Iron overload in acquired sideroblastic anemias and MDS: pathophysiology and role of chelation and luspatercept Free

A 69-year-old man, previously in good health, presents with anemia (hemoglobin 9.1  g/dL) and dyspnea on exertion. Three years ago, his hemoglobin level was 13  g/dL. His white blood count (WBC) is 4900/µL, WBC differential is inconspicuous, and platelet count is 195 000/µL. Although he never received transfusions, his serum ferritin (SF) level is elevated (1400 ng/mL). The transferrin saturation (TSAT) is 38%. Vitamin B12 and folate levels are normal, and renal function is not significantly impaired. His serum erythropoietin level is 140 IU/mL. The bone marrow biopsy and aspirate show expanded erythropoiesis and pathological ring sideroblasts. The karyotype is normal.

Not all patients with acquired sideroblastic anemia (SA) have a clonal bone marrow disorder. Reversible causes of SA include heavy metal poisoning (lead or arsenic), zinc overdose, copper deficiency, vitamin B6 deficiency, and alcohol abuse. Certain drugs like chloramphenicol, isoniazid, or linezolid may also cause SA by disturbing heme biosynthesis (see Table 1).

None of the reversible causes of SA are identified in this patient, whose age suggests that myelodysplastic syndrome with ring sideroblasts (MDS-RS) is the most likely diagnosis. The suspicion is confirmed by mutation analysis, revealing an SF3B1 (K700E) mutation with a variant allele frequency of 42%. The patient is not keen to receive blood transfusions and asks about treatment options addressing the underlying cause of his health problem.

While the above-mentioned reversible conditions can be treated by removing a noxious agent or ameliorating a deficiency, causal treatment for acquired SA in the context of MDS is not available, and our knowledge of the mechanism of ring sideroblast formation in MDS is still incomplete. The patient's systemic iron overload is easier to explain.

Ineffective erythropoiesis, a hallmark of MDS, is particularly pronounced in patients with ring sideroblasts.^1,2 By sending a signal to the liver to suppress hepcidin synthesis, ineffective erythropoiesis enhances intestinal iron absorption. This happens in all iron-loading anemias. The long-sought signal was eventually identified as erythroferrone,³ a protein hormone encoded by the ERFE gene and produced by proliferating erythroblasts under the influence of erythropoietin. If the signal is active over a long time, it precipitates systemic iron overload even before the patient becomes transfusion dependent. In the MDS patient population, the lowest hepcidin levels and hepcidin/ferritin ratios were found in patients with ring sideroblasts and SF3B1 mutation,^2,4 which may explain their propensity to parenchymal iron loading. While we understand that ineffective erythropoiesis, in concert with transfusion therapy, causes systemic iron overload in MDS-RS, the underlying pathophysiology of disturbed heme synthesis and mitochondrial iron accumulation is less well understood.

Erythroblast mitochondria take up large amounts of iron because this is where heme synthesis takes place. However, as soon as ferrochelatase has inserted iron into protoporphyrin IX (the end product of porphyrin synthesis), iron leaves the mitochondria again, now being part of the heme molecule. Heme synthesis is disturbed in MDS-RS. Although protoporphyrin IX is not deficient and ferrochelatase is neither mutated nor underexpressed,⁵ iron is not properly utilized for heme synthesis and thus accumulates in the mitochondrial matrix.

Comprehensive DNA sequencing revealed that about 80% of patients with MDS-RS have an acquired mutation of SF3B1, a component of the splicing machinery.⁶ Mutant SF3B1 causes aberrant splicing in a variety of genes and may thus contribute to ineffective erythropoiesis in several ways. Importantly, ABCB7, a transporter in the mitochondrial membrane, is affected by missplicing.⁷ Prior to the detection of mutant SF3B1, ABCB7 mRNA levels were found to be substantially decreased in MDS-RS.⁸ Another hint that ABCB7 is implicated in the formation of ring sideroblasts came from germline mutations of ABCB7 causing congenital X-linked SA with ataxia.⁹ Furthermore, an induced pluripotent stem cell model recently demonstrated that rescue of ABCB7 strongly reduces ring sideroblast formation.¹⁰ Apparently, mutant or deficient ABCB7 plays an important role in the pathogenesis of mitochondrial iron accumulation.

