Extramedullary haematopoiesis correlates with genotype and absence of cardiac iron overload in polytransfused adults with thalassaemia

Abstract

Background

The mechanisms responsible for the sporadic occurrence of extramedullary haematopoiesis in polytransfused thalassaemic patients have not yet been clarified. In this study we tried to elucidate the influence of genotype and other factors on the presence of extramedullary haematopoiesis.

Materials and methods

We performed a retrospective database review of our polytransfused thalassaemic patients between January 2006 and December 2011. Demographic, transfusional, genetic, radiological and biochemical data were collected and statistically analysed.

Results

Extramedullary haematopoiesis was found in 18 out of 67 patients (27%). All of them were splenectomised, had a higher nucleated red blood cell count and higher levels of the soluble form of transferrin receptor with respect to patients without extramedullary haematopoiesis; furthermore, patients with EMH had a lower transfusional iron intake and a higher pre-transfusion haemoglobin level as compared with those without extramedullary haematopoiesis. Ten out of the 18 patients with extramedullary haematopoiesis were compound heterozygotes for IVS 1–6/codon 39. A high frequency of thrombotic events was also recorded among all patients followed at our centre with this genetic profile.

Discussion

Among our cohort of thalassaemic polytransfused patients, extramedullary haematopoiesis was not such a rare event. Furthermore, we identified a group of patients, most of whom were compound heterozygotes for IVS 1–6/codon 39, with increased soluble transferrin receptor levels and excessive expansion of erythroid marrow probably responsible for the tendency to develop extramedullary haematopoiesis.

Introduction

Extramedullary haematopoiesis (EMH) is a physiological compensatory phenomenon occurring in patients affected by thalassaemia because of insufficient bone marrow function and consequent expansion of erythropoiesis¹. It may occur in all parts of the body but the paraspinal area of the thorax is the site most frequently involved. EMH is particularly common in patients with non-transfusion-dependent thalassaemia (NTDT) and rare in those with thalassaemia major, in whom the transfusion regimen is directed to suppress bone marrow expansion^2,3. Current transfusion regimen guidelines state that the pre-transfusion haemoglobin (Hb) should ideally be in the 9–10 g/dL range. This recommended transfusion scheme generally leads to the transfusion of 100–200 mL/kg/year of packed red blood cells, which is equivalent to 0.3–0.6 mg of iron per kg body weight per day⁴.

Previous seminal studies showed that the evaluation of the soluble form of transferrin receptor (sTfR) may be useful for better personalising the transfusion regimen^5,6. Recently, in a prospective study we found that in patients with NTDT sTfR may help to predict the presence of EMH⁷; to date, no study has investigated whether this association is also observed in regularly transfused patients or whether genotype and/or other factors may influence sTfR levels. In this study we retrospectively investigated these relationships in a cohort of regularly transfused thalassaemic patients.

Material and methods

This was a retrospective study and all medical records of patients on regular transfusion therapy followed at our unit from 2006 to 2011 were reviewed carefully. During the observation period extramedullary haematopoietic tissue was incidentally detected because most of our patients underwent computed tomography (CT) or magnetic resonance imaging (MRI) scanning for evaluation of respiratory function (high-resolution CT scan), assessment of pathological abdominal and mediastinal lymphadenopathy or investigation of neurological or orthopaedic dysfunction. Almost all had twice yearly measurements of sTfR. In the light of the preliminary results of our previously mentioned study⁷, five patients, in whom particularly high levels of sTfR were found, underwent MRI to assess the presence of paraspinal EMH. While most patients were regularly transfused to maintain pre-transfusion Hb levels in the 9–10 g/dL range, in accordance with the previously mentioned indications⁴, in patients with EMH the Hb pre-trasfunsion level was slightly increased to a target range of 10–10.5 g/dL in order to prevent further progression of EMH. However, at the time of its detection, most patients with EMH were already under this higher transfusion regimen because of the coincidental presence of cardiopulmonary disease or following a patient’s request to improve his or her sense of well-being.

