Background
Allogeneic hematopoietic stem cell transplantation (SCT) is a potentially curative approach in patients with acute myeloid leukemia (AML). Substantial improvement has been achieved in the last decades in SCT outcomes owing to improved supportive care and transplantation techniques and a larger proportion of SCT recipients are becoming long-term survivors [
1].
Reduced-intensity conditioning (RIC) has been widely introduced over the past 15 years to allow SCT in elderly and medically infirm patients not eligible for standard myeloablative conditioning (MAC) [
2]. Several studies have shown similar survival of AML patients after SCT with RIC or MAC [
3‐
9]. Most of these studies have shown that RIC is associated with reduced non-relapse mortality (NRM) but increased relapse rate, resulting in a similar leukemia-free survival (LFS) as MAC. However, due to the more recent introduction of RIC, there is paucity of data on the long-term outcome (beyond 10 years) after RIC.
Most deaths after SCT occur within the first 2 years [
10]. Long-term survivors remain at increased risk for late complications and late morbidity and mortality that is higher than their sibling donors or the age- and gender-matched general population [
11‐
13]. In the largest study of long-term survivors, the Center of International Blood and Marrow Transplantation Research (CIBMTR) has shown that the probability of patients who were alive and disease-free at 2 years after SCT to remain alive 10 years after SCT was 85 % (84 % among patients with AML) [
13]. Relapse was the most common cause of late death, but chronic graft-versus-host disease (GVHD), infections, organ toxicity and second cancers were also important causes of late mortality. These observations were limited to MAC recipients, and there is, similarly, paucity of data on the kinetics of late events after RIC and the expected outcomes of 2-year survivors after RIC.
In this study, we show that 10-year survival is similar after RIC and MAC, and that 2-year survivors after RIC can expect a similarly favorable outcome as 2-year survivors after MAC.
Discussion
The current study shows that with long-term follow-up LFS is similar after allogeneic SCT from HLA- matched siblings with RIC and MAC in patients with AML age >50 years. The role of dose intensity in SCT conditioning for AML has been explored in multiple retrospective studies [
18] (reviewed in 18). Most studies have shown that more intensive regimens control leukemia better, but LFS is not improved due to excess NRM. In a prior report the ALWP of EBMT has shown in a comparison of 315 RIC and 407 MAC recipients, age >50 years, that NRM was lower with RIC, relapse was higher, resulting in similar 2-year LFS [
3]. The current analysis includes the same group, extended with data accumulating from more patients, transplanted during the same period, now followed for almost 10 years. It shows that these shorter-term observations remained with long-term follow-up. The 10-year LFS was 31 % (95 % CI, 27–35) and 32 % (95 % CI, 28–35) after MAC and RIC, respectively (
P = 0.57). In addition, the GRFS, which is a surrogate for quality of life analysis, was similar between the two regimens. Several retrospective analyses and meta-analyses supported these observations [
4‐
9]. Luger et al. reported in the largest such comparison from CIBMTR, including 3731 MAC and 1448 RIC/nonmyeloablative (NMA) recipients, that the 5-year OS rates were 34, 33, and 26 % after MAC, RIC, and NMA conditioning, respectively [
5]. OS was similar after RIC and MAC but inferior after NMA. However, in this study NRM was lower after RIC only in the early post SCT period. By 3 years, late NRM negated this early advantage and NRM rates became equivalent. In the current analysis, NRM rate was lower after RIC throughout the post-transplant course up to 10 years after SCT. This is possibly explained by the selection of patient age ≥50 years in this analysis, compared to all adult patients in the CIBMTR study. Thus, the median age of MAC recipients was 54 and 42 years in the different studies respectively, while the median age of RIC recipients was similar. Older MAC recipients may be more prone to NRM in the late post SCT period than younger recipients. Advanced age is a predictor of NRM in many of these studies. Historically, only younger patients (<35–40) benefited from SCT with MAC in CR1 compared with chemotherapy, due to excess NRM [
19]. However, RIC has extended the benefit to older patients [
20,
21]. In the current analysis, RIC was associated with better long-term LFS than MAC in patients age >55 years.
