The Author Replies.

Dr. Rottembourg cites a 1981 paper by Brissot et al. stating that liver iron content (LIC) is related to the total body iron (TBI) by approximately each 130 μmol/g dry liver of LIC is equal to 1 g of TBI. Because Rostoker et al.3, 4 found a yearly decrease in LIC of 270 μmol/g per year, corresponding to 1.68 g of TBI, they conclude that the magnetic resonance imaging (MRI) estimate of LIC (MRI-LIC) fits well with their calculated annual iron losses in hemodialysis and with the relationship of MRI-LIC to TBI seen in hemochromatosis patients. However, if Rottembourg is correct, this would mean that severe iron overload begins at 1.5 g of iron (200 μmol/g divided by 130 μmol/g = TBI), which is a trivial amount of excess iron in the context of hemochromatosis-induced iron overload.

Rottembourg correctly quotes Brissot et al., but unfortunately, the statement by Brissot et al. is a mathematical or typographical error off by a factor of 10, which is easily proven by examining the data and regression equation of Brissot et al. provided in their Figure 1 of the same publication. Rostoker et al. and Issad et al. state that 130 μmol/g dry liver is moderate iron overload, whereas >200 μmol/g dry liver is severe iron overload. Brissot et al. present the Y axis as LIC in μmol/100 mg of dry liver, so we need to divide the MRI-LIC4, 5 by 10 to graph their results. It is readily apparent based on the figure that an LIC of 13 μmol/100 mg of dry liver is ∼7 g of mobilized excess iron (Figure 1, dashed line), whereas 21 μmol/100 mg dry liver is ∼14 g (Figure 1, solid lines). If we solve using the regression equation of Brissot et al. (LIC = (1.3 × mobilized excess iron) + 3.5), then the TBI is 7.3 g and 13.6 g, respectively, for these 2 examples. Based on Rottembourg’s expected hemodialysis annual blood losses, (1.68 g/yr), patients with severe iron overload by MRI-LIC would take at least 8.0 years to normalize their LIC, and yet Rostoker’s group reports that they did this in 10 to 12 months.

Figure 1

Comparison of liver iron concentration (LIC) values before venosections and mobilized excess iron in 29 cases of idiopathic hemochromatosis. The dotted lines on the graph demonstrate that 13 μmol/100 mg dry liver is 7.3 g of mobilized excess iron, whereas the solid lines demonstrate that 21 μmol/100 mg dry liver is 13.6 g of mobilized excess iron.

Additionally, Figure 2 of Brissot et al. demonstrates that semiquantitative histologic estimates of LIC, as Rostoker used in another publication, frequently overestimate the actual LIC.2, 6 At least one-half of the grade 2 LIC estimates had normal actual LIC, whereas ∼15% of grade 3 LIC estimates had normal LIC, and many others should have been categorized as grade 2.

Figure 2

The Y axis is the determined liver iron concentration (LIC) from a tissue biopsy, whereas the X axis is the estimated LIC based on a histologic estimation system. In groups II and III, all subjects below the line are misclassified by histology because their determined LIC is normal. Also in group III, subjects with between 4 and 10 μmol/100 mg dry liver are overclassified and should be in group II. Adapted from Brissot P, Bourel M, Herry D, et al. Assessment of liver iron content in 271 patients: a reevaluation of direct and indirect methods. Gastroenterology. 1981;80:557–565.

In summary, these data indicate that MRI-LIC measurement in dialysis patients overestimates TBI by a factor of 10 when applying Brissot’s equation, whereas I conservatively estimated that they were off by a factor of 3 to 6.2, 7 Brissot also demonstrates that histologic assessments of LIC are inferior to actual determinations. I could not have made my points any better.