Massive immune haemolysis after allogeneic peripheral blood stem cell transplantation with minor ABO incompatibility


Susan F. Leitman, National Institutes of Health, Department of Transfusion Medicine, Building 10, Room 1C711, Bethesda, MD 20892-1184, USA. E-mail:


Immune haemolysis as a result of minor ABO incompatibility is an underappreciated complication of haematopoietic transplantation. The increased lymphoid content of peripheral blood stem cell (PBSC) transplants may increase the incidence and severity of this event. We observed massive immune haemolysis in 3 out of 10 consecutive patients undergoing HLA-identical, related-donor PBSC transplants with minor ABO incompatibility. Non-ablative conditioning had been given in 9 of these 10 cases, including two with haemolysis. Cyclosporin alone was used as prophylaxis against graft-vs.-host disease (GVHD). Catastrophic haemolysis of 78% of the circulating red cell mass led to anoxic death in the first case seen, but severe consequences were avoided by early, vigorous donor-compatible red cell transfusions in the subsequent two cases. Haemolysis began 7–11 d after PBSC infusion and all patients with haemolysis had a positive direct antiglobulin test (DAT), with eluate reactivity against the relevant recipient antigen. However, neither the intensity of the DAT, the donor isohaemagglutinin titre, nor other factors could reliably be used to predict the occurrence of haemolysis. Our data indicate that haemolysis may be frequent and severe after transplantation of minor ABO-incompatible PBSCs when utilizing cyclosporin alone to prevent GVHD. Meticulous clinical monitoring and early, vigorous donor-compatible red cell transfusions should be practiced in all instances.

ABO incompatibility between donor and recipient occurs in 30–40% of subjects undergoing haematopoietic transplantation, owing to the fact that ABO blood groups are inherited independently from human leucocyte antigens (HLAs) (Klumpp, 1991; Petz, 1998). Immediate haemolysis as a result of ABO incompatibility is generally avoided by graft processing to remove incompatible red cells and plasma, and transfusion of donor-compatible blood components (Petz, 1998; Vengelen-Tyler, 1999a). Delayed immune haemolysis owing to isohaemagglutinin production by graft-associated passenger lymphocytes may take place when minor ABO incompatibility is present (Rowley & Braine, 1982; Hows et al, 1986; Gajewski et al, 1992), which occurs in 10–15% of cases, most commonly when group O donors are paired with group A or B recipients (Klumpp, 1991). The risk of haemolysis as a result of minor ABO incompatibility is increased with the use of an unrelated donor or with cyclosporin used as the sole agent to prevent graft-vs.-host disease (GVHD) (Hows et al, 1986; Gajewski et al, 1992). However, fatal haemolysis has not been described, and ABO incompatibility has not been found to adversely affect overall survival or the occurrence of graft rejection or GVHD in transplant procedures involving myeloablative conditioning and bone marrow as the source of stem cells (Klumpp, 1991; Petz, 1998).

The recent introduction of peripheral blood stem cells (PBSCs) as the haematopoietic graft source and low intensity non-myeloablative regimens for recipient conditioning has dramatically widened the applications of haematopoietic transplants (Khouri et al, 1998; Appelbaum, 1999; Carella et al, 2000). PBSCs are rapidly supplanting bone marrow as the preferred source of stem cells owing to their higher stem cell content, more rapid recovery of haematopoiesis, improved overall survival and relative ease of collection (Bensinger et al, 2000; Powles et al, 2000). Low intensity conditioning regimens have further reduced regimen-related toxicity and rely on the generation of graft-vs.-tumour immune effects rather than drug or radiation-induced cytotoxic effects to eradicate malignant disease (Khouri et al, 1998; Childs et al, 1999a). Non-myeloablative strategies employ anti-GVHD regimens balanced to favour donor engraftment and immune anti-tumour responses, and may utilize PBSC grafts with a high lymphocyte content to further enhance these effects (Childs et al, 1999a). PBSCs have recently been associated with an increased severity of passenger lymphocyte-associated haemolysis after traditional ablative transplants with minor ABO incompatibility (Toren et al, 1996; Bornhauser et al, 1997; Laurencet et al, 1997; Moog et al, 1997; Oziel-Taieb et al, 1997; Salmon et al, 1999). The potential thus exists for even greater minor ABO-mediated haemolysis in regimens utilizing PBSCs and non-myeloablative conditioning, owing to the immune-based design of such regimens. It is important that this complication be fully described and anticipated in view of the expectation that the overall toxicity of PBSC transplants will be decreased when compared with traditional approaches.

