1.1

Technology and tissue

Generally, these agents (as surgical tools) are products of the twentieth century. The devices which deliver these energies to tissue have developed through: a) technological mastery, and b) understanding of energy-tissue interactions which have both grown over that same time period. As two examples, consider that the first electrosurgical unit was introduced in the 1910s or that the first laser available for medical applications appeared in the 1960s (the ruby laser at 694 nm). In order to invent a suitable device, implement it and optimize its performance (in answering a clinical surgical need) the requisite knowledge base extends beyond producing the energy to the tissue-interaction.

2.1

RF for the treatment of trigeminal neuraglia

The approach to relieve facial pain due to ‘trigeminal neralgia’ by injuring the trigeminal nerve (at the appropriate division of its root) has been known for many years. The nerve has three divisions: the ophthalmic, maxillary, and mandibular. These divisions are responsible primarily for sensation (though the mandibular nerve does function for both sensation and motion). The ophthalmic division serves the cornea, ciliary body, lacrimal gland, mucous membrane of the nasal cavity, skin of the forehead, eyelid, eyebrow and nose. The maxillary division serves the dura, forehead, lower eyelid, orbit, upper lip, gums, teeth of the upper jaw, skin of the cheek and nose. The mandibular division serves the temple, auricle of the ear, lower part of the face, teeth and gums of the mandible, mastication muscles, and mucous membrane on the anterior part of the tongue. The fibers of the sensory root arise from the semilunar ganglion; the fibers of the motor root arise from nucleus in the pons.

Therapeutic techniques to injure the nerve have included chemical rhizotomy (injection of chemicals intracranially) and RF ‘ablation’ at the trigeminal ganglion. An alarming incidence of side effects from the procedures motivated Sweet and Wepsic to pose that a more controlled thermal dose could be delivered.⁹ They surmised from the evidence of the day, that this would be advantageous over the relatively poor control of the spread of injected chemicals and excessive electrocoagulation. Their solution was to use less current and to coagulate incrementally to effect. They cited key improvements which assisted in the task. Particularly, they noted the availability of more controllable heat sources than the then current RF generators at 2 MHz. The procedure was performed percutaneously with a 10 cm long probe (active electrode) with an exposed tip 5 or 10 mm in length; the dispersive pad was located on the shoulder. With improved drugs for sedation, via a perforation through the cheek, a freehand search for the foramen ovale was performed. Once located with the tip of the probe, an x-ray was taken to confirm positioning. Further, in order to confirm location of the active length of the probe tip, an electrical stimulation was performed at the threshold of sensation. A thermal “test” lesion was made for sensation; response here further confirmed targeting of the trigeminal branches. Temperature was monitored at a single point on the probe and watch was kept on the voltage or current to confirm the condition of the probe (i.e. an unusually high voltage to maintain a set temperature might indicate coagulum or a defective probe). Temperatures were kept at approximately 60 °C. Of 274 patients treated, 187 received relief from pain, but maintained sensation in the skin - an advance due to their conservative approach. Elementary to their hypothesis under which they worked was that the smaller myleniated or unmyelinated nerves were more susceptable to thermal damage than the more heavily myelinated nerves.

Around the same time, Hori and co-workers performed “controlled thermocoagulation” of the facial nerve for treating facial spasm using a different percutaneous approach.¹⁰ A flouroscopic technique was used to guide the RF probe through the stylomastoid foramen with confirmation by electrical stimulation. A first lesion was made at 55 °C for 10 sec. The temperature was then raised to between 60 - 65 °C, and repeated until a slight facial weakness was observed. Most patients experienced a transient facial weakness for up to 4 months, but 24 of 27 patients were spasm free at 1.5 years post-operatively.

In 1983, Cosman and co-workers in reviewing technology for “localized therapeutic destruction of nervous tissue” which was deemed successful as there were mechanisms 1) by which to monitor the temperature and thus to “quantify lesion size” and to prevent charring or boiling … and 2) which provided “excellent target localization using stimulation, impedance monitoring, and recording.”¹¹ The report stated that brain tissue is damaged at temperatures at and above 45 °C which is a lower threshold for effect than the typical coagulation threshold for proteins of 60 °C. The lesion size was most dependent on the temperature and size of the RF probe tip, and the size equilibrated after only 30 - 60 s. The lesions created were ellipsoidal and less than 1 cm in length and diameter. Some lesions showed evidence of tending toward the ventricles which was taken as an indication that a path of least resistance for the current and heat.

