the effect of iron chelation on haemopoiesis in mds patients with transfusional iron overload

The effect of iron chelation on haemopoiesis in MDS patientswith transfusional iron overload

P. D. JENSEN,1 L. HEICKENDORFF,2 BENT PEDERSEN,3 K. BENDIX-HANSEN,4 F. T. JENSEN,5 T. CHRISTENSEN,5

A. M. BOESEN1

AND J. ELLEGAARD1 1Department of Medicine and Haematology, 2Department of Clinical Biochemistry,

3Department of Cytogenetics, Danish Cancer Society and 4Institute of Pathology, Aarhus Amtsygehus, and5Centre for Nuclear Magnetic Resonance, Skejby Sygehus, Aarhus University Hospital, Denmark

Received 6 October 1995; accepted for publication 24 April 1996

Summary. Long-term follow-up data are presented onchanges in peripheral blood counts and Hb requirements of11 patients with myelodysplastic syndromes (MDS) duringiron chelation treatment with desferrioxamine for up to 60months. The erythroid marrow activity was indirectlyevaluated by repeated determinations of the serum transfer-rin receptor concentration. The efficacy of iron chelation wasevaluated by repeated quantitative determination of the liveriron concentration by magnetic resonance imaging.

Reduction in the Hb requirement (550%) was seen in7/11 (64%) patients. Five patients (46%) became bloodtransfusion independent. Platelet counts increased in 7/11(64%) patients and the neutrophil counts in 7/9 (78%)

evaluable patients. All patients in whom iron chelation washighly effective showed improvement of erythropoietic outputaccompanied by an increase in the serum transferrin receptorconcentration.

It is concluded that reduction in cytopenia in MDS patientsmay be accomplished by treatment with desferrioxamine, ifthe iron chelation is efficient and the patients are treated for asufficiently long period of time. Exactly how treatment withdesferrioxamine works remains a challenge for furtherinvestigation.

Keywords: iron overload, desferrioxamine, myelodysplasticsyndromes, transferrin receptor.

The myelodysplastic syndromes (MDS) are clonal disorders ofhaemopoiesis characterized by abnormalities in proliferationand differentiation, resulting in ineffective haemopoiesisinvolving one or more cell lineages and peripheral bloodcytopenia. The present management of MDS patients ismostly limited to supportive care for symptoms related toblood cytopenias. In many cases the patients develop ironoverload due to multiple blood transfusions, which may befollowed by serious organ damage (Schafer et al, 1981;Hoffbrand & Wonke, 1989). These complications may beprevented by iron chelation with desferrioxamine (DFO).

Recently, we have reported ( Jensen et al, 1992), thatremoval of the transfusional iron overload in MDS patients byiron chelation with DFO may reduce the blood transfusionrequirement. Thus, we found a significant reduction in the Hbrequirements from 12% to 46% in 3/6 MDS patients whowere treated with iron chelation for 6–17.5 months. In two ofthese patients, who had pancytopenia before iron removal,the platelet count also increased significantly as did the

neutrophil count in one patient. This report was basedexclusively on peripheral blood counts and the progress ofiron chelation was monitored only by repeated serum ferritindeterminations.

This study is an extension of our previous report and isbased on 11 MDS patients followed for up to 60 months(during and after treatment with DFO). The aim of this studyis to present long-term follow-up data on peripheral bloodcounts and the Hb requirements of these patients, and todescribe the morphological and cytogenetical bone marrowchanges observed during iron chelation, as well as changes ofthe erythroid marrow activity, as determined indirectly byrepeated measurements of the serum transferrin receptor(TfR) concentration. A further aim was to investigate therelationship between the efficacy of iron chelation asdetermined by repeated quantitative determinations of theliver iron concentration by magnetic resonance imaging(MRI) and treatment response.

MATERIALS

The study, approved by the local ethical committee, included11 consecutive patients with a confirmed diagnosis of MDS.

British Journal of Haematology, 1996, 94, 288–299

288 # 1996 Blackwell Science Ltd

Correspondence: Dr P. D. Jensen, Department of Medicine andHaematology, Amtssygehuset, Aarhus University Hospital, TageHansens Gade 2, DK-8000 Aarhus C, Denmark.

289Iron Chelation in MDS Patients

# 1996 Blackwell Science Ltd, British Journal of Haematology 94: 288–299

No patients were excluded from the study. According to thecriteria of the French–American–British (FAB) Co-operativeGroup’s classification (Bennett et al, 1982), two patients (nos.6 and 10) had refractory anaemia with ringed sideroblasts(RARS), two (nos. 2 and 8) refractory anaemia with excess ofblasts (RAEB), one (no. 11) stable AML developed from MDS,and six (nos. 1, 3, 4, 5, 7 and 9) refractory anaemia (RA). Allpatients were in a stable phase of the disease when enteringthe study. Only patients <70 years of age were included intothe study. No cytostatic treatment was given throughout thestudy. Three patients (nos. 4, 10 and 11) had splenomegaliaas determined by sonography. The clinical characteristics ofthe patients, iron store parameters, and transfusion historywhen entered in the study are summarized in Table I.

METHODS

DFO therapy. DFO was administered subcutaneously in theabdominal wall, by 12 h infusions 5 d a week by use of acomputer-assisted infusion pump (Deltec, CADD II) in fourpatients (nos. 5, 7, 9 and 10), and in seven other patients bysubcutaneous bolus injections (1 g twice daily). At thebeginning of the treatment a DFO dose–response curve wasassessed in order to tailor the dose. Increasing doses of DFO,from 1 g to 4 g, were given by 12 h subcutaneous infusions on4 successive days with 24 h collection of urine from thebeginning of each infusion (Pippard, 1989). The tailored DFOdose during the whole study was 2 g in all patients, except inpatient 1 who was treated by 2.5 g once daily. Supplementa-tion with vitamin C (200 mg once daily) was given to ninepatients, and was started 9–21 months after initiation of DFO

therapy. The patient compliance was optimized by regularpatient–doctor interviews, revealing intermittent non-com-pliance in only one patient (no. 8).

