Quantitation of cancer treatment response by 2-[18F]FDG PET/CT: multi-center assessment of measurement variability using AUTO-PERCIST™

Abstract

Background: The aim of this study was to assess the reader variability in quantitatively assessing pre- and post-treatment 2-deoxy-2-[¹⁸F]fluoro-D-glucose positron emission tomography/computed tomography ([¹⁸F]FDG PET/CT) scans in a defined set of images of cancer patients using the same semi-automated analytical software (Auto-PERCIST™), which identifies tumor peak standard uptake value corrected for lean body mass (SUL_peak) to determine [¹⁸F]FDG PET quantitative parameters.

Methods: Paired pre- and post-treatment [¹⁸F]FDG PET/CT images from 30 oncologic patients and Auto-PERCIST™ semi-automated software were distributed to 13 readers across US and international sites. One reader was aware of the relevant medical history of the patients (read_reference), whereas the 12 other readers were blinded to history but had access to the correlative images. Auto-PERCIST™ was set up to first automatically identify the liver and compute the threshold for tumor measurability (1.5 × liver mean) + (2 × liver standard deviation [SD]) and then detect all sites with SUL_peak greater than the threshold. Next, the readers selected sites they believed to represent tumor lesions. The main performance metric assessed was the percent change in the SUL_peak (%ΔSUL_peak) of the hottest tumor identified on the baseline and follow-up images.

Results: The intra-class correlation coefficient (ICC) for the %ΔSUL_peak of the hottest tumor was 0.87 (95%CI: [0.78, 0.92]) when all reads were included (n = 297). Including only the measurements that selected the same target tumor as the read_reference (n = 224), the ICC for %ΔSUL_peak was 1.00 (95%CI: [1.00, 1.00]). The Krippendorff alpha coefficient for response (complete or partial metabolic response, versus stable or progressive metabolic disease on PET Response Criteria in Solid Tumors 1.0) was 0.91 for all reads (n = 380) and 1.00 including for reads with the same target tumor selection (n = 270).

Conclusion: Quantitative tumor [¹⁸F]FDG SUL_peak changes measured across multiple global sites and readers utilizing Auto-PERCIST™ show very high correlation. Harmonization of methods to single software, Auto-PERCIST™, resulted in virtually identical extraction of quantitative tumor response data from [¹⁸F]FDG PET images when the readers select the same target tumor.

Keywords: Quantification; Response assessment; [18F]FDG PET/CT.

Fig. 1

Screen captures from Auto-PERCIST™. a The software detected all sites with SUL_peak higher than the computed or manually set threshold. b The reader than selected the true tumor lesions (shaded in green), excluding physiologic [¹⁸F]FDG activity

Fig. 2

Bland–Altman plot of the percentage change of tumor [¹⁸F]FDG uptake from baseline to follow-up. The plot is for the percentage changes of SUL_peak for all reads. Each dot represents a case (30 cases in total). The x-axis represents the average mean percentage change measurement by all readers. The y-axis represents the average difference between the 12 readers and the reference reader (read_reference). The solid line represents the average bias, and the dashed lines represent the corresponding bias ± 2 standard deviations (SD)

Fig. 3

Bland–Altman plot of the percentage change of tumor [¹⁸F]FDG uptake from baseline to follow-up. The plot is for the percentage changes of SUL_peak (%ΔSUL_peak) for only the reads with same lesion selected as the read_reference. Each dot represents a case (30 cases in total). The x-axis represents the average mean %ΔSUL_peak measurement by all readers. The y-axis represents the average difference between the 12 readers and the reference reader (read_reference) and the y-axis unit is one-tenth of one percent. The solid line represents the average bias, and the dashed lines represent the corresponding bias ± 2 standard deviations (SD)

Fig. 4

a PET maximum intensity projection (MIP) image of patient with right axillary node metastasis at baseline with SUL_peak of 1.62 and tumor volume of less than 1.00 cc. Though visually perceptible, Auto-PERCIST™ failed to detect the lesion due to small size. b On the follow-up MIP image, the number of metastatic nodes and the [¹⁸F]FDG uptake intensity are increased to SUL_peak of 2.84, allowing detection by Auto-PERCIST™