The eligibility criteria for entry into the retrospective analysis were the following: (i) full knowledge of the status of the transfusion regimen (blood consumption and pre-transfusion Hb level between 2006 and 2011); (ii) maintained on regular transfusion regimen for at least 15 years; (iii) the presence or absence of EMH had to be assessed by MRI or CT scan during the period of observation. Patients were considered positive at MRI or CT scan if they had at least one site of extramedullary haematopoietic tissue⁸; and (iv) sTfR had to be assayed at least once a year during the period of the study, without it being necessary that the determination coincided exactly with the EMH imaging examination.

The total amount of packed red blood cells transfused was calculated as the total amount of blood (mL) multiplied by the haematocrit of each unit (percentage) divided by 100 and expressed as mg Fe/kg/day (transfusional iron intake) according to the following formula:

$$\frac{[\text{whole\hspace{0.17em}blood}\hspace{0.17em}(\text{mL})\times \text{haematocrit\hspace{0.17em}}(\%)\times 100\times 1.08]}{356\times \text{kg\hspace{0.17em}of\hspace{0.17em}body\hspace{0.17em}}{\text{weight}}^9}$$

Patients who received pegylated-interferon and ribavirin to treat hepatitis C virus infection were excluded from the analysis.

Of the 105 patients on regular transfusion therapy monitored at our Centre 67 (64%) were evaluable for statistical analysis. Twenty patients were excluded because they did not undergo CT or MRI scanning.

All these patients had already been evaluated for their genetic defects as previously described and alpha thalassaemia mutations were identified by reverse hybridisation assay (alpha-globin StripAssay, Nuclear Laser, Vienna Lab, Vienna, Austria)¹⁰. Nucleated red blood cell count was determined on an ADVIA 2120i (Siemens Medical Solutions Diagnostics GmbH, Fernwald, Germany). sTfR was investigated with a commercially available kit using the N Latex sTfR and BN II System (Siemens Healthcare Diagnostics, Marburg, Germany) nephelometric technique. The reference ranges was 0.76–1.76 mg/L.

All patients underwent MRI within the Myocardial Iron Overload in Thalassaemia (MIOT) network¹¹. Images were acquired using a 1.5 T MR scanner (GE Signa/Excite HD, Milwaukee, WI, USA). The T2* (T2-star) technique was used to assess iron overload. The reproducibility of this technique and its transferability within the MIOT network had been previously demonstrated for the heart and for the liver¹¹. The global heart T2* value was obtained by averaging all segmental T2* values in the mid anterior septum and the mid inferior septum.

Serum ferritin was measured by the Imx ferritin assay, which is a microparticle enzyme immunoassay (Abbott, Illinois, USA). The reference range was 30–300 ng/mL.

The study was approved by the Ethics Committee of the Cardarelli Hospital in Naples.

Statistical analysis

Data from the two groups of patients (EMH+, EMH−) were analysed using a statistical software package (MedCalc^®, version 10.2.0.0, MedCalc Software bvba, Ostend, Belgium). Results of descriptive statistics are expressed as mean±standard deviation (M±SD). Fisher’s exact test was used to compare the incidences of different parameters between the two groups of patients. Student’s t test was used to compare differences in parametric data. A P-value below 0.05 was considered statistically significant. The area under the receiver operating characteristic (ROC) curve was used to determine whether any level of sTfR could be used to differentiate between patients with and without EMH.

Results

The genotype analysis revealed several different major groups of transfused patients with identical genetic defects: compound heterozygotes for IVS 1–110/ codon 39 (n=8; 12%), compound heterozygotes for IVS 1–6/codon 39 (n=12; 18%), homozygotes for codon 39/codon 39 (n=8; 12%), homozygotes for IVS 1–110/ IVS 1–110 (n=5; 7%), and compound heterozygotes for codon 39/ Hb Lepore (n=6; 9%). The remaining molecular defects were less frequent and highly heterogeneous (data not shown).

