Incidence and natural history of pure red cell aplasia in major ABO-mismatched haematopoietic cell transplantation

Summary

Major ABO mismatching is not considered a contraindication to allogeneic haematopoietic stem cell transplantation (HSCT). Modern reduced-intensity conditioning and reduced-toxicity regimens cause much less myeloablation than conventional myeloablative regimens, such as cyclophosphamide with busulfan or total body irradiation, which may affect the incidence of pure red cell aplasia (PRCA). We estimated the incidence and described the natural history of PRCA in patients with major ABO-mismatched donor stem cells. Between 2007 and 2008, 161 (27% of all patients undergoing HSCT) underwent allogeneic HSCT with major ABO-mismatched stem cells and 12 (7·5%) of these patients developed PRCA. Thirty and ninety day T-cell and myeloid cell chimerism and neutrophil and platelet engraftment did not differ between patients who developed PRCA and those who did not. The only risk factor associated with PRCA was the use of a fludarabine/busulfan conditioning regimen. All patients with PRCA needed red cell transfusion for several months after HSCT resulting in significant iron overload. Pure red cell aplasia resolved spontaneously in the majority (seven patients) but only resolved after stopping tacrolimus in three patients. Hence, after major ABO-mismatched HSCT, the incidence of PRCA was 7·5% and it resolved spontaneously or after withdrawal of immunosuppression in the majority of patients.

Major ABO mismatching is not considered a contraindication to allogeneic haematopoietic stem cell transplantation (HSCT). The human leucocyte antigen (HLA) system is inherited independently of the blood group system, thus blood group incompatibility between donor and recipient is common and may occur in up to 40% of all HSCTs (Bensinger et al, 1982; Barge et al, 1989; Bolan et al, 2001; Worel et al, 2003).

Modern reduced-intensity and reduced-toxicity regimens, such as busulfan with fludarabine, cause much less myeloablation than conventional myeloablative regimens, such as cyclophosphamide with busulfan or total body irradiation. Therefore, recipient cells producing donor red cell-specific antibodies are more likely to survive after a major ABO-mismatched transplant and cause pure red cell aplasia (PRCA).

The frequency and natural history of PRCA with current reduced-toxicity regimens is not known. We therefore, sought to study the incidence and describe the natural history of PRCA after HSCT in patients who received major ABO-mismatched haematopoietic stem cells.

Patients and methods

This study was conducted at The University of Texas MD Anderson Cancer Center and a retrospective study protocol was approved by the Institutional Review Board.

We reviewed the records of all patients who underwent allogeneic HSCT at the MD Anderson Cancer Center between January 2007 and December 2008. Patient demographics (age and sex) and clinicopathological characteristics [diagnosis, relationship to donor, graft source, cell dose, type of conditioning regimen, ABO compatibility with donor, incidence of acute graft-versus-host disease (aGVHD) and PRCA, chimerism, serum ferritin levels, time to engraftment and number of red cell transfusions] were obtained from the Laboratory Information Systems and the Department of Stem Cell Transplantation data base.

Conditioning regimens

Pre-HSCT conditioning regimens were classified according to Center for International Blood & Marrow Transplant Research (CIBMTR) guidelines (Bacigalupo, 2004; Giralt, 2005; Giralt et al, 2009) as myeloablative or reduced-intensity regimens.

Definitions

Pure red cell aplasia. Pure red cell aplasia was determined to be present when the bone marrow biopsy on posttransplant day 30 demonstrated adequate myeloid, lymphoid and megakaryocyte populations in the setting of absent or nearly absent erythroid precursors with absence of donor red cells on forward red cell typing of the recipient red cells and the recipient being red cell transfusion-dependent.

ABO incompatibility

For the purpose of this study, ABO incompatibility was divided into two main groups, major and minor.