In this report, we describe fatal and severe haemolysis as a result of PBSC transplants with minor ABO incompatibility, in the setting of both ablative and non-myeloablative conditioning. In the first patient, anoxic death owing to catastrophic haemolysis of more than 75% of the red cell mass occurred 7 d after infusion of a PBSC graft with minor ABO incompatibility. Two additional cases of haemolysis, one involving ablative and one non-ablative conditioning, were detected during prospective serological monitoring of the next nine consecutive minor ABO incompatible PBSC transplants. In both cases, severe consequences were avoided by prompt, vigorous donor-compatible red cell transfusions. A prospective laboratory monitoring schedule is described and guidelines for prompt intervention outlined as part of the strategy to reduce the toxicity of this increasingly common transplant complication.

Patients and methods

Patients and transplant procedures All patients were enrolled in institutional review board-approved protocols and gave informed consent for participation in transplant trials. Nine out of 10 patients received non-myeloablative conditioning, as described in the case reports for patients 1 and 9. A single patient received myeloablative conditioning, as described in case report 3. Donor and recipient HLA compatibility were determined using allele-level HLA typing (Childs et al, 1999b). All 10 patients received cyclosporin alone, without methotrexate or other anti-proliferative agents, as GVHD prophylaxis. No patients received prophylactic or therapeutic intravenous immunoglobulin preparations.

The target dose for PBSC grafts was a minimum of 5·0 × 106 CD34+ cells/kg to promote engraftment and potentiate graft-vs.-tumour effects. PBSCs were obtained following granulocyte colony-stimulating factor (GCSF) administration (filgrastim 10 μg/kg/d subcutaneously for 5 d), followed by two 15-l leukapheresis procedures, performed using a CS-3000 Plus cell separator (Baxter Healthcare, Deerfield, IL, USA) on d 5 and 6. All donors except for the index patient were HLA-identical siblings. CD34+ and CD3+ cell concentrations in PBSC components were measured using flow cytometry (Read et al, 1997). PBSC products were depleted of plasma by centrifugation, with resuspension of the cells in Plasmalyte-A (Baxter). The final product contained less than 15 ml of donor plasma. PBSC concentrates were infused fresh in four patients and were cryopreserved in liquid nitrogen prior to infusion in the remaining six cases (Table I).

Table I.  Clinical and laboratory data for consecutive PBSC transplants with Minor ABO incompatibility.
     Peak DAT   







  1. All patients except patient 3 received non-myeloablative conditioning with cyclophosphamide and fludarabine. Patient 3 received cyclohosphamide and irradiation. All patients received cyclosporin alone without methotrexate as anti-GVHD prophylaxis. CLL, chronic lymphocytic leukaemia; RCC, renal cell cancer; AML, acute myelogenous leukaemia; MM, multiple myeloma; CML, chronic myelogenous leukaemia; NHL, non-Hodgkin's lymphoma; CMML, chronic myelomonocytic leukaemia; D→R, donor to recipient; F, female; M, male; Donor pre titre, serum titre to the relevant recipient ABO antigen; PBSC, graft cell dose of CD34+ and CD3+ cells (106/kg); Fresh, infused fresh; Cryo, infused after cryopreservation; Peak DAT, strength of strongest positive direct antiglobulin test to IgG, complement (C3d), and day of strongest test; wk, weak; neg, negative; NT, not tested; Serum antibody screen, strength of antibody to relevant donor ABO group. Donors were HLA-identical siblings except for Patient 1.

