Repeatability of FDG quantification in tumor imaging: averaged SUVs are superior to SUVmax
Introduction
18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) has become an important tool in the evaluation of cancer patients, with an increasing use in staging as well as treatment response monitoring [1], [2], [3]. The latter is usually based on the percentage change of the standard uptake value (SUV), and its accuracy therefore depends on the reproducibility of the measured SUVs. The definition of the SUV as the activity per gram of tissue divided by the injected activity per patient weight already points to obvious factors determining the reliability of the SUV, namely, the accuracy of the measured tissue activity, of the patient's weight and of the injected activity. In this paper, we focused on the accuracy of the measured tissue activity. Tissue activity can be measured in various volumes of interest (VOIs) and with different approaches of voxel selection. The most common approach is to measure the value in the voxel with the highest activity (SUVmax). Alternatively, the value of several hottest voxels can be averaged (SUVmean). The European Association of Nuclear Medicine guidelines 2009 for PET/CT reporting recommend the use of SUVmax [4]. Several studies have emphasized the importance of using SUVmax instead of SUVmean because of the excellent interobserver agreement for SUVmax [5], [6], [7]. It is obvious that SUVmax is independent of the observer, while SUVmean potentially varies with the person who draws the VOI over the tumor. This effect can be minimized by automatic selection of the voxels within a preselected VOI using a fixed threshold from SUVmax. Decreasing the number of voxels also decreases the amount of counts recorded and therefore the accuracy of the measured values.
In this paper, we determined the repeatability of SUV values as a function of the number of voxels selected in a given VOI. For this purpose, the SUV was measured in two sequential FDG scans acquired from 35 to 40 and from 40 to 45 min. In addition, we evaluated the percentage change in SUVs in tumors before and after treatment for SUVmax and SUVmean.
Section snippets
Patients for calculation
Twenty patients with known malignant tumors in the chest were prospectively recruited between March 2009 and September 2009 for an institutional review board-approved study, after written informed consent, to analyze the autoradiographic method as a potential alternative in FDG quantification [8].
Evaluation of SUV change in treated tumors
We retrospectively analyzed 40 consecutively scanned patients with a histology of melanoma referred for FDG PET/CT before (scan 1) and after (scan 2) chemotherapy between September 2007 and September
Patient characteristics
Demographic data of the patients are summarized in Table 1.
Repeatability
The Bland–Altman plots in Fig. 1 demonstrate the reduction in the interscan difference from SUVmax to SUVmean. The S.D. of the absolute change was 1.01 for SUVmax, 0.53 for SUV5, 0.37 for SUV10 and 0.28 for SUVmean. The S.D. of the percentage change was 11.2 % for SUVmax, 5.8 % for SUV5, 4 % for SUV10 and 3.6 % for SUVmean.
The change of the several SUV values is illustrated for each patient in Fig. 2. While there was a decrease of SUVmax
Discussion
In this paper, we demonstrate a considerable increase of the repeatability of SUV values if they are derived from more than one voxel as is done in most studies. Several points regarding our study design should be mentioned. As opposed to phantom studies, the FDG uptake in the measured tumors between the two scans at 40 and 45 min is not constant, but slightly increasing. The percentage of the physiological change depends on the tumor activity. Due to this, the S.D. of the noise itself will be
Conclusion
The repeatability of SUV is markedly increased by deriving the value from multiple voxels. Compared to SUVmax, the variability is reduced by a factor of 2.7 if 10 voxels are pooled.
Acknowledgments
The authors would like to thank the PET technicians Josephine Trinckauf, Sabine Knoefel and Verena Weichselbaumer for their help with the data acquisition. There is no conflict of interest even though the authors acknowledge support by Medrad and one of the authors (F.B.) is owner of a patent involving the INTEGO injector.
References (14)
- et al.
Simplified quantification of FDG metabolism in tumors using the autoradiographic method is less dependent on the acquisition time than SUV
Nuclear Medicine and Biology
(2011) - et al.
Comparison of simplified quantitative analyses of FDG uptake
Nuclear medicine and biology
(2000) - et al.
Direct comparison of (18)F-FDG PET and PET/CT in patients with colorectal carcinoma
J Nucl Med
(2003) - et al.
Staging of non-small-cell lung cancer with integrated positron-emission tomography and computed tomography
N Engl J Med
(2003) - et al.
Integrated PET/CT: current applications and future directions
Radiology
(2006) - et al.
FDG PET and PET/CT: EANM procedure guidelines for tumour PET imaging: version 1.0
Eur J Nucl Med Mol Imaging
(2010) - et al.
Treatment monitoring by 18F-FDG PET/CT in patients with sarcomas: interobserver variability of quantitative parameters in treatment-induced changes in histopathologically responding and nonresponding tumors
J Nucl Med
(2008)