In current literature, Taha and Tew published on their experience with 500 patients treated with percutaneous RF rhizotomy for trigeminal neuralgia.¹² They proposed also that the technique is based on the premise that RF lesions are selective to the trigeminal “nociceptive pain fibers” (i.e. the unmyelinated C fibers and poorly myelinated A-delta fibers) and it preserves the trigeminal heavily myelinated fibers which carry tactile prorioceptive and motor functions.^13,14 They noted that current (1996) percutaneous techniques include: RF, glycerol injection and balloon compression.¹² The procedure was performed under conscious intravenous sedation. The goal of the therapy was the “dense hypalgesia in the painful trigger zone” which they defined as a 2/3 loss of pinprick perception providing pain relief to the patient. They recalled the possible side effects of the procedure: facial numbness, dysesthesia, corneal anesthesia, keratitis, trigeminal motor dysfunction, permanent cranial nerve deficit, intracranial hemorrhage and others. They reported that the highest rate of initial pain relief was for technically successful RF procedures with a 20% rate of pain recurrence within 9 years which is a comparable rate to surgical microvascular decompression and a more favorable rate than glycerol injection rhizotomy. With RF therapy, facial numbness was typically encountered though usually mild and well-tolerated. They reported having used a careful technique to target the specific trigeminal rootlets and to limit degree of sensory deficit (major dysesthesia incidence was 2%). If the patient could not assist during the procedure to “localize or quantify” the lesion, the surgeon needed to use personal observation, physiological monitoring or sensory evoked potentials. They cited “improved precision and reduced side effects” using a curved electrode (as opposed to straight). The lesion was better controlled with RF than glycerol or balloon compression, and the thermal damage perhaps more effective and not as morbid as the other modalities.

As a postscript to the technique which was presented in more detail in a subsequent review, the active length of the RF probe was 0 - 5 mm.¹³ Its diameter was 0.5 mm. The curved electrode tip carried a thermocouple, stimulator and a lesion generating element. The active length of the probe extended 0 - 5 mm beyond the cannula. Initial placement was by anatomical reference followed by a flouroscopic image. Final placement was determined by the patient’s response to stimulus which was obtained by applying a square wave at 0.2 - 0.8 V at 50 Hz for 1 msec. Paresthesia or pain on stimulation revealed the location. Also, a 40 °C test lesion evoked similar response. At this point, the awake patient was given more sedation. A lesion was induced by raising the probe temperature to 60 - 70 °C for 70 sec. Lesions were typically 5 x 5 x 4 mm. They reported 99% pain relief post-operatively.

Another recent report confirmed the imperative to adjust needle placement within the nerve by using the response to low threshold stimulation at 0.1 V at 50 Hz for 1 msec.¹⁵ The probe was repositioned if no response was obtained at 0.4-0.5 V. Initial settings for delivering the thermal lesion were low, at 10 V and 60 mA for 15 - 20 sec. Post some clinical response to this test lesion, final lesions were delivered at settings as high as 20V and 100 mA in 40 - 60 sec. A separate report on deleivering 70 °C for 90 sec for the thermal lesion noted proximity to the trigeminal of the abducens nervewhich affects the corneal response. This nerve is closest to the ophthalmic trigeminal branch, a target in about 25% of the patients treated with RF rhizotomy. They reported a case of reversible damage to the abducens. As the damage did reverse, they pose that perhaps this is a sign of preferential sensitivity to heat damage to small A delta and C fibers over more heavily myelinated fibers (i.e. the abducens.)¹⁴

2.2

RF for the treatment of movement disorders

There has been a history of clinical use of stereotactically-guided RF lesions for relief cf movement disorders such as those symptoms associated with Parkinson’s disease. These include dyskinesias, “on-off fluctuations”, dystonia, rigidity and bradykinesia. In the case of Parkinson’s disease, the hypothesis was that the symptoms of the disease were due to neuronal hyperactivity in the globis pallidus intemus (GPi). Interest in this “surgical” approach had waned with the advent of drugs which alleviated symptoms. For various reasons, the surgical technique is “back into the mainstream of treatment for otherwise intractable symptoms of Parkinson’s disease” as described in a 1996 article by Favre and co-workers.¹⁶ While there is still argument as to the exact location for a therapeutically optimal thermal lesion, the target is generally taken to be GPi. The hypothetical target is approximately a 1 x 1 x 3 mm³ rectangular volume.¹⁷ It is generally targeted with the same thermal dose parameters as descibed above for RF rhizotomy.¹⁶