Bone marrow examination. Thin sections (4�m) of paraffin-embedded, fixed bone marrow were investigated in all 11patients before treatment and re-investigated in 10 patients(not done in no. 9) after 13–37 months of treatment.Estimates of haemopoiesis in percent of total bone marrowvolume were performed using point counting on histologicalsections stained with chloroacetate-esterase. Furthermore,the amounts of iron in the bone marrow macrophages wereestimated semi-quantitatively. Smears from the bone marrowaspirates were reviewed; 300 nucleated bone marrow cellswere evaluated for each smear.

Peripheral blood counts. The haemoglobin (Hb) concentra-tion, the neutrophil count (NC) and the platelet count (PLC)was measured just before each transfusion episode using anautoanalyser (H-2, Technicon).

Blood transfusion requirement. The baseline value at start ofthe study was calculated as a mean value from the last threetransfusion periods before entering the study, based on thepre-transfusional Hb concentration (generally performed onthe day of blood transfusion) and the amount of Hb given ineach period. The requirement was calculated in g Hb permonth according to the following formula:

�A� BV� B� 75� ÿ �C� BV�D

� 31

where A = pre-transfusional Hb concentration [g/dl] at startof a transfusion episode; B = number of blood units transfusedthat time; C = pre-transfusional Hb concentration at start of

Table I. Demographic and laboratory findings in 11 MDS patients with transfusional iron overload at start of iron chelation.

Iron storeTransfusion history Erythroid marrow

Diagnosis* Liver Serum activitySex Age iron ferritin Duration Hb requirement Red cells Pre Hb

Pt FAB-type (F/M) (yr) (�mol/g) (�g/l) (months) (g Hb/month) (units†) (g/dl) S-TfR S-EPO

1 MDS, RA F 67 521 4340 22 261 77 8.6 0.53 1702 MDS, RAEB F 63 558 1780 34 337 116 8.9 0.50 2463 MDS, RA+ERD F 47 403 2510 28 350 105 8.7 0.69 14 MDS, RA F 64 540 5770 19 382 88 7.8 0.42 7505 MDS, RA F 46 558 3800 18 251 75 8.8 0.99 2356 MDS, RARS M 69 491 4790 30 339 90 9.6 2.36 677 MDS, RA F 41 623 8715 226 174 254 8.6 1.12 3348 MDS, RAEB F 18 378 870 15 186 51 8.9 1.36 919 MDS, RA F 45 370 4390 7 447 51 9.1 1.13 536

10 MDS, RARS M 58 501 3600 40 563 132 10.9 3.02 37211 MDS, AML F 65 626 7370 35 490 207 8.3 n.d. n.d.

Pre Hb� pre-transfusional Hb concentration. Pre Hb and Hb requirement given as means of last three determinations before initiation oftreatment.

Normal ranges: MRI determined liver iron concentration, 1–15 �mol Fe/g dry weight (mean�2 SD); S-Epo: 3.1–14.4 mIU/ml (2.5–97.5%percentiles); S-Trf: 1.14–2.85 �g/ml (2.5–97.5% percentiles).

* MDS�myelodysplastic syndromes; RA� refractory anaemia; RARS� refractory anaemia with ring sideroblasts; RAEB� refractory anaemiawith excess of blast cells; AML� acute myeloid leukaemia; ERD� end-stage renal disease; n.d.� analysis not done.

† 1 blood unit red cells equals 75 g Hb.

the following transfusion episode; D = number of daysbetween the transfusion episodes; 75 = mean amount Hbin g of one unit red cells; 31 = 31 days/month; BV = bloodvolume, calculated from the plasma volume and the actualhaematocrit according to: BV = plasma volume [litres]�100/(100ÿhaematocrit) (Ganong, 1979). The plasma volumewas calculated from the patient’s weight at start of DFOtreatment according to: plasma volume [litres] = weight[kg]�4.5% (Ganong, 1979).

Clinical response to iron chelation was defined as areduction of the blood transfusion requirements of at least50% when compared with the baseline values.

Iron store parameters in serum. Blood samples were, in allcases, drawn in the morning (8–10 a.m.). The serumconcentration of ferritin was measured by commerciallyavailable immunometric assays, during the first 2 years of thestudy by IRMA assay (Biorad, Quantum Ferritin), and duringthe following two years by an enzyme-immunometric assay(Amerlite, Ferritin assay, monoclonal), based on enhancedluminescence. Quality control experiments revealed coeffi-cients of variation from 5.1% to 8.1%. The reference rangewas 10–120�g/l for females and 15–300�g/l for males. Thedegree of correlation between the two applied methods is high(r = 0.99).

Serum transferrin receptor (TfR). The concentration wasmeasured by an enzyme immunoassay based on polyclonalantibodies against the human transferrin receptor (R&DSystems, Abington, U.K.). The serum samples were stored atÿ808C until assay. The samples were diluted 1:100 andanalysed in duplicates. Intra- and interassay coefficients ofvariation of 3% and 6% were found. A reference range basedon analysis of samples from 40 healthy individuals of 1.14–2.85�g/ml (2.5–97.5% percentiles) was found. In patient9 no serum sample had been collected, and in two patients(nos. 2 and 8) no serum sample was obtained before start oftreatment.

