Thesis

176 Part II Chapter 7 Dear Sir, We would like to respond to the interesting Letter to the Editor by colleagues Duarte and Sapienza [1], commenting on our recent original article [2]. To start, we would like to thank the authors to point out a method to determine healthy bone (HB) on Na18F-PET/CT that is biologically plausible, as indeed by far the highest amount of fluoride after intravenous injection first resides in erythrocytes and subsequently concentrates in the bone and not in soft tissue (with non-absorbed fluoride being excreted in the urine). It is, therefore, at least theoretically, an attractive option to use skeletal volume (SV) as a measure for the volume of distribution of Na18F, which could be superior to methods normalizing for either body weight (BW) or lean body mass (LBM). This method might be feasible, despite the fact that it introduces a new factor, measuring bone volume by segmentation of the CT, based on Hounsfield units (HU) thresholds. Preceding our measurements and analyses, we performed a structured PubMed search of existing literature. As reviewed in our work, the papers that we found used either max or mean SUV (normalized by BW) or SUL (normalized by LBM) to quantify (healthy) bone uptake [3, 4]. We specifically performed a PubMed search on normalization procedures, but this unfortunately did not result in work of the authors [5, 6], nor other articles that mentioned this technique to normalize for volume of distribution of Na18F. In hindsight our search should have been broader as replacing ‘normalisation’ in our search with ‘normalization’ indeed led to the articles referred to by the authors. We were therefore unaware of this method at time of our submission. We swiftly acquainted ourselves with the method presented by the authors, as we already observed and reported slightly suboptimal normalization of our HB SUVs as these were borderline positively correlated to the normalization factor for volume of distribution BW (HB mean SUV versus BW: Pearson’s correlation coefficient (r) = 0.37; p = 0.057, HB cutoff SUV versus BW: Spearman’s rho (ρ) = 0.45; p = 0.023). Alternatively, LBM normalization resulted in highly significant correlations between SUL and LBM for both HBmean SUL and HB cutoff SUL (r = 0.53; p = 0.009 and r = 0.59; p = 0.003, respectively). This underlines our preference of normalizing using BW instead of LBM. Based on the methods described by the authors, we measured the SV on the corresponding low-dose non-contrast enhanced CT’s with a lower threshold of + 120 HU (using Osirix as all our image analyses were performed with this software) and visually verified that this method only included cortical bone. In contrast to the population of the responding colleagues, however, (intramedullary) osteosynthesis

RkJQdWJsaXNoZXIy MjY0ODMw