Hi, my name is Manish Aghi from University of California, San Francisco. I'm going to be talking to you today about MRI-guided convection-enhanced delivery of therapeutics to glioblastoma. I have no disclosures. By way of overview, we'll start with an introduction describing the rationale for convection-enhanced delivery or CED as we call it. I'll then describe setup number one, which is using a ClearPoint MRI interventions platform for CED of therapeutics to brain tumors. These procedures occur purely in the MRI suite, and I will then give two examples of clinical trials that use that platform in our system. Setup number two is done using brain lab catheters, can be done in a conventional operating room followed by infusion in an MRI suite. I will also give an example of a clinical trial using that platform. The rationale for CED for GBM is that chemotherapy for brain tumors when given systemically is limited by compromised delivery across the blood-brain barrier and the ensuing systemic toxicity that occurs when a high dose is given systemically in order to achieve a therapeutic concentration in the central nervous system. For intratumoral drug delivery, CED offers several advantages over diffusion. For example, CED improves delivery of drugs by utilizing a pressure gradient to distribute macromolecules, achieving a larger distribution volume than what's achieved with diffusion alone on the order of centimeters rather than millimeters. Unfortunately, despite these advantages, prior clinical trials have reported failure of convection-enhanced delivery to meaningfully impact patient survival in glioblastoma. Most notably, a phase three randomized trial published a decade ago compared CED of interleukin-13 conjugated pseudomonas exotoxin to implantable gliadel wafers and revealed no clinical benefit. However, subsequent evaluation of the data revealed several limitations that we can use to build future CED platforms, including accuracy of catheter placement was an issue, suggesting a role for preoperative computer planning and better training of surgeons. The agent itself, the IL-13 receptor was found to be not ubiquitously expressed on the surface of glioblastoma and the durability of responses proved to be quite limited and suggested a need for a setup that would allow repeat treatment. Here you see an example proving that accuracy was an issue in which a patient with a tumor shown in orange underwent stereotactic implantation of a catheter with the intent to treat the orange tumor, but the drug agent in green was infused a few millimeters away from the intended target and so the real-time visualization revealed inaccurate placement of the catheter. As a result, there is a considerable rationale for real-time imaging during convection-enhanced delivery. Real-time CED is important because previous studies utilizing CED showed promised but limited activity, as I mentioned, due to several variables, including non-uniform distribution in cells nearest the injection site, high but transient exposure with corresponding toxicity and inability to confirm efficient intracranial drug exposure. Real-time CED of a drug mixed with gadolinium infused in an MRI suite provides us the ability to have guidance and confirmation of cannula placement and detection of reflux, leakage, hemorrhage, etc. with real-time adjustments then made to achieve accurate and comprehensive drug delivery throughout the tumor. The sequence of steps in this integrated CED platform for image-guided delivery of therapeutics to brain tumors is we start with a preoperative planning MRI done a few days before the intended treatment. We then do our software planning, typically the day before the treatment, in which we use computerized algorithms for shape-fitting and inverse planning to determine the optimal site for our catheters and the optimal number of catheters to achieve maximal tumor coverage. We then, having planned the infusion on the day of the procedure, will insert the cannula or cannulas and then perform the infusion in the MRI suite with real-time adjustments made to catheters to ensure that the delivery goes off as efficiently as intended. Here you see an example of the trajectory planning software, planning a trajectory to infuse a tumor near the anterior commissure in front of the third ventricle, a right frontal burl and catheter placement trajectory is prescribed that will obtain about 65% coverage of the tumor. And the decision is then made as to whether a second catheter at a slightly different angle would maximize that degree of infusion. So the first platform I wanted to describe comes from ClearPoint or MRI interventions. The patient is in an MRI suite throughout this infusion. We start by placing a grid to localize the entry point. And you see here that this patient had a prior craniotomy that is marked out. There are two frames. Historically, we used a skull-mounted frame that would screw directly into the bone. But because it had occupied sufficient real estate, it often required larger incisions and even opening old craniotomies to expose sufficient amount of bone. More recently, we have switched to a scalp-mount frame that screws directly through the skin and therefore requires minimal incisions, just a stab wound through the center of the frame through which the infusion occurs. We then mount a tower to the frame, whether it's the skull-mount frame as shown on top or the scalp-mount frame shown down below. The tower is connected through to a series of knobs that dangle outside the MRI machine and the software calculates adjustments in each direction which are made in order to get the tower angled so that it'll achieve the correct cannula angle into the tumor. Once that angle is achieved with the scalp-mount frame, we would make a stab wound in the center of the tower and then drill the burr hole through the tower at the correct trajectory to ensure that the catheter goes into the tumor at the correct trajectory. We then pass the catheter into the tumor at that trajectory and the catheter is connected. It's a rigid cannula that is connected to an infusion line. The infusion occurs at rates up to 50 microliters per minute. The first example of a treatment that was delivered through this Clairpointe MRI Interventions platform that I wanted to describe was actually TOCA511, a replicating retrovirus expressing yeast cytosine deaminase that converts the prodrug 5-fluorocytosine to the chemotherapy 5-fluorouracil. This agent has gone on to be the subject of a negative randomized phase 3 study, but nonetheless was used to treat via convection enhanced delivery 18 patients followed by 5-fluorocytosine treatment. You can see that with the infusions, we were able to make real-time adjustments. Here you can see a patient who received most of the initial infusion into the top of the half of the tumor. The catheter was then advanced into the bottom half of the tumor in real time. An additional drug was infused to cover the inferior aspect of the tumor. From these studies, we can calculate a volume of infusion and a volume of distribution, VD and VI. The VD to VI ratio is a way of measuring the comprehensiveness of the infusion. A high VD to VI ratio shows a successful infusion. There was an example of a patient shown here who experienced both radiographic and clinical improvement. The patient presented with impaired dexterity and ability to play the piano due to some left-hand weakness. Infusion of a tumor in the area of the right motor strip corresponding to hand control led to some necrosis and improved manual dexterity in the patient's left hand. 18 patients were treated using convection enhanced delivery of TOCA511 in that study. The average tumor volume was 10.5 cubic centimeters, a range of 2.6 to 25 cubic centimeters, so nearly a 10-fold range in tumor volume. Initial patients had poor coverage of tumor, but as the technique improved, we got up to 70% tumor volume coverage with an average of 45%. The flow rates of 50 microliters per minute were tolerated well without reflux. Patients received three milliliters delivered to one to three targets during these infusions. The VD to VI ratios, which I mentioned earlier, are an important parameter, averaged 1.6 with a range of 1 to 3.5.

MRI-guided convection-enhanced delivery

therapeutics

glioblastoma

blood-brain barrier

systemic toxicity