The clinically and instrumentally relevant findings in the whole population and in patients divided in those with EMH and those without EMH are summarised in Table I. Overall, 67 patients were analysed and the population included 34 males and 33 females. Of these 67 patients, 17 (25%) and 50 (75%) patients were evaluated with CT or MRI, respectively. Fifty-five patients were splenectomised and the age at first transfusion varied from 0.25 to 46 years. EMH was found in 18 (27%) patients, all of whom had been splenectomised; the patients affected were mainly older than 20 but younger than 50. In all patients without EMH, both Hb pre-transfusion level and transfusional iron intake were stable and comparable in each year in the 2006–2011 interval; a similar unchangeable trend was observed for sTfR level (data not shown). Overall, the concentration of sTfR varied from 0.7 to 14 mg/L (4.7±2.5 mg/L), but in patients with EMH it varied from 1.3 to 14 mg/L (6.6±3.1 mg/L) while it varied from 0.7 to 8.9 mg/L (4.1±1.9 mg/L) in patients without EMH, with a statistically significant intergroup difference (p<0.01). A statistically significant difference was also found when comparing data from splenectomised patients with EMH (100%) and those without EMH (76%) indicating that the different splenectomy rate among both groups did not influence sTfR level (p<0.05) (data not shown). Overall, males had significantly higher sTfR levels than had females (5.4±2.8 mg/L vs 4.1±2.1, respectively, p<0.05) (data not shown). Figure 1 shows the distribution of sTfR levels for each patient divided into those with EMH and those without EMH.

Table I.

Characteristics of regularly transfused patients overall and divided according to whether they had extramedullary haematopoiesis or not (EMH+, EMH−).

	Overall (n=67)	EMH+ (n=18)	EMH− (n=49)	p
Male/Female	34/33	11/7	23/26	0.41
MRI (+/−)/CT (+/−)	17 (9/8)/50 (9/41)	9 (9/0)/9 (9/0)	8 (0/9)/41 (0/41)	<0.01
sTfR mean (mg/L) (range)	4.7±2.5 (0.7–14)	6.6±3.1 (1,3–14)	4.1±1.9 (0.7–8.9)	<0.01
NRBC∘×103/μL	7.0±9.7 (0–48.5)	15.3±12.8 (0.7–48.5)	4.1±6.3 (0–30.4)	<0.01
Median age at first transfusion (months)	18 (3–552)	27 (3–552)	12 (3–480)	0.23
Median age (years) (range)	41.2 (19.4–74.7)	43.4 (36.1–70.9)	40.7 (19.4–74.7)	0.07
10 years ≤ age ≤20 years	1 (1%)	0 (0%)	1 (2%)	1.0
20 years < age ≤ 50 years	58 (87%)	13 (72%)	45 (92%)	0.05
Age >50 years	8 (12%)	5 (28%)	3 (6%)	<0.05
Splenectomy (%)	55 (82%)	18 (100%)	37 (76%)	<0.05
Haemoglobin (g/dL) mean±SD	9.7± 0.3	9.8± 0.3	9.6±0.4	0.07
mg Fe/Kg/day	0.28± 0.08 (0.11–0.49)	0.24±0.06 (0.16–0.36)	0.29±0.09 (0.11–0.49)	<0.05
Global heart T2*(ms) mean±SD (range)	32.2±10.1 (8.5–47)	39.0±3.4 (34–46)	29.7±10.8 (8.5–47)	<0.01
Liver T2*(ms) mean±SD (range)	8.1±6.8 (0.9–32.1)	6.5±4.7 (1.0–17.4)	8.7±7.4 (0.9–32.1)	0.29
Ferritin (ng/mL) mean±SD (range)	1,582±1312 (109–5,780)	1,443±1161 (109–4,060)	1,631±1,368 (182–5,780)	0.61

Results are given as mean ± SD and range (minimum, maximum). P values <0.05 (tested by Student’s t-test or Fisher’s exact test) are considered statistically significant. MRI: magnetic resonance imaging; CT: computed tomography; sTfR: soluble form of transferrin receptor; Fe: iron; T2*: T2-star;

^∘Nucleated red blood cells, i.e. erythroblasts.

Figure 1.

sTfR level in patients with EMH and without EMH.

sTfR: soluble form of transferrin receptor; EMH: extramedullary haematopoiesis.

