Mimicking liver iron overload using liposomal ferritin preparations

Abstract

Close monitoring of liver iron content is necessary to prevent iron overload in transfusion-dependent anemias. Liver biopsy remains the gold standard; however, MRI potentially offers a noninvasive alternative. Iron metabolism and storage is complicated and tissue/disease-specific. This report demonstrates that iron distribution may be more important than iron speciation with respect to MRI signal changes. Simple synthetic analogs of hepatic lysosomes were constructed from noncovalent attachment of horse-spleen ferritin to 0.4 microm diameter phospholipid liposomes suspended in agarose. Graded iron loading was achieved by varying ferritin burden per liposome as well as liposomal volume fraction. T1 and T2 relaxation times were measured on a 60 MHz NMR spectrometer and compared to simple ferritin-gel combinations. Liposomal-ferritin had 6-fold stronger T2 relaxivity than unaggregated ferritin but identical T1 relaxivity. Liposomal-ferritin T2 relaxivity also more closely matched published results from hemosiderotic marmoset liver, suggesting a potential role as an iron-calibration phantom.

FIG. 1

R2 as a function of iron concentration for unbound ferritin dispersion in 1.5% agarose gel (filled square symbols). Linear regression for these data demonstrate a slope of 4.5 sec⁻¹/mg(Fe)/g gel and an R2-intercept of 14.2 sec⁻¹. Dotted line reflects regression line calculated from iron overloaded marmoset liver, having a slope of 16.9 sec⁻¹/(mg(Fe)/g wet wt) and R2-intercept of 10.6 sec⁻¹ (12). Other symbols indicate the relationship between R2 and iron concentration for liposomally bound ferritin dispersion in 1.5% agarose gel. Iron was increased by increasing liposomal iron loading (+ signs) at fixed liposomal volume fraction or by varying liposomal volume fraction with fixed liposomal iron burden (circles). Both mechanisms yield similar, near-linear relationships with slopes of 24–27 sec⁻¹/mg(Fe)/g, nearly 6-fold stronger relaxation than unaggregated ferritin and better agreement with marmoset data.

FIG. 2

R1 as a function of iron concentration for “free” and liposomal-ferritin dispersions in gel. No significant R1 difference is seen between free and liposomally bound ferritin gels. Pooled regression line is also shown.

FIG. 3

R2 as a function of interecho spacing for six different iron concentrations.