A major ABO incompatibility existed when the recipient had naturally occurring isohaemagglutinins against the red cell antigen(s) present on the surface of the donor’s red blood cells. This occurred between groups A, B, or AB donors and group O recipients (A→O, B→O, AB→O) and between group AB donors and group A or B recipients (AB→A, AB→B). Patients with bidirectional mismatch (A→B, B→A) were also included with the major group.

A minor ABO incompatibility existed when the donor had naturally occurring isohaemagglutinins against the red cell antigen(s) present on the surface of the recipient’s red blood cells. This occurred between group O donors and group A, B and AB recipients (O→A, O→B, O→AB).

Stem cell processing

For all major ABO incompatible HSCTs (A→O, B→O, AB→A and AB→B) the stem cell product was red cell-depleted. For bidirectional ABO incompatible HSCTs (A→B and B→A), the stem cell product was red cell- and plasma-depleted.

Blood group serology

ABO forward/reverse typing was determined serologically using whole blood collected in ethylenediamineteteraacetic acid (EDTA). ABO typing was performed according to routine methods established at the blood bank at the MD Anderson Cancer Center, including both the solid-phase technique and manual tube methods. Anti-donor and anti-host isohaemagglutinin titres were not routinely performed at our institution during the study period,

Engraftment

White cell and platelet engraftment

The day of neutrophil engraftment was defined as the first of 3 consecutive days on which the patient’s absolute neutrophil count was >0·5 × 10⁹/l. The day of platelet engraftment was defined as the first of seven consecutive days on which the platelet count was >20 × 10⁹/l without platelet transfusion.

Red cell engraftment

The day of red cell engraftment was defined as 30 days after HSCT for patients who did not require red cell transfusion or for patients who were red cell transfusion-dependent, the day on which donor red cells appeared on forward red cell typing, or the last day of red cell transfusion for patients who did not undergo follow-up by red cell type and screen. Red cell type and screens were performed every third day for those who needed red cell transfusions.

Statistical analysis

Descriptive statistics were used for patient demographics; age at transplant, sex, disease (myeloid or lymphoid), graft source (bone marrow or peripheral blood), donor (sibling or unrelated), cell dose (CD34), intensity of conditioning regimen (myeloablative or reduce-intensity), conditioning regimen (fludarabine/busulfan or non-fludarabine/busulfan), aGVHD (grade II–IV) and serum ferritin. The Wilcoxon rank-sum test was used to compare variables between the PRCA and non-PRCA patients. A logistic regression model was used to determine which variables affected the risks of developing PRCA. The backward elimination method was used to derive the final multivariate model. Time to red cell engraftment was calculated and plotted using the Kaplan Meier method and the log-rank test was used to compare the differences in time to engraftment between PRCA and the non-PRCA patients. All P-values less than 0·05 were considered statistically significant and analyses were performed using SAS 9.2 (SAS Institute Inc, Cary, NC) and r version 2.11.1 (R foundation for statistical computing; http://www.r-project.org)

Results

Patient characteristics

During the study period, 596 patients underwent allogeneic HSCT of which 308 patients (52%) received ABO-matched HSCTs, 127 (21%) received minor ABO-mismatched stem cells and 161 patients (27%) received major ABO-mismatched haematopoietic stem cells. PRCA was not seen in the patients who had an ABO-matched donor or a minor ABO-mismatched donor. Further analysis was therefore limited to 161 patients with major ABO mismatched donors (Table I). Of these major ABO mismatched patients, 101 were males and 60 were females. Most patients (86/161; 53%) had a diagnosis of acute myeloid leukaemia.

Table I.

Patient characteristics.

