Site specific measurements of bone formation using [18F] sodium fluoride PET/CT

Abstract

Dynamic positron emission tomography (PET) imaging with fluorine-18 labelled sodium fluoride ([¹⁸F]NaF) allows the quantitative assessment of regional bone formation by measuring the plasma clearance of fluoride to bone at any site in the skeleton. Today, hybrid PET and computed tomography (CT) dual-modality systems (PET/CT) are widely available, and [¹⁸F]NaF PET/CT offers a convenient non-invasive method of studying bone formation at the important osteoporotic fracture sites at the hip and spine, as well as sites of pure cortical or trabecular bone. The technique complements conventional measurements of bone turnover using biochemical markers or bone biopsy as a tool to investigate new therapies for osteoporosis, and has a potential role as an early biomarker of treatment efficacy in clinical trials. This article reviews methods of acquiring and analyzing dynamic [¹⁸F]NaF PET/CT scan data, and outlines a simplified approach combining venous blood sampling with a series of short (3- to 5-minute) static PET/CT scans acquired at different bed positions to estimate [¹⁸F]NaF plasma clearance at multiple sites in the skeleton with just a single injection of tracer.

Keywords: [18F] sodium fluoride ([18F]NaF); bone blood flow; bone formation; bone plasma clearance; metabolic bone disease; osteoporosis; osteoporosis therapy; positron emission tomography (PET); quantitative imaging; standardised uptake value.

Figure 1

Images of bone seeking tracers used for quantitative imaging. (A) [^99mTc]TcMDP whole body planar gamma camera bone scan image as used by Moore et al. (27) to measure whole body and regional bone plasma clearance; (B) [¹⁸F]NaF PET sagittal image of the lumbar spine (L1-L4) as used in studies based on the Hawkins method (26). Reproduced with permission from (40). (C) [¹⁸F]NaF PET coronal image of the proximal femur as used by Frost et al. to study regional bone plasma clearance in the hip and femoral shaft (41). Both PET images are two-dimensional (2D) projection views of the complete three-dimensional (3D) scan data, and are restricted by the 15-cm axial field of view of the PET scanner. The [¹⁸F]NaF activity collecting in the urinary bladder during the 1-hour dynamic scan has been masked to give a clearer view of the uptake in bone. Reproduced with permission from (40). [¹⁸F]NaF, fluorine-18 labelled sodium fluoride; PET, positron emission tomography.

Figure 2

Quantitative analysis using the Hawkins model. (A) Representative curves showing the arterial input function measured by direct blood sampling and corresponding bone time activity curve (TAC) for a [¹⁸F]NaF dynamic PET scan of the lumbar spine. Both curves have been corrected for radioactive decay. Reproduced with permission from (40). (B) The Hawkins compartmental model used for the analysis of [¹⁸F]NaF PET dynamic bone scans (26). The rate constant K₁ describes the effective bone plasma flow to the unbound bone pool, k₂ the reverse transport of tracer from the unbound bone pool back to plasma, k₃ the forward transport from the unbound bone pool to bone mineral, and k₄ the reverse flow. Bone plasma clearance K_i is calculated using Eq. [1]. Reproduced with permission from (40). (C) Results of fitting the bone TAC and arterial plasma input function to the Hawkins compartmental model. As well as the 4 parameters K₁, k₂, k₃ and k₄ the model also fits the fractional volume of blood within the bone ROI, F_BV. The plasma clearance to bone mineral K_i is calculated using Eq. [1]. The figure shows time activity plots of the amount of tracer in each compartment of the Hawkins model and the resulting fit of the summed curves to the measured bone TAC. Reproduced with permission from (40). [¹⁸F]NaF, fluorine-18 labelled sodium fluoride; PET, positron emission tomography.

Figure 3

Precision and treatment response of the different parameters in the Hawkins model. (A) The precision errors of the parameters in the Hawkins compartmental model (Figure 2B) used for the analysis of [¹⁸F]NaF PET dynamic scans expressed as the coefficient of variation. The data comes from the analysis of lumbar spine scans in 20 postmenopausal women who had scans at baseline, 6- and 12-month after stopping alendronate and had no changes in biochemical markers of bone turnover in that period (68). (B) The treatment response expressed as the percentage change from baseline of the parameters in the Hawkins model measured by [¹⁸F]NaF PET dynamic scans of the lumbar spine in 18 postmenopausal women after 6 months treatment with teriparatide (68). (C) The ratio of the absolute value of the treatment response in (B) divided by the precision error in (A) for each parameter. Parameters with a large treatment response and a small precision error have the highest ratios and are likely to be the most sensitive parameters for measuring response to treatment. [¹⁸F]NaF, fluorine-18 labelled sodium fluoride; PET, positron emission tomography.

Figure 4

Quantitative analysis using the Patlak plot. (A) Standard Patlak analysis (i.e., assuming k_loss =0) of [¹⁸F]NaF PET data obtained during a 60-minute dynamic scan. The graph is a plot of normalized bone uptake ($\frac{C_b(T)}{C_p(T)}$) against normalized time ($\frac{{\displaystyle {\int }_0^T{C}_p(t)dt}}{C_p(T)}$) (see Eq. [2]). Bone plasma clearance (K_i) is found from the slope of the straight-line fit to the 10–60 minutes data points. The intercept on the vertical axis gives the volume of distribution V. Reproduced with permission from (40). (B) Evaluation of k_loss by the modified Patlak analysis of Holden et al. (72). Triangles, schematic plot of the standard analysis (k_loss =0) applied to 10–60 minutes dynamic scan data. Circles, modified analysis (Eqs. [3] and [4]) with the value of k_loss optimized to give the best straight-line fit to the 10–60 minutes data points. Squares, modified Patlak analysis over corrected for kloss. Reproduced with permission from (37). (C) Derivation of the semi-population input function. The population residual function is scaled for injected activity and the time of peak counts adjusted to agree with a region of interest drawn over the aorta for dynamic PET scan data or to the time of injection for static scans. This curve is added to the terminal exponential fitted to the 30-, 40-, 50- and 60-minute venous plasma data. The terminal exponential is rolled off using a ramp function at the time of peak counts so as not to affect the early rise of the bolus. Reproduced with permission from (65). (D) Derivation of bone plasma clearance using the simplified static-scan method (36,37). The right upper point is based on a measurement of bone uptake from a single 5-minute static scan acquired around 45–75 minutes after injection of tracer. The left lower point is the intercept of the graph and represents the population average volume of distribution V. The value of K_i is obtained from the slope of the straight line through the two points. Reproduced with permission from (40). [¹⁸F]NaF, fluorine-18 labelled sodium fluoride; PET, positron emission tomography.