wkneg11Anti-A2 +Yes

Serial serological monitoring Prospective serologicalal monitoring was initiated after the index case for all subsequent PBSC transplants with minor ABO incompatibility. Direct antiglobulin tests (DATs) were performed daily and red cell antibody screens were assayed every 3 d for the first 14 d following PBSC infusion in patients 2–10. When a positive DAT was detected, or when the intensity of the DAT increased, red cell eluates were evaluated for the presence of the relevant ABO isohaemagglutinin. Complete blood counts were monitored daily, and twice daily on d 7–14; lactate dehydrogenase (LDH) was measured daily. Haemolysis was defined as an LDH increase of more than 50% above the previous day's values (normal range 113–226 U/l) and a haemoglobin (Hb) decrease of more than 2 g/dl over 24 h in the absence of bleeding, or a Hb that did not rise in response to at least two consecutive red cell transfusions. The extent of haemolysis was calculated using the patient's haematocrit, weight and blood volume (estimated at 70 ml/kg). The percentage of host-derived red cells remaining in the circulation (the residual recipient red cell fraction) following red cell exchange was calculated using known mixtures of group O and B cells in a buffered-gel card system (Micro Typing Systems, Pompano Beach, FL, USA). A standard curve of mixed field agglutination test results was constructed from these cell mixtures.

Transfusion practices All patients received leucoreduced and irradiated blood components. Transfusions consisted of donor-compatible red cells and recipient-compatible platelets and plasma, starting 2 weeks prior to conditioning and continuing throughout the transplant period. Red cell units were collected in additive solution to minimize plasma content. Platelet support was entirely from plateletpheresis donations. A single pretransplant 10-unit red cell exchange was performed in patient 2. Red cell exchanges were not performed in other patients.

Laboratory tests Blood samples were evaluated using standard immunohematological techniques for blood type, red cell antibody screen and DAT, utilizing polyspecific and monospecific reagents (Vengelen-Tyler, 1999b c, d). Cold acid eluates were prepared from red cell samples using an ELU-KIT II (Gamma Biologicals, Houston, TX, USA). Titration studies of isohaemagglutinins were performed by serially diluting sera in saline and testing with commercially obtained reverse grouping cells incubated at room temperature for 30 min (Vengelen-Tyler, 1999b). Post-transplant studies for the degree of donor–recipient chimaerism in patient 1 were performed using polymerase chain reaction analysis of informative minisatellite regions, as previously described (Childs et al, 1999b).


Patient characteristics

The case reports for the three patients with haemolysis are listed below. The clinical and serological data for all 10 patients in this series are listed in Table I.

Patient 1 The index patient was a 55-year-old man with B-cell chronic lymphocytic leukaemia, with persistent diffuse marrow involvement (Rozman et al, 1984) despite six cycles of fludarabine therapy. During fludarabine therapy, the DAT remained normal, without clinical or laboratory evidence of haemolysis. He received non-myeloablative conditioning with cyclophosphamide (60 mg/kg/d for 2 d) and fludarabine (25 mg/m2/d for 5 d). The patient's blood group was A, Rh-positive, and he received a PBSC graft containing 6·1 × 106 CD34 and 5·1 × 108 CD3 cells/kg from his HLA-matched, group O, Rh-positive daughter.

The immediate post-infusion course was uncomplicated and no blood components were administered during the first week. On d 7, fever, hyperbilirubinaemia and increasing LDH occurred, followed on d 8–9 by a precipitous decrease in the haematocrit from 24·9% to 5·6% (Fig 1). Rapid recovery of the absolute neutrophil count to 920 × 109/l also occurred on d 9. The LDH rose to 1364 U/l and the total bilirubin to 149 μmol/l (indirect bilirubin 91 μmol/l), while the creatinine remained within normal range. The peripheral blood smear demonstrated spherocytes and red cell autoagglutination. Transfusion with 6 units of group O red cells resulted in a steady increase in haematocrit. However, the patient experienced a cardiorespiratory arrest while still profoundly anaemic. Respiratory and renal failure ensued, neurological function did not recover and the patient died on d 16 post transplant.