Characteristics	Total, n (%)	Patients with PRCA, n (%)	Patients without PRCA, n (%)
Total	161	12 (7·5)	149 (92·5)
Sex
Male	101 (63)	6 (6)	95 (94)
Female	60 (37)	6 (10)	54 (90)
Median age (range), years	54 (2–74)	52 (24–63)	54 (2–74)
Diagnosis
AML/MDS	86 (53)	8 (9)	78 (91)
ALL	10 (6)	1 (10)	9 (90)
Lymphoma or myeloma	54 (34)	1 (2)	53 (98)
Other haematological disorders	11 (7)	2 (18)	9 (82)
Type of donor stem cells
Matched related (10/10)	46 (29)	5 (11)	41 (89)
Matched unrelated	84 (52)	7 (8)	77 (92)
Mismatched related/unrelated	31 (19)	0 (0)	31 (100)
Graft source
Bone marrow	39 (24)	6 (15)	33 (85)
Peripheral blood	100 (62)	6 (6)	94 (94)
Cord blood	22 (14)	0 (0)	22 (100)
Median CD34 cell dose (range), (× 106/kg)	4·33 (0·03–53·88)	4·64 (1·59–53·88)	4·07 (0·03–32·11)
Intensity of preparative regimen
Myeloablative	67 (42)	6 (9)	61 (91)
Reduced-intensity	94 (58)	6 (6)	88 (94)
Preparative regimen
Fludarabine/melphalan	48 (30)	3 (6%)	45 (94)
Fludarabine/busulfan – myeloablative	37 (23)	6 (16)	31 (84)
Fludarabine/busulfan – reduced-intensity	12 (7)	3 (25)	9 (75%)
BEAM	14 (9)	0 (0)	14 (100)
Fludarabine/cytoxan/alemtuzumab/rituximab	29 18)	0 (0)	29 (100)
Others	21 (13)	0 (0)	21 (100)
ABO incompatibility
A→O	86 (53)	5 (6)	81 (94)
B→O	23 (14)	2 (9)	21 (91)
AB→O	8 (5)	3 (38)	5 (62%)
A→B	21 (13)	2 (10)	19 (90%)
B→A	11 (7)	0 (0)	11 (100)
AB→A	9 (6)	0 (0)	9 (100%)
AB→B	3 (2)	0 (0)	3 (100)
Acute graft-versus-host disease (Grade II–IV)	41 (25)	0 (0)	41 (100)

PRCA, pure red cell aplasia; AML/MDS, acute myeloid leukaemia/myelodysplastic syndrome; ALL, acute lymphoblastic leukaemia; BEAM, camustine/etoposide/cytarabine/melphalan.

Incidence of PRCA

Pure red cell aplasia developed in 12 (7·5%, CI 3·4–11·5%) of the 161 patients that received major ABO-mismatched HSCTs (Table II). The remaining 149 of the major ABO-mismatched patients (93%) without PRCA were used as a control group. Characteristics of the two groups are summarized in Tables I and III.

Table II.

PRCA patient characteristics.

