Kinetic analysis of cardiac dynamic 18F-Florbetapir PET in healthy volunteers and amyloidosis patients: A pilot study

Abstract

Objectives: This study aimed to explore the potential of full dynamic PET kinetic analysis in assessing amyloid binding and perfusion in the cardiac region using ¹⁸F-Florbetapir PET, establishing a quantitative approach in the clinical assessment of cardiac amyloidosis disease.

Materials & methods: The distribution volume ratios (DVRs) and the relative transport rate constant (R₁), were estimated by a pseudo-simplified reference tissue model (pSRTM2) and pseudo-Logan plot (pLogan plot) with kidney reference for the region of interest-based and voxel-wise-based analyses. The parametric images generated using the pSRTM2 and linear regression with spatially constrained (LRSC) algorithm were then evaluated. Semi-quantitative analyses include standardized uptake value ratios at the early phase (SUVR_EP, 0.5-5 min) and late phase (SUVR_LP, 50-60 min) were also calculated.

Results: Ten participants [7 healthy controls (HC) and 3 cardiac amyloidosis (CA) subjects] underwent a 60-min dynamic ¹⁸F-Florbetapir PET scan. The DVRs estimated from pSRTM2 and Logan plot were significantly increased (HC vs CA; DVR_pSRTM2: 0.95 ± 0.11 vs 2.77 ± 0.42, t'(2.13) = 7.39, P = 0.015; DVR_Logan: 0.80 ± 0.12 vs 2.90 ± 0.55, t'(2.08) = 6.56, P = 0.020), and R₁ were remarkably decreased in CA groups, as compared to HCs (HC vs CA; 1.08 ± 0.37 vs 0.56 ± 0.10, t'(7.63) = 3.38, P = 0.010). The SUVR_EP and SUVR_LP were highly correlated to R₁ (r = 0.97, P < 0.001) and DVR(r = 0.99, P < 0.001), respectively. The DVRs in the total myocardium region increased slightly as the size of FWHM increased and became stable at a Gaussian filter ≥6 mm. The secular equilibrium of SUVR was reached at around 50-min p.i. time.

Conclusion: The DVR and R₁ estimated from cardiac dynamic ¹⁸F-Florbetapir PET using pSRTM with kidney pseudo-reference tissue are suggested to quantify cardiac amyloid deposition and relative perfusion, respectively, in amyloidosis patients and healthy controls. We recommend a dual-phase scan: 0.5-5 min and 50-60 min p.i. as the appropriate time window for clinically assessing cardiac amyloidosis and perfusion measurements using ¹⁸F-Florbetapir PET.

Keywords: 18F-florbetapir; Cardiac amyloidosis; Pseudo-reference tissue; Quantitative analysis; pSRTM.

Fig. 1

Bull's eye plot of DVR from pSRTM2, SUV_50–60min, and SUVR_50–60min in HC (a, c, e) and CA group (b, d, f) respectively.

Fig. 2

Kinetic modeling (pSRTM2(a), Logan plot(b)) using kidney reference tissue for representative CA patient and HC subject.

Fig. 3

R₁ generated by pSRTM2 (a) model fitting, SUVR_EP (b), DVR generated by pSRTM2 (c) and Logan plot (d) model fitting and SUVR_LP (e) of total myocardium region for CA and HC groups. Asterisks denote significant differences with HC group (*P < 0.05).

Fig. 4

Positive linear correlation between DVR from pSRTM2 and DVR from Logan plot (a). Positive linear correlation between ROI estimates DVR from pSRTM2 and SUVR_LP (b). Positive linear correlation between ROI estimates R₁ from pSRTM2 and SUVR_EP (c).

Fig. 5

Representative DVR parametric images for cardiac amyloidosis subject and healthy control subject. Parametric images of DVR were generated from ¹⁸F-Florbetapir dynamic PET using the pSRTM2 with linear regression (LR), and linear regression with and without spatially constrained (LRSC) algorithm with 3-dimensional Gaussian filters with FWHM varying from 3 to 12 mm, where the spatial resolution of reconstructed PET is 2.9 mm FWHM.

Fig. 6

The change of mean SUVR (myocardium to the kidney) in the CA and HC group for every 10 min (a). The correlation of DVR and SUVR over time (b).

Fig. 7

The Kinetic modeling (2-Tissue Compartment Model with metabolite-uncorrected image-derived input function (a), 2-Tissue Compartment Model with metabolite-corrected image-derived input function (b)) for representative CA patient and HC subject.