Analysis of chimaerism in thalassaemic children undergoing stem cell transplantation


Professor I. A. G. Roberts, Department of Haematology, Imperial College School of Medicine, Ducane Rd, London W12 0NN, UK. E-mail:


We have prospectively assessed the relative contribution of host and donor to haemopoiesis following stem cell transplantation (SCT) in children with β-thalassaemia major (n = 35), using karyotype analysis or Southern blot/polymerase chain reaction analysis of variable number tandem repeats on genomic DNA from peripheral blood. Early haemopoiesis was fully donor in origin in 24 out of 35 cases and remained so throughout the post-transplant course in all but one patient, who evolved to stable mixed chimaerism. The remaining 11 cases (31%) initially showed mixed chimaerism: four of these rejected, one eventually eradicated host haemopoiesis to become fully donor haemopoietic, and the remaining six had persistent mixed chimaerism, with 5–38% host haemopoiesis. The risk of graft rejection was high when > 15% host haemopoiesis was present at 3 months post transplant: four out of six such patients rejected their grafts; conversely, zero out of 29 patients with < 15% host haemopoiesis at 3 months rejected (P < 0·0001). There was a higher incidence of significant acute and chronic graft-versus-host disease in patients with full donor chimaerism. These studies confirm that the mixed chimaeric state is common following SCT for thalassaemia, often persists (with up to 4 years follow-up) and is compatible with long-term cure. Analysis of chimaerism in patients undergoing SCT for β-thalassaemia enables monitoring of engraftment in the early post-transplant period, provides insight into the biology of engraftment and may be useful in identifying patients at high risk of rejection.

Haemopoietic stem cell transplantation (SCT) from an HLA-identical sibling is curative for the majority of children with β-thalassaemia. With good-risk patients (Pesaro class 1), 3 year event-free survivals (EFS) of up to 94% have been reported (Lucarelli et al, 1990; Galimberti et al, 1997). Unfortunately, although still an important treatment option, it is significantly less successful in those with acquired organ damage (3 year EFS 77% for Pesaro class 2 and 53% for class 3), owing both to an increased incidence of autologous reconstitution and treatment-related mortality (Galimberti et al, 1997).

These considerations have led to the suggestion that SCT with non-myeloablative conditioning regimens may be a useful therapeutic approach in thalassaemic patients with acquired organ damage. Using such regimens, reliable myeloid and lymphoid engraftment can be achieved with apparently reduced short-term toxicity, even in patients with severe organ toxicity (Giralt et al, 1997; Slavin et al, 1998; Amrolia et al, 2000; Kottaridis et al, 2000). Given the reduced toxicity of SCT with non-myeloablative conditioning, it seems an attractive approach for children with non-malignant disorders, in whom myeloablation per se plays a limited role in disease eradication and a mixed chimaeric state may be sufficient to correct the disease phenotype.

Previous studies have demonstrated that mixed chimaerism is not uncommon after SCT for thalassaemia with conventional conditioning regimens and is compatible with long-term transfusion-independence (Nesci et al, 1992; Kapelushnik et al, 1995; Andreani et al, 1996). There is also evidence in thalassaemic patients that the rate of mixed chimaerism may vary with the conditioning regimen used (Nesci et al, 1992), suggesting that mixed chimaerism might be even more common following SCT with non-myeloablative conditioning regimens. However, these studies have either been small or carried out on highly selected patient populations. As a first step in understanding the incidence and significance of mixed chimaerism following SCT for β-thalassaemia with standard conditioning regimens, we prospectively assessed the relative contribution of host and donor to haemopoiesis following SCT in our cohort of patients.

Between September 1993 and May 2000, 46 consecutive children with β-thalassaemia major underwent SCT from an HLA-identical sibling/parent at our three institutions. With a median follow-up of 53 months, 41 patients (89%) are currently asymptomatic from their original disease and transfusion-independent. We prospectively assessed the relative contribution of host and donor to haemopoiesis following transplantation in 35 transplants (33 patients) using karyotype or molecular analysis of variable number tandem repeat sequences (VNTRs) in order to determine the frequency of the mixed chimaeric state in our patient population, whether early residual host haemopoiesis was predictive of subsequent rejection and whether chimaerism status was associated with cell dose, ABO-mismatch between host and donor, the thalassaemic status of the donor and graft-versus-host disease (GvHD).