Age (years)/ sex	ABO (D/R)	Diagnosis	Graft/ Donor	Preparative regimen	ANC >0·5 × 109/l	Plt > 20 × 109/l	App of donor RBCs (days)	Resolution	Serum ferritin μg/l (pre-post Tx)	Liver iron (mg/g d.w.)	Day-30 chimerism T/Myeloid cells %	Day-90 chimerism T/myeloid cells %
36/F	A/B	AML	M/MUD	Flu 40 mg/m2 × 4 /  Bu 130 mg/m2 × 4 /  ATG 4 mg/kg	12	12	376	After tacro taper	385/1918	13·0	100/100	100/100
63/F	AB/O	AML	M/MUD	Flu 25 mg/m2 × 4 /  Mel 70 mg/m2 × 2 /  ATG 4 mg/kg	12	16	83	Spontaneous	791/1943	Not done	100/100	100/100
25/F	AB/O	AML	M/MUD	Flu 40 mg/m2 × 4 /  Bu 130 mg/m2 × 4 /  ATG 4 mg/kg	14	18	51	Spontaneous	1146/2437	Not done	Not done	68/100
54/M	A/O	AML	PB/Sib	Flu 40 mg/m2 × 4 /  Bu 130 mg/m2 × 4	12	13	236	After tacro taper	913/3102	Not done	31/100	100/100
56/M	B/O	Myeloma	PB/Sib	Flu 25 mg/m2 × 4 /  Mel 70 mg/m2 × 2 /  ATG 4 mg/kg	12	16	364	After tacro taper	92/1959	17·0	95/99	Not done
60/F	A/O	ALL	PB/Sib	Flu 25 mg/m2 × 4 /  Mel 70 mg/m2 × 2 /  ATG 4 mg/kg	12	28	98	Spontaneous	1452/2773	14·5	100/100	Not done
64/M	AB/O	AML	M/MUD	Flu 40 mg/m2 × 4 /  Bu AUC 4000 /ATG 4 mg/kg	20	24	Not  detected	No resolution*	1640/6392	23·5	85/100	17/100
44/M	A/B	CML	M/MUD	Flu 40 mg/m2 × 4 /  Bu 130 mg/m2 × 2 /  ATG 7·5 mg/kg	13	17	47	Spontaneous	375/Not done	Not done	99/99	7/99
51/F	A/O	MDS	M/MUD	Flu 40 mg/m2 × 4 /  Bu 130 mg/m2 × 2 /ATG 7·5 mg/kg	13	29	211	Spontaneous	3202/4275	25·0	49/100	89/99
43/F	A/O	AML	PB/MUD	Flu 40 mg/m2 × 4 /  Bu AUC 6000 /ATG 4 mg/kg	12	14	226	Spontaneous	623/3512	12·0	74/99	100/100
48/M	A/O	MDS	PB/Sib	Flu 40 mg/m2 × 4 /  Bu AUC 5000	19	35	Not  detected	No resolution**	3307/6756	Not done	21/99	Not done
59/M	B/O	AML	PB/Sib	Flu 40 mg/m2 × 4 /Bu AUC 6000	12	12	268	Spontaneous	2753/2723	Not done	99/99	63/99

F, Female; M, Male; D/R, Donor/Recipient; AML, Acute Myeloid Leukaemia; ALL, Acute Lymphoblastic Leukaemia; CML, Chronic Myeloid Leukaemia; MDS, Myelodysplastic Syndrome; M, Marrow; PB, Peripheral Blood; MUD, Matched Unrelated Donor; Sib, Sibling; Flu, fludarabine; Bu, busulfan; Mel, melphalan; ATG, Anti-Thymocyte Globulin; AUC = Area under Curve; ANC, Absolute Neutrophil Count; Pit, Platelet; App of Donor RBCs, Appearance of donor red cells in recipient; After Tacro Taper, After Tacrolimus Taper; Pre-Post Tx, Pre-Post Transplant.

^*Patient received a second transplant due to graft failure.

^**Patient died from sepsis.

Engraftment

Donor T-cell and myeloid cell chimerism at 30 and 90 days after HSCT, as well as neutrophil and platelet engraftment did not differ significantly between patients who developed PRCA and those who did not. However, none of the 12 patients with PRCA had aGVHD compared with 41 of 149 (27%) patients without PRCA (Table III).

Table III.

Haematopoietic recovery.

Variable	PRCA	No PRCA	P-value
Days to ANC >0·5 × 109/l, median (range)	12 (12–20)	12 (0–43)	0·87
Days to platelet count >20 × 109/l, median (range)	17 (12–35)	14 (0–65)	0·48
Median 30-day donor chimerism, range (%)
T cells	95 (21–100%)	100 (0–100%)	0·15
Myeloid cells	100 (99–100%)	100 (0–100%)	0·88
Median 90-day donor chimerism, range (%)
T cells	89 (7–100%)	100 (0–100%)	0·26
Myeloid cells	100 (99–100%)	100 (0–100%)	0·41
Mean serum ferritin, μg/l
Before transplant	1390	1677	0·90
Day 100 posttransplant	3435	2708	0·04
Median days to appearance of donor red cells	218*	30	<0·0001
Mean red cell units transfused	47	7	<0·0001
Liver iron, median (range), mg/g dry weight	15·8 (12·5–25)	5·0 (2·0–15)	0·03

PRCA, pure red cell aplasia; ANC, absolute neutrophil count.