ABCB7 is usually characterized as a transporter that exports an unknown product of the iron-sulfur (Fe-S) cluster assembly from the mitochondrial matrix to the cytosol. A defect of ABCB7, by depleting cytosolic aconitase of its Fe-S cluster and converting the enzyme into iron regulatory protein 1 (IRP1), may be sensed as cellular iron deficiency, thus stimulating increased cellular iron uptake. However, that is only part of the truth because ABCB7 is also involved in the biosynthesis of heme in erythroid cells through interaction with ferrochelatase.¹¹ Both ferrochelatase and ABCB7 are part of the mitochondrial heme metabolism complex,¹² which comprises several proteins that facilitate substrate channeling and help to coordinate iron and porphyrin metabolism.

ABCB7 is required to stabilize ferrochelatase, which must form a homodimer to be functional. Mammalian ferrochelatase possesses an Fe-S cluster that is needed for stability and function. Posttranslational stability of newly formed ferrochelatase is dramatically decreased during iron deficiency or disturbed Fe-S cluster assembly in the mitochondria.¹³ Pull-down assays revealed that ABCB7 protein interacts with the carboxy-terminal region of ferrochelatase containing the Fe-S cluster. Importantly, knockdown of ABCB7 leads to significant loss of mitochondrial proteins containing an Fe-S cluster and to decreased stability of ferrochelatase.¹⁴ 

The role of Fe-S cluster incorporation into ferrochelatase is underscored by rare cases of HSP9A and HSCB mutations that cause congenital SA.^15,16 Both heat shock proteins are mitochondrial chaperones that facilitate the transfer of nascent [2Fe-2S] clusters to recipient mitochondrial proteins. There are also rare cases of congenital SA with GLRX5 mutations.^17,18 Tetrameric glutaredoxin 5 is an Fe-S cluster carrier that binds 4 glutathions and 2 Fe-S clusters, and mutations of GLRX5 decrease the stability of ferrochelatase.¹⁹ Interestingly, ABCB7 has a potential glutathione binding site and a proposed role of transporting a glutathione-coordinated Fe-S cluster [2Fe-2S](GS)₄.²⁰ As summarized in Figure 1, SA may thus be attributable to problems at different stages of mitochondrial Fe-S cluster assembly and Fe-S cluster delivery to ferrochelatase.

The evidence mentioned above suggests a mechanism for mitochondrial iron accumulation in acquired (MDS-related) SA as depicted in Figure 2. The diagram indicates that ABCB7 deficiency leads to impaired stabilization of ferrochelatase because ABCB7 fails to provide sufficient support for Fe-S cluster incorporation into ferrochelatase. Consequently, dysfunctional ferrochelatase cannot utilize the iron that is imported into mitochondria for heme synthesis. Iron crosses the inner mitochondrial membrane as Fe²⁺. If not immediately accepted and processed by ferrochelatase, Fe²⁺ will autoxidize to Fe³⁺, thus becoming useless for heme synthesis. According to this model, impaired ferrochelatase protein function plays a pivotal role in producing the sideroblastic phenotype, even though the ferrochelatase gene is neither mutated nor underexpressed.

Due to ineffective erythropoiesis, the patient has already developed a modest degree of systemic iron overload. Is it time to start iron chelation therapy (ICT)? Since most guidelines suggest starting ICT when SF exceeds 1500  ng/mL, the patient first receives epoetin alfa, which succeeds in avoiding transfusion therapy and raises the hemoglobin to around 10.5  g/dL for 15 months. After relapse, luspatercept is given, achieving another transfusion-free period of 1 year, with hemoglobin levels around 10  g/dL. Thereafter, anemia worsens again, the patient requires transfusion therapy, and SF quickly rises to 1800  ng/mL. Iron chelation with deferasirox (DFX) is started.