Figure 1.

Course of severe haemolysis in patient 1. On d 8 and 9 post transplant, fever, haemodynamic instability and renal insufficiency occurred, the haematocrit dropped precipitously and the LDH doubled. Although the haematocrit responded to transfusions with group O red cells, cardiopulmonary arrest occurred, renal failure ensued, neurological function did not recover and the patient died on d 16. Serological testing revealed that donor-type anti-A isohaemagglutinins were the cause of massive immune haemolysis.

Red cell serological testing was performed on samples obtained before and after haemolysis. The direct and indirect antiglobulin tests were negative on d 4. On d 7, the DAT was positive, showing moderate coating of the red cells with IgG (1+) and stronger coating with complement C3d (2+). Anti-A1 was detected in the red cell eluate. Isohaemagglutinin activity against A1 cells was not detected in the serum until 2 d later and peaked at a titre of 256 on d 11. The donor's anti-A1 titre was 128. Molecular chimaerism analysis on d 11 demonstrated 60% donor T lymphocytes, 10% donor B lymphocytes and 0% donor myeloid cells.

Patient 3 Patient 3 was a 38-year-old, group B, Rh-positive woman in second relapse of acute myelomonocytic leukaemia. Myeloablative conditioning consisted of cyclophosphamide (60 mg/kg/d for 2 d) and total body irradiation (1260 cGy). She received a PBSC graft containing 5·3 × 106 CD34 and 2·7 × 108 CD3 cells/kg from her HLA-identical, group O, Rh-positive sister.

Serological monitoring was initiated on the first day after PBSC infusion, and donor-compatible group O packed red cell transfusions (2 units on d 3 only) were administered to maintain the Hb concentration greater than 8·5 g/dl. On d 6, the DAT became weakly positive, with C3d coating of the red cells and a non-reactive red cell eluate. On d 7, the DAT increased in intensity to 3+ with anti-C3d, anti-B reactivity was present in the eluate and blood samples appeared grossly haemolysed. The haematocrit decreased from 27% to 22% over several hours, but responded promptly to transfusion of 4 units of group O red cells. Laboratory values the next day were consistent with haemolysis of more than 80% of the patient's estimated red cell mass over the prior 36 h. A forced saline diuresis was administered and the creatinine remained normal. The LDH peaked at 2134 U/l and the total bilirubin at 58 μmol/l, both on d 9. Additional transfusions were required only on d 10 and 13 (2 units of red cells each day). Acute GVHD developed on d 36. Relapsed leukaemia was detected on d 84, and the patient died after unsuccessful salvage chemotherapy.

Patient 9 Patient 9 was a 38-year-old man with refractory non-Hodgkin's lymphoma in relapse after an autologous PBSC transplant. Conditioning for non-myeloablative transplant consisted of cyclophosphamide (1 g/m2/d) and fludarabine (25 mg/m2/d) for 4 d. The patient's blood group was A, Rh-positive, and he received a PBSC graft containing 8·25 × 106 CD34 and 3·63 × 108 CD3 cells/kg from his group O, Rh-positive, HLA-identical sister.