^*Two patients had persistent PRCA.

Red cell engraftment and red cell transfusions

The appearance of donor red cells was markedly delayed in patients with PRCA at a median of 231 days after HSCT compared with a median of 30 days after HSCT in the non-PRCA group (P < 0·0001; Table III). Patients with PRCA also required more red cell transfusions, with a mean of 47 red cell units compared with seven red cell units in the non-PRCA group (P < 0·0001; Table III).

Serum ferritin and liver iron

Serum ferritin levels did not differ between the PRCA and non-PRCA groups before HSCT. However mean (standard deviation) serum ferritin levels 100 days after HSCT were elevated in the PRCA group compared with the non-PRCA group [PRCA 3435 (1710) μg/l; non-PRCA, 2708 (2672) μg/l; P = 0·04; Table III]. There was a strong correlation with serum ferritin 100 days after HSCT, red cell transfusion and liver iron overload as demonstrated by liver magnetic resonance imaging (Majhail et al, 2008; Sucak et al, 2008; Ozkurt et al, 2009; Busca et al, 2010; Dahl et al, 2010; Lim et al, 2010). In six patients who developed PRCA, the median liver iron levels, as measured by magnetic resonance imaging, were 15·8 mg/g of dry weight (d.w.) (range, 12·5–25·5 mg/g d.w.) at last follow-up. Liver iron concentration was significantly higher in six PRCA patients than in five control ABO-mismatched patients without PRCA and high (>1000 μg/l) ferritin, 15·8 mg/g d.w. compared to 5 mg/g d.w. (P = 0·03).

Resolution of PRCA

Pure red cell aplasia spontaneously resolved at a median of 98 days after HSCT (range 47–268 days) in seven patients (58%); none of these patients received any intervention or treatment for PRCA except red cell transfusions. However, three patients later relapsed and died from their underlying malignant disease, and another died of post-transplant lymphoproliferative disorder. Among the five patients (42%) who did not experience spontaneous resolution of PRCA, the PRCA resolved in three, 60 to100 days after Tacrolimus was stopped (236, 364 and 376 days after HSCT). One patient developed graft failure and experienced persistent PRCA despite a second transplant and died from sepsis and renal failure (547 days after the first HSCT). The remaining patient did not achieve red cell engraftment despite treatment with plasma exchange, rituximab and intravenous immunoglobulin and died of overwhelming sepsis and multi-organ failure 549 days after transplant (Fig 1). Six patients who developed PRCA are alive and in remission at a median follow-up time of 29 months (range 22–44) after HSCT. At 3 years after transplant, overall survival was not significantly different from the control patients (Fig 2): 58·3% for the PRCA group and 45·5% for the non-PRCA group (P = 0·37).

Fig 1.

Kaplan–Meier analysis showing time to red cell recovery. PRCA, pure red cell aplasia.

Fig 2.

Overall survival for PRCA patients versus non-PRCA patients. PRCA, pure red cell aplasia.

Risk factors

The only significant factor identified in univariate and multivariate analyses for the development of PRCA was the conditioning regimen used. Patients who received fludarabine/busulfan regimen had a significantly higher risk (odds ratio, 8·18; 95% confidence interval, 2·07–31·14) of developing PRCA than those who received an alternative regimen. The higher risk was seen in both the fludarabine/busulfan myeloablative conditioning group (P = 0·008) and the reduced intensity fludarabine/busulfan group (P = 0·01) when considered separately. Other variables including age, sex, disease, donor type, graft source, intensity of preparative regimen and aGVHD were not associated with PRCA (Table IV).