Patients and methods

Patient characteristics Forty-six consecutive patients with thalassaemia major (43 β-thalassaemia major and three with Eβ thalassaemia) underwent a total of 49 stem cell transplant procedures at our three institutions between September 1993 and May 2000. The donor was an HLA-identical sibling in 46 transplants and a fully HLA-matched parent in three transplants. The source of stem cells was the bone marrow in 48 transplants and cord blood in one transplant. Five patients underwent a second transplant procedure for rejection/autologous reconstitution: in three cases both transplants were in this period and are included in our analysis. The median age of our patients at the time of transplantation was 5·7 years (range 2–17 years). According to the Pesaro grading system (Lucarelli et al, 1990), 13 patients were class 1, 24 were class 2 and nine patients were class 3.

Conditioning and GvHD prophylaxis Thirty-six patients received the ‘standard’ conditioning regimen consisting of busulphan 14 mg/kg over 4 d (d −9 to d −6) and cyclophosphamide 200 mg/kg over 4 d (d −5 to d −2). One patient required two additional doses of busulphan on the basis of low plasma levels. Two patients with Pesaro class 3 grading received the same regimen but with a reduced dose of cyclophosphamide of 120 mg/kg. Two patients undergoing a second transplant received 16 mg/kg busulphan: one of these had previously failed a cord blood transplant after the standard conditioning regimen. Seven patients transplanted since 1998 received fludarabine 50 mg/m2 in addition to the busulphan 14 mg/kg and cyclophosphamide. One of these seven had previously failed a transplant after standard busulphan/cyclophosphamide conditioning and another received an unconditioned peripheral blood stem cell infusion for incipient rejection. Forty-six transplants utilized serotherapy, with either Campath 1G (5 mg/d, d −4 to d −2, or 5 mg or 10 mg/d, d −14 to d −10) in 30 cases, Campath 1H (5 mg/d, d −4 to d −2, in three patients, and 0·1 mg/kg, d −9 to d −7, in one patient) in four cases or anti-lymphocyte globulin (ALG; 12·5 mg/kg/d for 3–5 d) in 12 cases. Additional GvHD prophylaxis consisted of cyclosporine alone in 24 transplants or cyclosporine plus methotrexate (10 mg/m2 on d +3 and d +6) in 24 transplants.

Analysis of chimaerism Chimaerism data was collected prospectively for 39 of the 49 transplants described above. Chimaerism was not analysed in 10 transplants because VNTR assays were not available at one site until 1998. Thirty-five patients were evaluable for both early and late chimaerism; three patients were not evaluable for late chimaerism because they died and one had insufficient follow-up. Chimaerism analysis was performed using XY-FISH (fluorescence in situ hybridization) on whole blood in some sex-mismatched transplants (n = 8), or using Southern blot (n = 22) or polymerase chain reaction (PCR) (n = 5) analysis of VNTRs in peripheral blood genomic DNA using the YNH, M27β or TBQ probes at 1, 3, 6 and 12 months and then yearly post-bone marrow transplantation (BMT). In order to assess the relative contributions of donor and recipient, these assays were quantified by comparing the percentage of male and female cells for FISH analysis or using densitometry of autoradiographs for VNTR analysis. Assay sensitivities to 5% for both techniques were confirmed by mixing experiments of known amounts of patient and donor cells/DNA. Early chimaerism was defined as status up to 3 months post transplant and late chimaerism as the status beyond 1 year. Mixed chimaerism (MC) was defined as the presence of > 5% host haemopoiesis. Full donor chimaerism (FDC) was defined as < 5% host haemopoiesis. Rejection was defined as > 90% host haemopoiesis and in every case was associated with relapse of thalassaemia and red cell transfusion-dependence. Mononuclear cell (MNC) doses for each transplant were collected and their influence on chimaerism status analysed. Statistical analysis was carried out using Fisher's exact test.