Daily serological monitoring and transfusions of donor-compatible group O red blood cells (1 unit on d 4, 2 units on d 5 and 2 units on d 8) to maintain the Hb concentration greater than 10 g/dl were initiated. On d 10, tea-coloured urine and brown-coloured plasma were noted, the LDH increased from 300 to 696 U/l, and 5 units of red cells were transfused over 24 h, with no increase in the Hb until after the fifth unit was given. The creatinine remained stable, and there were no schistocytes on the peripheral blood smear. The LDH peaked at 1205 U/l on d 12. The patient's DAT became weakly positive on d 11, with IgG coating of the red cells; anti-A1 was recovered in the eluate. Anti-A1 isohaemagglutinins became detectable in the patient's sera on d 8 and rapidly increased in titre from 8 to 64–256 on d 9, 11 and 13 respectively. The patient's red cells showed 4+ agglutination with anti-A1 typing sera on d 3, but were only weakly (1+) positive to anti-A1 with mixed-field agglutination by d 8. The volume of red cells transfused between d 5 and d 11 was equivalent to the patient's entire estimated red cell volume. The patient developed diffuse pulmonary haemorrhage on d 13 and thrombotic microangiopathy on d 26, requiring numerous additional red cell transfusions. He is currently alive and well without recurrent disease.

Clinical and laboratory findings

The clinical and laboratory characteristics of the 10 patients in this series are summarized in Table I. Median PBSC graft content was 6·7 × 106 CD34 and 3·2 × 108 CD3 cells/kg, with no significant differences in these cell doses for grafts associated with haemolysis compared with those without haemolysis (Table I). In all cases with haemolysis, a positive DAT was detected within 0–2 d of haemolysis (IgG plus C3d in one case, and either C3d alone or IgG alone in the other two), with red cell eluates containing the relevant isohaemagglutinin. The DAT was also positive in five out of seven cases without haemolysis (IgG alone in three cases and weakly reactive IgG and C3d in two cases), with the red cell eluate containing relevant isohaemagglutinins in one of these cases. Isohaemagglutinins with activity against the relevant recipient antigen were detectable in the sera of all three cases with haemolysis, as well as in one patient without haemolysis. Coagulation studies and renal function were normal at the onset of haemolysis in all three cases.

No statistically significant relationships could be established in this small series between donor-recipient ABO types, donor isohaemagglutinin titres, qualitative or quantitative aspects of the direct antiglobulin test, donor sex, CD34 or CD3 cell content of the graft, cryopreservation prior to infusion and the occurrence of massive immune haemolysis (Table I). Isohaemagglutinins were measured with assays that did not distinguish between IgG and IgM isotypes. Grafts from three out of six female donors were associated with haemolysis. In female donors, there was a history of prior pregnancy in two out of three cases (patients 3 and 9) with haemolysis and in three out of three cases without haemolysis. None of the grafts from the four male donors were associated with haemolysis. There was no concurrent or prior PBSC transplant cohort at our institution receiving post-transplant methotrexate for comparison. There were no cases of haemolysis in 14 minor ABO incompatible PBSC transplants in which the graft underwent lymphocyte depletion (positive CD34+ selection) prior to infusion, even when cyclosporin alone was used for GVHD prophylaxis (Barrett et al, 1998). A comparison of the cases with haemolysis in this series with prior published reports of haemolysis after minor ABO incompatible transplants is presented in Table II.

Table II.  Acute haemolysis in minor ABO incompatible PBSC transplants.

Blood group
Donor titre
HLA, Sex
cell dose
× 106/kg
Onset (total)

Peak LDH,
HGB nadir

  • *

    Cases followed with serial monitoring.

  • Blood group, ABO group for donor into recipient; Donor titre, serum titre against the relevant recipient ABO antigen; MTX, methotrexate; ARF, acute renal failure; ALL, acute lymphoblastic leukaemia; CML AP, chronic myelogenous leukaemia, accelerated phase; Bu/Cy, busulfan, cyclophosphamide; MM, multiple myeloma; Cy/TBI, cyclophosphamide, total body irradiation; CLL, chronic lymphocytic leukaemia; Cy/Flu, non-myeloablative cyclophosphamide, fludarabine, as described in text; Id Sib, HLA identical sibling; MUD, matched unrelated donor; MRD, matched related donor; MNC, mononuclear cells; d, day after infusion of cells; DIC, disseminated intravascular coagulation; NS, not stated in report; Gr, grade; ICH, intracerebral haemorrhage; MOSF, multiorgan system failure; TMA, thrombotic microangiopathy.

