The National Institute of Biomedical Imaging and Bioengineering (NIBIB), the National Cancer Institute (NCI), and SNMMI organized a virtual workshop titled “Engineering New Instrumentation for Imaging Unsealed Source Radiotherapy Agents” on August 16 and 17, 2021. The impetus for this workshop was introduced by an earlier Newsline article, “Time for a next-generation nuclear medicine γ camera?” (2020;61[7]:16N), asking whether we need to reconsider instrumentation used for theranostic methods with electron- and α-emitting unsealed sources for recently emerging cancer therapies.
The workshop convened physicians and scientists from relevant fields to investigate the clinical challenges of treating cancer and to discuss possible technical developments in imaging instrumentation for improving outcomes. Panel sessions focused on the clinical applications of radiopharmaceutical therapy (RPT), as well as isotope production and dosimetry’s roles in delivering safe and effective therapies. The workshop was moderated by George Zubal, PhD, Program Director for Nuclear Medicine (NIBIB), and Jacek Capala, PhD, DSc, Program Director, Radiation Research Program, Division of Cancer Treatment and Diagnosis (NCI).
Kris Kandarpa, MD, PhD, Director of NIBIB’s Research Sciences and Strategic Directions, opened the workshop with a welcome message outlining NIBIB’s mission to develop imaging methods that lead to personalized precision medicine. He encouraged participants to evaluate recent achievements in PET and consider how similar future improvements could be achieved with α-emitter imaging.
Session 1: Overview of Diagnostic/Therapy Practice
The first workshop session started with a talk by Steven Larson, MD (Memorial Sloan Kettering Cancer Center; New York, NY), describing the therapeutic advantages of α-emitting isotopes. Alpha particles are characterized by double-strand break triggering, high linear energy transfer, relatively short range, greater relative biological effectiveness, and a low oxygen enhancement ratio. Dr. Larson and his team have demonstrated 3 separate therapy protocols that work quite well, with good therapeutic indices associated with cures. Reporting from the same institution, John Humm, PhD, covered specific classes of α-emitting radionuclides, which can be simple (those that decay to stable or non–α-emitting progeny) or complex (those with radioactive α-emitting progeny). He pointed to 211At as possibly the best α emitter that does not exhibit problematic progeny. This has great potential if concerns about its challenging radiochemistry can be resolved. He emphasized dosimetry problems with imaging α-emitting radionuclides, related to the limited resolution of current SPECT cameras, where images of α sources cannot provide accurate information at the relevant cellular target level for microdosimetry.
Daniel Pryma, MD (University of Pennsylvania; Philadelphia) explained current RPT practices using 131I-MIBG, which was developed in 1980 and received therapeutic FDA approval in 2018. Several dosimetric challenges are associated with these studies, including the fact that dosimetry image acquisition requires several visits to the imaging department. Work is being done to use population inputs and information about typical patient kinetics to simplify this process. Michael King, PhD (University of Massachusetts Medical School; Worcester), reviewed manufacturers who have developed dedicated cardiac SPECT systems. These have smaller heads to get closer to patients or 2 heads oriented at 90° for efficient acquisition over 180°. Other advancements (new radiopharmaceutical developments, imaging system design optimization, software advances, and guidelines/standards) facilitated by medical societies have played a role in establishing and enhancing the clinical utility of cardiac SPECT. The hope is that SPECT will also play a role in similar developments for unsealed source radiotherapy agents. The last speaker in this session, Robert Mach, PhD (University of Pennsylvania; Philadelphia), detailed his research findings indicating that SPECT is comparable to PET in studies with high target density or in studies with lower target density in a control group. He further noted that SPECT can be used to separate the photopeaks between 123I and 99mTc to quantify uptake of the 2 isotopes and measure both terminal density and cerebral blood flow. This could be applied to imaging parent and progeny isotopes in RTP, which is not possible with PET because of the 511-keV emissions of all PET isotopes.
Panel Discussion: Sarah Cheal, PhD (Memorial Sloan Kettering; New York, NY), Robert Miyaoka, PhD (University of Washington School of Medicine; Seattle), Emilie Roncali, PhD (University of California Davis), and Vikram Bhadrasain, MD (NCI) participated in a discussion and Q&A with session presenters. Several interesting topics were raised, including 213Bi dosimetry, comparison to external-beam doses, and development of cameras for specific tasks. Other topics included antibodies in theranostics and sensitivity of SPECT for dosimetry imaging. This led to a discussion of possible new γ detectors and collimator-less cameras. Concluding comments addressed the SNMMI Dosimetry Challenge and ways in which customizing doses to specific patients could improve outcomes.