Serum erythropoietin (EPO). The concentration was mea-sured by an enzyme-linked immunoabsorbent assay based onmonoclonal and polyclonal antibodies against human ery-thropoietin (R&D Systems, Abington, U.K.). The standardsused were recombinant human EPO calibrated against W.H.O.standard 67/343. The samples were analysed in duplicatesundiluted and when necessary repeated at a dilution of 1:10.Intra- and interassay coefficients of variation of 3% and 8%were found. A reference range based on analysis of samplesfrom 40 healthy individuals of 3.1–14.4 mIU/ml (2.5–97.5%percentiles) was found.

Liver iron concentration. The quantitative determination ofthe liver iron concentration was performed by MRI of the liveras previously described ( Jensen et al, 1994). In short, spinecho images were obtained using a Philips Gyroscan, S 15-HP,operating at 1.5 tesla. Time of echo was 25 ms, the repetitiontime was dependent on heart rate (cardiac gating), between480 and 800 ms. Signal intensity measurements wereperformed on one or two oblique images, right liver lobeand posterior vertebral muscles in the same slice ( Johnston etal, 1989). The signal intensity ratio between liver tissue andskeletal muscle was calculated and adjusted to a constantvalue of the repetition time (684 ms). The daily calibration of

the MRI system comprised control of resonance frequency ofthe system and test of the signal to noise ratio using aperformance phantom. The established reference range was1–15 (mean�2 SD)�mol Fe/g dry weight (investigated in 15healthy controls). The inter-recording variation investigatedin normal controls was 2.9�2.7 (mean� SD) �mol Fe/g dryweight.

Cytogenetics. Karyotyping was carried out at start ofDFO treatment and two to six times between 2 and44 months later. Bone marrow cells were culturedaccording to the high-resolution technique described byYunis (1981). After 3–5 h of culture, methotrexate (10ÿ5

M)was added. The S-phase block was released with thymidine(10ÿ3

M) after 17 h. Colchicine was added 10 min beforetermination of the culture period. The cells were harvestedwith conventional methods, and the slides were ‘aged’ byheating at 608C for 17 h before banding. Giemsa bands wereproduced with Wright’s stain as described previously(Pedersen & Kerndrup, 1986).

Primed in-situ labelling (PRINS). Aspirated bone marrows(2–3 ml aliquots) were transferred into a 10 ml test tubecontaining 2–3 ml 0.9% saline and 100 IU of heparin.Mononuclear cell suspensions were prepared by densitycentrifugation using Lymphoprep (1.077g/ml, NycoMed,Norway). After washing, the cells were cryopreservedin a mixture of 0.5 ml 20% FCS and 0.5 ml 20% DMSO(Merck, Darmstadt, Germany) in 1.8 ml polypropylenetubes (Nunc, Roskilde, Denmark), and stored at ÿ1808C inliquid nitrogen. Before analysis the samples were thawedrapidly and transferred to RPMI containing 20% FCS,centrifuged at 200 g for 5 min, decanted, and washedonce in 20% FCS/RPMI medium. The single cell suspensionwas transferred to a 50 ml plastic tube containing 30 ml KCl(5 g/l). After 10 min at room temperature the cells werefixed four times by adding 5 ml methanol/glacial acetic acid(3/1) followed by centrifugation (200 g, 8 min) and aspirationof the supernatant.

PRINS was performed with an oligonucleotide probespecific for centromeric alpha satellite DNA of chromosome8 as previously described (Koch et al, 1995). A few �l of fixedcells were pipetted directly on to the slide. The site ofapplication was premarked at the undersurface of the slide toallow up to eight different samples to be handled on thesame slide. Within 2 h of the slide preparation, the PRINSsignals could be analysed under a microscope. Withineach sample 100 nuclei were analysed. The 100 nucleichosen were the first 100 found in which the number ofsignals in the nuclei could be unequivocally determined.

Statistical methods. All data sets were analysed for com-patibility with normal distribution by the Kolmogorov-Smirnov goodness-of-fit test (Lilliefors’ modification). Incase of compatibility the two-sample t-test (pooled variance)was used when comparing mean values between differentgroups of patients. Means before and after treatment onthe same patients were compared by paired t-test. In case ofnon-compatibility, Wilcoxon’s signed ranks test was per-formed. Regression analysis was performed according tothe least-squares method. The level of significance was setat 5%.


290 P. D. Jensen et al



RESULTS

Hb requirementsDuring iron chelation the Hb requirement fell clearly in 7/11(64%) patients (Fig 1a, c, e–i), who had been on ironchelation for 6–60 months. Four of these patients (36%)became blood transfusion independent after 18–26 monthsof treatment and have now remained so for from 3 months to3 years. In another patient (Fig 1g) the transfusionrequirement was reduced by 70% when she stopped ironchelation treatment for personal reasons, although shebecame transfusion independent after 15 months withouttreatment. In two further patients (nos. 1 and 9) thetransfusion requirement was reduced by 88% and 71%respectively at the time when DFO was stopped, in patient9 due to bone marrow transplantationand in patient 1 due tonormalized iron stores. Four patients (nos. 2, 4, 10 and 11)showed no clear reduction or even increase (no. 10) of Hbrequirement. The baseline Hb requirement was very high inthese patients (5–7.5 blood units per month), and, interest-ingly, three patients (nos. 4, 10 and 11) showing the highestHb requirements had spleen enlargement, which was notencountered in any of the other patients.

