Interobserver and intraobserver variability for the... : Nuclear Medicine Communications (original) (raw)

ORIGINAL ARTICLES

Interobserver and intraobserver variability for the assessment of brown adipose tissue activity on 18F-FDG PET-CT

Bahler, Lonneke; Holleman, Frits; Booij, Jan; Hoekstra, Joost B.; Verberne, Hein J.

Departments of aInternal Medicine

bNuclear Medicine, Academic Medical Center, Amsterdam, The Netherlands

Correspondence to Lonneke Bahler, MD, Department of Internal Medicine, Academic Medical Center, F4-255, PO Box 22660, 1100DD Amsterdam, The Netherlands Tel: +31 205 665 981; fax: +31 206 914 904; e-mail: [email protected]

Received October 15, 2015

Accepted October 30, 2015

Abstract

Objective

Measurement of brown adipose tissue (BAT) activity is the focus of intensive research, among others as a potential target for weight-lowering strategies. In this, BAT activity is visualized and quantified using 18F-fluorodeoxyglucose (18F-FDG) PET-CT. The aim of this study was to determine the interobserver and intraobserver variability for detecting and quantifying BAT on 18F-FDG PET-CTs.

Methods

Three observers retrospectively independently assessed 55 18F-FDG PET-CTs (performed between April 2013 and January 2014) for BAT activity parameters: BAT volume, the maximal and mean standardized uptake value (SUVmax and SUVmean) obtained in healthy male controls. One observer reassessed the scans after 2 months for the intraobserver variability. Interobserver and intraobserver variability were expressed using Lin’s concordance coefficient (LCC) and Bland–Altman plots. Correlations between the three parameters were assessed using Spearman’s correlation.

Results

The LCCs for the interobserver and intraobserver concordance for SUVmax were the highest (LCC SUVmax varied between 0.998 and 0.999, for SUVmean between 0.989 and 0.991 and for volume between 0.947 and 0.972). The Bland–Altman analysis showed a small absolute mean difference between all observers for both SUVmax and SUVmean, but the differences for volume were markedly higher. All parameters correlated statistically strongly and positively.

Conclusion

The SUVmax showed the lowest interobserver and intraobserver variation. Although SUVmean and BAT volume had a higher interobserver and intraobserver variation, the variation is still within acceptable limits. Therefore, all parameters can be used to describe BAT activity. However, for an adequate comparison between studies, we recommend the use of SUVmax.