Italian Society of Hematology guidelines for thalassemia and non-invasive iron measurements

A recent article by Angelucci et al.¹ offers recommendations for the diagnosis, monitoring, and management of thalassemia major and related disorders. The authors’ recommendations were based on a literature survey that is itself a valuable resource for clinicians and researchers. By comparing non-invasive liver iron quantification by biomagnetic liver susceptometry (BLS) and magnetic resonance imaging (MRI), it was concluded that BLS using superconducting quantum interference devices (SQUID) was inaccurate and systematically underestimated liver iron concentration (LIC). Two studies were cited with extremely different correlations reported as coefficients of determination (R²).^2,3 In one of the references, BLS had a very high correlation with wet-weight liver biopsy(R²=0.98).² In the second reference, the correlation between LIC from dry-weight biopsy and by BLS is alleged to be extremely poor (R² = 0.21).³ However, this value cannot be found in that citation nor can it be derived from any given data elsewhere.⁴ Most likely, the regression coefficient of 0.46 (functional relationship between LIC from BLS and biopsy) was mistaken for a correlation coefficient R.

In fact, from the mentioned larger study group (blind substudy within the CICL670A0107 trial) R²=0.74 for BLS and R²=0.75 for MRI-R2 could be calculated (Pearson’s correlation). Both methods rely on the large magnetic susceptibility of the hemosiderin/ferritin iron complex,⁴ directly measured by BLS or via its interference with the transversal proton relaxation rate R2 or R2* by MRI. The authors should have wondered about such a correlation difference between these two methods. Consequently, a high correlation (R²=0.86) was directly observed between BLS and MRI-R2 using the same MRI sequence as above.⁵ Although we previously clarified the problem of under- or overestimation of LIC by BLS or from dry-weight liver biopsy,⁶ the underestimation aspect appears again as a critical factor. In the CICL670A0107 substudy, a relationship of 0.46 was observed between LIC from in vivo BLS converted by a factor of 3.33 to a dry-weight equivalent LIC from deparaffinized biopsies.³ Both BLS and MRI are non-invasive probes that determine the volume concentration of iron in living tissue. Thus, LIC from wet-weight biopsies would be the adequately corresponding parameter for calibration or validation.

Despite the similarities between MRI and BLS, there is a marked technical distinction between them, namely, the physics of the BLS signal is simple and well understood,⁴ but the MRI signal is subject to a number of complicating physical factors, which make it impractical to calculate the iron concentration in an analytic way. As a result, MRI is forced to rely on biopsy data for calibration. In marked contrast, the BLS signal can be reliably calculated from magnetic flux integrals and the instrument can be calibrated without reference to biopsy data. This BLS calibration has been validated by comparing to LIC from wet-weight biopsies.^2,7 Problems with validation have certainly arisen in the past when non-invasive iron measurement methods like BLS were incautiously compared with LIC from dry-weight biopsies. On the other hand, liver iron is primarily measured in heat-dried biopsies either from fresh-tissue or paraffin-blocks. A critical investigation of the relationship between different liver biopsy preparations was recently performed resulting in a wet-to-dry weight ratio of 5.7±1.4 to 6.3±0.8 for deparaffinized, heat-dried liver samples depending on the drying temperature.⁸ As the wet-weight of liver samples may be affected by different biopsy techniques and analysis in different laboratories, a blinded direct comparison between LIC by BLS and LIC from fresh-tissue heat-dried biopsies excised by cutting needle has been performed for the first time and a conversion factor of 6.1±0.3 (R²=0.86) was obtained.⁹ Without going into further details, this is in good agreement with the CICL670A0107 substudy.

We are running two of the 6–7 biosusceptometer facilities worldwide on a daily and routine basis, having used SQUID biosusceptometry in the diagnosis and follow-up of more than 3,000 patients, and were also involved in MRI-R2 measurements.^3,10 All iron measurement methods (liver biopsy, MRI and BLS) can be reliably used to monitor changes in patients with iron overload under treatment if applied with the necessary expertise. Especially in young patients and in repeated annual monitoring of iron stores, non-invasive methods have a clear advantage. Unfortunately, SQUID-BLS depends on dedicated equipment which is available only in a few specialized centers, although this may change in the future with the incorporation of high-temperature superconducting technology into BLS.⁴ MRI technique would be available in many more locations worldwide, however, it still requires special and long-standing interest of radiologists, MRI imager time in competition with the hospital’s schedule, and stable long-term machine performance, which together may turn out to be a restricting factor in practice. In summary, while Angelucci et al. provide a valuable general survey of the field, the recommendation with regard to non-invasive iron measurements has serious shortcomings. In addition, thorough analysis of available evidence and technical expertise of used methods are also needed before judging the accuracy of specific methods. Although serum ferritin reflects the relative change of body iron in response to iron chelator doses in large study groups,¹¹ it fails to reliably define absolute liver iron concentration or monitor changes of iron stores on an individual basis in patients with thalassemia.¹² Non-invasive techniques for iron quantification are clearly needed to optimally follow patients under iron chelation treatment. BLS or MRI in specialized centers might be the best way to achieve this goal.