Favre presented a survey of 28 neurosurgical centers engaged in this practice cf “pallidotomy” for the treatment of tremor in patients with Parkinson’s disease. Magnetic resonance imaging (MRI) and computed tomography (CT) were used at the various centers to determine the target coordinates. This is a minimally invasive procedure requiring intracranial access in the absence of a suitable natural foramen. Access is achieved using a burr or twist dill to put a hole in the skull. A stereotactic frame established the entry point and trajectory with the ‘standard’ coordinates for the target. These were: 1-3 mm anterior to the midcommissural point, 18-23 mm lateral to the midline, and 2-6 mm inferior to the intercommisural line. Some centers reported using recording electrodes to define the pallidal base and outline the pallidum. Nearly all centers reported using stimulating electrodes to identify the target. Response to this “macrostimulation” assured suitable probe placement. Some observed patients’ visual response (via the recording electrodes) as a measure of proximity to the optic tract as evoked by flashing lights. Most performed a test lesion over a reported range of 40 - 84 °C; final lesions were generally made at 75 °C for 60 sec (none reported treating at less than 60 °C or for less than 30 sec). None of the responding centers relied on radiologic imaging alone for final determination of the target. Physiological confirmation was deemed essential. Generally, the region was treated with three lesions separated from each other along the trajectory of the probe by 2 mm. This lesion distribution assisted in avoiding thermal damage to the optic tract. The overall lesion was intended to be a cylinder of coagulation which encompassed the medial GPi. The survey reported that the overall lesions increased in size post-operatively and were maximal on the third day. They noted ongoing controversy as to the proper location and size for a lesion. As a related note on overall lesion size, Lozano and co-workers reported lesions which were approximately 4 x 4 x 6 mm (range 80 - 150 mm³).¹⁷

In a recent study, Tsao and co-workers performed RF lesioning of the GPi with extensive pre-therapeutic mapping of the site by microelectrode recordings of cellular activity.¹⁸ The GPi was targeted initially with the same standard coordinates noted above, and approached (initially with the microelectrode) by a stereotactic frame using CT image data. By visualization of the microelectrode readings on an oscilloscope, cellular activity was correlated to physiologic observations. This refined the target to localize the lesion to the GPi as it was characterized by activity, not anatomic location. This also was used to identify the non-targeted optical tract. The lesion electrode was 1.1 mm diameter with a 3 mm exposed tip, or 1.6 mm diameter with a 2 mm tip. A thermocouple in the probe measured the temperature. The lesion probe was used for a final stimulation (to confirm no unwanted motor activity in the patient’s contralateral side and no visual field abnormalities). The report mentions two methods for producing a lesion: one, in which 3 - 6 discrete lesions were made along a tract at 2 mm increments of pull-back on the probe, starting temperature of 75 °C for 60 sec which was increased to 80, 85, and 90 with an increase also in duration over a range from 60 to 120 sec; the other method used a constant temperature of 75 °C for 60 sec with pull-back on the probe in 1 or 2 mm increments along one or two tracts. They report good MRI visualization of the post-operative lesions revealing a central necrosis and a surrounding edema. The adjustments in targeting based on microelectrode recordings did impact the placement of the lesions.

2.3

RF for the treatment of brain tumors

Interstitial thermal therapy is a technique in which the thermal energy source (usually a thin probe to render the therapy minimally invasive) is introduced into the targeted tissue directly. It is associated with an increase in temperature proximal to the probe: a) to levels that are sufficient to induce hyperthermia (~ 40-50 °C) for which the temperatures are sustained for several tens of minutes the actual duration of which depends on the temperature, or b) temperatures are raised to coagulative levels (> 60 °C) for a duration cf minutes. In using RF as the thermal agent, we can refer to the therapy as radiofrequency-induced thermal therapy (RFITT). Actually, the techniques described in 2.1 for treating trigeminal neuralgia and 2.2 for motor disorders are applications of RFITT in which the intent is to coagulate a small volume of tissue using stereotactically placed probes for a minimally invasive coagulation of small targets by probes held to ~ 70 °C for 70 sec. Typically, tumors present targets with dimensions on the order of centimeters. The goal is often to treat this large volume with a technique such as RFITT. While RFITT has a presence in surgical literature for treating liver tumors, there is little on the targeting cf larger volumes intracranially with RFITT.

