activator (tPA) administered intravenously, but the fact that it must be given within 3 hours of symptom onset makes this a very limited approach. Based on more recent data, Europe has now extended the time window to 4.5 hours after symptom onset. tPA may take hours to have an effect, and many patients do not gain benefit from tPA treatment. Some patients develop side effects of neurotoxicity or hemorrhage.

There are some recent clot retrieval devices that are FDA-approved, but none have been shown to have efficacy in clinical studies. However, their use may constitute the standard-of-care in some larger centers.

Patients with carotid artery occlusion have a particularly poor prognosis, and often do not benefit from tPA administration. If FUS is shown to be an effective noninvasive mechanism for thrombolysis, it may represent a major step forward in treating these difficult patients.

Issues in Study Design
If the MRI in a stroke patient demonstrates an ischemic penumbra, or area of diffusion-perfusion “mismatch,” then one might successfully proceed with FUS dissolution of the clot even out to eight hours after symptom onset. Given the critical need to reopen the clotted vessel as quickly as possible, bypassing the CT scan and relying entirely on MRI for determination of hemorrhage, as well as FUS treatment-related skull correction and guidance of the FUS procedure, is desirable.

First experimental data show that transcranial sonothrombolysis with FUS can be achieved within seconds and without the use of a lytic agent, such as tPA. However, in selected cases FUS may be utilized as an adjunct to tPA administration, which might shorten the time to recanalization of the vessel. A corresponding protocol would assess FUS as a facilitator of tPA clot lysis, and not compare the two treatments against each other.

Alternatively, FUS could be evaluated as an alternative to mechanical thrombectomy. One would expect them to carry similar risks of hemorrhage into the reperfused region if too much time had elapsed before recanalization. It seems reasonable that FUS might allow restoration of flow sooner than mechanical thrombolysis, given the logistics of the procedures, but this remains to be seen. The entire FUS treatment could probably be accomplished in just under an hour. Time to administration of intra-arterial tPA takes more than an hour just to get the drug on board.

FUS and mechanical thrombectomy require visualization of the clotted vessel with MR Angiography (MRA) and Digital Subtraction Angiography (DSA), respectively. tPA administration does not demand anatomic clot localization prior to treatment. Visualization of the occlusion point is clearly a prerequisite for FUS and mechanical thrombectomy.

FUS also offers great potential for modulation of the microcirculation. There remains much in the way of basic investigation of the effects of different acoustic pressures and other non-thermal effects of FUS that influence the behavior of the microvasculature. FUS may also offer direct neuroprotective effects through modulation of the endothelial nitric oxide synthase (eNOS) system.

Relevant Models
Much of the current understanding of sonothrombolysis comes from in vitro analysis of the effects of the brain unit on various clot models. Early results indicate that clot resolution will be more rapid in vivo than in vitro, especially using high intensity FUS. There is some controversy over the relative utility of high intensity vs. low intensity. Current studies are assessing the patterns of fragmentation of large clots, yielding preliminary information on the risk of distal emboli of the fragmentation material. These could also compare the toxicity of the hemolysate from FUS with that found after administration of TPA.

The specific question indicates which model would be best. The mouse model would be sufficient for evaluation of synergistic effects of short pulses of FUS on TPA activity. The baboon stroke model reported from France might be better to assess the risk of bleeding with FUS.

Initial animal studies have involved rats, dogs, pigs, sheep and baboons, depending one which variable was being examined. Short transcranial FUS pulses could be performed (in mouse or rat, but not rabbit) using a thrombotic clot in an artery to show that it can be opened with FUS, and to assess the optimal parameters to do so and minimize injury to the artery. Preliminary work might even be performed in a peripheral artery, not necessarily requiring a cerebral artery. The nature of the dissolved clot, including the risk of smaller fragments embolizing to distal vascular territories, could be evaluated in this model.

The stroke research community will clearly require demonstration of FUS safety in the brain with these sonothrombolysis treatment parameters. Safety studies evaluating FUS effect on brain could be performed in a porcine model, using the same sonication parameters that are effective for clot dissolution in the rat or mouse arterial occlusion model.