Table IV.

Univariate and multivariate analysis of risk factors.

Variable	Total patients (patients with PRCA)	P-value	OR	95% CI for OR	Multivariate P-value	OR	95% CI for OR
Age at transplant	161(12)	0·083	1·004	0·97–1·04
Sex
Male	101(6)	–	1·00	–
Female	60(6)	0·35	1·76	0·54–5·72
Disease type
Myeloid	94(10)	–	1·00	–
Lymphoid	67(6)	0·09	0·26	0·06–1·22
Donor type
Matched unrelated	83(7)	–	1·00	–
Alternative*	35(0)	0·15	0·14	0·008–2·69
Matched sibling	43(5)	0·13	1·463	0·45–4·73
Graft
HPC-A	99(6)	–	1·00	–
HPC-C	23(0)	0·25	0·31	0·02–5·97
HPC-M	39(6)	0·06	2·79	0·87–8·97
Intensity of preparative regimen
Myeloablative	67(6)	–	1·00	–
Reduced-intensity	94(6)	0·54	0·69	0·21–2·25
aGVHD
Yes	120(12)	–	1·00	–
No	41(0)	0·12	0·11	0·01–1·87
Conditioning regimen
Non-fludarabine/busulfan	112(3)	–	1·00	–	–	1·00	–
Fludarabine/busulfan	49(9)	0·002	8·18	2·07–31·14	0·002	8·18	2·07–31·14

PRCA, pure red cell aplasia; OR, odds ratio; Cl, confidence interval; HPC-A, haematopoietic progenitor cells, apheresis; HPC-C, haematopoietic progenitor cells, cord; HPC-M, haematopoietic progenitor cell, marrow; aGVHD, acute graft-versus-host disease.

^*Cord blood and mismatched unrelated or related donors.

Discussion

This study of patients undergoing major ABO-mismatched HSCT found a 7·5% incidence of PRCA, which was associ ated with a fludarabine and busulfan conditioning regimen.

In previous smaller studies (Bolan et al, 2001; Helbig et al, 2007; Zhu et al, 2007) the incidence of PRCA ranged from 0 –29%, with 0% in 12 patients receiving myeloablative conditioning with cyclophosphamide and total body irradiation and 29% in 14 patients receiving reduce-intensity conditioning with fludarabine and cyclophosphamide (Bolan et al, 2001). A second study of 42 patients showed a 26% incidence of PRCA (Zhu et al, 2007). Yet in another study, six of 44 patients (13%) receiving mainly a myeloablative conditioning regimen developed PRCA (Helbig et al, 2007). Hence, conditioning regimens and sample size may account for the differences between these studies and our much larger study. In their retrospective analysis of 11 cases with PRCA from a series of 42 patients undergoing major ABO mismatch transplants from 1997 to 2005, Zhu et al (2007) concluded that blood group A/O donor/recipient pairs was significantly associated with the occurrence of PRCA. However, this was not the case in our study.

Our multivariate regression analysis showed that the only significant risk factor for developing PRCA was the use of fludarabine/busulfan conditioning regimen, suggesting that isoagglutinin-producing plasma cells survive this conditioning regimen better than other regimens. Nine of 49 (18%) patients receiving this conditioning regimen developed PRCA, including 6 of 37 (16%) receiving myeloablative therapy and 3 of 12 (25%) receiving a reduced intensity regimen. Although we studied a large number of patients over a 2-year period, only 12 PRCA were identified, and this low event rate limits our ability (power) to identify risk factors for PRCA. For example, no patient in the PRCA group developed aGVHD (Grade II–IV), but 27% of patients without PRCA had prior GVHD (P = 0·12), a statistically insignificant result possibly due to inadequate power. Similarly, GVHD was absent in all 6 patients with PRCA in a previous study (Helbig et al, 2007).

