The Author Replies.

Rostoker and colleagues do not contest my conclusion that applying the ratio of magnetic resonance imaging (MRI) estimate of liver iron content (LIC) to total body iron observed in hereditary and transfusional overload overestimates total body iron by a factor of 3 to 6 in dialysis patients. Unfortunately, this letter title states MRI-LIC is accurate for “evaluating iron load in dialysis patients.” A high MRI-LIC score is not the same as iron overload in dialysis patients.1, 2 Based on the rate of decline in high MRI-LIC scores after halting iron therapy, most dialysis patients appear to have normal or mildly elevated total body iron.

In this letter, Rostoker et al. contend the MRI-LIC estimates are validated against liver biopsy–based LIC by pointing to their January 2017 publication in Heliyon Journal, which they did not reference in their Kidney International Reports article.1, 3 There are major problems with this study, including all patients had hepatitis C, and you recategorized 45% of patients’ results. Although you compare MRI-based LIC scores to semiquantitative histological assessments, you do not measure iron content per gram of liver dry weight, which is “considered as the reference method for quantifying iron in the liver” according to Deugnier and Turlin. Determination of LIC “on deparaffinized tissue should be the rule,” as iron distribution may be “heterogeneous as in the cirrhotic liver.”

The choice of patients with hepatitis C is problematic, as chronic hepatitis, especially hepatitis C–related injury, results in hepatic iron deposition in 35% to 56% of patients, and this is not reflective of total body iron overload. The brief pathologic description suggests iron was a bystander in at least 9 and possibly all 11 patients.

Although you highlight a strong relationship of MRI-LIC estimates to the semiquantitative methods, you lowered the MRI-LIC result in 3 patients by 53.7 to 107.4 μmol/g of dry liver based on the time between scan and liver biopsy, which aligned the MRI-LIC score with the histological estimate. You justify this by stating you observed this degree of decline in other patients. I think this is speculative at best and, in my opinion, improper. Two other patients had their histological scores altered from group 0 (possibly no iron present) to group 1 (normal iron content), which effectively aligned the histology result with the MRI-LIC result. In these 5 patients, without adjustment, the MRI-LIC appears to have overestimated liver iron content. Although there is undoubtedly a good relationship of MRI-LIC to histological semiquantitative and quantitative LIC, there is a degree of miscategorization into higher categories, such as the patient I highlighted in my commentary, and as others have noted.2, 5, 6

I disagree that research-related liver biopsy is contraindicated, as you are claiming that some dialysis patients have severe iron overload. There is a clinical and research need to know, hemochromatosis is an indication for liver biopsy for staging, and the risks of liver biopsy appear overstated according to the position paper by the American Association for the Study of Liver Diseases.

In the Heliyon article, the authors refer to a controversies conference on iron use in dialysis, and a Dialysis Advisory Group of the American Society of Nephrology commentary on iron use in dialysis.3, 8 I was a participant in the conference and a coauthor on the commentary. In neither was there agreement that iron overload in dialysis patients is a significant problem. Rather, we raised fundamental issues with the claims, including overestimating the actual total body iron, and lack of proof of consequent disease or end-organ damage, and therefore called for further research.