Notably, a recent report from the neurosurgical literature reported on application cf RFITT in an animal intracranial glioma tumor model.¹⁹ The experiment combined RFITT tissue heating and the release of a chemotherapeutic agent carried by heat-sensitive liposomes. The liposomes, microscopic vesicles made of phospholipid bilayers, had been shown to be able to carry the drugs, and to release approximately 80% of their contents at or above 41 °C. The heating element was a RF electrode. A separate thermocouple at the edge of the tumor provided feedback. The desired effect was to maintain the temperature at the tumor edge at 41 °C for a total of 30 minutes. The effect of this hyperthermia were seen on histology 5 days post-operatively. The results showed a higher concentration of the chemotherapeutic agent in the tumor after heating combined with liposomal drug delivery (higher than controls, heating alone, liposomes alone, and drug alone) with a delay in tumor growth. The authors suggested that the hyperthermia opened blood brain barrier (BBB) and increased the transport of drug with specific potentiation of the effect by use of the liposomes. This was not a clinical study, and there were concerns expressed about liver and spleen toxicity, but containment in the liposomes may be a factor in this and requires study.

3.1

A new laser wavelength for brain tissue

In evaluating a new laser wavelength for neurosurgical applications, Martiniuk and co-workers reported on the use of a “frequency-shifted” Nd:YAG laser with a modified wavelength of 1440 nm.²⁰ The laser was used in normal rabbit brain as a test for the wavelength to both coagulate and evaporate at the same time. The wavelength was short enough to allow transmission through a quartz fiber, but long enough to have water absorption (~ 10 x that of 1064 nm) greater than the Nd:YAG. Thus, this would have less penetration due to the increased absorption, but would have better coagulative properties than the highly absorbing CCF laser. Their results showed a higher absorption for 1440 nm than for 1064 nm in tissue; there was more disruption with the new wavelength and less peripheral damage. They proposed that the laser had promise as a tool for resecting tumors with optimized hemostasis.

3.2

Pulsed laser for ventricular shunt clearance

In an in vitro experiment to model occluded ventricular shunts, Christens-Barry and co-workers reported on the use of the pulsed holmium:YAG (Ho:YAG) to break up tissue.²¹ In a clinical setting, the shunts are often occluded by ingrown tissues from the choroid plexus. The laser operated at 2090 nm which is a wavelength highly absorbed by water (~ 100 x the absorption of 1064 nm). Special fibers will transmit this wavelength. The laser emitted pulses of light 300 microseconds in duration. The strong absorption and short pulse duration provided high peak powers for a photodisruptive effect in which small, short-lived vapor bubbles were explosivley generated. The pulses were delivered in groups of 5 - 10 pulses at 200 - 450 mJ per pulse through 200 and 300 micron fibers. The results showed disruption of the tissues tested which included human specimens. For shunt clearance, while there was damage to the tissues, there was no damage to the shunt material. Though, seaparately, direct hits on the material, rather than oblique hits as used for shunt clearance, did cause some slight marring of the surface.

