Radiopharmaceutical therapy (RPT), an alternative to chemotherapy, has worked well in patients with lymphoma, late-stage, metastatic prostate cancer, and neuroendocrine tumors. It is effective at delivering pinpoint radioactivity specifically to metastatic tumor cells distributed throughout the body. Patients who are treated with RPT agents typically receive the same amount of radioactivity even though the unique physiology of each patient impacts biodistribution of the radioactive drug over time and can affect treatment outcome. Alternatively, by imaging the radiation emitted by the RPT agent within the body, it is possible to calculate how much radiation energy is deposited in tumors and normal tissues within an individual patient (?dosimetry?). This information affords personalized medicine because the amount of radioactivity can be adjusted to avoid underdosing (not enough tumor radiation to kill the tumor) or overdosing (too much radiation to normal tissue that leads to side effects) the patient. From experience with external beam radiation therapy (EBRT), we know that patient-specific prescriptions based on absorbed dose (treatment planning) lead to better patient outcomes. Like EBRT, patient-specific treatment planning for RPT requires sophisticated dosimetry tools that Voximetry Inc (?Vox?) has developed. As part of a previous Phase I SBIR grant, Vox has developed a Monte Carlo dosimetry algorithm which leverages the enormous computing power of graphics processing units (GPUs) to perform voxel-based dosimetry. Our approach will make treatment planning faster and more accurate, so that it can be used clinically to compute patient-specific dosimetry within minutes as opposed to tens of hours required on central processing units (CPUs). Vox will ultimately benefit cancer patients by making available a personalized treatment that targets metastatic cancer that in many cases is more efficacious and has fewer side effects than chemotherapy. In this proposal, we aim to integrate our fully benchmarked and IP- protected dosimetry algorithm into an automated, cost-effective RPT treatment planning solution, Torch, by adding additional features such as image registration, contour propagation, and voxel-based pharmacokinetic (PK) modeling. Torch will not only be the most accurate product on the market, it will be 1/3 of the cost of competitors? offerings.
The specific aims that will be accomplished in the proposal are to (1) develop GPU- accelerated deformable image registration and contour propagation within the Torch workflow, (2) develop GPU-accelerated pharmacokinetic modeling for voxel-level time activity curve integration, and (3) validate Torch through beta testing using computational phantoms and patient data. The successful completion of these aims will support a commercially viable product that is ready for clinical use. This product will be proven safe and effective in a retrospective clinical trial which will be followed by a 510(k) application to the FDA.
Millions of patients are treated with radiopharmaceutical therapy each year. Unfortunately, current prescription methods do not consider absorbed dose to individual patients but instead use empirically derived or standard methods leading to suboptimal treatment outcomes in some patients. Our solution, Torch, will be the first commercialized Monte Carlo radionuclide dosimetry software that will permit patient-specific treatment planning.