The mean Hb requirements for all 11 patients decreasedfrom 343 (SD 123) g Hb per month before iron chelation to166 (SD 221) g Hb/month (P = 0.005, Wilcoxon’s signedrank test) at the last time of follow-up, or when treatment wasstopped (patients 1 and 9). This reduction was accompaniedby an increase of the mean pre-transfusional Hb concentra-tion in 8/11 cases (Fig 2a, c–i) and for the whole group from8.8 (SD 0.57) g Hb/dl at baseline to 10.5 (SD 1.8) g Hb/dl atlast follow-up (P = 0.02, Wilcoxon’s signed rank test).

Peripheral blood countsFollow-up of the peripheral blood counts (Fig 3) revealed anincrease in platelet count (PLC) and/or neutrophil count (NC)in 9/11 (82%) patients (nos. 1–9). PLC increased clearly in 7/11 (64%) patients (nos. 2, 3, 5–9), of whom four had severethrombocytopenia (<20� 109/l) at baseline (nos. 5, 7–9).The NC increased in 8/9 (89%) evaluable patients (nos. 1, 4–9), in four cases from baseline values <1.0� 109/l. In fivepatients (nos. 5–9) a trilineage haematological improvementwith a simultaneous rise in haemoglobin level, PLC and NCwas observed. Three patients (nos. 5, 7 and 8) with trilineageresponse were followed after discontinuation of DFO. In thesepatients the PLC increased further but the NC remainedessentially unchanged, and all three patients remained bloodtransfusion independent. One patient (no. 11) progressed toovert AML after discontinuation of DFO (Fig 3k).

Bone marrow examinationThe FAB-type of the patients at the start of treatment is listedin Table I. A re-examination was performed in 10 patients(not done in patient 9) after 13–37 months of treatment.In six of these cases no further bone marrow examination wascarried out because the treatment was terminated; in fivecases (nos. 3, 5–8) the patients had become blood transfusionindependent and in one case (no. 1) the iron stores hadbecome normal. During treatment the FAB-type changed only

in two patients, in both cases from RA to RAEB, accompaniedby an increase of the percentage of myeloblasts from 2% to15% in one case (no. 4), and from 3% to 20% in another case(no. 1). After 21 months without iron chelation the latter hadprogressed to AML. Before start of treatment the bonemarrow showed erythroid hypercellularity in 9/11 patients.At the time of re-investigation the cellularity was changed inonly two patients, in patient 5 from hypo- to hypercellularand in patient 2 from normo- to hypercellular. The bonemarrow iron content was strongly increased in all 11 patientsbefore treatment and remained unchanged in all patients atfollow-up, except in patient 1 in whom marrow iron wasnormalized after 27 months of iron chelation.

Chromosome analysisAll 11 patients were karyotyped. Seven (nos. 2, 3, 5 and7–10) had a normal bone marrow karyotype before the startof iron chelation, whereas structural and/or numericalaberrations were observed in four patients (nos. 1, 4, 6 and11). Six cases with a normal initial karyotype werere-examined (not done in patient 9) between two and six(mean 3.5) times. In five cases the karyotype remainednormal, and three of these patients (nos. 3, 5 and 8) becameblood transfusion independent, whereas another two (nos. 2and 10) did not. The sixth patient (no. 7) remainedcytogenetically normal after 6 and 16 months of DFOtreatment and her Hb requirement decreased by 70%. 5months after termination of the treatment she presented with46,XX,der(22)t(1;22)(p11;p11)in 2/10 metaphases, and thesame karyotype was found 13 months later, when she hadbecome transfusion independent. Two (nos. 6 and 11) of thefour patients with karyotype abnormalities before the start ofDFO treatment had trisomy 8 in 60% and 90%, respectively,of the metaphases. In the former patient (no. 6) the abnormalcell clone showed a steep decrease in size from 60% at start oftreatment to 10% after 21 months of treatment (Fig 4);meanwhile the patient became transfusion independent. Inorder to verify this reduction of the abnormal cell clone weperformed primed in-situ labelling on frozen bone marrowcells which had been collected in parallel with the con-ventional chromosome analysis. The decrease of the trisomy8 clone as observed by conventional cytogenetics wasparallelled by the fall in interphase cells showing threealpha satellite centromeric spots. In the latter case (no. 11)clonal evolution took place: the abnormal clone persisted anddeveloped a 5q deletion, so that at treatment termination47,XX,del(5)(q13q33),+8was observed in 9/10 metaphases.In the third of the four cytogenetically abnormal patients(no. 1), 46,XX,del(11)(q22q24) was found in 40% of themetaphases 6 months prior to DFO treatment. After 6 monthsof treatment 90% of the metaphases had this karyo-type and at termination all cells showed this aberration.Interestingly, however, the transfusion requirement fellfrom 261 to 32 g Hb/month during the period of ironchelation. After termination of DFO treatment and normal-ization of iron store parameters the Hb requirement grad-ually rose to the original baseline and the patient progressedto AML. The fourth patient in the group (no. 4) presenteda complex abnormal karyotype at the start of DFO



Fig 1. Follow-up of the Hb requirement (k) and the TfR serum concentration (l) (not performed in patient 11) in 11 MDS patients during ironchelation and in five patients (nos. 1, 5, 7, 8 and 11) also after termination of DFO treatment (indicated by black horizontal bars). The brokenhorizontal lines represent the lower and upper limits of the reference range of the serum TfR concentration from 1.14 to 2.85�g TfR/ml(2.5–97.5% percentiles).



Fig 2. Pre-transfusional Hb concentration during iron chelation in 11 MDS patients. Broken lines represent mean concentration of the last threevalues before start of iron chelation. Horizontal black bars represent follow-up after termination of iron chelation.



