Cappellini and colleagues1 claim that the use of a superconducting quantum interference device (SQUID) biosusceptometer underestimates liver iron concentration (LIC) in their phase 3 study of deferasirox (DFX). LIC was measured either in deparaffinized liver samples excised by various biopsy techniques (Menghini with saline flushing, cutting needles) or in an anterior position above the right liver lobe by biomagnetic liver susceptometry (BLS) using low TC-SQUID biosusceptometers. In vivo wet-weight LICs measured by BLS were converted by a factor of 3.33 into dry-weight values. This approximate conversion factor2 has been uncritically adopted throughout the literature, even by ourselves, although there were strong data available supporting a higher factor for the ratio of wet to dry weight3,4 and a significant difference between LIC from fresh tissue and from deparaffinized samples.5,6 Thus, the conversion factor between LIC as determined by BLS and from deparaffinized liver samples would have been at least 5.5 ± 1.0 (calculated factor ± uncertainty).7 Related to activities around this phase 3 study program of DFX, the authors have developed more direct knowledge of ratios of wet to dry weight by various biopsy processing techniques (eg, a conversion factor of 5.8 ± 0.6 for deparaffinized liver samples) and their “paramount importance” for comparison of LICs.8
Consequently, the authors should have corrected their LICs measured by SQUID-BLS in order to analyze their data more accurately in this important publication on a novel oral chelator. We think that it is allowed to correct an initially false study assumption in a scientific paper. Measurements by BLS would have the highest impact especially in the LIC group of 7 mg Fe/g dry weight or less, although the final outcome may not change significantly. Moreover, we would hope to avoid giving potential readers the wrong impression that BLS underestimates LIC per se. One could, in fact, claim the opposite, as in our title, particularly in the case of deparaffinized samples.
As part of this discussion, it should be emphasized that the different conversion factors also have a strong impact on the LIC safety thresholds2 in iron-overloaded patients with thalassemia. These recommended thresholds were based in part on LICs measured by BLS with an approximate conversion factor of 3.33. For example, the threshold for increased risk of cardiac failure of LICs equaling 80 μmol/g wet weight (about 15 mg/g dry weight1 (p 3455)) would convert to 26 ± 5 mg/g dry weight using the conversion factor of 5.8 for deparaffinized liver biopsies. Thus, dry-weight LICs could be very different depending on the selected biopsy techniques and processing methods.7
Deferasirox (ICL670) effectively reduces liver iron concentration, whether assessed by biopsy or SQUID
We thank Drs Fischer, Harmatz, and Nielsen for their letter and interest in our paper. In this study, liver iron concentration (LIC) was assessed either by biopsy (84% of patients) or superconducting quantum interference device (SQUID; 16%, primarily children and adults with contrain-dications to biopsy). SQUID assessments were performed at 3 centers (in Italy, Germany, and the US); baseline and end-of-study assessments were always conducted at the same site.
We acknowledge that the discrepancy between LIC values obtained from biopsy and SQUID was primarily due to the accepted conversion factor of 3.33 that was used to convert wet-weight into dry-weight values.1 As discussed in the paper, we opted to present prenormalized LIC values because the discrepancy between values obtained from biopsy and SQUID was only identified late in the trial via a validation study. By this time, most patients had been assigned doses according to baseline LIC and most had completed a large proportion of the study. Subsequently, the more appropriate correction factor noted by Drs Fischer, Harmatz, and Nielsen (5.83 ± 0.60) was determined through an independent study.2
The effect of deferasirox for reducing iron burden, as measured by LIC, was potentially underreported in our paper because we elected to use prenormalized SQUID data. Although these values are different in absolute magnitude from the prenormalized values, they do not differ in the relative magnitude of response. The inclusion or exclusion of patients whose LIC was assessed by SQUID does not alter the primary outcome of the study, which was to demonstrate that deferasirox is generally well tolerated and effective for reducing body iron burden.
Despite the technical limitations of the application in this study, SQUID has made, and is continuing to make, important contributions to the field of iron chelation research. As with all methods that are used in longitudinal assessments, consistency of technique is of utmost importance. For the patients who have access, SQUID will remain an important tool to monitor iron overload and response to chelation therapy.
Correspondence: Maria Domenica Cappellini, Università di Milano, Fondazione Ospedale Maggiore Policlinico, Mangiagalli, Regina Elena IRCCS, Milan, Italy; e-mail: maria.cappellini@unimi.it.