Fundamental questions that must be answered include:

  • What is the exact nature of the clots to be treated in humans?
  • How can we expedite treatment, given that time is of the essence? Can ultrashort TE (UTE) sequences provide fast simulation and obviate the need for CT scan in accomplishing skull modeling for the MRgFUS treatment? Radiation force imaging might totally bypass the need for CT simulation.
  • How sensitive is MRA at identifying the locus of occlusion?
  • Should FUS sonothrombolysis begin proximally and move distally, or vice versa?

Clinical Pilot Studies
An initial feasibility and safety study of MRgFUS will consist of 30 patients who present within the three-hour window, but are not candidates for tPA (eg. those who are on coumadin, have any kind of tumor, or any source of bleeding including recent surgery, etc.). The occlusion site would need to be well-visualized on MRA (preferably 3D), and ultrashort TE (UTE) sequences would provide skull modeling in order to bypass the time in CT. Patients demonstrating perfusion-diffusion mismatch on MRI might also be included, if presenting within eight hours of symptom onset. Those with occlusion of the distal ICA, M1 and M2 would all be candidates. This study would include patients with carotid occlusion. Thrombolysis would be performed from proximal to distal if that is shown to be free of generating particulate emboli.

In some major centers in the US, clot retrieval devices have become standard, even though they have not been proven effective. Interventional radiologists at some of those centers might be unwilling to participate in a trial. However, a Randomized Clinical Trial might be performed to directly compare one of the clot retrieval devices against MRgFUS sonothrombolysis. The randomization would need to occur very quickly, because, in the event of mechanical thrombectomy, time is used gaining arterial access. Randomization might not be required if the study was performed in Canada or Germany.

One would need to demonstrate that the risk from hemorrhage was not unacceptably high. Given that 20% of TPA cases experience hemorrhage outside of the original infarct zone, avoidance of systemic anticoagulation would be desirable.

Patients with atrial fibrillation and embolic stroke might also make a good study population.

Trying to treat patients who have failed tPA would take too long, and should not be pursued in the initial feasibility studies.

In those that do present within the time window, another simple study would be to see if MRgFUS can accelerate the time to recanalization in those patients receiving tPA.

The majority of stroke patients present too late to be candidates for tPA, and those that do not have an ischemic penumbra in the first eight hours after symptom onset might be recruited into a later study of MRgFUS sonothrombolysis.

Centers interested in pursuing the early studies include the Brigham & Women’s Hospital in Boston, the University of Virginia, The University of California at San Diego, and the University of Cincinnati.

Technology Roadmap: Tissue Disruption

Thermal Ablation
The same temperature (57-60 deg C over a few seconds) causes tissue ablation in a wide variety of tissues due to the similarities in protein content. The FUS parameters and energy requirements to achieve that target ablation temperature can vary widely.

The treatment temperature for ablation is 45-60 degrees C for less than 1 minute, and is based on cell biology, not frequency. Temperatures below 45 deg C are probably safe for most tissues. Even brief exposure of 2-4 seconds to temperatures above 56 deg C will guarantee ablation in virtually all mammalian tissues.

The thermal dose, describing the temperature history over time, is an important parameter, more relevant than temperature alone. It should be provided when describing an experiment or treatment. Reaching the same target temperature at different speeds can make significantly different results.

Gray Matter vs. White Matter Vs Tumor
Even 43 deg C for 5 minutes can affect brain tissue and cause functional changes, such as seizures. Temperature is an indirect measure of tissue death, given the importance of time of exposure, energy, tissue factors, etc. The temperature history over time is a better indicator. Even 45 deg C will yield 100% coagulation of tissue over 500 seconds. A more direct indicator of tissue death would be PET imaging. One must use caution above 45 deg C due to the variability of response in different tissue types in that range.

It is possible to measure temperature using T1 on MRI, but it is not practical, because the trajectory during cooling may differ from that due to heating because of the possibility of a phase transition.