Fig 3. Platelet (l) and neutrophil (k) counts during iron chelation in 11 MDS patients. In patient 11 the neutrophil fraction contained a variablenumber of blasts. This patient was treated with hydroxyurea during the last 3 months of follow-up after termination of DFO treatment due toprogression to overt AML. Horizontal black bars represent follow-up after termination of iron chelation.



treatment: 46,XX[2]/46,XX,ÿ1,+del(1) (p22),+del(1)(q22),ÿ4[4]/47,XX,idem,+19[2]/ 47,XX,idem, +21[2], and furtherclonal evolution took place during iron chelation.

Efficacy of iron chelationThe efficacy of iron chelation was monitored by repeatedquantitative MRI-determinations of the liver iron concentra-tion and measurements of serum ferritin in parallel, as shownin Fig 5. The serum ferritin concentration decreased clearly in9/11 patients, but increased in two patients (nos. 8 and 10).The MRI-determined liver iron concentration decreased in allpatients, but the reduction was only minor in five patients,including patient 8 in whom intermittent compliance

problems were identified. The mean efficacy of iron chelation,was only 2.9 (SD 1.9) �mol/g/month in this subgroup of fivepatients but was 6 times higher in the subgroup of six patientswith high efficacy chelation of 17.7 (SD 7.2)�mol/g/month.When investigating the relationship between the efficacy ofiron chelation and the treatment response (reduction of Hbrequirement at least 50%) we found that a treatmentresponse was seen in all patients with high efficacy of ironchelation, whereas non-responders only showed minorreductions (Fig 5), with exception of the patient (no. 8) whohad intermittent compliance problems. This distinction couldnot be made by looking only at the changes of the serumferrritin concentrations (Fig 5c).

Relationship between transfusion requirements and the erythroidmarrow activityThe erythroid marrow activity was evaluated indirectly in 10patients (not done in patient 11) by repeated measurementsof the TfR concentration in serum (Fig 1). At baseline (inpatients 2 and 7 after 6 and 10 months of treatmentrespectively) the TfR serum level was found below the normalreference range in seven patients and low in the normal rangein another patient. These patients had MDS of subtype RA orRAEB. In the remaining two patients who had RARS (nos. 6and 10) the baseline values were high in, or above, thereference range, respectively. During iron removal the TfRlevels increased in all seven patients (nos. 1, 3, 5–9) whoshowed simultaneously decreasing Hb requirements. In thepatients who showed no improvement in the Hb requirement,the TfR concentration in serum remained unchangedthroughout DFO treatment. In patient 1 the increased TfRlevel was accompanied by reduction in Hb requirements untiliron chelation was stopped. During follow-up without ironchelation both the transfusion requirement and the TfR levelreturned to initial values. Only in patient 10 (who initiallyhad the highest transfusion requirement of all patients) the

Fig 4. Follow-up of the fraction of bone marrow cells with +8 (patient6) before and during iron chelation with desferrioxamine byconventional chromosome analysis (k) and by primed in situ labelling(PRINS) (l). Vertical broken line denotes start point of treatment.

Fig 5. Relationshipbetween treatment response and changes of the iron store parameters during iron chelation: (a) blood transfusionrequirements;(b) MRI-determined liver iron concentration; (c) serum ferritin concentration, in 11 MDS patients. The data represent baseline values and valuescollected in connection with the last determination of the liver iron concentration. Full lines denote responders (reduction of blood transfusionrequirement550% compared with baseline value); broken lines denote non-responders.

TfR concentration decreased during iron chelation in parallelwith an increase of the transfusion requirements to around563 g Hb per month (7.5 blood units/month). The relation-ship between the transfusion requirements and the TfR levelsfor all 10 patients together is demonstrated by a scatterdiagram (Fig 6) of all corresponding values of repeatedmeasurements of TfR and the transfusion requirementsduring follow-up. Except for patient 10 with RARS, wefound an inverse linear relationship between the bloodtransfusion requirement and the serum TfR concentration.The calculated mean slope of the three demonstratedregression lines is 285 (SD 30) g Hb per month/�g TfR/ml.From this mean slope it could be calculated that an increase ofthe serum TfR level by 0.26�g/ml corresponds to a reductionin the Hb requirements by approximately 1 blood unit permonth.

Serum EPO concentrationIn seven patients we investigated the immediate effect of ironchelation on the serum EPO concentration, when the DFOdosage was increased from 1 to 4 g daily, given by 12 hcontinuous subcutaneous infusions on 4 successive days. Theblood samples were in all cases drawn in the morning, 12 hbefore start of the DFO infusion. As seen in Fig 7, the serumEPO concentration increased in 5/7 patients. No increase wasfound in patient 3 who had MDS and severely impaired renal

function, and a slightly decreased level in patient 4 who hadthe the highest baseline level. A follow-up of serum EPOconcentration throughout the study revealed no clear pattern(data not shown). The clinical response regarding the effect ofiron chelation on the Hb requirement was not predictablefrom the initial serum EPO concentration (Table I), nor did theimmediate EPO response to DFO dosage escalation predict theclinical response, as increase of serum EPO was observed inboth responders and non-responders (Fig 7).

DISCUSSION

Treatment responseThe present data confirm our earlier observations ( Jensenet al, 1992) made in three MDS patients with transfusionaliron overload, showing that iron chelation with DFO reducedthe Hb requirements by up to 46% in two of the casesaccompanied by a significant increase of the platelet countand of the neutrophil count in one of the cases. The presentdata show an even more beneficial effect of iron chelation onhaemopoiesis in MDS patients, if the patients are treated for asufficiently long period of time. Thus, we found that DFOtreatment not only alleviated the anaemia but induced bloodtransfusion independence in almost one half of our patients.However, the maximal improvement of erythropoiesis wasfirst seen after at least 1.5 years of iron chelation. None of thepatients in our previous study had been treated for so long.A rise in the platelet and/or neutrophil counts was found in 9/11 patients, but a trilineage response was seen in only fivepatients who initially had pancytopenia. The treatment effectin these five patients is of special clinical importance as it mayreduce the risk of bleeding and infection.