Toren et al, 1996ALLO→A
Id Sib, F→M
NO540 MNCd 8
(7 d)
Steroids, diuresis
RBC exhange
Bornhauser et al, 1997CML AP
YES6·0 CD34
270 CD3
d 5
(7 d)
Grade IV GVHD d 19
Died D 58 ICH
Laurencet et al, 1997MM
Id Sib, F→M
NO5·0 CD34
430 MNC
d 12
(7 d)
Grade II GVHD d 25
3 U A RBC d 11–12
Oziel-Taieb et al, 1997MMO→ANO3·5 CD34d 840002 U A RBC d 7
Id Sib, F→M
 410 CD3
(7 d)4–5Died MOSF d 20
Salmon et al,1999AraC/Cy/TBIO→A
NO12·2 CD34d 6
(5 d)
Died GVHD d 35
Present series
 Patient 1
NO5·6 CD34
610 CD3
d 8–9
(2–5 d)
Yes; DIC
Cardiac arrest d 9
Died MOSF d 16
 Patient 3
Id Sib, F→F
NO5·3 CD34
270 CD3
d 7–8
(3–5 d)
Vigorous hydration and
transfusion. Recovered.
 Patient 9
Id Sib, F→M
NO8·25 CD34
323 CD3
d 10–11
(3–5 d)
Vigorous transfusion.
Pulmonary haemorrhage
TMA. Recovered.

Efficacy of red cell exchange

A 10-unit red cell exchange was performed on an investigational basis prior to transplant in patient 2, who weighed 90 kg (Table II). More than 40% of the recipient's original red cell mass persisted in circulation following the exchange and typing with anti-B sera demonstrated significant residual recipient cells (3+ mixed field agglutination) for three weeks post transplant. Red cell exchanges were not performed for the other patients in this series


Immune haemolysis as a result of red cell isohaemagglutinins produced by graft-associated passenger lymphocytes occurs in both solid organ and bone marrow transplants (Starzl et al, 1964; Klumpp, 1991; Petz, 1998). The use of cyclosporin alone as anti-GVHD prophylaxis has been associated with an increased risk of haemolysis in this setting (Hows et al, 1986; Ramsey, 1991; Gajewski et al, 1992). The reason for the possible association of haemolysis with cyclosporin is unknown, but may be related to the fact that cyclosporin more potently inhibits T- rather than B-cell function (Shevach, 1985), retards the induction of tolerance against foreign antigens (Webster et al, 1987), and is less effective in preventing secondary rather than primary immune responses (Lindsey et al, 1980). The absence of an agent cytotoxic to B cells, such as azathiaprine or methotrexate, may further imbalance suppression of T- vs. B-cell activity (Hows et al, 1986; Gajewski et al, 1992).

In solid organ transplantation, the incidence and severity of haemolysis owing to minor ABO incompatibility are directly related to the lymphoid content of the grafted organ (Ramsey, 1991). Recent reports in haematopoietic transplantation also indicate that the severity of haemolysis may be dramatically increased when peripheral blood rather than bone marrow is used as the stem cell source. PBSCs contain a 16-fold increase in CD3+ T lymphocytes and an 11-fold increase in CD19+ B lymphocytes compared with conventional marrow harvests (Korbling et al, 1995). PBSC grafts in this series and other reports also contained high doses of donor stem cells and lymphocytes (Tables I and II). We observed no relationship between the stem cell dose or lymphocyte content and the occurrence of haemolysis in this series, however, further analysis of lymphocyte and dendritic cell subsets in PBSC grafts might be of interest in future studies. Lymphocytes in PBSC grafts may also be specifically primed towards antibody production owing to a switch from a type 1 to a type 2 helper response induced by the administration of granulocyte colony-stimulating factor (Pan et al, 1995).