Can Cavitation be Controlled?
There is always some level of gas in the fluid that potentially, under ultrasound pressure, can create oscillating microbubbles . When the pressure level grows beyond some threshold, sudden collapse of bubbles can release enough energy to damage cells. This is known as inertial cavitation. Bubbles may be also used to enhance the heating effect at the target. When they oscillate only (without collapsing) they trap much more acoustic energy and cause local heating. This can be beneficial if it occurs within a target tissue, or problematic if it results in excessive heat deposition away from a target tissue. In long bursts ultrasonic excitations, bjerknes forces between bubbles can induce attraction between microbubbles and result in bubble coalescence.

The cavitation threshold is strongly dependant on tissue characteristics and on the acoustic parameters (pressure, frequency etc”). Some people (i.e. Jean-Francois Aubry) claim that the cavitation threshold is reduced by a factor of 10 in the presence of standing waves. Others, (i.e. Shuki Vitek), claim to have proven that it is absolutely not true. It is reduced by a non-negligible factor in the presence of standing waves. More experimental work and cross-validation is needed to quantify this factor.

Inertial cavitation cannot be easily controlled, other than to avoid the threshold conditions which lead to its occurrence. The presence of microbubbles in tissues is strongly correlated with the cavitation phenomenon. The threshold varies with the square root of the frequency, and is lower with lower frequencies. Nuclei of small particles enhance the propensity towards cavitation. Local energy deposition can be quite high due to cavitation, which can alter local membrane permeability. Several research groups are focusing on controlling cavitation. Blood brain barrier opening is a good example of the general relationship of microbubbles, wherein increased energy is associated with more negative effects.

Cavitation allows enhancement of ablation, greater effect with less energy delivery. Cavitation can be detected immediately, and has a well-defined acoustic signature. It is more difficult to determine its exact origins, and some techniques need to be developed to confirm its location. Vascular thrombolysis is based in cavitation. Lithotripsy and histotripsy are good examples of therapeutic use of cavitation to obliterate kidney stones and atrial cardiac tissue, respectively.

Thermal Ablation vs. Cavitation
Where possible using thermal effects is better controlled and predicted. However cavitation might be essential in some applications in the brain. To provide a solution for these applications the effects need to be predictable and controlled. The technology must be better mastered before it can be widely used.

Cavitation would typically create larger effects in tissue than thermal mechanisms. At 220kHz, the spot dimensions can be as large as ~1cc where as in 650kHz each sonication is ~0.2cc. Each sonication cycle requires cooling of the skull (~2min) so the rate of ablation can be roughly between 0.1-0.5cc per minute. A 3 cm diameter meningioma (volume 14 cc) would require between 28 (at 220kHz) and 70 (at 650 KHz) sonications. Treatment time would average approximately 60min @ 220khz and 140min @ 650khz for a 3 cm diameter tumor. Cavitation could increase the sonication ablative efficiency by a factor of 2-3, or more. The challenge is to control the cavitation process.

Treatment involving large volumes, such as widespread opening of the blood brain barrier, or delivery of gene fragments or drug to the whole brain, would necessitate cavitation for acceptable treatment times. Three to four hours is a reasonable upper limit to the amount of time an awake, partially sedated patient can comfortably remain in one position in the MRI for treatment. The awake patient provides the grea safety factor in terms of avoidance of inadvertent heat injury, given the sensitivity of dura and skin. Pain would usually be felt prior to the point of irreversible damage. Dural damage might be associated with venous thrombosis if heating occurred near a venous sinus. For volumes much in excess of 22 cc (3.5 cm diameter), one might consider staged treatment in multiple sessions.

Treatment Envelope

Treatment Envelope Adjacent to Calvarium
There are different considerations for the cranial vault, as opposed to the skull base. Significant thermal energy is deposited on first penetrating the calvarium. Cavitation may need to occur closer to the skull. Acoustic reflections off of the skull base are observed. There is less energy per square cm just under the cranial vault, away from the deep focus. If the US focus is near the skull base, there is more concentrated thermal energy adjacent to that focus than one encounters widely over the calvarium, which can yield up to 10X as much thermal energy deposition.

