Systemic Iron Overload in Patients Undergoing Allogeneic Stem Cell Transplantation – a Magnetic Resonance Imaging Based Study in 81 AML and MDS Patients

Abstract 489

Systemic iron overload (SIO) can be frequently encountered in AML and MDS patients (pts) and predominately occurs as a consequence of recurrent red blood cell transfusions (RBC). SIO has been associated with an increased risk for infectious complications and acute graft-versus-host disease (aGvHD) as well as with excessive early non-relapse mortality (NRM) after allogeneic stem cell transplantation (allo-SCT). However, most of these studies relied on surrogate markers like ferritin and transfusion burden, which might be interference prone in this patient population. Consequently, the prognostic impact of these parameters is under considerable debate and a variety of different thresholds for risk stratification have been proposed. Studies using improved and objective assessments of SIO like magnetic resonance imaging (MRI) are rare and limited in patient number. We aimed at determining prospectively the influence of SIO as quantified by MRI on post-transplant outcome in a large cohort of AML and MDS patients undergoing allo-SCT.

From 2009 on, MRI-based assessment of liver iron content (LIC) according to the methods described by Gandon (Lancet 2004) and Rose (Eur J Haematol 2006) was routinely performed prior to conditioning in all AML and MDS patients at risk for SIO undergoing allo-SCT at our center. Further, serological parameters of SIO were determined at the same time. Post-transplant outcome including the occurrence and extent of GVHD as well as NRM were correlated with variables of SIO. Categorial variables were assessed using Fisher's excact test, while competing events statistics was used to compare cumulative incidences of NRM and aGvHD. Correlations are reported as Spearmans rank correlation coefficient (r).

Over a period of 30 months 59 AML- and 22 MDS patients with a median age of 57 years were prospectively screened with liver MRI. Median LIC was 140 μmol/g (range: 40 – 350 μmol/g). Ferritin was elevated in all patients with a median of 2204 ng/ml (range: 305 – 45049 ng/ml) and patients had received a median of 24 units of packed red blood cells (RBC, range: 4 – 127).

There was a strong positive correlation between transfusion burden and LIC (r = 0.702, p<0.001) as well as between ferritin and LIC (r = 0.594, p < 0.001). A threshold of 20 or more RBC, which is widely accepted as a good marker for SIO, was found to predict an elevated LIC (>=125 μmol/l) with a sensitivity and specificity of 70.0 % and 81.8 %, respectively. Contrasting, the commonly used criterion of ferritin above 1000 ng/ml albeit, being very sensitive (95.5%), provided only very poor specificity (27.0%). Increasing the ferritin threshold to 2500 ng/ml, a level known to correlate with increased NRM, lead to increased specificity 81.1% at the price of moderately reduced sensitivity (63.6%).

None of the three SIO parameters was associated with an increased risk of aGvHD or infections after allo-SCT, while preexisting aspergillosis was more common in iron overloaded patients (LIC >= 125 μmol/g: 33.3% vs. 0.0 %, p <0.001; RBC >=20: 27.1 vs. 3.4%, p = 0.013; Ferritin >= 1000 ng/ml: 21.2% vs. 0.0 %, p = 0.199). A moderate correlation between LIC (r = 0.494, p < 0.001) as well as transfusion burden (r = 0.478, p < 0.001) and the time between diagnosis and allo-SCT was noted, while ferritin was not associated with that parameter. Interestingly, both transfusion burden (r = 0.248, p = 0.026) and ferritin (r = 0.329, p = 0.003) but not LIC showed a weak but significant correlation with hematopoietic transplantation comorbidity scores.

Regarding NRM we were able to show only a modest trend for transfusion burden of 20 or more RBC as a predictor for an adverse prognosis (100 day CI of NRM: 26.2% vs. 7.7%, p = 0.080). Furthermore, ferritin above 2500 ng/ml (CI: 26.0 vs. 13.4%; p = 0.231) did not correlate with NRM. In contrast, an LIC of 125 μmol/g or more, which is known to be associated with organ toxicity in thalassemia patients, predicted for a significantly increased risk of NRM (CI: 30.8 % vs. 6.3%; p = 0.016). Multivariate analysis confirmed LIC but not transfusion burden or ferritin as an independent risk factor for an increased NRM (HR 1.007 for every 1 μmol/g increase, p = 0.022).

We conclude that systemic iron overload is an independent negative prognostic factor for post-transplant outcome in AML and MDS patients but definition of SIO should be based on reliable parameters like MRI- measured LIC.