3.3

Contact tip for neurendoscopy

In a clinical study of 49 patients, Vandertop and co-workers reported on the experimental application of a contact laser tip for neuro-endoscopic procedures.²² The authors expressed that while a “ffeebeam “ use of a laser (for non-contact ablation/coagulation) is suitable for work on tumors or the choroid plexus, it may not be a controlled enough technique for neuro-endoscopy. The clinical goal is to enter the third ventricle and remove obstruction to the flow of cerebrospinal fluid (CSF). The tissues encountered are relatively white, non-pigmented, ependymal tissues and arachnoid membranes which are scattering and not highly absorbant. This necessitates the use of relatively high powers when using visible wavelengths or near 1R; they report powers on the order of 15 W for several seconds. In such cases, tissue can undergo carbonization and be instantly become highly absorbing and the deposition of heat becomes unpredictable. Notably, in the vicinity of the third ventricle are non-targeted sensitive structures like the basilar artery. They designed a blunt ball-shaped tip, 0.8 - 1.0 mm in diameter, which was pretreated with a 90% absorbing layer of carbon particles. This was placed at the end of a 400 micron optical fiber transmitting either 810 nm (diode laser) or 1064 nm (Nd:YAG) light. The carbon coated tip absorbs the light and becomes hot enough with low power to penetrate tissues which may poorly absorb the wavelength of light. Preclinical testing of the probe which was placed in a contact fashion had shown that the probe could penerate tissue 0.3 - 0.5 mm thick for an exposure of 0.5 sec at 1 -3 W; the collar of damage peripheral to the penetration was 0.2 -0.3 mm wide. Thus, the laser power was low, and the thermal effect remained proximal to the probe with short exposure times. Forty-nine patients underwent procedures for symptoms attributed to obstructed flow of CSF. The fiber was passed through a neuro-endoscope to access the targeted tissue. An overall success rate was reported for powers ranging from 1 - 5 W. They reported that the desired level of control was realized in the clinical setting.

7.1

Advances toward MRl-guided intracranial CITT

To appreciate the principles of image-guidance for control of interstitial thermal therapies, consider the Gilbert report cited at the end of the previous section. Experimental cryotherapy was performed on the cortex of a living rabbit using a liquid nitrogen probe.³⁸ The probe was placed on the surface of the brain through a hole in the skull. The targeted tissue was the full thickness of the cortex inferior to the probe; the requirement was that the thalamus on the other side of the cortex be spared. These structures were not visible from the surface. MRI provided the visualization of the tissues and of the interaction. MRI is a non-invasive radiologic technique with notably high contrast for soft tissues and inherent sensitivity under certain protocols for visualization of thermal events. In this case, three types of imaging protocols (sequences) were used: a T1-weighted (T1) spin echo sequence, a T2-weighted (T2) spin echo sequence, and a T1-weighted spoiled gradient recalled echo (SPGR) sequence. Imaging was performed on a small bore, non-clinical imaging unit with a field strength of 2.35 Tesla (T) and a bore of 30 cm. The anatomy of the cortex and underlying thalamus were evident in coronal slices (each 3 mm thick) by all three sequences (though with different contrast for different structures). The intra-operative visualization of the freezing process was seen on T1 and SPGR imaging. The advance of the ice in tissue was seen as a black, hypointense zone in contrast to the adjacent normal tissue. This decrease in signal was due to the freeze which crystallizes the tissue and ‘zeros’ the relaxation time associated with the water protons allowing no observable signal in the imaging process. Post-operative contrast-enhanced T1 and T2 imaging post-operatively showed evidence of disruption of the BBB and vasculature as well as edema formation. Previous reports had presented similar results in providing preclinical evidence of MRI to monitor thermal therapies in tissue and provide feedback for the control.^39-42

A difficulty for the application of MRI-guided CITT has been the availability of suitable probes, probes which are MRI-compatible.^43,44 Notably, the probes used in the studies cited above were made of stainless steel or pyrex.^36-38 The stainless steel probes would create a substantial susceptability artifact in the MR images so as to obscure the anatomy. The glass pyrex probe would not have this problem, but would not be sufficiently robust for clinical use. Contemporary probes used in general surgery are stainless steel and are large (~ 4 mm diameter). Also, the clinical cryotherapy units are typically not compatible with the high magnetic fields. Cryotherapy devices have recently been adapted for use in the MRI environment using nitrogen or argon via the Joule-Thompson gas expansion effect for cooling. MR-compatible probes on the order of 1 - 2 mm have been manufactured for non-intracranial applications (prostate, liver). Also, a recent report in the German literature presented intra-operative imaging of in vivo experiments using a 2.7 mm diameter MR-compatible cryotherapy probe in normal porcine brain.⁴⁵

7.2

MRI-guided intracranial LITT

MRI-guided LITT of brain tissues has been studied and implemented clinically since first proposed.^39,46-51 Realtively deeply penetrating red and near infrared laser light can be transmitted down quartz optical fibers with diameters as small as 0.2 mm. The fibers are MRI-compatible and present no susceptability artifact in the images. The fibers can be manufactured 10-12 m long so that, even though the laser unit may not be in the MRI procedure room, the light can be transmitted to the surgical field. Thus, laser light was readiliy implemented into the MRI environment.

