Reproducibility and Sensitivity in Assessing Liver Iron Content (LIC) in Highly Transfused Patients by Routine Magnetic Resonance Imaging(MRI).

Intoduction Due to the poor signal to noise ratio of MRI, there are no strong data pertaining to the validity of MRI techniques in the assessment of LIC in high post-transfusionnal iron overload. Moreover, reproducibility (from center to center) of this technique is not clearly established. We have compared chemically determined high LIC and two different MRI protocols of measuring LIC in patients transfused monthly: the first, to evaluate the exact reproducibility of a simple and widely recognized procedure used in hepatic diseases and the second, to improve sensitivity at higher LIC.

Patients and Methods

MRI method: MRI studies were performed using a 1.5 T system (Philips intera, The Netherlands). Transverse images were recorded, visualizing the right liver lobe (L) and posterior vertebral muscles (M) in the same slice. The signal intensity (SI) for hepatic tissue and the posterior vertebral muscle were determined. The L/M SI ratio (SIR) was then calculated. First procedure: Four magnetic resonance gradient echo sequences (T2++/t2+/PD/T1), exactly as previously published in hepatic disease studies (Gandon,Y Lancet 2004), were performed and the same software for LIC determination was used in order to evaluate the center to center reproducibility. Second procedure: EGT0.5 and EGT0 sequences using shorter echo times with the aim of improving accuracy at higher levels of iron overload were performed.

Patients and liver biopsy: 21 liver biopsies were performed, after written consent, in 16 patients (MDS=4, b thalassemia=12). In the case of repeated biopsy, an interval of a minimum 12 months was respected between the two biopsies. Results: 21 biopsies, each more than one mg, in 16 adult patients (8 males, 8 females) with a median age of 41 (range 23–73 ). LIC ranged from 2.6 to 54mg/g liver dry weight (mg/gdw), (median 12.6 mg/gdw). Hepatic fibrosis and cirrhosis were histologically found in 4/21 and 4/21 cases, respectively. First procedure: In biopsies with LIC below 375 micromol/g dw (n= 14), a strong correlation was demonstrated between LIC and L/M SIR (r = 0.86), for the best sequence, similar to that previously reported (r =0.91). For the entire group, including cases with LIC > 375 micromole/g dw (n= 21), the correlation was not as good (r=0.72) as the previous group using the same sequence. Second procedure: EGT0.5: 4 cases having low LIC and SIR >1.5 were excluded. Among the 17 remaining cases, linear agreement between LIC and L/M SIR was excellent (r=0.89). EGT0: 16 cases had SIR<1.5 and were analyzed with similar results (r=0.83).

Conclusion In mild post-transfusionnal iron overload, our results show: measurement by routine MRI produces an accurate and reproducible estimation of LIC from center to center, with the same level of correlation as previously reported, independent of the machine used. At high liver iron concentration (LIC> 375 micromol/gdw) (33% of our cases), some simple and rapid specific sequences can also provide an accurate measurement of LIC. These results suggest that, whatever the level of iron overload encountered, the LIC determination by simple and rapid routine MRI is more a question of the standardization than of the sensitivity of the method.