Radiation Dosimetry of Whole-Body Dual-Tracer ¹⁸F-FDG and ¹¹C-Acetate PET/CT for Hepatocellular Carcinoma

DISCUSSION

Although risk from radiation exposure to the individual may be of less impact for the population known to have cancer because of their reduced life expectancy (23), the information is still of importance to the evaluation of risk/benefit for justification, and relevant to protocol optimization and personnel protection (24). The use of ¹⁸F-FDG and ¹¹C-acetate for the evaluation of patients with HCC is known to provide complementary biochemical sensitivities to identify tumor cells with various degrees of differentiation. Currently, to our best knowledge, we are not aware of published studies formally assessing radiation-absorbed dose from whole-body dual-tracer PET/CT to HCC patients. Our study assessed the dual-tracer PET/CT-based radiation dosimetry of ¹⁸F-FDG and ¹¹C-acetate for all organs based on patient-specific data and the latest recommendation from ICRP publications. The derived overall effective dose from dual-tracer PET/CT scans with a mean administered ¹⁸F-FDG and ¹¹C-acetate activity of 328 and 481 MBq, respectively, was calculated to be up to about 29 mSv, and up to about 57 mSv if a 1-stop triple-phase contrast-enhanced abdomen CT was included at the same setting. These results assist in evaluating justification and optimization of protection of the procedure.

For internal absorbed dose assessment, OLINDA/EXM was used to perform dosimetry calculations for the various body organs, which is currently unique in the field of nuclear medicine dosimetry in that it has received approval from the Food and Drug Administration to be distributed after 510(k) premarket notification (15). This code allows calculations for 814 radionuclides and a wide variety of adult, pediatric, and pregnant female phantoms; furthermore, it also allows users to modify organ masses in the phantoms for more patient-specific dose calculations when it may be known that an organ is larger or smaller than that assumed in the reference phantom. Moreover, organ doses calculated by OLINDA/EXM based on reference phantoms representing the average patient were found to be in good agreement with patient-specific Monte Carlo mean dose estimates (25). We also compared organ equivalent dose and effective dose calculated by OLINDA/EXM with those by the dose coefficients from ICRP publication 106 (19). The dose estimate difference between these 2 methods may be due to the dose coefficients provided in publication 106, which are computed by age- and sex-averaging, and moreover ICRP publication 106 still uses tissue-weighting factors defined in ICRP publication 60 (26), instead of the latest ICRP publication 103 (3) applied in the present study. Patient-specific organ masses were derived from previous reports on variations in the mass of different body organs in relation to stature and BMI (16–18,27,28). Marine et al. and Clark et al. (16,17) described phantoms that model different body types in a series of percentile height phantoms to evaluate how specific absorbed fractions may vary with height and weight differences across the human population. As Stabin stated (29), the biokinetic model used to calculate the dose is one of the major uncertainties in the evaluation of radiation doses for radiopharmaceuticals. If careful patient-specific dosimetry is performed, with attention paid to accurate measurement of individual organ volumes, many of the biokinetic model uncertainties can be minimized, and the total uncertainty in the individual dose estimate can be reduced to perhaps ±10%–20% (29).

With patient-specific data, the effective dose from ¹⁸F-FDG PET scanning was 6.08 mSv for male and 6.62 mSv for female patients, respectively. These values are consistent with the dose estimates reported in the literature (30–33) and also with the dose estimates from ICRP publication 106 (19). The measured radiation-absorbed doses from ¹¹C-acetate PET scanning differ markedly from the previously determined estimates of Seltzer et al. (11). Because the dosimetry methodologies used by us and by Seltzer et al. (11) were different, the difference between the dose estimates obtained is mainly due to the difference in the biokinetic data used to create the kinetic models. Because the kinetic data that we used were from ICRP publication 106 (19), which are more complete, representing all major source organs, more accurate dosimetry estimates are likely possible. We, therefore, believe our estimates of internal doses to be reasonably accurate and to represent the best estimates of such internal absorbed doses for our HCC patient population based on the models used.

Currently, a few CT dose-calculation tools, including ImPACT Dose (http://www.impactscan.org/ctdosimetry.htm) (34), CT-Expo (Georg Stamm, http://www.sascrad.com/attachments/File/Leaflet_CT-Expo_v2_1(E).pdf) (35), eXposureTM (http://www.radiologysolutions.bayer.com/products/ct-dosemanagement/rep/), and methodologies (36), report organ doses based on their calculations of simplified stylized anatomic models that are anatomically crude but widely used for several decades for practical applications with the standard mathematic representations of the reference man (37) and other representative phantoms in radiation protection, nuclear medicine, and medical imaging (38,39). Recently, more realistic models, developed by Xu et al. (40), based on imaging data from human subjects, including updated anatomic information from reference data on adults and children (18), can be used in place of traditional stylized models for more accurate dose estimates (41).

The use of VirtualDose improves on existing dose calculation tools by considering the latest CT scanners based on the newer realistic models developed by Xu et al. (40): 25 available models, including adult male and female models of various body mass, pediatric models from newborn through adolescent, and pregnant models of 3 gestational ages. Standard CT models have been modeled and patient dose calculated using Monte Carlo radiation transport methods, with corrections applied to represent the behavior of most commonly available CT scanners. The Monte Carlo results suggest that the organ equivalent dose estimates can be different by a margin as great as 277% between our calculations and those derived from the use of earlier stylized MIRD-type phantoms (42). VirtualDose enabled us to assess organ equivalent dose under TCM schemes, which has now been frequently implemented in clinical body CT examinations for a decade. Quantifying the magnitude of organ dose under TCM, however, is practically challenging. The main challenge relates to the fact that the x-ray radiation is dynamically altered over the patient habitus for TCM examinations. VirtualDose provides accurate estimates of organ equivalent dose, based on modeling change of irradiation condition and integrating it with the anatomic feature of the patients. Therefore, the use of this dose tool is expected to improve both the accuracy and the usability in reporting CT doses.

The limited whole-body CT effective dose to reference phantoms determined by k factor (0.0019 for chest-abdomen-pelvis, the ICRP publication 110) was overestimated in the present study compared with the patient-specific effective dose calculated by VirtualDose. This finding is not surprising because k factor is an age- and sex-averaged parameter, based on data averaged over many scanner makes and models; moreover, effective dose is defined by ICRP as a single parameter to reflect overall risk averaged over all ages and both sexes for a reference patient (3) and is not a physical parameter that can be measured, which is always computed through multiple steps and approximations. OLINDA/EXM and VirtualDose were used in the present study to calculate organ equivalent doses, and then effective doses were calculated using ICRP publication 103 tissue-weighting coefficients (3).

The present study is limited because it was performed for a small sample size cohort (n = 14) that was recruited consecutively in the first 4 mo of the implementation of such scans for a specific population with HCC in a single center. This limitation should be considered in the generalization of the conclusions. However, as noted above, a review of our results against literature values shows general consistency.