Fig 6. Scatter diagram of the relationship between correspondingvalues of the TfR serum concentration and the calculated Hbrequirement during iron chelation (data from Fig 1) from 10 MDSpatients: k, nos. 1–5, 7 and 8 (RA, RAEB); l, no. 9 (RA); h, no. 6(RARS);T, no. 10 (RARS). Except that of patient 10, all data could befitted by three linear regression lines (data obtained after achievementof transfusion requirement (=0 g Hb per month) are excluded fromregression analysis). FAB-types and patient numbers of fitted data areindicated. The regression equation of the left regression line wasy = 502ÿ293x. Arrow indicates outlier (excluded from analysis).

Fig 7. Changes of the serum EPO concentration (d-S-EPO) incomparison to baseline value during escalation of DFO dosage from1 to 4 g on four successive days at initiation of iron chelation in sevenMDS patients. Full lines denote responders and broken lines denotenon-responders.



The treatment response did not correlate to the amount ofiron overload or the duration of transfusion history beforetherapy, nor did an achievement of normal iron stores seem tobe essential for treatment response. Thus, only one patienthad a normal liver iron concentration, a normal serumferritin concentration and normal marrow iron when thetreatment response was maximal.

The efficacy of iron chelation on the other hand, calcu-lated as the ratio between the reduction of the liver ironconcentration and the corresponding treatment time, seemedto be of greatest importance for achievement of optimaltreatment response. Thus, a reduction of the Hb require-ments by at least 50% was, with only one exception, seen inthe group of patients in whom iron chelation was highlyeffective. In the non-responders a high blood transfusionrequirement (>4 blood units/month), partly due tohypersplenism in three patients, could possibly explain thelack of treatment success. Therefore, in case of severehypersplenism, splenectomy may be considered followed byDFO treatment.

Altogether we found a treatment response after ironchelation in approximately two-thirds of our consecutiveseries of MDS patients. It should, however, be acknowledgedthat our group of patients was not randomly selected, as onlypatients below 70 years of age and without signs of diseaseprogression were included. Therefore it may not be possibleto extrapolate our findings directly to an unselected groupof MDS patients.

CytogeneticsA chromosome analysis was performed in all cases to assess apossible effect of DFO on karyotype abnormalities. Tworesponding and three non-responding patients had a normalkaryotype at initiation of DFO therapy. Numerical and/orstructural chromosome aberrations occurred in one responderand four non-responders. However, whereas the karyotype ofthe responder (47,XY,+8) did not undergo further evolution,additional abnormal clones developed in all four non-responders. A progressive cytogenetic normalization accom-panied by an improvement in the anaemia was observed inone patient who presented with +8. Although a spontaneousreduction in size of the dysplastic cell clone cannot beexcluded in this single case, the finding is of interest as it hasbeen shown that in cultured lymphocytes from a patientwith Fanconi anaemia (Porfirio et al, 1989) addition of DFOto the culture reduced the spontaneous chromosomebreakage by 50%. Moreover, in vitro studies (Estrov et al,1987; Dezza et al, 1989) have shown that clonogenic cells fromHL-60 and U-937 are more sensitive to the antiproliferativeeffect of DFO than normal human CFU-GM and BFU. Inaddition, it was also shown by Dezza et al (1989) that DFOhad a selective inhibitory effect on a hyperploid blast cellpopulation from a patient with acute leukaemia treated withDFO.

One clinically responding patient in our study demon-strated cytogenetical evolution during treatment withDFO, which suggests that the cytogenetical findings arenot always connected with the efficacy of erythropoiesis.

Erythroid marrow activityThe effective erythroid output is reduced in MDS patients dueto various degrees of defective proliferation and/or differen-tiation (Lintula, 1986). The erythroid progenitor growth canbe evaluated in progenitor cell assays (Chui & Clarke, 1982).The ineffectiveness of the erythropoiesis in MDS can bequantified morphologically, or more accurately by ferrokineticstudies (Ricketts et al, 1975). However, ferrokinetic techni-ques assume stable erythropoiesis throughout the studyperiod and cannot be performed repeatedly during treatmentdue to ethical reasons. Recently, an assay for soluble TfR hasbeen introduced as a quantitative measurement of the totalerythropoietic activity (Beguin et al, 1988; Huebers et al,1990). Concerning the origin of circulating TfR, it has beenshown in rats (Chitambar et al, 1991) that the loss oftransferrin receptors from reticulocytes during terminalmaturation in vitro involves shedding of the receptor invesicles, and the release of a soluble form of TfR receptor fromthese vesicles. For human sera it has been shown (Shih et al,1993) that the predominant form of serum TfR is a truncatedmonomeric form of transferrin receptor of about 85 kD. Theserum TfR concentration correlates well with the erythrontransferrin uptake (ETU), a precise measure of the erythroidprecursor mass (Huebers et al, 1990; Beguin et al, 1993). Inorder to follow the changes of erythropoiesis during ironchelation we repeatedly determined the serum TfR concen-tration. Before the start of iron chelation we found the serumTfR level below the normal range in patients with RA andRAEB and around the upper limit of the normal range inpatients with RARS. This is in agreement with the study ofBowen et al (1994). During iron chelation the TfR levelsnormalized in patients who showed a progressive reduction ofthe Hb requirement but remained unchanged below thereference ranges in non-responders. A rise in serum TfR hasalso been described in MDS patients responding to EPOtreatment by Hb increase (Ghio et al, 1993), in anaemia ofchronic renal failure and in various other haematologicaldiseases (Cazzola et al, 1992). Monitoring the effect of EPO,the serum TfR concentration appeared to be the earliest andmost reliable predictor of treatment response (Cazzola et al,1992), which is also in agreement with our data. Further-more, we found an inverse linear relationship between theserum TfR level and the Hb requirement. Interestingly, theindividual slopes of the regression lines were independent ofthe stage of the disease as expressed by the FAB-type, but haddifferent x-intercepts. The distance between the intercepts forthe regression lines for RA/RAEB patients and the RARSpatients was approximately 2.4�g TfR/ml, but an increase ofserum TfR by only 1�g/ml equals a reduction of thetransfusion requirement from 4 units blood/month to zero.This indicates that the relationship between the effectiveerythroid output and the serum TfR concentration duringiron chelation is unaffected by the baseline TfR level. Thebaseline TfR levels were high in our RARS patients, which isin accordance with findings by others (Bowen et al, 1994).These patients always have an increased erythroid mass and ahigh degree of ineffective erythropoiesis (Cazzola et al, 1988),which seems to lead to high serum TfR concentrations. Thus,an increase of serum TfR during improvement of the