The area under the ROC curve (AUC_ROC) was used to determine whether any level of sTfR could be used to differentiate between patients with and without EMH. The highest value of the AUC_ROC in predicting the occurrence of EMH was found for sTfR concentrations with a threshold value of 4.6 mg/L [AUC_ROC=0.76 (95% CI: 0.64–0.86, p=0.0003)]. The sensitivity and specificity for this variable were 72.2% and 71.4, respectively (data not shown).

Patients with EMH also had a higher (p<0.01) nucleated red blood cell count in the peripheral blood with respect to that found in patients without EMH. While the age at first transfusion was comparable between both subgroups, the percentage of splenectomised patients among those with EMH was statistically significantly higher than that observed in patients without EMH. The Hb pre-transfusion level among patients with EMH was higher that that among patients without EMH, but in this case the difference was not statistically significant. On the other hand, the transfusional iron intake was lower (p<0.05) in patients with EMH than in patients without EMH. Interestingly, while both the liver T2* and the mean ferritin level were comparable between the two groups, the global heart T2* value was significantly higher in the group with EMH than in group without EMH.

The 49 patients without EMH included all eight IVS 1–110/codon 39 (16%) compound heterozygotes, two out of the 12 IVS 1–6/codon 39 compound heterozygotes (4%), seven out of the eight 39/codon 39 (14%) homozygotes, all five IVS 1–110/ IVS 1–110 homozygotes (10%), five out of six codon 39/Hb Lepore compound heterozygotes (10%), four CD39/IVSI-1 compound heterozygotes (8%), three IVSI-6/IVSI-6 compound heterozygotes (6%), two CD39/IVSII-1 compound heterozygotes (4%), two IVSI-110/IVSI-6 compound heterozygotes (4%); the remaining 11 (22%) molecular defects were highly heterogeneous (data not shown).

The detailed clinical and genetic parameters of patients with EMH are shown in Table II. Nine out of 18 patients were compound heterozygotes for IVS 1–6/ codon 39, while the remaining molecular defects were less frequent and highly heterogeneous. Interestingly, two cases of spinal cord compression were recorded (patients n. 13 and n. 16). Table II also shows the clinical and instrumental characteristics of the remaining patients with IVS 1–6/codon 39 attending at our centre on regular transfusion but without EMH (patients n. 19 and n. 20) or with a NTDT phenotype (patients n. 21 to n. 24). Patient n. 21 was started on transfusion therapy because of the occurrence of spinal cord compression; one patient with an accessory spleen and a high frequency of thrombotic events was also recorded among this class of patients. As shown in Table II, among the seven polytransfused patients who were started on a more intensive transfusion regimen (increased transfusion volume/frequency), the levels of sTfR were considerably reduced. The increase in pre-transfusion Hb did not correlate exactly with the decrease in sTfR but required an increase of 0.09±0.02 mg fe/kg/day in transfusional iron intake per patient (data not shown).

Table II.

Clinical and genetic parameters of the patients with EMH.