Incidence and patterns of mixed chimaerism post SCT

As shown in Fig 1A, early chimaerism studies (before 3 months) demonstrated fully donor haemopoiesis in 24 out of 35 evaluable transplants (69%). One patient subsequently evolved to stable mixed chimaerism with a low level of host haemopoiesis, the remaining 23 cases (66%) maintained full donor haemopoiesis throughout the post-transplant course. Eleven cases (31%) initially showed mixed chimaerism. The evolution of these cases is illustrated in Fig 1B. Four of these patients rejected their grafts between 2 and 18 months post SCT, with recovery of autologous haemopoiesis. One patient (UPN 7) eventually eradicated host haemopoiesis, at least to the limits of detectability. The remaining six cases had persistent mixed chimaerism, with 5–38% host haemopoiesis at the most recent follow-up (8–46 months post SCT). Currently, 24 out of 35 cases (69%) show full donor chimaerism, 7 out of 35 (20%) show mixed haemopoietic chimaerism and 4 out of 35 (11%) have rejected with autologous reconstitution. Figure 2 shows a representative Southern blot analysis of VNTRs in patients with full donor chimaerism throughout the post-SCT period (A), stable mixed chimaerism (B) and a patient with early mixed chimaerism who eventually rejected his graft (C).

Figure 1.

(A) Incidence of graft rejection and mixed haemopoietic chimaerism at three time points after stem cell transplantation for thalassaemia major. FDC, full donor chimaerism (precise definition in text); MC, mixed (donor–recipient) chimaerism. (B) Evolution of mixed haemopoietic chimaerism in 11 patients.

Figure 2.

Representative Southern blot analysis of variable number tandem repeats in three patients post SCT. (A) Full donor chimaerism. (B) Stable mixed chimaerism. (C) Early mixed chimaerism progressing to graft rejection. The first lanes in each panel show the VNTR pattern of the donor (D) or host (H) pre-SCT, the remaining lanes show the pattern at varying time points, shown in months, post transplant.

Influence of mononuclear cell dose, ABO mismatch and donor haemoglobinopathy status on chimaerism

In the evaluable patients, there was no significant difference in the median MNC dose of the graft between patients who subsequently became full donor chimaeras (4·6 × 108/kg) and those who developed mixed chimaerism (3·5 × 108/kg) (P = 0·67). Twenty-one transplants were ABO blood group matched, 14 were mismatched. There was no significant difference between the incidence of early mixed chimaerism in transplants that were ABO matched (7 out of 21 or 33%) compared with those that were ABO mismatched (4 out of 14 or 29%). In 21 out of 35 evaluable transplants the donor had β-thalassaemia trait. There was also no difference in the frequency of mixed chimaerism in the children transplanted from a haematologically normal donor (5 out of 14 or 36%) compared with those transplanted from donors with the β-thalassaemia trait (6 out of 21 or 29%). Although only one out seven patients transplanted using fludarabine as additional conditioning developed mixed chimaerism, formal assessment of the influence of conditioning regimens on the incidence of chimaerism post transplant will require greater patient numbers.

Mixed chimaerism and rejection

Four of the 35 (11%) transplants were rejected at 2 (UPN 17), 5 (UPN 23), 6 (UPN 33) and 18 months (UPN 4) post SCT. In each case, reconstitution of host haemopoiesis was observed. The earliest rejection/autologous reconstitution occurred in the patient who underwent cord blood SCT (UPN 17). In all cases, rejection was preceded by early mixed chimaerism. The risk of rejection was significantly higher when > 15% host haemopoiesis was present within the first 3 months post SCT: four out of six such patients have rejected their grafts. Conversely zero out of 29 patients with < 15% host haemopoiesis during the first 3 months post SCT have rejected (P < 0·0001).

Chimaerism and GvHD

As shown in Table I, patients with early FDC showed a trend to a higher incidence of acute GvHD ≥ Grade II (38%) than patients who were early MC (9%) (P < 0·09). Similarly, the incidence of chronic GvHD was also higher in late FDC (27%) than late MC (0%), although this difference did not reach statistical significance (P < 0·12) because of the small number of patients with this complication.

Table I.  Chimaerism and graft-versus-host disease.
Acute GvHD
≥ Grade 2
  • *

    P < 0·09.

  • P < 0·12.

FDC9/24 (38%)FDC6/22 (27%)
MC1/11 (9%)*MC0/7 (0%)

