Donor-derived myelodysplastic syndrome/acute leukaemia (DD-MDS/AL) is a rare life-threatening complication of allogeneic haematopoietic stem cell (HSC) transplantation. However, it is unknown whether the risk differs by HSC source. Therefore, we evaluated the incidence of DD-MDS/AL in 2390 engrafted patients. With a median follow-up of 7·1 years (1–20·8), the incidence of DD-MDS/AL was 0·53% (95% confidence interval (CI), 0·01–1·41%], 0·56% (95%CI, 0·01–1·36%) and 0·56% (95%CI, 0·01–1·10%) in recipients of bone marrow (n = 1117), peripheral blood (n = 489) and umbilical cord blood (UCB, n = 784), respectively. While follow-up is shorter in recipients of UCB and peripheral blood, incidence of DD-MDS/AL is, thus far, similar between HSC sources.
Donor-derived myelodysplastic syndrome and acute leukaemia (DD-MDS/AL) are rare adverse events after allogeneic haematopoietic stem cell (HSC) transplantation. Originally identified in 1971 (Fialkow et al, 1971), nearly 90 cases have now been reported in the literature (Hertenstein et al, 2005; Ruiz-Argüelles et al, 2007; Wang et al, 2011; Wiseman, 2011; Fraser et al, 2005; Matsunaga et al, 2005; Sevilla et al, 2006; Yamazaki et al, 2011; Gustafsson et al, 2011; Rodriquez-Macias et al, 2013; Nakamizo et al, 2011). Based on several reviews, the incidence of DD-MDS/AL has been estimated to be between 0·12% and 5% (Hertenstein et al, 2005; Ruiz-Argüelles et al, 2007; Wang et al, 2011; Wiseman, 2011). While speculative in most cases, proposed mechanisms for the development of DD-MDS/AL include transplantation of donor haematopoietic progenitors with a cryptic translocation or a preleukaemic clone, presence of an underlying stromal defect either inherent in the host or acquired by preceding chemoradiotherapy, transformation of donor cells by viral or host antigenic stimulation, impaired immune surveillance, oncogene transfection by fusion of donor cells with residual leukaemic cells, prolonged use of granulocyte-colony stimulating factor (G-CSF, NEUPOGEN®; Amgen, Thousand Oaks, CA, USA) and replicative stress during the period of marked HSC expansion after transplantation, or the presence of occult leukaemia in the donor (Hertenstein et al, 2005; Ruiz-Argüelles et al, 2007; Wang et al, 2011; Wiseman, 2011).
Since the first documented cases in 2005, reports of DD-MDS/DD-AL after umbilical cord blood (UCB) transplantation have been more frequent than with other HSC sources (Fraser et al, 2005; Matsunaga et al, 2005; Sevilla et al, 2006; Yamazaki et al, 2011; Gustafsson et al, 2011; Rodriquez-Macias et al, 2013; Nakamizo et al, 2011), causing some investigators to speculate that the incidence of DD-MDS/AL may be higher after UCB relative to bone marrow or mobilized peripheral blood transplantation (Greaves, 2006; Greaves et al, 2011). Greaves et al (2011) raised the important concern that UCB might carry an increased risk of DD-MDS/AL due to the presence of these pre-leukaemic clones previously shown to exist in a proportion of UCB samples (Greaves, 2006; Greaves et al, 2011). Because of the proliferative stress that invariably occurs after transplantation, it was hypothesized that a ‘second hit’ could enhance the risk of DD-MDS/DD-AL in cells already predisposed to leukaemogenesis.
Therefore, we sought to determine the relative incidence of DD-MDS/AL between the three HSC sources in patients transplanted at the University of Minnesota after 1992, when molecular methods for evaluating chimerism were routine. In addition, we describe the characteristics and outcomes in eight patients with proven DD-MDS/AL.
A total of 2582 consecutive recipients of allogeneic HSC transplanted between January 1992 and December 2010 were identified (time points after 1 January, 2011 were excluded due to insufficient follow up). Patients with graft failure (n = 106) or who died prior to engraftment (n = 86) were excluded. For the 2390 evaluable patients, the donor was related (total n = 996; 502 marrow, 461 peripheral blood, 23 UCB and 10 mixed UCB and marrow) or unrelated (total n = 1394; 615 marrow, 19 peripheral blood and 760 UCB). The median follow up was 9·9 years [95% confidence interval (CI), 9·0–10·3], 6·3 years (95%CI, 5·9–7·3) and 5·5 years (95%CI, 5·1–6·0) for marrow, peripheral blood and UCB recipients, respectively (P < 0·01). Recipients of allogeneic HSC had marrow and/or peripheral blood examined for chimerism routinely at 1, 3, 6, 12 and 24 months after transplant and at the time of relapse when possible, and always when there was a change in the morphological and cytogenetic characteristics or phenotypic profile from the original diagnostic specimen. Chimerism was assessed by molecular methods as previously described (Fraser et al, 2005), using quantitative polymerase chain reaction (PCR) of informative polymorphic variable number tandem repeat or short tandem repeat regions in the recipient and donor on whole marrow or peripheral blood.