Iron-related oxidative stress is toxic, and widespread subclinical organ dysfunction can result from transfusional iron overload developing in adulthood.²¹ However, many MDS patients do not live long enough to develop clinical complications of iron overload. Furthermore, it is difficult to determine the extent to which iron overload contributes to morbidity and mortality in older patients with MDS, because iron-related complications overlap with age-related medical problems (Figure 3).

Nevertheless, an SF >1000  ng/mL showed a clear dose-dependent impact on overall survival in patients with low-risk MDS,²² and SF was an independent prognostic factor even if transfusion burden, reflecting the degree of bone marrow failure, was included in multivariate analysis. Similarly, data from the European LeukemiaNet prospective MDS registry indicated that besides transfusion burden, which was the most important prognostic factor, increasing levels of SF had an independent impact on overall survival of transfusion-dependent patients with lower- risk MDS. This registry also found inferior survival rates when labile plasma iron was detectable in patients with lower-risk MDS, whether the patients were transfusion dependent or not.²³ An update with a larger number of patients confirmed that transfusion dependency is associated with the presence of toxic iron species and inferior survival.²⁴ 

Since ICT provides several potential benefits to patients with transfusion-dependent MDS (Figure 4), it may extend overall survival. This is suggested by multiple retrospective studies, many of which are not totally convincing because patient populations were usually well characterized regarding disease-related parameters and risk factors but not stratified according to performance status. Therefore, fitter patients may have been more likely to receive ICT, thus inflating its benefit. Two registry studies went to great lengths to avoid such bias. The Canadian MDS Registry, which carefully documents performance status and comorbidities, showed a significant survival benefit of ICT when roughly comparable cohorts were analyzed. This finding was corroborated by harnessing several important matching criteria.²⁵ Another thorough analysis comes from the prospective European LeukemiaNet MDS (EUMDS) Registry, which similarly adjusted for relevant prognostic factors and confirmed a significant survival benefit of ICT.²⁶ To address potential problems with time-varying confounding, a further investigation used marginal structural outcome modeling to study the association between ICT using DFX and important clinical outcomes, including survival, in a large US Medicare-based MDS patient cohort. Among patients who received a minimum of 20 blood transfusions, DFX therapy was associated with a significantly reduced risk of death that was proportional to the duration of DFX therapy.²⁷ In a meta-analysis of 9 observational studies, ICT was associated with an overall lower risk of mortality compared with no ICT (adjusted hazard ratio 0.42; 95% CI 0.28-0.62; P  <  0.01); however, there was significant heterogeneity across the studies.²⁸ Overall, these studies all point in the same direction, which makes it hard to ignore their results.

The only prospective randomized trial of ICT in MDS is the Telesto study, which showed a significant advantage for DFX vs placebo.²⁹ The primary endpoint of the study was event-free survival, which was measured using a composite primary endpoint of time from randomization to time of first documented nonfatal event (related to cardiac and liver function and transformation to AML) or death. The Kaplan-Meier curves for event-free survival separate after about 2 years, which is plausible because it takes time for iron-related organ damage to develop and, accordingly, for a clinical benefit of ICT to manifest itself.

Some “rules of thumb” are offered in Table 2. An excellent, more comprehensive treatise on iron chelation in MDS was recently published in Blood.³⁰ SF levels are generally used to detect iron overload. In addition, TSAT is important, since TSAT levels > 70%-80% can serve as surrogate parameter for the presence of toxic labile plasma iron, which cannot be measured in routine clinical practice. TSAT levels > 70%-80% were found to be a prerequisite for parenchymal iron loading,³¹ and TSAT > 80% predicts inferior clinical outcomes in MDS.³² Cardiac magnetic resonance imaging (MRI) should be considered in patients with cardiac comorbidities, since clinical outcomes for MDS may be aggravated by myocardial iron overload, with adverse effects on prognosis (Figure 5). The heart is more vulnerable to iron overload than the liver, with clinically relevant cardiac dysfunction occurring at much lower tissue iron concentrations than clinically relevant liver dysfunction.³³ About 18% of MDS patients have cardiac iron overload detectable by MRI.³⁴ 

In general, MDS patients should have a life expectancy of at least 1-2 years to be likely to benefit from ICT. If persistent ineffective erythropoiesis or transfusion need is anticipated, most guidelines recommend starting ICT when SF levels exceed 1500  ng/mL.