erythroid output, as found in our patients during ironchelation, does not indicate if this is due to an increasedeffective erythropoiesis, or an increased erythroid mass withan unchanged rate of ineffectivity. Therefore, furtherelucidation of the changes of erythropoiesis induced by ironchelation requires other techniques, e.g. ferrokinetic studiesat the start and termination of treatment, and erythroidprogenitor assays, which also may be performed repeatedlyduring iron chelation. Changes of erythroid proliferationand differentiation may also be evaluated by multipara-meter flow cytometry as demonstrated recently ( Jensen et al,1993).

It has recently been shown (Wang & Semenza, 1993) thatthe EPO gene transcription is not only activated by hypoxiabut may also be enhanced by desferrioxamine. Thus, it wasdemonstrated in kidney cells and in Hep3B cells that thehypoxia-inducible factor, a nuclear factor that binds to thehypoxia-inducible enhancer of the EPO gene at a site that isrequired for transcriptional activation, is not only induced byhypoxia but also by DFO. Furthermore, it has been shown(Ghio et al, 1993) that a Hb increase in MDS patientsresponding to EPO treatment is associated by a rise of thenumber of circulating progenitors. Therefore an increasederythroid output in our MDS patients during iron chelationmay, at least in part, be mediated by an increased EPOproduction. We could demonstrate increasing EPO serumconcentrations in most patients when escalating the DFOdosage at the beginning of treatment. However, the increasein serum EPO at 2 g DFO daily as used in this study was onlymoderate, and the follow-up data on the EPO serumconcentration were non-informative, possibly due to vari-ability of the pre-transfusional haematocrit levels influencingthe EPO serum concentration (Beguin et al, 1993).

Although the improvement of the erythroid output was themost prominent change of haemopoiesis during iron chela-tion, five patients showed a trilineage response. This findingmay indicate iron chelation action at stem cell level.

CONCLUSIONS

Iron chelation of MDS patients with transfusional ironoverload not only alleviates the anaemia but may alsoinduce blood transfusion independence if the patients aretreated for a sufficiently long period of time. The improvementmay persist after termination of iron chelation. The reductionof Hb requirement is accompanied by an increase of the pre-transfusional Hb concentration and inversely correlates withthe serum TfR level. Trilineage responses may be seen in somepatients. High efficacy of iron chelation as determined bymonitoring the liver iron concentration is of the greatestimportance for achievement of an optimal treatmentresponse. The most obvious reasons for lack of response totreatment are a high blood transfusion requirement even-tually combined with hypersplenism or insufficient ironchelation. Our data confirm and expand earlier observationsand may encourage to a more widespread use of ironchelation in MDS patients. Unfortunately, DFO treatmentdemands parenteral administration and is expensive. There-fore an investigation of the iron chelation efficacy and the

effect on haemopoiesis of oral chelators in MDS patients isneeded.

ACKNOWLEDGMENT

The author thanks Hanne Hjorth Eriksen for excellenttechnical assistance (PRINS).

REFERENCES

Beguin, Y., Clemoris, G.K., Pootrakul, P. & Fillet, G. (1993)Quantitative assessment of erythropoiesis and functional classifica-tion of anaemia based on measurements of serum transferrinreceptor and erythropoietin. Blood, 81, 1067–1076.

Beguin, Y., Huebers, H.A., Josephson, B. & Finch, C.A. (1988)Transferrin receptors in rat plasma. Proceedings of the NationalAcademy of Sciences of the United States of America, 85, 637–640.

Bennett, J.M., Catovsky, D., Daniel, M.T., Flandrin, G., Galton, D.A.G.,Gralnick, H.R. & Sultan, C. (1982) Proposals for the classification ofthe myelodysplastic syndromes. British Journal of Haematology, 51,189–199.

Bowen, D.T., Culligan, D., Beguin, Y., Kendall, R. & Willis, N. (1994)Estimation of effective and total erythropoiesis in myelodysplasiausing serum transferrin receptor and erythropoietin concentra-tions, with automated reticulocyte parameters. Leukemia, 8, 151–155.

