Toward Individualized Voxel-Level Dosimetry for Radiopharmaceutical Therapy

Radiation is used to treat approximately half of all patients with cancer, either alone or in combination with other treatment modalities such as surgery or chemotherapy.¹ The 3 primary modalities of radiation currently used are, in order of prevalence, external beam radiation therapy, brachytherapy, and radiopharmaceutical therapy (RPT). External beam radiation therapy primarily encompasses photons from radioactive sources or Bremstrahlung photons generated from a linear accelerator, electrons also generated from linear accelerators, and more recently particle therapy using hadrons (eg, neutrons, protons, and heavy ions) from accelerators. Brachytherapy involves the introduction of sealed radiation sources to the disease site by precise placement with applicators, and RPT involves a biochemical pathway to deliver a radionuclide to the site of cancer. RPT is also called targeted radionuclide therapy, endo-radiotherapy, or in vivo targeted radiation therapy.

RPT is a unique radiation therapy modality lying between external beam radiation therapy and chemotherapy. It can be administered with radiolabeled molecules or particles that may be associated with targeting agents such as antibody or antibody fragments, peptides, or low-molecular-weight ligands designed to deliver therapeutic doses of ionizing radiation to specific disease sites while minimizing damage to normal cells. RPT is based on the concept of delivering cytotoxic levels of radiation to disease sites with site-specific accumulation, targeting diseases at the cellular level with short-range biological effectiveness as opposed to the gross anatomic level (external beam).

In recent years, we have seen a tremendous acceleration in the development of therapeutic radiopharmaceuticals with a wide variety of radionuclides, adding to the treatment of diseases such as polycythemia, thyroid malignancies, metastatic bone pain, radiation synovectomy, hepatocellular carcinoma and other hepatic metastases, neuroendocrine tumors, non-Hodgkin lymphoma, and others.

Radionuclides used in RPT emit radiation either via alpha or beta decay with half-lives on the order of days. RPT is currently administered through intravenous or locoregional injection, with the amount of administered activity either fixed or based on the patient’s body weight or body surface area. Severe limitations to prescribing the dosage of pharmaceuticals based on patient weight or body surface prevent the realization RPT’s full potential. Inevitably, most patients are underdosed or overdosed, some quite dramatically. A growing body of evidence shows that fixed administered activity can result in a huge variability in the dose delivered to critical and dose-limiting organs (up to an order of magnitude). From external beam experience, tumor response to radiation therapy is typically sigmoidal, indicating that a threshold of absorbed dose must be obtained. An increase in administered activity of 50% to 200% depending on the patient, with a resulting increase in tumor dose, could vastly improve the efficacy of RPT.

The study of the dose distribution resulting from the systemic or locoregional administration of radiolabeled molecules is termed “dosimetry.” Dosimetry methods that provide dose estimates to individual patients rather than reference geometries are needed for assessment of RPT therapeutic efficacy and for adjustment of the RPT dose for maximum tumor control and minimal normal tissue toxicity.

More and more clinical trials with RPT are submitted and approved by the National Cancer Institute (NCI) clinical trial network (NCTN). Including accurate dosimetry in these trials becomes essential to mitigate related uncertainties,^2,3 enable unification and comparison across the 3 modalities of radiation therapy, and facilitate data collection on RPT dosimetry.

Dosimetry for RPT has evolved from a standard anatomic model-based approach that provides the mean absorbed dose in an organ to one that provides voxel-level absorbed dose distribution (in Gy). This type of voxel dosimetry is achieved with functional 3D imaging (positron emission tomography or single-photon emission computed tomography) combined with accurate structure delineation from anatomic images (computed tomography or magnetic resonance imaging) and advanced dose calculation algorithms (eg, Monte Carlo or dose point kernel). In addition, the use of imaging allows for individualization of the dosimetry. It could ultimately even incorporate radiobiological modeling to better predict biological outcome based on the absorbed dose distribution.^4–8 NCTN has initiated pilot studies within approved clinical trials to assess the feasibility of incorporating individualized RPT voxel dosimetry in future clinical trials and collecting data to evaluate dose uncertainties.