Session 2: Overview of Imaging
The next workshop session began with an overview from Ben Tsui, PhD (Johns Hopkins University School of Medicine; Baltimore, MD), who summarized SPECT development over the past decades. He noted that 2 major breakthroughs have made quantitative SPECT more practical: maximum-likelihood expectation maximization and ordered-subset expectation maximization algorithms. In addition, quantitative SPECT has several important implementation requirements, including good quality of SPECT images, good quality of CT images, and accurate alignment of SPECT and CT images to reduce misregistration of image artifacts and to apply attenuation corrections. Also from Johns Hopkins, Eric Frey, PhD, reminded attendees that, when conducting targeted radionuclide therapy, the main objectives are to avoid significant toxicity in normal tissues and to deliver a lethal dose to tumors; hence, it is necessary to image both large and small organs to obtain accurate voxels for 3D dosimetry. He explained issues that must be addressed to achieve these improvements: new detector materials with improved energy resolution and better intrinsic spatial resolution, novel collimation geometries, improved intrinsic resolution, and more detector areas with larger axial fields of view.
The third speaker in the session, Ling-Jian Meng, PhD (University of Illinois at Urbana-Champaign), reviewed a proposed camera design applying the concept of hyperspectral SPECT imaging, which could allow multiisotopic, multifunctional molecular imaging using various combinations of radiotracers. Based on his innovative sensor work, his team has been developing preclinical and clinical imaging systems that show promise for routine imaging in humans. He concluded that, given improvements in sensor spatial resolution and sensitivity, it may be time to revisit Compton cameras as a possibility to improve SPECT imaging. Todd Peterson, PhD (Vanderbilt University; Nashville, TN), presented his unique camera design integrating a high-purity germanium detector camera with a MicroCAT II CT scanner. Using this system, he was able to demonstrate multiisotope capabilities and compared his camera to other systems, demonstrating that germanium could set very narrow photopeak energy windows. His team is also working on mechanical cooling to reduce power consumption. Lars Furenlid, PhD (University of Arizona; Tucson), oversees a lab that develops technologies to image γ rays, principally for SPECT, which are either semiconductor- or scintillator-based. His current designs include a third-generation cross-strip cadmium telluride detector, a hybrid photomultiplier tube/silicon photomultiplier (SiPM) scintillation camera, and a third-generation large-area ionizing radiation quantum imaging detector camera. He summarized the many opportunities for fundamental advancements in SPECT imaging of α emitters, noting that development is needed in the theory of mathematics as well as new mathematical observers, estimation methods, and spectrum-aware reconstruction methods. Also needed are larger area high-Z semiconductors; high-Z, high-light-output scintillators; large-area gaseous or solid-state electron multipliers; and advances in SiPMs.
Panel Discussion: The panel discussion for this session focused on topics concerning germanium detectors for scatter rejection, higher sensitivity imaging systems, and associated reconstruction methods. Additional discussions covered microdosimetry and high-energy photon imaging with camera geometries positioned close to the patient.
Session 3: Overview of Isotopes: Dosimetry and Future Directions
Session 3 opened with an overview from Jehanne Gillo, PhD (U.S. Department of Energy [DOE]; Germantown, MD), of the DOE Isotope Program (DOE-IP), which has a mission to produce and distribute radioisotopes that are not commercially available. She described a dramatic increase in the numbers of reactor- and accelerator-based isotope production facilities at DOE National Laboratories and their academic partners, providing diagnostic and therapeutic radioisotopes, including α emitters (223Ra, 221At, and 225Ac, as well as 227Ac used by Bayer to obtain 223Ra for production of Xofigo). In addition to production and distribution of currently used isotopes, the DOE-IP also operates a discovery arm focused on identifying new radioisotopes that might be of interest to the RPT community. A funding opportunity announcement to support research on new isotopes entering preclinical and clinical trials was recently issued (https://www.isotopes.gov/FOA-Advancing-Novel-Medical-Isotopes-for-Clinical-Trials). DOE has also started an isotope traineeship to apply advanced manufacturing techniques to isotope production.
Douglas Van Nostrand, MD (Georgetown University Medical Center; Washington, DC), provided a comprehensive overview of radioiodine as a paradigm of theranostics. He pointed out its diagnostic (123I and 124I) and therapeutic (131I) utility, facilitating: (1) definition of maximal safe administered therapeutic activity to minimize unacceptable side effects; (2) determination of minimal administered activity to achieve desired therapeutic outcomes; and (3) assessment and mitigation of altered genomic cancer molecular biology (redifferentiation). These permit 131I therapy in patients with negative scans and enhance therapeutic results in patients with positive scans. Because MEK-inhibitors increase iodine accumulation in tumor cells (which can be monitored by PET) the 124I/131I theranostic pair allows successful treatment of non–iodine-avid tumors.