It has been proposed that treatment of lesions adjacent to the calvarium will require the low-frequency unit (230 kHz), though that may not be necessary for lesions at the skull base. However, without cavitation, working at low frequency will require more power at focus as the absorption coefficient is lower. Although absorption in the skull is higher at higher frequency, the focal spot is smaller (so antenna gain is better). Moreover, the question of absorption coefficient value in the skull bone does not reach a consensus. Finally, control of cavitation near the skull was not addressed properly today. So, the question is much more complex.

Microbubbles (either natural or injected) can be utilized, with shorter pulses (by two orders of magnitude), to decrease skull base heating and could be tailored to create apoptosis. Depending on the gain, one can encounter gas bubbles in animal brains with 1-3 cycles in short bursts, which can be visualized with ultrasound.

With homogeneous amplitude, one can generate a bubble cloud using low frequency pressure waves at a 4 mm focus. Ultrasound energy is absorbed, creating a reflecting wall, which absorbs energy and is associated with further energy reflection. The specific interaction of the beam with the bubble cloud is related to the vector of flow propagation. One should be able to prevent expansion by controlling the amount of bubbles in the cloud. Bubble behavior may be easier to control adjacent to the skull base than near the cranial vault, because of the smaller focal zone. The phenomenon has not been well studied. Animal studies are ongoing at the University of Toronto, with advances in treatment planning accomplished at the Brigham & Women’s Hospital in Boston.

Thermal lesions may be too difficult to attain close to the calvarium and skull base. Cavitation may offer a better option to ablate in these regions. Further 3D and skull phantom simulations (best performed with supercomputers, given their complexity), and confirmatory animal experiments are required. Simulations would compare various frequencies, focal shapes, reflections, etc.

There was consensus among the technical teams to collaborate on ing the simulations. Thilo Hoelscher at UCSD mentioned his proposal to FUSF for funding of a study to generate a database of human skulls analyzed with high resolution CT scanning. The group goal is to enable practical clinical treatment of these tumors within five years (by 2014), with FUSF providing the appropriate start-up funding.

Through a separate breakout session, the FUSF organized collaboration between four centers to assess remote effects within the skull, especially in response to a case of hemorrhage in a brain tumor case. These investigators include Shuki Vitek and Eyal Zadicario of Insightec in Israel, Mickael Tanter in Paris, France, Greg Clement at the Brigham & Women’s Hospital in Boston, all led by Kullervo Hynynen at the University of Toronto.

Sensitivity of Cranial Nerves
The thermal history over time, length-heating histogram, and specific microvascular relationships are all-important parameters. Assessing temperature alone is not very revealing. Even in the case of defined target temperature, the actual temperature of the nerve is difficult to determine exactly. Cranial nerve thermal sensitivity is dependent on the proportion of myelinated fibers, and to nerve size, although heat over 60 deg C even for seconds can permanently destroy nerves. The acoustic nerve has different thermal sensitivity in the internal auditory canal, versus outside the canal, due to changes in myelination. A stretched or compressed nerve will have less thermal tolerance, related in part to lower fiber number and demyelination.

The optic and olfactory nerves represent two extremes. The optic nerve is relatively easy to monitor using stimulation and recording (Visual Evoked Responses). Sagi Harnof ed optic nerve sensitivity to thermal effects of FUS and found the exposure to 45 deg C and less was tolerated, and that temperatures of 45-55 deg C were safe based on histology (functional studies were not performed). Temperatures above 55 deg C caused histologic changes, and above 65 deg C a clear lesion in the nerve was seen (reported at the FUSF Symposium, Oct 2008, Tyson’s Corner, VA).

Experience from surgery and radiosurgery suggests that the optic nerve is extraordinarily sensitive to damage. Empirical observations in clinical cases may provide the best data, as they did in assessing cranial nerve tolerance to stereotactic radiosurgery in the early 1990’s.

Between 45 and 60 deg C, non-myelinated fibers are rendered nonfunctional. Most nerves will at least suffer temporary dysfunction in that range. This principle is utilized by anesthesiologists for temporary anesthesia. Short exposures under 45 deg C should be well-tolerated. Animal models of non-cranial nerves are not relevant to cranial nerve conclusions.