This report documents three additional cases of immune haemolysis associated with minor ABO-incompatible PBSC infusions, including two that received low intensity non-myeloablative conditioning (Table I). Serological monitoring revealed the development of a positive DAT a few days before or at the time of haemolysis in all cases. However, a positive DAT, including strong complement (C3d) coating of red cells, was also detected in some patients who did not develop haemolysis. Serum isohaemagglutinins and red cell eluates with specificity for the relevant ABO antigen were detected in all cases with haemolysis, but were also occasionally observed in patients who did not experience haemolysis. The standard blood banking assays utilized in this study did not distinguish between IgG and IgM, however, prior studies have not revealed a relationship between haemolysis and either the titre or isotype of donor isohaemagglutinins (Hows et al, 1986). Thus, serological findings alone have not been useful for predicting the occurrence of severe haemolysis in this setting.

Non-myeloablative regimens are being increasingly utilized in haematopoietic transplantation (Khouri et al, 1998) and have facilitated the performance of transplants in the outpatient setting and in subjects without malignant disease, owing to the reduced acute toxicity of such regimens (Childs et al, 1999b). Non-ablative transplant protocols in our facility utilized cyclosporin as a sole agent for GVHD prophylaxis and deliberately avoided methotrexate, mycophenylate or other anti-proliferative agents, and also avoided lymphocyte depletion of the graft, with the goal of enhancing immune-mediated donor anti-tumour effects (Childs et al, 1999a). Non-ablative transplants produce mixed donor/recipient haematopoietic chimaerism and the relative engraftment kinetics of myeloid, T-lymphocyte and B-lymphocyte lineages are highly variable compared with ablative transplants (Childs et al, 1999b). Chimaerism analysis is not routinely performed for B lymphocytes during the period of haematopoietic recovery, nor for myeloid or T-cell lineages during the very early stages of engraftment when haemolysis is observed. The intriguing early donor B-cell chimaerism detected in patient 1 suggests that this assay might prove useful in predicting which patients are at greater risk of haemolysis.

A cumulative summary of all adequately reported cases of severe haemolysis as a result of minor ABO-incompatible PBSC transplants, including the three patients from the present series and five prior cases (Toren et al, 1996; Bornhauser et al, 1997; Laurencet et al, 1997; Oziel-Taieb et al, 1997; Salmon et al, 1999), is shown in Table II. This listing includes the only previous fatal case (Oziel-Taieb et al, 1997), as well as the only case described in a patient given post-transplant methotrexate (Bornhauser et al, 1997). Haemolysis began 5–12 d after PBSC infusion and at least 1 week before the onset of GVHD in patients in whom this complication occurred. Three out of eight cases involved phenotypically matched donors who were not HLA-identical siblings. Seven of the eight cases in which information was available regarding donor sex involved female donors (including all three in this series), and six of these seven cases involved male recipients (Toren et al, 1996; Laurencet et al, 1997; Moog et al, 1997; Oziel-Taieb et al, 1997; Salmon et al, 1999). In an earlier study, recipients of bone marrow from female donors did not have an increased risk of haemolysis owing to minor ABO incompatibility (Hows et al, 1986). However, transplant recipients of bone marrow from female donors have experienced an increased risk of acute GVHD following marrow transplantation, particularly when the donors have had a prior pregnancy or blood transfusions (Atkinson et al, 1986; Gale et al, 1987). We did not observe a relationship between PBSC donor parity and haemolysis in this series (Table I), while information on donor parity and prior transfusions was not available in the majority of other reports (Table II).