In contrast to external-beam radiotherapy, where the absorbed dose can be precisely inferred from measurements, tissue absorbed dose in nuclear medicine must be approximated using different models. Wesley Bolch, PhD (University of Florida; Gainesville), presented 3 principal methods to compute tissue absorbed dose: (1) direct Monte Carlo (MC) radiation transport simulation; (2) dose-point kernel (DPK) convolution; and (3) the Medical Internal Radiation Dose (MIRD) S-value formalism. Because of the high degree of accuracy, MC simulations are the reference standard for tissue dosimetry and the most reliable tool for computing radionuclide S-values. DPK convolution is commonly used at the voxel level, between the application regimes of S-value and direct MC methods. S-values, the most practical method because of the link to the MIRD schema, are applicable at any scale, although the underlying approximations of the method mostly limit their use to the organ and suborgan levels.
Yuni Dewaraja, PhD (University of Michigan; Ann Arbor), presented advances in SPECT and PET imaging for patient-specific dosimetry, focusing on imaging methods used for 90Y and 177Lu. 90Y can be imaged by both SPECT and PET. SPECT detects bremsstrahlung that has continuous energy spectra. MC reconstruction, model-based scatter estimation, and deep learning–based scatter estimation are used to enable quantification of the bremsstrahlung. The main challenges in PET imaging of 90Y are low positron yield and coincidences with bremsstrahlung photons. These challenges can be addressed by dedicated reconstruction algorithms and new instrumentation, such as time-of-flight, digital, and whole-body PET. A relatively low intensity of γ rays is a challenge for SPECT imaging of 177Lu, which requires efficient counting methods, application of deep learning, and joint dual-photopeak reconstruction or the option of combining SPECT with data from 68Ga PET.
Panel Discussion: The panelists addressed the limitations of current dosimetry methods and the latest progress in development. Uncertainty in dose estimates was deemed the major problem. Contributing factors, particularly for bone marrow and small structures (metastases), include definition of the region of interest, low signal, and reliability of data obtained using standard partial-volume correction methods. Several methods to improve the reliability of dose estimates and the need to define standards for the whole dosimetry workflow were mentioned. Barriers to wider adoption of radioiodine treatment and challenges and opportunities in combining RPT with conventional radiation therapy were also discussed.
Keynote Overview: Peering into the Black Box of Cancer Therapy
In the keynote lecture, George Sgouros, PhD (Johns Hopkins University School of Medicine; Baltimore, MD) addressed RPT efforts in the context of conventional systemic cancer therapies. He described the latter using a black box analogy, in which inputs are mechanism, target validation, preclinical model toxicity, patient selection, genomics, and theranostics. Treatment is the black box itself, and outputs are tumor response, time to progression, overall survival, quality of life, and clinical toxicity. In this scenario, researchers can understand mechanisms by changing inputs and looking at responses; this is the long-standing process for agents that cannot be imaged. This approach, unfortunately, is not effective. In a paper published in 2018, Wong et al. examined the success rate of oncologic drugs and found that 97% of cancer drugs evaluated in humans fail (Biostatistics. 2018;20:273–286). Many of these agents are targeted therapies blocking signaling pathways that control cancer cell growth, division, and spread. In many cases, targeted therapies miss their target; a 2019 paper by Lin et al. found that many cancer drugs work as a result of off-target effects (Sci Transl Med. 2019;11[509]:eaaw8412). Consequently, over the past decade the cost of cancer drugs has gone up, but the clinical benefits of those drugs have not increased proportionately, adding to financial concerns.
RPTs present a promising alternative. They are administered systemically and regionally and can target metastatic cancer, leading to radiation-induced DNA damage and killing cells rather than controlling cell behavior. Their efficacy depends on differential delivery of radiation, which can be assessed by imaging. Dr. Sgouros presented several examples of such approaches in clinical trials using 131I, 90Y, 213Bi, 223Ra, 227Th, 212Pb, and 225Ac. In 1 example, implementation of personalized dosimetry in hepatic artery infusion of 90Y microspheres for hepatocellular carcinoma doubled patient survival time, without changing the agent or the patient population. These examples showed that imaging and individualized dosimetry-based treatment planning can further improve RPT outcomes.
Keynote Overview Q&A: The panel discussion focused on obstacles to implementing dosimetry for RPT, including: a need for more examples (preferably randomized clinical trials) showing that dosimetry has a huge impact; a need for consistent, well-validated, and standardized dosimetry methodologies; lack of knowledge of radiation and radionuclide therapy; reimbursement challenges; and the need for multiple scans. Panelists noted that some of these problems can be addressed by simplifying dosimetry procedures and enhancing education of both patients and physicians.
Conclusion
This workshop represented a first step in evaluating and combining physicians’ needs for cancer treatment and imaging scientists’ knowledge of instrumentation and dosimetry calculations to improve cancer treatment outcomes. We invite the community to view the recorded sessions of this workshop under “Engineering New Instrumentation for Imaging Unsealed Source Radiotherapy Agents” at the NIBIB events page (https://www.nibib.nih.gov/NIBIB-Webinars-and-Conferences). NIBIB looks forward to continuing this workshop in early 2023 with additional discussions and a review of progress made based on insights and discussions from the workshop reported here.
- © 2022 by the Society of Nuclear Medicine and Molecular Imaging.