Microbubbles might allow enhanced safety of treatment adjacent to cranial nerves, by avoidance of the thermal threshold limit. However, destructive tissue effects are far less controllable with microbubbles, compared to purely thermal effects.

Remote Effects of Intracranial FUS
Cytotoxic, blood brain barrier, and vasodilatory effects in the near and far fields

Vasodilatation can occur without invoking thermal effects. Thermal effects are related to vascular changes through BBB opening, but these are limited to a few mm from the focus using the high and mid-frequency Insightec system. Spreading depolarization occurs at lower energy threshold than BBB disruption. Vasodilation occurs in the myocardial vasculature with low-pressure ultrasound. Edema is transient and limited to the immediate boundary around the target. Within the thermal target zone one finds mainly necrosis, with a thin rim of apoptotic cells.

Nature of the transient penumbra around thermal ablation
The transient penumbra around a region of thermal ablation is likely vasogenic edema extending out for 2 mm or so, although it is not well visualized on diffusion-weighted imaging. Vasogenic edema extends beyond the region of BBB opening. Animal studies do not appear to reveal gliosis later in the edematous zone.

Standing waves and Ringing
In cadaveric studies and by computer simulations, intracranial acoustic reflections cause an increased pressure, which can be as high as 5% of the pressure in the focus, which is less than 1% of the energy (thermally insignificant, from an energy standpoint). The pressure change can result in cavitation, although one should be well below the cavitation threshold established by Kullervo Hynynen.

In 2007, based on the Umemura publication, Jean-Francois Aubry hypothesized that, as the cavitation threshold in standing waves was less than that seen in propagating waves, it could explain the dramatic effects that occurred in the TRUMBI study at low frequency (300 kHz). He proposed that the cavitation threshold in standing waves is lower by a factor of 10x. This is a key question that requires further analysis and confirmation in experimental work.

In standing waves, the pressure can be doubled, in theory. Larry Crum showed that cavitation is increased with long bursts. Iron deposition in vessel walls can alter the cavitation threshold. Knowing the cavitation threshold elucidates the safety margin.

If identified as a potential risk, standing waves can be minimized by adjusting the frequency. This leaves the question of whether there are areas in the brain where ~5% of the pressure change encountered in the focus leads to significant risk, such as tissue interfaces or regions of calcification. This still needs evaluation and further research.

The most important point is to know the cavitation threshold for monochromatic sonifications in the case of waves traveling in one direction, waves traveling in counter directions and finally standing waves. The case of waves traveling in opposite and parallel directions is as important as the case of standing waves.

Dealing with phenomena involving less than 1% of the energy is insignificant from a thermal standpoint, but not in terms of potential bioeffects of the ultrasound energy (such as cortical spreading depression). Animal models are still important, although skull shape and thickness differences limit applicability of the results to humans. The overall safety margin is still poorly defined.

The in vivo model is far from perfect, and interpretation of results in cadavers is hampered by the lack of blood flow and post mortem bubble formation. Study in the closed skull phantom will likely yield useful results. This should then be followed by appropriate in vivo animal studies for verification of the cavitation phenomenon, then raise the power by a factor of 3 to perform the requisite safety studies. Even then, one might expect additional empirical data from experience in diseased patients.

A collaborative work group was assembled to further investigate standing waves and hot spots using 3-D whole brain acoustic simulation via a common database. This work will be supported financially by the Foundation and is expected to take 6 months to complete. The work will be led by Kullervo Hynynen of the University of Toronto, and will involve Greg Clement in Boston, Mickael Tanter in Paris and Shuki Vitek in Israel.

MRI volume rendering for treatment monitoring
Work is proceeding on 3-D volumetric thermal mapping, which generates a slab of adjacent MR slices around a region of interest. Current models enable construction of 20 slices each 10 mm thick. One needs to select the sensitive regions for analysis. The technique requires imaging as quickly as possible, which is limited by STN. One generally requires a single shot image, and EPI bandwidth is critical. This might necessitate thinner slice thickness. Planar thermometry is inferior, often resulting in disorientation.

