Somatic mutations of RNA splicing machinery were identified in myelodysplastic syndromes (MDS), and a strong association was found between SF3B1 mutation and disease phenotype with ring sideroblasts. Recent studies showed that this mutation identifies a distinct group of MDS with ring sideroblasts, characterized by isolated erythroid dysplasia with a high degree of expanded but ineffective erythropoiesis, resulting in inappropriately low hepcidin levels. In this work we analyzed the relationship between SF3B1 mutation, body iron status, hepcidin level and liver and cardiac iron overload detected with MRI T2* in patients with MDS.

We studied 34 patients with MDS diagnosed according to WHO criteria 2008 at the Department of Hematology, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy and at Ospedale Galliera, Genova, Italy (8 patients with RA or RCMD, 20 with RARS or RCMD-RS, 6 with RAEB-1 or -2). Serum hepcidin-25 measurements were performed by a validated Mass Spectrometry based method (SELDI-TOF MS). SF3B1 mutations were analyzed on DNA from circulating granulocytes on an Illumina HiSeq. Magnetic Resonance Imaging (MRI) examinations were performed with a validated scanner using a gradient echo T2* technique.

Somatic mutations of SF3B1 were detected in 19 of 34 patients, with a median variant allele frequency of 0.40 (range 0.12-0.54). Ten of 19 SF3B1-mutated and 10 of 15 SF3B1 wild-type patients (67%) were RBC transfusion-dependent (median number of RBC transfusions 32, range 7-490).

Variable hepcidin levels were found in the patients studied (median value 13.24 nM, range 3.25-75.56, versus 29.19 nM, range 3.26-66.15, in SF3B1-mutated and SF3B1 wild-type patients, respectively). We calculated the hepcidin to ferritin ratio, as a measure of adequacy of hepcidin levels relative to body iron stores, which was inversely related to the SF3B1 mutation (median value 0.015 nmol/mcg, range 0.004-0.035, versus 0.035, range 0.002-0.17, in SF3B1-mutated and SF3B1 wild-type patients, respectively, P=.04). In a multivariable regression model adjusted for RBC transfusion requirement, the hepcidin to ferritin ratio was independently associated with SF3B1 mutation (P=.042), indicating a blunted hepcidin response to iron overload in SF3B1-mutated patients.

We then investigated the relationship between SF3B1 mutation status and parenchymal iron overload estimated by liver and cardiac MRI. Median hepatic T2* in the study population was 7 ms, ranging from 1.49 to 30.08. Focusing the analysis on patients with RBC transfusion-dependency, a higher prevalence of hepatic iron overload (defined as MRI T2* values less than 6.3 ms) was observed in SF3B1-mutated compared with SF3B1 wild-type patients (9/10 versus 5/10 respectively). In a multivariable regression analysis adjusted for RBC transfusion history, SF3B1 mutation was an independent predictor of hepatic T2* value (P=.038). When the analysis was limited to patients without RBC transfusion need, two SF3B1-mutated patients showed a liver iron overload, associated with mild increased in AST and ALT values. Cardiac iron overload defined as MRI T2* values less than 10 ms was detected in two out of 10 patients receiving more than 50 RBC units, both of them carrying SF3B1 mutation.

In conclusion, our study shows that MDS patients carrying a somatic mutation of SF3B1 have inappropriately low hepcidin levels, resulting in parenchymal iron loading due to excessive iron absorption and reticuloendothelial release. These results suggest that SF3B1 mutation may be associated with liver iron overload even in untransfused patients, as observed in congenital iron loading anemias. These results may be relevant for decision-making concerning treatment of transfusion iron overload in MDS patients.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.