Immunosuppression can influence levels of mixed chimaerism post SCT

In the patients with long-term mixed chimaerism, the level of host contribution to haemopoiesis was stable in four out of seven patients, but fluctuated in three patients. During erythropoietin therapy, which raised the haemoglobin level from 7 to 14 g/dl, UPN 9 experienced a marked rise in host chimaerism from a previously stable level of 5% to 15%, which promptly declined to the baseline level on discontinuation of erythropoietin (data not shown). UPN 20, a 16-year-old with Pesaro Class II β-thalassaemia, was transplanted from her HLA-identical non-affected brother, had neutrophil engraftment at d 13 post SCT and was transfusion-independent from d 23. Three months post SCT, at a time when cyclosporine levels were low, her haemoglobin and leucocyte counts dropped and she became transfusion-dependent again, with a reduced reticulocyte count, a negative direct Coomb's test and no evidence of a viral or autoimmune aetiology. VNTR analysis at that time showed 10% host haemopoiesis in the peripheral blood and a bone marrow aspirate showed decreased erythropoiesis and myelopoiesis, with reduced erythroid blast-forming units (BFU-E) [14/105 bone marrow (BM) MNCs; normal 40–200]. She was presumed to have incipient late rejection and was treated with a 1-week course of prednisolone 1 mg/kg with stabilization of her haemoglobin level and improvement in her leucocyte count, in association with a decline of host contribution to haemopoiesis to undetectable levels (Fig 3). When steroids were stopped and erythropoietin therapy was instituted to decrease her transfusion requirement, the haemoglobin level declined and the host contribution increased to 10%. Steroids were therefore reintroduced, with a marked improvement in haemoglobin levels and bone marrow erythropoiesis within 3 weeks, so that she was transfusion-independent again. In association with this, the host contribution to haemopoiesis fell to undetectable levels once again. As shown in Fig 3, her subsequent course continued to show a marked correlation between immunosuppression with steroids, improvement in haemoglobin and fall in host chimaerism until her steroids were weaned slowly at 15 months post transplant. She is currently asymptomatic and transfusion-independent 6 years post SCT and, although she remains a mixed chimaera, her levels of host haemopoiesis have stabilized at 5%.

Figure 3.

Association between mixed chimaerism and immunosuppression for GvHD in patient UPN 20. Hb levels are in g/dl.


Using Y-chromosome FISH and VNTR analyses, we have shown that analysis of chimaerism may be useful to monitor engraftment following SCT for β-thalassaemia, especially in the early post-transplant period, when haemoglobin electrophoresis cannot provide useful information owing to red cell transfusions. We have confirmed that mixed chimaerism is common post SCT for β-thalassaemia and that a low level of residual host haemopoiesis is compatible with long-term transfusion-independence. Thus, eradication of the host haemopoietic system is not mandatory for correction of disease phenotype. This suggests that non-myeloablative conditioning regimens may have a role to play in SCT for thalassaemia, as long as stable, predominantly donor haemopoiesis can reliably be achieved.

The frequencies of mixed chimaerism observed in our patient cohort (31% early, 20% late) are similar to those observed in the study of Nesci et al (1992) in thalassaemics and aplasia (reviewed by Socie et al, 1995), although somewhat higher than those seen after unmanipulated HLA-identical SCT for haematological malignancies (Petz et al, 1987). It is unclear whether the higher rate of mixed chimaerism observed in thalassaemia is as a result of differences in the conditioning regimens used (in particular the use of total body irradiation in leukaemic patients and in vivo T-cell depletion in thalassaemics) or is intrinsic to the thalassaemic state (e.g. as a result of an expanded haemopoietic compartment). In support of the former, particularly with the advent of non-myeloablative regimens, there is an increasing body of evidence suggesting that the conditioning regimen may influence the incidence of mixed chimaerism post SCT (Frassoni et al, 1990; Giralt et al, 1997; Childs et al, 1998; Slavin et al, 1998) and Nesci et al (1992) have confirmed these findings in SCT for thalassaemia. Although T cell-depleted SCT is associated with high rates of mixed chimaerism in patients with chronic myeloid leukaemia (Offit et al, 1990), our use of Campath/ALG serotherapy may not be a critical factor (perhaps because host as well as donor T cells were depleted), as similar rates of mixed chimaerism were seen in thalassaemic patients in the absence of serotherapy (Nesci et al, 1992). Although our incidence of mixed chimaerism is similar to that observed by the Pesaro group (Nesci et al, 1992; Andreani et al, 1996), the absolute levels of host haemopoiesis we have observed in our long-term mixed chimaeras are, in general, much lower than those seen by the latter group. This may reflect differences in pre-BMT preparation (e.g. our use of hypertransfusion in the peritransplant period), our use of Campath and methotrexate as GvHD prophylaxis, ethnic differences in our patient populations and differences in the sensitivities and quantification of the assays used.