Patient’s number	Genotype	Year of detection and type of examination	Change in Hb (g/dL)		mg Fe/ kg/day	Change in sTfR (mg/L)		Global heart T2*	Liver T2*	Other events	Chelation therapy⋄
			Basal	Final		Basal	Final
1	Cd6(−A)/IVSII-1	2009 CT	9.2	9.6	0.27	8.0	5.3	38.5	6.8		DFP
2	IVSI-6/Cd39	2009 CT	-		0.26	-		38	13.2		DFO
3	IVSI-6/Cd39	2006 MRI	-		0.27	-		39	5.9		DFO
4	IVSI-6/IVSI-1	2008 CT	-		0.24	-		-	-		DFP
5	IVSI-6/Cd39	2009 MRI	9.6	10.6	0.16	17.7	11.0	34	4.0		DFP/DFO
6	HbSouthern Italy/ −alpha 20.5	2008 CT	-		0.36			36	4.5		DFO
7	IVSI-6/Cd39	2008 CT	-		0.22	-		46	4.6	pulmonary artery thromboembolism	DFO
8	CD39/Cd39	2009 MRI	-		0.28	-		34	12.5		DFO
9	IVSI-6/Cd39	2009 CT	-		0.25	-		43	5.6		DFP
10	IVSI-6/Cd39	2009 CT	-		0.27	-		36	3.5		DFO
11	IVSI-6/IVSI-6	2006 CT	-		0.18	-		-	-		DFO
12	aaa/IVSI-654	2006 MRI	-		0.17	-		-	-		DFP/DFO
13∘	IVSI-6/Cd39	2009 MRI	9.5	9.8	0.30	8.5	6.0	43	17.3		DFO
14	IVSI-6/Cd39	2009 MRI	9	10.1	0.16	18.2	14.0	40	1.0		DFP
15	Cd39/-87	2009 MRI	9.5	10.0	0.26	8.6	7.0	38	1.3		DFO
16∘	Cd39/Hb Lepore	2008 MRI	9	9.8	X	10.6	6.0	-	-		DFO
17	IVSI-6/IVSII-1	2007 CT	-		0.3	-		40	5.4		DFP
18	IVSI-6/Cd39	2008 MRI	9.1	9.9	0.16	17.5	7.9	40	4.9		DFO
19	IVSI-6/Cd39	2009 CT (negative)	-		0.51	-		39	4.5	accessory spleen	DFO
20	IVSI-6/Cd39 (−a3.7/aa)	2008 CT (negative)	-		0.13	-		-	-	pulmonary artery thromboembolism	DFO
21∘∘	IVSI-6/Cd39	2008 MRI	8.2**	9.9	0.22	20.0	10.0	32	11.6		DFX
22∘∘∘	IVSI-6/Cd39	2011 CT (negative)	9		-	14.5		-	-	pulmonary artery thromboembolism	DFO
23∘∘∘	IVSI-6/Cd39	2008 MRI	10.2		-	8.7		36	10.9	intestinal infarction	DFO
24∘∘∘	IVSI-6/Cd39 (−a3.7/aa)	2010 MRI (negative)	10.5		-	8.2		44	6.2		DFP

Patients 1 -18: regularly transfused patients with EMH (extramedullary haematopoiesis); patients 19 and 20: regularly transfused patients who were compound heterozygotes for IVS 1–6/codon 39 but without EMH; patients 21–24: NTDT (non-transfusion-dependent thalassaemia) phenotype.

^∘Patients with spinal cord compression;

^∘∘Patient 21 had NTDT with spinal cord compression and the baseline Hb value was recorded in the absence of transfusion;

^∘∘∘NTDT patients. For the few patients who had increased Hb pre-transfusion, both the Hb pre-transfusion and sTfR (soluble form of transferrin receptor) level observed at the time of its detection (baseline) and following the increase in the transfusion regimen (final) are reported. The final value was that observed following 2 years of change in Hb pre-transfusion regimen. Fe: iron;

^⋄the most frequent therapy used in their chelation history;

CT: computed tomography; MRI: magnetic resonance imaging; DFO: deferoxamine: DFP: deferiprone; T2*: T2-star.

Interestingly, none of these patients had evidence of cardiac iron overload even if hepatic iron accumulation was significant; most of them had been treated with deferoxamine-based regimens for most of their life.

In a further analysis of the data, all the IVS 1–6/codon 39 clinical compound heterozygotes on transfusion therapy were compared with codon 39/codon 39 homozygotes and with a control group of NTDT patients from our previously mentioned study⁷. As shown in Table III, the age of the patients and the percentage of those splenectomised were comparable among the three groups. The comparison between the groups of patients with IVS 1–6/codon 39 and those with codon 39/codon 39 mutations showed results similar to those observed comparing patients with EMH to patients without EMH thus underlining that codon 39/codon 39 is representative of patients without EMH. Differently, patients with the IVS 1–6/codon 39 genetic compound mutation, despite having significantly higher haemoglobin values had comparable levels of sTfR but higher prevalence of EMH with respect to that observed in the NTDT group.

Table III.