The perceived obstacles in wide adoption of high-quality individualized RPT dosimetry and treatment planning include the following: (1) cost and effort associated with nuclear imaging examinations necessary for dose calculation, image registration, and calculation of dose distribution or dose-volume histograms; (2) lack of availability of standard treatment planning methods and software systems; and (3) no well-documented correlations between the dose and patient outcome. To review state-of-the-art individualized RPT dosimetry and planning software systems, the NCI organized a workshop on RPT in conjunction with the NRG Oncology (Oncology Network Group) semiannual meeting on July 18, 2019. This workshop was a collaboration among the NCI, the Imaging and Radiation Oncology Core services (IROC, NCTN), and the Center for Innovation in Radiation Oncology (CIRO, NRG Oncology, NCTN).

The invited representatives from the dosimetry software industry and institutions included Voximetry Inc, Varian, MiM Software, Dosisoft, MIRADA Medical, HERMES, Phillips, PLANET Dose, VoxelDose, MIRDcalc, IDAC, and Rapid Inc. Workshop attendees included representatives from IROC, the RPT workgroup of CIRO/IROC, NRG Oncology, the Food and Drug Administration, and the American College of Radiology.

The extensive capabilities offered for individualized RPT dosimetry were presented by MIM Software, Mirada Medical, MIRDCalc free dosimetry software, ABX-CRO, Voximetry, and Rapid Inc. Organ-level, planar/3D hybrid, and 3D image-based dosimetry for multiple theranostic agents have been developed and some have received Food and Drug Administration clearance (eg, Y-90 and Ra-223). Dosimetric calculation methods used by these software programs rely on pharmacokinetic modeling, radiation transport algorithms, and Monte Carlo algorithms and leverage graphics processing units’ computing power for fast implementation.

Because image-based RPT dosimetry relies on quantitative nuclear imaging, camera-specific quantitative reconstruction for single-photon emission computed tomography/computed tomography is included. Tools include

• Automated target and normal volume generation for improved accuracy

• Deformable image registration

• Time activity curve fitting and integration

• 3D dosimetric evaluation functions for metrics such as dose-volume histogram and isodose displays

• Assessment and tracking of therapeutic response

• Functions for pretreatment, posttreatment workflows, and multidepartment collaboration

• Statistical analysis tools

• Services including imaging protocol development and associated phantom study

Early experiences in RPT individualized dosimetry showed that clinical implementation was often difficult owing to lack of familiarity with and trust in the techniques and results, much of it justified as dosimetry had very little standardization and a correspondingly high level of imprecision, often using planar imaging for quantification and an arbitrary number of activity time points. The lack of standardization is still an issue that requires greater attention; it is therefore imperative that community guidelines be created to avoid the adoption of suboptimal practices.

Because we are in the early phase of individualized RPT dosimetry, we have the great opportunity to offer technical guidance and standardization as we develop the pilot programs associated with NCTN clinical trials. These efforts are exceptionally important because these therapies are included in various future clinical trials and eventually will become a routine choice for certain types of disease.

The following critical questions were asked at the workshop:

• Should we use terminology such as “dosimetry,” which is more accurate than the widely accepted term “biomarker,” for radiology and/or nuclear medicine, and does it affect its adoption?

• Is it imperative to accumulate outcome evidence for dosimetry similar to the formats adopted by external beam radiation therapy, such as the Kaplan-Meier curve?

• What is the best way to introduce the concept of the dose-volume histogram to the nuclear medicine community not familiar with that reporting concept?

• Are the dosimetric quantities expressed in Gy equivalent from patient to patient, disease site to disease site, or nuclide to nuclide?

• What are the best ways to standardize quantitative RPT dosimetry, from calibration, phantom evaluation, reconstruction, postprocessing, and segmentation to dose calculation?

The assessments from the workshop confirm the appropriateness of the quality assurance approach to be adopted by IROC for upcoming pilot studies with individualized RPT dosimetry. The quality assurance will use IROC’s infrastructure and make the service operation support available to participating institutions. The approach will enable centralized quality assurance, centralized reconstruction, independent validation, and standardized image analysis and dosimetry methodologies.

The increased availability of radioisotopes with therapeutic and imaging potential will further lead the field of RPT. More precise image-based dose calculations⁹ and reduced uncertainties in the doses to targets and organs at risk will aid clinical trials integrating RPT dose escalation, combination therapy (eg, RPT combined with chemotherapeutic agents), and radiobiological treatment planning, with data collection for potential outcome correlation.¹⁰ These comprehensive clinical trials treating RPT as a synergy of biology, chemistry, physics, and medicine are the way forward to establish individualized dosimetry as an integral component of the treatment process. This will make RPT reach its potential as a powerful tool in cancer therapy.