Normal values for 18F-FDG uptake in organs and tissues measured by dynamic whole body multiparametric FDG PET in 126 patients

Abstract

Background: Dynamic whole-body (D-WB) FDG PET/CT is a recently developed technique that allows direct reconstruction of multiparametric images of metabolic rate of FDG uptake (MR_FDG) and "free" FDG (DV_FDG). Multiparametric images have a markedly different appearance than the conventional SUV images obtained by static PET imaging, and normal values of MR_FDG and DV_FDG in frequently used reference tissues and organs are lacking. The aim of this study was therefore to: (1) provide an overview of normal MR_FDG and DV_FDG values and range of variation in organs and tissues; (2) analyse organ time-activity curves (TACs); (3) validate the accuracy of directly reconstructed MR_FDG tissue values versus manually calculated K_i (and MR_FDG) values; and (4) explore correlations between demographics, blood glucose levels and MR_FDG values. D-WB data from 126 prospectively recruited patients (100 without diabetes and 26 with diabetes) were retrospectively analysed. Participants were scanned using a 70-min multiparametric PET acquisition protocol on a Siemens Biograph Vision 600 PET/CT scanner. 13 regions (bone, brain grey and white matter, colon, heart, kidney, liver, lung, skeletal muscle of the back and thigh, pancreas, spleen, and stomach) as well as representative pathological findings were manually delineated, and values of static PET (SUV), D-WB PET (K_i, MR_FDG and DV_FDG) and individual TACs were extracted. Multiparametric values were compared with manual TAC-based calculations of K_i and MR_FDG, and correlations with blood glucose, age, weight, BMI, and injected tracer dose were explored.

Results: Tissue and organ MR_FDG values showed little variation, comparable to corresponding SUV variation. All regional TACs were in line with previously published FDG kinetics, and the multiparametric metrics correlated well with manual TAC-based calculations (r² = 0.97, p < 0.0001). No correlations were observed between glucose levels and MR_FDG in tissues known not to be substrate driven, while tissues with substrate driven glucose uptake had significantly correlated glucose levels and MR_FDG values.

Conclusion: The multiparametric D-WB PET scan protocol provides normal MR_FDG values with little inter-subject variation and in agreement with manual TAC-based calculations and literature values. The technique therefore facilitates both accurate clinical reports and simpler acquisition of quantitative estimates of whole-body tissue glucose metabolism.

Keywords: Dynamic whole-body PET; Multiparametric imaging; Normal values; Patlak.

Fig. 1

Example of SUV and Patlak images analysed in our study. In this case, a 34-year-old man with mediastinal sarcoidosis. Left column: static SUV images; Middle column: MR_FDG images; right column: DV_FDG images. The actual FDG uptake in the inflammatory sarcoid lymph nodes is evident on the MR_FDG image, whereas the free FDG in the circulation is clearly visible on the DV_FDG image

Fig. 2

Distribution plots of SUV_mean, MR_FDG,mean, K_i,mean and DV_FDG,mean for the DM and Non-DM populations in 13 different types of tissue and organs. Plotted are the mean and standard deviation values. As seen, population variation was remarkably similar for SUV and MR_FDG values. Calculated p-values between patients with DM and without DM are displayed above each organ

Fig. 3

Time-activity curves for the first 6 min of dynamic scanning over the chest area. Represented values are mean and SEM of all 126 patients. The panels on the left side correspond to the patients without diabetes; the panels on the right side to the patients with diabetes

Fig. 4

Time-activity curves for the remaining 6–70 min of dynamic WB scanning. Represented values are mean and SEM of all 126 patients. The panels on the left side correspond to the patients without diabetes; the panels on the right side to the patients with diabetes

Fig. 5

A Correlation between multiparametric MR_FDG values and manual TAC-based MR_FDG estimates using PMOD’s PKIN module. N = 10, corresponding to 110 correlation points. B Bland–Altman analysis of %Difference (100*(Multiparametric MR_FDG − TAC-based MR_FDG)/average) vs average

Fig. 6

Correlation of A SUV_mean and B MR_{FDG mean} with glucose for brain grey matter, pancreas, liver, and skeletal paravertebral muscle. C Correlation of MR_{FDG mean} and age. Plotted are the VOIs for brain grey matter and liver tissue

Fig. 7

Pearson’s correlation analysis of patients with and without diabetes, of SUV_mean and MR_{FDG mean} with glucose levels, tracer activity, sex, age, BMI, and weight (A); as well as between paired sets of organs (B). Cells labelled “p” showed statistically significant correlation (p < 0.05) without correction for multiple comparisons. The full correlation results can be found in the Additional file 1: figure S1 and S2, table S4A, S4B, S5A and S5B