Characteristics of all regularly transfused patients homozygous for codon 39 (Cd39) (n=8/67) or with IVS1–6/ Cd39 compound heterozygosity (n=12/67) and of a control group with a NTDT phenotype (n=14/14).

	p	Cd39/Cd39 (n=8)	Cd39/IVS1–6 (n=12)	NTDT (n=14)	p
Male/Female	0.36	5/3	4/8	7/7	0.45
Median age (years) (range)	0.09	38.3 (19.4–43.1)	41.4 (31.0–61.6)	42.9 (28.4–68.2)	0.57
Splenectomy (%)	1.0	7 (88%)	10 (77%)	9 (64%)	0.39
sTfR mean (mg/L) (range)	<0.01	3.6±1.5 (2.0–6.2)	7.3± 3.3 (0.7–8.9)	9.6±2.1 (6.4–13)	0.10
NRBC×103/μL	<0.05	3.13±4.4 (0–13.6)	16.57±14.4 (0.25–48.5)	14.6±14.9 (0–45.48)	0.74
Haemoglobin (g/dL) mean±SD (range)	0.32	*9.7±0.5 (9.1–10.5)	*9.9±0.3 (9.5–10.6)	9.0±0.6 (8.5–9.8)	<0.01
EMH+	<0.01	1(13%)	10(83%)	3 (21%)	<0.01

^*Hb (haemoglobin) pre-transfusion. The NTDT (non-transfusion-dependent thalassaemia) group was composed of six IVS1–6/IVS1–6 homozygotes, four −87/IVS1–6 compound heterozygotes, two IVS1–6/Hb Lepore compound heterozygotes and two IVSI-6/dβ compound heterozygotes. The p-values are relative to the comparison between Cd39/Cd39 and Cd 39/IVS1–6 groups (left) and between Cd 39/IVS1–6 and NTDT groups (right). sTfR: soluble form of transferrin receptor; NRBC: nucleated red blood cells; EMH: extramedullary haematopoiesis.

The mean±SD sTfR levels in IVS 1–110/codon 39 compound heterozytes, IVS 1–110/ IVS 1–110 homozygotes and codon 39/Hb Lepore compound heterozygotes were 3.6±1.44, 3.54±1.11 and 4.69±3.01 mg/L, respectively. These three groups of patients had clinical and biochemical characteristics very similar to those observed among patients homozygous for codon 39/codon 39 mutations; likewise, the occurrence of EMH in these three groups (0%, 0% and 17%), respectively; data not shown) was comparable to that in the codon 39/ codon 39 group and consequently they were excluded from further comparative analysis.

Discussion

Currently, as a consequence of an optimal transfusion regimen, erythroid marrow stimulation and expansion is reduced in polytransfused thalassaemic patients causing a lower incidence of EMH among these patients than in NTDT patients¹². Nonetheless, the exact mechanisms underlying the sporadic and the frequent occurrence of EMH in transfused thalassaemic patients and NTDT patients, respectively, have not yet been clarified. In this study we first tried to define the genotypic-phenotypic characteristics of patients with EMH even if they had been under a regular transfusion regimen for a long time; secondly, as sTfR is a valuable index of marrow erythropoietic activity in patients receiving chronic transfusions and a good predictor of EMH in NTDT patients, we tested the validity of this relationship in our transfused patients.

We found EMH in 18 (27%) of the 105 regularly transfused patients monitored at our Centre, of whom 67 (64%) were evaluable for statistical analysis. Twenty patients were excluded because they did not undergo CT or MRI scanning. MRI was performed in only a few patients because of the detection of particularly increased sTfR levels. Therefore, in our study 36% of patients were not excluded on a voluntary basis, EMH was mainly an incidental finding and specifically studying five patients (8% of evaluable) by MRI should not have introduced a selection bias or rendered the analysis incorrect. Furthermore, it is worth noting that two of these patients were affected by spinal cord compression despite the theoretically proper maintenance transfusion regimen. Apart from these two cases, we do not know for most of our patients the precise time of onset of EMH and it was not clear whether they had EMH as an active phenomenon or as consequence of a past insufficient transfusion regimen as could be hypothesised in the case of the patients with lower sTfR levels. Nevertheless, in this group of polytransfused patients, we confirmed the ability of sTfR level to predict the presence of EMH, as previously observed in NTDT patients⁷.