It is unclear whether the increased incidence of severe haemolysis in patients whose blood group was type A is related to factors intrinsic to the red cell, such as the increased complement-binding activity of anti-A-sensitized cells (Bakacs et al, 1993) or is simply because group A is approximately four times more common than group B in transplant recipients. The possible role of cryopreservation in influencing the development of post-transplant haemolysis is also unknown. Cryopreservation does not appear to adversely affect the pace or durability of neutrophil and platelet recovery, but may disrupt the integrity of dendritic cell membranes that function in antigen presentation (Taylor et al, 1990). Data on cryopreservation was not included in most of the published reports in Table II.

In some centres, prophylactic red cell exchanges have been used to reduce the number of circulating recipient red cells prior to transplant. In one series, prophylactic 8-unit exchanges performed in four consecutive recipients of minor ABO-incompatible unrelated-donor marrow did not prevent haemolysis or decrease the incidence of acute renal failure secondary to massive immune red cell destruction (Gajewski et al, 1992). We confirmed this finding by documenting that 40% of the circulating red cells were still of recipient type following a 10-unit red cell exchange. Because as little as 40–100 ml of ABO-incompatible blood can result in acute, severe transfusion reactions (Mollison et al, 1997), it may be technically impossible to reduce residual circulating recipient red cells to clinically insignificant levels. It is possible that a red cell exchange might provide increased numbers of donor-compatible red cells that would be less susceptible to haemolysis, however, a policy to prophylactically perform red cell exchanges in all subjects at risk seems unwarranted. In contrast, vigorous transfusion of donor-compatible red cells at the earliest signs of haemolysis will also reduce the fraction of recipient red cells in the circulation, as observed in patients 3 and 9, and in prior cases (Greeno et al, 1996).

The results of this study indicate that haemolysis associated with the use of minor ABO-incompatible PBSC transplants can be brisk and potentially life-threatening following either myeloablative or non-myeloablative recipient onditioning. Use of cyclosporin alone in the absence of an anti-proliferative agent for GVHD prophylaxis, utilization of non-genotypically HLA-matched donors and, possibly, use of a female donor, appear to be associated with an increased risk of this event. Immune haemolysis in the post-transplant setting may be difficult to diagnose and be confused with other haemolytic conditions such as thrombotic thrombocytopenic purpura. Red cell serological assays, examination of the peripheral blood smear and assessment of renal function may be useful to clarify the diagnosis. Our data suggest that the aetiology of the haemolysis is multifactorial, and that serological and clinical factors alone are insufficient to reliably predict or exclude patients at risk of haemolysis. We do not recommend that conditioning or anti-GVHD regimens be altered to reduce the incidence of haemolysis because these regimens heavily influence engraftment, GVHD, tumour eradication and relapse prevention, thus being the prime determinants of overall transplant outcome. Rather, we recommend increased awareness and vigilant monitoring during the first 2 weeks post transplant, to include (i) daily LDH, (ii) daily direct antiglobulin tests and twice daily complete blood counts on d 5–12, (iii) transfusions with donor-compatible red cells to keep the Hb above 9·5 g/dl, and (iv) red cell eluates and isohaemagglutinin screens as indicated by the DAT and complete blood count. Clinicians involved in the care of patients receiving haematopoietic transplantation should be aware of the potential for severe, catastrophic haemolysis with the use of PBSC-vs.-marrow grafts in the setting of minor ABO incompatibility. Appropriate transfusion practices with donor-compatible red cells and recipient-type plasma and platelets, with prompt hydration and aggressive red cell transfusions at the earliest signs of haemolysis, should be practiced in all cases.


The authors gratefully acknowledge the expert serological guidance of David Stroncek, the cell processing expertise of E. J. Read and Charles Carter, the clinical contributions of Michael Bishop and Claude Kasten-Sportes, and the support of the staff in the Transfusion Services Laboratory, Department of Transfusion Medicine and the Bone Marrow Transplant Laboratory, Hematology Branch, National Heart Lung and Blood Institute.


  1. The views expressed in this paper are those of the authors and do not represent the official position of the United States Department of Health and Human Services, the United States Public Health Service, the Department of Defense or the United States Army.