LAUSR

Dear Editor, Treatment with alpha-emitting Ra-dichloride may be considered as a palliative treatment option in castrationresistant prostate cancer patients showing bone metastastic spread. In a small group of patients, Murray et al. recently showed that baseline F–fluoride uptake in bone metastases prior to Ra-dichloride treatment was significantly correlated with corresponding Ra uptake, associated Ra absorbed dose and subsequent lesion response to treatment [1]. Twenty-nine F–fluoride-positive bone lesions were investigated in repeated F–fluoride PET imaging (baseline, 6and 12-week time points), and a lesionby-lesion analysis provided each FLAB-derived volume (mL) and corresponding average SUV (SUVmean: g.mL). Percentage of injected activity was calculated as 100 × SUVmean × lesion volume/patient weight (W; g). Percentage change in SUVmean relative to the baseline PET allowed the authors to define the response to Radichloride therapy, and SUVmean measurement uncertainty (MU) at the lesion level by Lin et al. was used [2]. We would like to emphasize that the percentage of injected F–fluoride activity trapped in an arbitrary hottest volume of the whole metastatic spread (PPat), which is obtained at the patient level from F–fluoride PET volumetric data, might be used for predicting and assessing response to Radichloride treatment. In F–FDG PET imaging, we previously proposed to use an average SUV from volumetric data that was obtained by pooling several (N) hottest voxels possibly located in several separate places over the whole body (SUVmax-N) [3]. The SUVmax-N was subsequently applied to assessing treatment response in Takayasu disease [4]. It is characterized by a reduced MU because the larger the number of pooled hottest voxels N, the more reduced its MU. We therefore suggest that SUVmax-N is also suitable for F– fluoride PET-CT imaging in the current palliative framework of castration-resistant prostate cancer patients who usually show a large total bone metastatic volume (V = N × voxel volume). Furthermore, MU of the related PPat metrics (= 100 × SUVmax-N × V/W) is that of the SUVmax-N because (i) for an arbitrary value of N, and hence V, the same coefficient of variance (CV) applies to both SUVmax–N and SUVmax–N × V, (ii) there is no MU for W that is also involved in the SUV numerator and thus cancels [1]. Reproducibility percentages (R) for different values of N, i.e., the minimal relative change between two outcomes assessed from two different scans that is required to consider a significant difference, are presented in Table 1. These percentages were previously obtained in F–FDG PET-CT imaging performed in lung cancer and we assume they may apply to F–fluoride PET-CT imaging, because PET-CT imaging can identify neither the F chemical linkage nor disease type underlying F decay. As supporting evidence, a very close CV has been reported at the patient level for the maximal SUV (N = 1) in F–fluoride and F–FDG PET imaging: 12.0% versus 11.0%, respectively [2, 5]. * Eric Laffon [email protected]