Cases of DD-MDS/AL were confirmed by chimerism analysis. Charts of all identified cases were reviewed for details of natural history and treatment response. Cumulative incidence curves were created, treating non DD-MDS/AL death as a competing risk. Median follow-up by reverse censoring was 7·1 years (95%CI, 6·5–7·6). Statistics were performed using sas 9.2 (SAS Institute Inc., Cary, ND, USA) with the dataset closed on 10 May, 2013. P-values <0·05 were considered statistically significant.
In this cohort, 73% of patients had a malignant disease, 61% were male, 58% were aged ≥18 years, 75% received a myeloablative conditioning and 56% were transplanted between 2001 and 2010. Importantly, the median follow up for recipients of bone marrow was longer than that for recipients of peripheral blood and UCB [8·6 years (range, 1–20·8), 5·9 years (range, 1–16·2) and 5·2 years (range, 1–19·2), respectively (Kruskal–Wallis test, P < 0·01)]. Eight cases of DD-MDS/AL were diagnosed between 1 month and 15 years (median 30 months) after transplant. Characteristics of the eight patients with DD-MDS/AL are detailed in Table 1. Four patients had cytomegalovirus reactivation that required antiviral therapy during their early post-transplant course and four had poor haematopoietic recovery, defined as the need for haematopoietic growth factor therapy after day-100, which was associated with multiple viral reactivations in three. Five had monosomy 7, including all four recipients of UCB. Interestingly, one patient in this series with DD-MDS had sustained complete remission with loss of the cytogenetic clone after haematopoietic growth factor discontinuation. The patient with DD-acute lymphocyte leukaemia was treated with chemotherapy after relapse according to the Children's Oncology Group protocol 1961 and remains disease-free 43 months later. Two with DD-acute myeloid leukaemia were treated with high dose cytosine arabinoside-based regimens and one patient went on to a second HSCT. Two with DD-MDS were treated with a second HSCT, while one was given decitabine as a single agent. Only two of the eight patients survive; the remaining six have died secondary to DD-MDS/AL.
Table 1. Donor-derived myelodysplastic syndrome and acute leukaemia case characteristics.
At 15 years, the incidence of DD-MDS/AL was 0·53% (95% CI, 0·01–1·41%) in recipients of marrow, 0·56% (95%CI, 0·01–1·36%) in recipients of mobilized peripheral blood transplant and 0·56% (95%CI, 0·01–1·10%) in recipients of UCB (Fig. 1) and 0·40% (95%CI, 0·01–0·86%) and 0·31% (95%CI, 0·01–0·61%) for those with related and unrelated donors, respectively (P = 0·19).
This analysis illustrates the rarity of DD-MDS/AL regardless of HSC source. While the results fail to discern a statistical difference in DD-MDS/AL at 15 years after transplant between the various HSC sources, it could be argued that recipients of peripheral blood and UCB might have a higher earlier incidence and potentially a higher overall occurrence if longer length of follow up was similar for all groups. Still, the median follow up was >5 years in all groups, which is probably the highest risk period for the majority of patients based on this and other reports (Hertenstein et al, 2005; Ruiz-Argüelles et al, 2007; Wang et al, 2011; Wiseman, 2011; Fraser et al, 2005; Matsunaga et al, 2005; Sevilla et al, 2006; Yamazaki et al, 2011; Gustafsson et al, 2011; Rodriquez-Macias et al, 2013; Nakamizo et al, 2011). Even if the follow-up period is restricted to an earlier time point, the incidence of DD-MDS/AL after bone marrow transplant [0·10% (95%CI, 0–0·30%)] is not sufficiently different from that observed in recipients of peripheral blood or UCB transplant [0·56% (95%CI, 0–1·36%) and 0·56% (95%CI, 0–1·10%)] to reach statistical significance (P = 0·16). Importantly, abnormalities in chromosome 7 appear to be a common cytogenetic finding and some patients are curable simply by haematopoietic growth factor withdrawal, consistent with observations reported by others (Socié et al, 1999). In conclusion, these results provide a measure of reassurance that the incidence of DD-MDS/AL is low (1% or less) regardless of HSC source. However, longer follow up, particularly in recipients of UCB and peripheral blood, may be needed before making a definitive conclusion on the relative risks of DD-MDS/AL between the three HSC sources.
Supported by a Public Health Service grant P01-CA6549376518 (TED, CBG, JEWJ) from the National Cancer Institute, and the Children's Cancer Research Fund (ACD, JEWJ).
ACD, TED, CGB and JEWJ contributed equally to study design, interpretation of data and drafting the manuscript. TED performed the statistical analysis. ACD, TED, JEWJ prepared the data file. ACD, TED, CGB, and JEWJ contributed to interpretation of data and critically reviewed the manuscript. All authors approved the final manuscript.