It is also worth noting that, in previous studies, aGVHD was associated with a faster decline in anti-donor isohaemagglutinins (Bolan et al, 2001; Griffiths et al, 2005), which is the cause of PRCA in patients receiving major ABO-mismatched transplants (Griffiths et al, 2005). Therefore, the association between the absence of GVHD and PRCA has a biological basis and a much larger study may be required to prove this. Moreover, previous studies have shown that reduced intensity (Couriel et al, 2004) and reduced toxicity regimens like busulfan and fludarabine (Andersson et al, 2008) are associated with lower incidence of GVHD. Therefore, we cannot exclude the possibility that, in our study, busulfan and fludarabine conditioning is truly a confounding variable and that patients who received busulfan and fludarabine may also have a lower incidence of GVHD and a lower graft-versus-plasma cell effect, which in turn results in a slower decline in donor-specific isohaemagglutinins and increases the risk of red cell aplasia. Likewise, busulfan and fludarabine was only used for patients with myeloid malignancies and not lymphoid malignancies. Patients with lymphoid malignancies are likely to have previously received rituximab, which may lower isoagglutinin titres and thereby the risk of PRCA. In our study the odds ratio for development of PRCA for lymphoid malignancies was 0·26 (P = 0·09). A much larger study will be needed to clarify the impact of busulfan and fludarabine, GVHD and prior rituximab.

For most patients (7/12; 58%) who developed PRCA in our study, the PRCA resolved spontaneously with only red cell transfusion support. This is in agreement with other studies (Bolan et al, 2001; Helbig et al, 2007). Three (25%) of the patients with PRCA in our study achieved remission 60–100 days after withdrawal of tacrolimus without specific intervention other than transfusion support. This may be due to ‘unchecking’ of the graft-versus-plasma cell effect after stopping tacrolimus. A judicious treatment strategy would therefore be transfusion support pending spontaneous resolution. Early withdrawal of immunosuppression may be effective, but could result in life threatening GVHD. Therefore, it would be prudent to withdraw immunosuppression as per standard institutional practice.

Other treatment modalities for PRCA that have been tried are plasma exchange with donor-type plasma replacement (Damodar et al, 2005), treatment with a CD20 monoclonal antibody (rituximab) directed at residual recipient B lymphocytes (Helbig et al, 2005), removal of persistent isohaemagglutinins with Ig-Therasorb® immunoadsorption (Rabitsch et al, 2003), donor lymphocyte infusions (Musso et al, 2004) and mesenchymal stem cell infusions (Fang et al, 2009). One of our patients did not achieve resolution of PRCA despite intervention treatment with erythropoietin, rituximab, plasma exchange, and intravenous immunoglobulin. The efficacy of these anecdotal approaches remains to be proven in a larger study.

Our findings indicate that PRCA is a significant problem following major ABO-mismatched transplantation, resulting in a significant need for long-term transfusion and morbidity due to iron overload. It is therefore prudent to select a donor without a major ABO-mismatch if more than one fully matched donor is available. Patients who develop PRCA should receive regular transfusions pending spontaneous resolution. Withdrawal of tacrolimus does result in resolution of PRCA in the majority of patients who do not have spontaneous resolution. However, the benefit of early tacrolimus withdrawal needs to be balanced against the risk of potential GVHD. Further, larger studies are needed to better define the natural history of PRCA, its impact on morbidity and mortality and to identify other risk factors for PRCA, such as GVHD, associated primary diseases (acute myeloid leukaemia versus lymphoid malignancy) other conditioning regimens or a particular donor recipient mismatch (O recipient and A or B donor, versus A or B recipients and AB or A or B donors). Lastly, potential therapeutic approaches, such as rituximab or bortezomib (Poon & Koh, 2012) need to be validated in a prospective trial.

In summary, major ABO-mismatch may lead to PRCA, which resolves spontaneously or after tacrolimus withdrawal in most patients. These patients require transfusion support for longer periods and are at risk of significant iron overload.