However, all these analyses may be associated with a selection bias. Several randomized comparisons have been reported over the last years. Bornhauser et al. randomized patients with AML in CR1 to standard MAC (with 12Gy TBI) or RIC with an intermediate dose of TBI (total 8Gy) [
22]. The 3-year LFS was similar among the regimens. The GITTMO group randomized patients to BuCy versus fludarabine and high-dose busulfan (FB4) [
23]. NRM was reduced with the FB4 regimen but LFS was similar. It should be noted that the RIC arms in both these studies would be considered MAC according to the registry criteria used in the current analysis. These regimens are better defined as reduced-toxicity myeloablative regimens (RTC). Scott et al. randomized patients to conventional RIC versus MAC [
24]. The study was stopped early as relapse rates were markedly higher in the RIC group. The reduction in NRM was not sufficient to compensate for this elevated risk and LFS was higher after MAC. The conclusion from these randomized studies is that MAC is still the standard regimen for younger patients. RIC can be a suitable alternative in patients who are older or those not eligible for MAC. The new RTC regimens may prove to be as effective but safer regimens that may ultimately replace MAC and may even be acceptable in MAC-ineligible patients.
Patients seek consultation regarding their prognosis not only prior to SCT but also as time elapses afterwards. Many of the clinical factors predictive of LFS in the early post-transplant period are no longer predictive later on as the risk for early events declines. Most events after MAC occur within the first 2 years [
10]. In the largest study of long-term survival including 10,632 patients reported to the CIBMTR as having been alive and disease-free at the 2 year time-point, the probability of remaining alive at the 10-year time-point was 85 % [
12]. Older age and chronic GVHD were the main risk factors in the entire population, while advanced disease at SCT was an additional risk factor in patients with leukemia. Relapse and NRM occurred in 10 and 9 % of AML patients surviving alive and disease-free 2 years after SCT. The CIBMTR has also designed an online calculator for estimation of subsequent LFS [
25]. However, these data apply only to MAC, and data regarding the kinetics of post RIC events are scarce. The current analysis shows that most events after RIC also occur in the first 2 years. The 10-year OS of patients alive and disease-free 2 years after SCT was 73 and 74 % after MAC and RIC, respectively, with advanced disease at SCT been the major prognostic factor. These rates are mildly lower than those reported in the CIBMTR study; however, in that study, the median age of AML patients was 28, with only 6 % over age 50 years, while all patients included in this analysis are over 50 years.
These data can serve to reassure patients given RIC at the 2 year time-point that their subsequent survival is favorable and not significantly different than among those given MAC. However, the causes of subsequent deaths are somewhat different between RIC and MAC. While relapse is the major cause of late death in both, it is a more prominent cause of death after RIC. Chronic GVHD and second cancers are more prominent causes of late death after MAC.
Second cancers are the cause of 5–10 % of late mortality after MAC [
11]. Data of the incidence of second cancers after RIC are limited as extended follow-up is required. In a single center report from the Tel HaShomer group, the 10-year incidence of second cancers was 1.7 % after MAC, 7.4 % after RIC and 5.7 % after fludarabine-based RTC regimens [
26]. After adjusting for patient characteristics, it was shown that the incidence of second cancers is not reduced in the RIC/RTC era. A larger CIBMTR study found that the overall risk of second cancers is reduced after NMA/RIC although there was an increase of cancers of specific sites such as head and neck. Among patients aged 40–60 years with MDS and AML, there was no difference between RIC/NMA and MAC [
27]. In the current report, death due to second cancers was more frequent after MAC. The Tel HaShomer study speculated that fludarabine may have an important role in the pathogenesis of second cancers. Fludarabine-based RTC regimens were included with MAC in the CIBMTR and current EBMT reports. The current analysis is only on deaths and among older patients which may also explain part of these differences. The surveillance for second cancers remains an important task in long-term patient education and follow-up [
28].
This study has several limitations. This is a retrospective analysis that compared two not well-matched cohorts. However, the retrospective design of this study was the only way to try to answer the question regarding outcome of patients receiving either RIC or MAC in clinical practice at the present time. A randomized study of this size with this long-term follow-up is unlikely to be performed. While there are differences between the two groups, especially regarding age at the time of transplantation, the overlap is large enough for a comparison and allows adjusting for these differences. Other factors such as source of stem cells and use of in vivo T cell depletion are in part inherent to the conditioning regimen used and they reflect clinical practice. The objective was to assess the outcome of the strategies as they have been used until now including such factors. We focused on match-sibling donors and therefore the conclusions cannot be extended to other settings such as SCT from matched unrelated donors or alternative donors, which are increasingly been used.
Acknowledgements
The authors would like to thank all EBMT centers for contributing patients to the study and data managers for their great work. A complete list of the members of the European Blood and Marrow Transplantation Group appears in Additional file
1.