The fact that all the patients with EMH were splenectomised further stimulates the speculation on the role of splenectomy in the development of EMH; in fact, as erythropoiesis/erythroid expansion occurring in the spleen results in splenomegaly, the probability of being splenectomised may have depended on a past tendency of the spleen to expand. Splenectomy could, therefore, be a risk factor for the development of EMH because it had previously selected patients more prone to have erythroid expansion which, as long as the stimulus persisted (such as in the case of a transfusion regimen unable to suppress it), may had led to its development outside the bone.

Taken together our data identified a group of patients with a particularly high frequency of EMH among our cohort of regularly transfused patients. The detailed analysis of this group of patients revealed that they had some clinical characteristics of transfusion-dependent thalassaemia intermedia, as assessed by the age at first transfusion as well as a high prevalence of the genetic compound IVS 1–6/codon 39. The increased level of sTfR in this group of patients showed that adequate inhibition of ineffective erythropoiesis was not achieved, despite a currently optimal transfusion strategy; further characteristics associated with massive expansion of the erythroid marrow were an increase in the nucleated red blood cell count, the presence of an accessory spleen in one patient and thromboembolic events in some patients¹³.

Accordingly, our data showed that in chronically transfused IVS 1–6/codon 39 patients the average sTfR level was significantly higher than that observed in all other transfused patients but comparable to that observed in a control group of NTDT patients. Furthermore, the evaluation of the compound heterozygotes for IVS 1–6/ codon 39 but with a NTDT phenotype confirmed the presence of EMH, thrombosis and increased number of erythroid precursors also in this category of patients. Taken together, these data suggest that, despite the maintenance of a high pre-transfusion Hb, there was an excessive expansion of erythroid marrow responsible for the tendency to develop EMH and the persistence of a NTDT-like phenotype. Such a mechanism may be also responsible for observed pattern of cardiac and hepatic iron loading in this subgroup of patients^14,15, although the influence of previous chelation therapy was not clearly assessed. However, a similar lack of propensity to extrahepatic iron distribution was observed in multiply transfused patients with homozygous sickle cell disease¹⁶ or sickle/β⁰-thalassaemia¹⁷ in whom splenectomy or autosplenectomy is frequently encountered, suggesting a potential role of splenectomy also in modifying the kinetics of iron storage and elimination¹⁸. Further studies are needed to confirm this hypothesis at a molecular level and should include evaluations, in this category of patients, of the levels of non-transferrin-bound iron and of plasma hepcidin and growth differentiation factor 15, which may be involved in the reduction of extrahepatic iron distribution¹⁹. Furthermore, it could be argued that expanded erythropoiesis, wherever it occurs, could serve as a sink for iron, in the absence of which, iron loading of cardiac parenchyma is more likely to occur. On the other hand, data on the genetic profile of patients without EMH highlight the prevalence among this group of carriers of β⁰-thalassemia alleles with the typical phenotype of thalassaemia major patients, characterised by an earlier age at first transfusion and higher transfusional iron intake compared with patients with EMH.

Thus, although this study lacks information on other genetic modifiers of the clinical severity of thalassaemia, our data identified a genotype/phenotype association characterised by a tendency to excessive erythroid hyperplasia and massive expansion of the erythroid marrow. Further studies in a larger series of patients are needed to confirm these findings.

Overall, considering the 67 evaluable patients in this retrospective study, the prevalence of IVS 1–6/ codon 39 compound heterozygotes (18%) was higher than that of all other genetic compounds; although there are no other similar studies with which to compare this finding, it is not surprising because codon 39 and IVS 1–6 alleles are among the more frequently genetic defects involved in the wide spectrum of thalassemia mutations in southern Italy²⁰.

In conclusion, our data led us to hypothesise that in EMH in thalassaemia could be at least in part genetically sustained and may suggest that compound heterozygotes for IVS 1–6/codon 39 have a higher probability of developing serious complications. Further studies are needed to demonstrate whether a proper transfusion regimen at an earlier age is able to reduce the risks associated with this genetic background.