Choosing the Right Metrics for Evaluation of Radiopharmaceutical Therapy Dosimetry Methodologies

In this issue of The Journal of Nuclear Medicine, Chicheportiche et al. (1) compare radiation dosimetry calculated from single-time-point (STP) imaging to that obtained with multiple-time-point (MTP) imaging and conclude that they are highly correlated and that, therefore, the simpler method can be used “with confidence.” The justifications underpinning their desire to simplify dosimetry image acquisition are valid, namely the resource intensiveness of MTP dosimetry and the potential inconvenience and discomfort for patients. Conversely, they highlight the improved efficiency and patient experience by moving to STP. Currently in most nuclear medicine departments, scanner and staff availability are scaled to diagnostic imaging volumes, with little incremental bandwidth for time-intensive dosimetry acquisitions. Additionally, returning over several days for multiple imaging sessions is not a common patient experience in radiology or nuclear medicine. Thus, we are currently in a self-fulfilling prophecy wherein dosimetry is not widely used because of the complexity, and because it is not widely used, it is thought to be not particularly impactful. Studies exploring ways to simplify and streamline the dosimetry process and improve the patient experience are important for increasing the likelihood that dosimetry will be adopted by more institutions for more therapeutic procedures. However, it is of vital importance to be clear on the intended use of dosimetry and to ensure that the testing methodology matches the use case. Radiopharmaceutical dosimetry can be used to estimate average radiation exposure across a population (common with diagnostic radiopharmaceuticals), in which situation individual outliers are of relatively little consequence and comparing the average of results from one method to another is perfectly adequate and appropriate. For radiopharmaceutical therapy (RPT), dosimetry could be applied to the individual patient to make treatment decisions, and in this case, outlier or discordant measurements can have considerable impact, potentially even leading to life-threatening or fatal adverse events.

Although the work by Chicheportiche et al. is interesting, rigorous, and compelling, it is important to underscore that it shows that the 2 methods get the same answer on average. Because the error bars are still relatively wide, the answer can be discrepant or inaccurate in any individual. In external-beam radiotherapy, the expectation is that absorbed dose estimates are accurate to within 5% or better (2). That is, every individual measurement—not the average of the measurements across a population—is accurate to within 5%. These treatments, whether external beam or radiopharmaceutical, have a relatively narrow therapeutic window, and inaccurate dosimetry puts patients at risk for serious side effects if overdosed and at risk for ineffective or suboptimally effective treatment if underdosed. Therefore, to be used to make clinical decisions in individual patients, dosimetry needs to be both accurate and precise. Unfortunately, even for MTP dosimetry many studies highlight that the uncertainty is likely more than 5% (3–5); therefore, even if MTP and STP agree perfectly, it is not clear whether the answer has sufficient precision to alter clinical decisions based on the results. So, it is difficult to make the claim that STP dosimetry data can be used to make treatment safer or more effective when the comparator MTP dosimetry results are unlikely sufficiently precise for that task. Even when provided with the same data, users achieve results outside the 5% threshold when performing MTP dosimetry; greater standardization and rigor are therefore required (6).

In the article, the authors indicate an expectation that STP dosimetry will result in the same clinical choice as MTP dosimetry in at least 90% of patients. This likely falls short of what is needed for adoption as a standard patient management strategy (assuming for the moment that MTP dosimetry could or should be used to make clinical choices). The outliers are potentially very important. If up to 10% of the patient population will be “mismanaged” with STP dosimetry, it would be helpful to understand whether the patients in whom STP will be less reliable (or come to a different answer from MTP) can be identified and undergo MTP instead. The authors indicate that in some cases STP dosimetry significantly underestimated the absorbed dose compared with MTP dosimetry; thus, using a threshold for organ-at-risk dose to flag potentially discordant STP dosimetry results may not be adequate for some of the population. As the authors point out, this concern likely is not at issue for the currently accepted toxicity thresholds but may be more significant should those toxicity thresholds be raised. Said another way, these discordances do not matter today for ¹⁷⁷Lu-peptide receptor radionuclide therapy because we likely underdose a majority of patients. Indeed, especially for the current kidney dose threshold derived from external-beam radiotherapy (7)—a threshold that appears to substantially underestimate the renal tolerance to RPT (and in particular RPT given over multiple cycles) (8)—a spuriously low STP dosimetry result is unlikely to result in patient harm. However, if we were to incorporate dosimetry into dose selection (whether administered activity per treatment or number of cycles per patient) along with a future empirically derived organ limit, these underestimates could be quite clinically significant.

As the authors discuss, because there is currently very little renal toxicity resulting from clinical ¹⁷⁷Lu-peptide receptor radionuclide therapy, dosimetry may not be an important patient management tool in the current treatment setting. However, increased high-quality dosimetry within the field as a whole will allow for a better understanding of absorbed doses across larger populations and for the development of toxicity thresholds based on absorbed dose from RPT, rather than trying to adapt toxicity thresholds developed from fractionated external-beam radiation therapy. If there is a better understanding of the true absorbed dose levels at which we can expect a 5% or 15% complication rate, this would allow for potential dose escalation, which may increase the likelihood of a satisfactory therapeutic response. As a corollary to gaining a better understanding of the organ-at-risk toxicity thresholds, high-quality dosimetry on larger patient populations will allow for a better assessment of absorbed doses to tumors and, potentially, the development of expected therapeutic response as a function of RPT absorbed dose. This may allow for dosimetry to be used not only to assess the likelihood of complications due to RPT but also to determine an expected number of cycles to get the desired response. It could help us move away from empiric dosing and toward individualized treatment planning. At first glance, one may infer that at least at the population level, STP could be used for this purpose, but this use would be valid only if toxicity or tumor response is also aggregated at a population level to mitigate the effects of outlier measurements. Ideally, future development of normal-tissue complication probability curves and tumor control probability curves will rely on high-quality dosimetry data that are unlikely to be obtained from STP methodologies (or maybe even from SPECT/CT dosimetry in general).

Finally, the idea that MTP is too inconvenient or uncomfortable for a patient should be revisited. In radiation oncology, patients generally undergo several diagnostic studies before their decision to receive external-beam radiotherapy. They get at least a CT simulation as part of the treatment planning process and potentially additional PET, MRI, or ultrasound studies over one or more days. Then, they return several days or weeks later for daily treatments that can last for 8 wk or longer. This is certainly inconvenient for the patient but important for their anticipated successful outcome (and less inconvenient than cancer death). If RPT were approached similarly to fractionated external-beam therapy in radiation oncology, where the patient’s expectations are set from the initial discussions about the treatment planning and delivery process, MTP dosimetry may not be viewed as an inconvenience but as an important part of the treatment planning and administration process that is integral to ensuring the best possible outcome. Improving scanning and reconstruction technologies to reduce the imaging time required while maintaining high-quality image data will be important for improving the patient experience.

Certainly, the issues around dosimetry imaging acquisition are complex and include reimbursement, standardization, reproducibility, and accessibility; they will need to be addressed. We will then need evidence that patient-specific dosing, informed by dosimetry, is superior to population-level dose administration. To realize that goal, RPT dosimetry must be accurate and precise not only across a population but also at the individual level. The simpler the better, but simplification cannot be achieved at the expense of reliability in a significant subset of patients.

• © 2023 by the Society of Nuclear Medicine and Molecular Imaging.

• Received for publication August 22, 2023.
• Revision received August 28, 2023.