Cazzola, M., Barosi, G., Gobbi, P.G., Invernizzi, R., Riccardi, A. &Ascari, E. (1988) Natural history of idiopathic refractory sidero-blastic anemia. Blood, 71, 305–312.

Cazzola, M., Ponchio, L., Beguin, Y., Rosti, V., Bergamashi, G.,Liberato, N.L., Fregoni, V., Nalli, G., Barosi, G. & Ascari, E. (1992)Subcutaneous erythropoietin for treatment of refractory anemia inhematologic disorders: result of a phase I/II clinical trial. Blood, 79,29–37.

Chitambar, C.R., Loebel, A.M. & Noble, N.A. (1991) Shedding oftransferrin receptor from rat reticulocytes during maturation invitro: soluble transferrin receptor is derived from receptor shed invesicles. Blood, 78, 2444–2450.

Chui, D.H.K. & Clarke, B.J. (1982) Abnormal erythroid progenitorcells in human preleukemia. Blood, 60, 362–367.

Dezza, L., Cazzola, M., Danova, M., Carlo-Stella, C., Bergamaschi, G.,Brugnatelli, S., Invernizzi, R., Mazzini, G., Riccardi, A. & Ascari, E.(1989) Effects of desferrioxamine on normal and leukemic humanhematopoietic cell growth: in vitro and in vivo studies. Leukemia, 3,104–107.

Estrov, Z., Tawa, A., Wang, X., Dube, I.D., Suhl, H., Cohen, A.,Gelfand, E.W. & Freedman, M.H. (1987) In vitro and in vivo effectsof desforoxamine in neonatal acute leukemia. Blood, 69, 757–761.

Ganong, W.F. (1979) Review of Medical Physiology, 8th edn, pp. 10–11. Lange Medical Publications, Los Altos, California.

Ghio, R., Ballerari, E., Ballestrero, A., Gatti, A.M., Mareni, C., Massa,G., Patrone, F., Sessarego, M. & Timitilli, S. (1993) Subcutaneousrecombinant human erythropoietin for the treatment of anaemiain myelodysplastic syndromes. Acta Haematologica, 90, 58–64.

Hoffbrand, A.V. & Wonke, B. (1989) Results of long-term subcuta-neous desferrioxamine therapy. Bailliere’s Clinical Haematology, 2,345–362.

Huebers, H.A., Beguin, Y., Pootrakul, P., Einspahr, D. & Finch, C.A.(1990) Intact transferrin receptors in human plasma and theirrelation to erythropoiesis. Blood, 75, 102–107.

Jensen, I.M., Hokland, M. & Hokland, P. (1993) A quantitativeevaluation of erythropoiesis in myelodysplastic syndromes usingmultiparameter flow cytometry. Leukemia Research, 17, 839–846.





Jensen, P.D., Jensen, F.T., Christensen, T. & Ellegaard, J. (1994)Non-invasive assessment of tissue iron overload in the liverby magneticresonance imaging. British Journal of Haematology, 87, 171–184.

Jensen, P.D., Jensen, I.M. & Ellegaard, J. (1992) Desferrioxaminetreatment reduces blood transfusion requirements in patients withmyelodysplastic syndrome. British Journal of Haematology, 80, 121–124.

Johnston, D.L., Rice, L., Vick, G.W., Hedrick, T.D. & Rokey, R. (1989)Assessment of tissue iron overload by nuclear magnetic resonanceimaging. American Journal of Medicine, 87, 40–47.

Koch, J., Hindkjaer, J., Kølvraa, S. & Bolund, S. (1995) Construction ofa panel of chromosome specific oligonucleotide probes (PRINS-primers) useful for the identification of individual human chromo-somes in situ. Cytogenetics and Cell Genetics, 71, 142–147.

Lintula, R. (1986) Ferrokinetic abnormalities and red cell life span inmyelodysplastic syndromes: a review. Scandinavian Journal ofHaematology, 36, 48–52.

Pedersen, B. & Kerndrup, G. (1986) Specific minor chromosomedeletions consistently occurring in myelodysplastic syndromes.Cancer Genetics and Cytogenetics, 23, 61–75.

Pippard, M.J. (1989) Desferrioxamine-induced iron excretion inhumans. Bailliere’s Clinical Haematology, 2, 323–343.

Porfirio, B., Ambroso, G., Giannella, G., Isacchi, G. & Dallapiccola,B. (1989) Partial correction of chromosome instability inFanconi anemia by desferrioxamine. Human Genetics, 83, 49–51.

Ricketts, C., Jacobs, A. & Cavill, I. (1975) Ferrokinetics anderythropoiesis in man: the measurement of effective erythropoiesis,ineffective erythropoiesis and red cell lifespan using 59Fe. BritishJournal of Haematology, 31, 65–75.

Schafer, A., Cheron, R.G., Dluhy, R., Cooper, B., Gleason, R.E.,Soeldner, J.S. & Bunn, H.F. (1981) Clinical consequences ofacquired transfusional iron overload in adults. New England Journalof Medicine, 304, 319–324.

Shih, Y.J., Baynes, R.D., Hudson, B.G. & Cook, J.D. (1993)Characterization and quantitation of the circulating forms ofserum transferrin receptor using domain-specific antibodies. Blood,81, 234–238.

Wang, G.L. & Semenza, G.L. (1993) Desferrioxamine induceserythropoietin gene expression and hypoxia-inducible factor 1DNA-binding activity: implications for models of hypoxia signaltransduction. Blood, 82, 3610–3615.

Yunis, J.J. (1981) New chromosome techniques in the study of humanneoplasia. Human Pathology, 12, 540–549.

the effect of iron chelation on haemopoiesis in mds patients with transfusional iron overload

Documents