Background

R₂* relaxometry is a common method to determine liver iron concentration in vivo. However, different iron compounds show different R₂* dependence on iron concentration, i.e. different relaxivities.
This work was performed to study concentration-dependent R₂* of mixtures containing two different iron compounds.

Methods

Polymer coated Fe₃O₄ superparamagnetic nanoparticles (NP) functionalized with COOH, 6 nm in diameter (type UMC 3001, lot #13573), were purchased from Bangs Laboratories (Warrington, PA) and diluted in purified water. Together with FeCl₃, NPs were mixed with agarose in 16 ml vials to obtain probes suitable for MRI. Total concentrations were adjusted to seven different levels ranging from 0 to 0.1 mg/g for NP and six concentrations, 0 to 3 mg/g, for FeCl₃. All probes were placed in a holder filled with agarose gel to enable GRE scanning. MR scans were performed on a 1.5 T Somatom Avanto scanner (Siemens Healthcare, Erlangen, Germany). Multi-slice multi-echo GRE was performed at TR/FA of 120 ms / 50°, a minimum TE of 1.1 ms and 12 echos at an echospacing of 1 ms. Additional acquisitions were performed with minimum TEs of 1.3, 1.5, 1.7 and 1.9 ms at the same echospacing to achieve an effective echospacing of 0.2 ms when combining all acquisitions. GRE signals as a function of TE were measured by ROIs manually placed on each probe which were copied to all TEs and acquisitions. For each combination of concentrations, 8 slices were analyzed. Median values of ROIs were fitted to a monoexponential decay function, while MR image noise was accounted for, as proposed by Jensen et al.¹ by calculating the square root of sum-of squares of ideal signal and noise signal S_noise

$$S_{fit} = \sqrt { (S_{0} e^{(TE*{R_{2}}^{*})} )^2 + {S_{noise}}^2 }$$

with scaling factor S₀, transversal relaxation rate R₂* and S_noise as fit parameters using a nonlinear Leverberg-Marquardt least-squares fitting algorithm. Fits exceeding predefined thresholds of residual errors or S_noise were excluded from further analysis. Transversal relaxation rates R₂* as function of concentration of one compound were plotted with the concentration of the other compound as parameter. Coefficients of determination R² and regression parameters for all linear combinations between compound concentration and R₂* were determined. Relaxivities of each compound, i.e. slope of the regression line R₂* vs. concentration were determined, as well as regression line intercepts, and dependencies on the concentration of the other compound were studied.

Results

Fits performed best in the R₂* range from 100 s^-1 to 800 s^-1. Uncertainties increased for R₂* above 800 s^-1, seen from the spread of data points in Figs. 1 and 2. Fig. 1 shows R₂* of iron compound mixtures as a function of nanoparticle concentration, Fig. 2 the correspondent values as a function of FeCl₃ concentration. All dependencies of R₂* on concentration were linear with R² close to 1. Fig. 3 shows nanoparticle relaxivities as a function of FeCl₃ concentration. Unexpectedly, nanoparticle relaxivity was not constant but depended on concentration of added FeCl₃. On the other hand, as Fig. 4 shows, the same was observed for relaxivity of FeCl₃ in the presence of nanoparticles. Both dependencies could be approximated by a quadratic function, however, in the case of nanoparticles only if omitting their relaxivity without added FeCl₃. Intercepts showed a linear dependence on concentration of added compound, cf. Fig. 5.

Discussion

Excessive body iron is the cause for several serious diseases. Iron accumulates in the liver, where it is stored as hemosiderin in immobilized form. Nanoparticles have been suggested previously as a model for this form of storage iron¹.
In our study, iron concentrations of nanoparticles were chosen lower than for FeCl₃ since Fe²⁺/Fe³⁺ ions in nanoparticles are immobilized in a solid phase, causing an increase in relaxivity by about a factor of about 50 compared to Fe³⁺ ions in aqueous solution in the case of FeCl₃. Also, concentrations were chosen not to exceed transverse relaxation rates R₂* of 1500 s^-1 assumed as maximum for acquisition parameters used. To reliably quantify larger R₂*, the minimum TE, limited by scanner hardware, was too long.
The chemical more aggressive FeCl₃ was used instead of the less reactive FeCl₂ since Fe²⁺ ions show much lower relaxivity than Fe³⁺ ions².
It is generally known that transversal relaxation rate R₂* depends linearly on the concentration of a substance, e.g. contrast agent, with the relaxivity of this substance as slope of the regression line. Our work showed the dependency of one iron compound's relaxivity on the concentration of another iron compound. The linearity of R₂* as function of a single compound’s concentration was excellent, therefore we regard our results reflect the truth. An approach to explain nonconstant relaxivity is that Fe³⁺ ions may absorb to the surface of the nanoparticles functionalized with COO^-. At low Fe³⁺ concentrations, this might reduce NP relaxivity, but increase it at high concentrations. Computer simulations may be helpful to study this behavior in more detail.

Conclusion

It has been shown that in a mixture of different iron compounds relaxivities of one compound varied with the concentration of the other. This suggests that interactions of both compounds are more complex than previously assumed.

Acknowledgements

No acknowledgement found.