When full donor chimaerism was established within 3 months of transplantation, it was maintained with time in all but one patient, suggesting that host haemopoiesis had been permanently eradicated. In contrast, as reported previously (Andreani et al, 1996), chimaeric status was less stable for patients with early mixed chimaerism, with progression to rejection in some patients and eradication of host haemopoiesis in others. This emphasizes the importance of serial studies of chimaerism at least for the first year. In the long-term mixed chimaeras, the level of host contribution was stable in the majority, but fluctuated in three patients. In patient UPN 20, a clear correlation between immunosuppressive therapy with steroids and levels of mixed chimaerism was observed. This has not previously been reported and suggests that steroids were suppressing an alloreactive host response against donor haemopoiesis. While it is often difficult to distinguish an immunologically mediated rejection from autologous reconstitution, in the former situation careful monitoring of chimaerism may be useful in determining the response to anti-rejection therapy. Another long-term mixed chimaera, UPN 9, experienced a significant rise in host chimaerism during erythropoietin therapy; while it is possible that this hormone exerted differential effects on host and donor erythroid cells, it is unclear why it should affect the host:donor contribution of the non-erythroid cells measured in our peripheral blood chimaerism assays.

Similar to the studies of Offit et al (1990) and Ortega et al (1999) in SCT for leukaemia, mixed chimaerism was common and did not predispose to subsequent rejection/autologous reconstitution unless an early, high level host contribution was observed. In our patient cohort, early mixed chimaerism with a host contribution of > 15% at 1–3 months was strongly predictive of subsequent rejection. This may be useful in identifying patients who would benefit from more intensive immunosuppression or donor lymphocyte therapy (Aker et al, 1998). We therefore propose that chimaerism should be monitored monthly for the first 3 months following SCT for β-thalassaemia. If the host contribution is less than 15% at all time points, using methodology similar to that used in this study, and the haemoglobin level is stable, we feel rejection is improbable and chimaerism should be checked again at 1 year post transplant. However, if a host contribution of greater than 15% is observed within the first 3 months, such patients are at high risk of rejection and chimaerism should continue to be monitored monthly. In these patients, the optimal strategy to prevent rejection/autologous reconstitution is unclear. In the setting of a fully conditioned SCT in which immunosuppression has not been optimal, our experience suggests that an initial trial of more intensive immunosuppression (usually by increasing cyclosporine A levels) is often adequate to arrest progression. In contrast, in the setting of non-myeloablative conditioning there is increasing evidence to suggest withdrawal of immunosuppression with or without donor lymphocyte infusion may be effective (Childs et al, 1998). The level of host contribution that was predictive of rejection/autologous reconstitution is lower in our patient population than that determined by Nesci et al (1992); again this may reflect differences in the sensitivities and quantification of the assays used. In both these studies, the levels of host haemopoiesis predictive of rejection/autologous reconstitution were well above the sensitivity of FISH/VNTR analyses, so that assays with higher sensitivity, e.g. PCR analysis for β-globin point mutations (Kapelushnik et al, 1995), may not add further predictive power.

In our study, patients with early full donor chimaerism showed a trend to a higher incidence of significant acute GvHD than patients who were early mixed chimaeras. A similar trend to association was observed between late full donor chimaerism and chronic GvHD. If these results are confirmed in larger cohorts with greater statistical power, this may reflect immunological tolerance between donor-derived and host cells in mixed chimaeras. Previous studies on this issue have given conflicting results (reviewed in Socie et al, 1995). An association between acute GvHD and full donor chimaerism has previously been reported after SCT for aplastic anaemia (Huss et al, 1996) and leukaemia (Bertheas et al, 1991). Nesci et al (1992) did not observe such an association after SCT for thalassaemia; however, in this study the incidence of acute GvHD was much lower than in our cohort. The inter-relationship between chimaerism and GvHD warrants further study, ideally with large numbers of patients and multivariate analysis to exclude confounding variables.

In summary, we have confirmed that the mixed haemopoietic chimaerism is both common and curative following fully conditioned allogeneic BMT, providing a rationale for the use of non-myeloablative conditioning regimens in this disorder, as long as stable, predominantly donor haemopoiesis can reliably be achieved. Analysis of chimaerism in patients undergoing BMT for β-thalassaemia enables early monitoring of engraftment and may be useful in identifying patients at high risk of rejection and perhaps also GvHD.


Campath antibodies were kindly provided by Dr G. Hale, Oxford, UK.