In current use, intracranial targets are tumors or vascular malformations ranging from 1 - 4 cm in diameter. In order to create a large enough volume of cell kill, various strategies can be used: single bare fibers (0.6 - 1.0 mm diameter) with repositioning as needed, multiple bare fibers, or diffusing tip fibers.

Important to the implementation of LITT has been the sensitivity of MRI to changes in temperature. We have used one of two types MRI techniques for thermal monitoring:^49-51

These MRI pulse sequences also provide images in a time frame suitable for the devloping lesion. The T1 FSE can acquire a full field-of-view image in 13 sec; the SPGR can do so in 5 sec. The latter pulse sequence provides data on the change in the phase of the chemical shift which is a temperature senstive parameter and from which temperature maps can be derived.^50,52,53 The greyscale MRI images are supplemented with computer processing to enhance the visualization of the information in the images.^49-51,54,55

While considerable clinical experience has been gained on conventional closed-bore MRI scanners. An essential development for our program was the introduction of an interventional ‘open’ MRI scanner.^56-58 Two key elements of such a system allow for direct access to the patient by health care personnel and dynamic control of the image plane by the physician. The latter assists in surgical planning and targetting of the tumor. Siefert and co-workers recently reported on neurosurgical applications of a similar system. In that report, they refer to having treated three patients with LITT with MRI-guidance.⁵⁹

7.3

MRI-guided intracranial RFITT

RF technology has been considered for use in the MRI environment. However, the technical challenges to making a device operate in the MRI environment have been substantial. The problems of RF interference with the MRI images and the materials choice for compatible probes is being solved at present and prototype systems exist. While the probes are designed currently for abdominal use, modifications for intracranial applications are possible as indications warrant.

While not used intra-operatively, MRI has contributed to RFITT in evaluating thermal lesions post-operatively. Tomlinson and co-workers imaged 21 patients who had undergone RF thalotomy for tremor control.⁶⁰ Single lesions were made with a 1.6 mm diameter electrode with a 3 mm active length. The tip had been heated to 78 °C for 60 sec; sometimes with a second ‘dose.’ Post-operative MRI assessed the lesion’s development over time. With both T1 and T2 imaging, “hemorrhagic necrosis” and edema were clearly identified and characterized the lesion. It also provided a measure by which to confirm the success of targeting the thalamus.

In a 1995 report, DeSalles and co-workers performed RFITT for thalotomy and pallidotomy in treating 11 patients for severe tremor, chronic pain, dystonia.⁶¹ They cited others who had used MRI to assess RF-induced lesions. They focused on the initial 72 hours post-therapy. MRI images were taken pre-operatively and the targeting of the lesion was done with a stereotactic frame and the MRI images. The surgery involved 4 +/- 2 RF lesions at 80 °C for 60 sec from a 2.5 mm long probe. They did T1, T2 and contrast-enhanced images. The lesion was evident immediately after therapy with variations in MRI appearance per the different sequences over time. In a study of MRI of RFITT-treated patients for pallidotomy, Krauss and co-workers observed 36 patients 1-3 days post-operatively with follow-up out to six months.⁶² Zones of effect were evident. MRI assessed the mean volume of the middle zone of tissue change which represented hemorrhagic necrosis at 44 mm³; if including the edemamatous response, this volume measurement extended to 262 mm³. MRI revealed additional edema spreading toward the internal capsule and optic tract. At 6 months, total lesions measured ~ 22 mm³. The report makes no correlation between lesion size & location and clinical outcome. The study did use the MR imaging to confirm lesion overlap with the GPi. Overall, the thermal lesions (guided to the target by physiologic response/monitoring) were entirely or mostly in the GPi.

7.4

MRI-guided intracranial FUS

There has been considerable effort in the use of FUS for non-invasive image-guided therapy. Current clinical tests are ongoing for use in the breast for both benign and malignant disease. As mentioned in section 7 above, there have been advances in the approach to targeting in the brain which compensate for the difficulties of sound transmission through the skull.³³ Further, in vivo studies in experimental animal models have been used to monitor the effects of FUS in normal brain tissues.^63,64 It is certainly appealing that a solution to the problem of focusing ultrasound through the skull would make this non-invasive therapy available for clinical research in human disease.