The potential of volumetric imaging is a key safety element in this technology, and efforts should be invested in developing imaging tools.

Microcalcifications not appreciated on CT
Microcalcifications may cause local heating, but the overall effect depends on size and location. The physics is similar to that in bone, but the density is much less. Calcium close to the thermal focus can undergo significant heating. Blocking phased array transducers that aim through regions of calcium can reduce the heating. Within a tumor, this excessive heating might even enhance therapy. However, it can cause complications near critical structures. Excessive calcification near sensitive neural structures might be an absolute contraindication to treatment. Ultrashort TE imaging (UTE) with MR (~ 200 usec) can visualize calcium, and might replace CT for detecting calcifications. Besides, it might also replace CT for skull modeling.

Further work and analysis needs to compare, in human brain imaging, that MR techniques can be just as good (or better) than CT in identifying local calcifications.

MRI Determination of Bone Density – Potential Obviation of CT Scan
Ultrashort TE Sequences

Bone contains much less free water than soft tissue, and the resultant T2 in bone ranges from 10s to 100s of usecs. Ultrashort TE imaging (UTE) may be useful for defining bone and calcium, and detecting bound water (protons). GE scanners can measure 8 usec T2 yielding high resolution bone images. It requires the readout sequence in addition to the short echo time, which will add to overall scanning time.

This issue is of practical importance in the management of stroke, where time to treatment is of the essence. If one is able to save time by bypassing the CT scan (used for detecting hemorrhage and performing skull modeling for phased array corrections), then clinical results might be improved.

Acoustic force radiation imaging
Acoustic force radiation imaging is another technique offering significant benefit. It may be used to change the phasing of the transducer elements to optimize the sharpness of the acoustic focus. The principle is similar to auto-focusing in a camera. Displacement varies with beam intensity. The algorithm is based on the energy at the focus to recover the phases for the array. At low frequency the skull base can be transparent, and radiation force imaging may be useful. Acoustic radiation force imaging works even better at high frequency as the force is proportional to the brain absorption coefficient. The system ed in Paris uses 1000 shots over a few minutes to correct the beam focusing. It has been patented and ed in vitro within a 1.5T MRI system and will be incorporated in the Brain System being developed there by the end of the year.

Bone Correction Algorithm
Bone correction algorithms in current use, based on CT model, can take anywhere from seconds to hours to generate results. Bone absorption as well as generating aberrations vary with frequency. Complexity and sensitivity to the quality of the focus increases at higher frequencies. A different (non CT) method is based on measuring signals reflected from bubbles at the focal spot. Thus the complexity of the algorithm may depend on the overall system design (geometry, frequency, etc.) and the clinical application requirements.

Summary: Quo vadis?

The Brain Workshop marked a milestone in the evolution of MRgFUS in the brain. Several concrete initiatives were started with collaborations formed at the Workshop. These include the Remote Effects Task Force, formation and determination of a collaborative plan from the neuromodulation Research Program Planning Group (RPPG), refinement of a research plan from the ischemic stroke RPPG, and consolidation of results to date for the intracerebral/intraventricular hemorrhage RPPG. In addition, momentum from the workshop was used to begin work on clinical protocols for the treatment of malignant gliomas, metastatic brain tumors, meningiomas, pituitary macroadenomas, and essential tremor.

In his closing remarks, Dr. Kassell reflected on our original Metrics of Success:

  • We have some answers to burning questions, and many new questions.
  • We have an outline of a comprehensive work plan.
  • We have an actionable plan for 4-5 pilot clinical studies.
  • We have created a climate and a structure for collaboration. Academic institutions are cooperating with each other and with companies, and companies collaborating with each other.

He also made the following summary statements:

  • Constraints of time eliminated some important topics. We did not have time to deal with all of the important questions.
  • We will initiate discussions with leaders at the NIH to help move this effort forward.

Feedback from participants was very enthusiastic and positive. Due to the rapid progress in this field, we anticipate hosting another Brain Workshop March 21-24, 2010 [tentative].