Reproducibility of 18F-fluoromisonidazole intratumour distribution in non-small cell lung cancer; methodological issues to avoid mismanagement of the patients

Excerpt

I was interested to read the paper by Grkovski M and colleagues published in the Dec 2016 issue of EJNMMI Res [1]. They aimed to assess the reproducibility of 18F-fluoromisonidazole (FMISO) positron emission tomography (PET) as a non-invasive, quantitative imaging technique, spatiotemporal intratumour distribution in patients with non-small cell lung cancer (NSCLC) [1]. The Pearson correlation coefficient r was calculated for mean standardized uptake values (SUV) within investigated volumes of interest and for voxels within tumour volumes (r _TV). The reproducibility of FMISO voxelwise distribution, SUV- and tumour-to-blood ratio (TBR)-derived indices was assessed using correlation and Bland-Altman analyses [1]. Although they correctly used Bland-Altman, they reported Pearson’s correlation r which in reproducibility (precision, repeatability, reliability, or interchangeability) is one of the common mistakes [2‐6]. Pearson’s correlation r only assesses the linearity between two continuous variables. Any shift in the location and/or scale of the regression line which leads to non-reproducibility cannot be detected by this correlation coefficient [2‐6]. Therefore, for quantitative variables, Intra Class Correlation Coefficient single measure is the best statistical test to evaluate reproducibility [2‐6]. …