A short characterization of approved iron chelators is given in Table 3. Due to its short half-life and need for continuous parenteral infusion, deferoxamine (DFO) is not widely used any more. However, it can be combined with the oral iron chelator deferiprone (DFP) to achieve enhanced removal of cardiac iron, based on a shuttle effect that employs the superior capacity of DFP to mobilize cardiac iron,³⁵ which is then transferred to DFO for excretion. When using DFP, regular blood counts must be carried out to detect agranulocytosis, though the incidence of DFP-induced agranulocytosis is not increased in MDS compared to patients with nonmalignant bone marrow disorders.³⁶ The most widely used iron chelator in patients with MDS is DFX, owing to its effectiveness³⁷ and convenient oral administration. In a fraction of MDS patients, iron chelation with DFX leads to hematologic improvement, probably through diminished oxidative stress in the bone marrow.³⁸ Since DFX decreases renal plasma flow and glomerular filtration rate (GFR), dosage must be adjusted in patients with impaired renal function. Furthermore, patients on DFX therapy should be monitored for visual disturbances and hearing loss.

Although a recent case report suggests that luspatercept also works in congenital X-linked SA,³⁹ its main role is to improve erythropoiesis in patients with MDS-RS. Luspatercept is a recombinant fusion protein that binds transforming growth factor β (TGF-β) superfamily ligands to reduce the SMAD2/SMAD3-dependent signaling implicated in suppression of erythropoiesis. Luspatercept was approved by the US Food and Drug Administration (FDA) after the MEDALIST trial,⁴⁰ a randomized, placebo-controlled, phase 3 clinical trial, demonstrated that patients with very low- to intermediate-risk MDS-RS or MDS/MPN-RS-T (myeloproliferative neoplasia with ring sideroblasts and thrombocytosis) achieved higher rates of transfusion independence when treated with luspatercept (38%) compared to placebo (13%). FDA approval was extended to erythropoiesis stimulating agent-naive patients with very low- to intermediate-risk MDS, irrespective of ring sideroblast status, after the COMMANDS trial⁴¹ showed that luspatercept induced higher rates of red blood cell (RBC) transfusion independence for at least 12 weeks with concurrent mean Hb increase of ≥1.5  g/dL (weeks 1-24) than did epoetin alfa (59% vs 38%). Noteworthy, luspatercept was superior to epoetin alfa only in patients with ring sideroblasts (72.5% of the study population) or SF3B1 mutation (60.5% of the study population). The high proportion of such patients in the study explains why the trial met its primary endpoint.

Luspatercept treatment may obviate the need for RBC transfusion for more than a year, thus diminishing further iron loading. However, luspatercept cannot be expected to substantially reduce the existing iron overload. If luspatercept induces a rise in hemoglobin by 2  g/dL, the associated increase in red cell mass, roughly equivalent to 2 RBC units, requires mobilization of about 2  ×  200  mg of iron, mainly from macrophages. This amount of storage iron recycled into improved erythropoiesis is small compared to >10  g of iron that accumulates, for instance, over 2 years in a patient with a moderate transfusion requirement of 2 RBC units per month. Accordingly, the MEDALIST trial showed that the mean change from baseline in SF was negligible (−2.7  µg/L averaged over weeks 9-24, and −72.0  µg/L averaged over weeks 33-48).⁴⁰ 

By augmenting erythropoiesis and decreasing hepcidin, luspatercept may redistribute iron from macrophages to hepatocytes, as recently shown in patients with transfusion-dependent thalassemia, necessitating the comcomitant use of iron chelators for the effective management of iron overload.⁴² In the best case, luspatercept increases hemoglobin sufficiently to enable phlebotomies, which are clearly the most efficient way to remove excess iron. If successful in diminishing oxidative stress, they provide an additional opportunity to improve erythropoiesis. If phlebotomies are not feasible, the same goal may be pursued with iron chelation.

Norbert Gattermann: Research funding by Takeda. Lecture honoraria by Novartis and Bristol-Myers-Squibb.

Norbert Gattermann: nothing to disclose.