Unrelated Donors are Associated with Improved Relapse Free Survival Compared to Related Donors in Patients with Myelodysplastic Syndrome Undergoing Reduced Intensity Allogeneic Stem-Cell Transplantation

Abstract

Reduced intensity allogeneic stem-cell transplantation (RI alloSCT) is a potentially curative treatment approach for patients with myelodysplastic syndrome (MDS). It is currently unclear if older related donors are better than younger unrelated donors for patients with MDS undergoing RI alloSCT. We retrospectively studied 53 consecutive MDS patients who underwent RI alloSCT between April 2007 and June 2014 and evaluated associations between donor type and outcomes with adjustment for significant covariates. 34 patients (median age: 64 years) and 19 patients (median age: 60 years) received allografts from unrelated and related donors, respectively. Unrelated donors were younger than related donors (median age: 32 vs 60 years, p<0.0001). There were no significant differences in baseline disease characteristics of patients receiving allografts from related or unrelated donors. Patients who received allografts from unrelated donors had a lower relapse risk (adjusted hazard ratio [aHR]=0.35, p=0.012) and improved relapse-free survival (aHR=0.47, p=0.018). HLA mismatched unrelated donors were associated with a higher risk of grade 2–4 acute graft versus host disease (GVHD) (HR=4.64, p=0.002) without an accompanying increase in the risk of non-relapse mortality (p=0.56). Unrelated donors provided a higher mean CD8 cell dose (p=0.014) and were associated with higher median donor T cell chimerism at day 60 (p=0.003) and day 100 (p=0.03). In conclusion, patients with MDS who received allografts from unrelated donors had a lower risk of relapse and improved relapse-free survival when compared to patients who received allografts from related donors. These findings should be confirmed in a prospective study.

BACKGROUND

Reduced intensity allogeneic stem cell transplantation (RI alloSCT) is a potentially curative approach for the treatment of myelodysplastic syndrome (MDS). The median age at diagnosis of MDS is between 68 and 76 years [1–4] and prognosis is related to disease-specific variables such as the International Prognostic Scoring System (IPSS) and World Health Organization (WHO) classification.

The current practice in most hematologic malignancies is to identify an HLA matched sibling donor before considering an unrelated donor. As patients with MDS are generally older, sibling donors tend to be significantly older than unrelated donors.

It is currently unclear if outcomes are different with an older HLA matched sibling donor versus a younger unrelated donor for patients with MDS in the setting of RI alloSCT. Some studies suggest an association between advanced donor age and worse outcomes [5–8] while others report worse outcomes with young unrelated donors.[9] However, most of these studies included patients with a variety of hematologic malignancies, conditioning regimens and graft sources, and some are confounded by the use of T cell depleting antibodies in unrelated donor transplants, limiting their potential to inform clinical practice in donor selection.

We previously showed that certain young unrelated donors might be better than older related donors in RI alloSCT due to differences in graft CD8 T cell content [10]. In this study, we aimed to determine the optimal donor type for patients with MDS undergoing RI alloSCT.

PATIENTS AND METHODS

Patient Cohort and Study Design

We retrospectively studied 53 consecutive MDS patients who underwent a first peripheral blood alloSCT with fludarabine-busulfan conditioning between April 2007 and June 2014 at the University of Pennsylvania. Patients received fludarabine 120mg/m2 intravenously (IV) and busulfan 6.4mg/kg IV, followed by an infusion of peripheral blood stem cells (PBSCs) from either a related or an unrelated donor without T-cell depletion. Participants received standard GVHD prophylaxis with tacrolimus or cyclosporine and IV methotrexate. Some patients (n=8) also received maraviroc on clinical trials of graft versus host disease (GVHD) prophylaxis [11]. All participants received standard antimicrobial prophylaxis and daily granulocyte colony-stimulating factor until neutrophil engraftment.

Baseline clinical information and follow up data were collected through 11/25/2015 via a retrospective review of our electronic medical records and transplant database. This study was approved by the University of Pennsylvania Institutional Review Board and patients provided informed consent for data collection before transplantation.

Response Assessment and Time-to-Event End points

Time to disease relapse, grade 2–4 acute GVHD (aGVHD), grade 3–4 aGVHD, chronic GVHD (cGVHD), non-relapse mortality (NRM), relapse free survival (RFS) and overall survival (OS) were defined as the time from transplantation to the event. For all outcomes, patients were censored at the time of last contact or second transplantation, whichever occurred first. Additionally, for GVHD outcomes, patients were censored at the time of donor lymphocyte infusion. Disease relapse was defined as morphologic or cytogenetic evidence of disease demonstrating pre-transplantation characteristics. Restaging evaluation, including bone marrow biopsies, was routinely performed at day 100 or earlier in patients with signs indicating early relapse. Acute and chronic GVHD were graded according to the Consensus Conference criteria and National Institutes of Health criteria, respectively [12, 13]. Donor T-cell chimerism levels were measured after immunomagnetic positive selection of CD3+ cells from peripheral blood samples (STEMCELL Technologies, Vancouver, BC, Canada).

Statistical Analysis

Correlations between donor type and clinical variables were assessed by Fisher’s exact test for categorical variables. For continuous variables the t-test for means or the Wilcoxon rank-sum test for medians were used as appropriate. Competing risks regression analyses were conducted to determine the impact of donor type on time to relapse, NRM and GVHD outcomes. Death without the event was a competing risk for relapse and GVHD outcomes while relapse was a competing risk for NRM. Cox regression was used to determine the impact of donor type on RFS and OS. Univariable and multivariable analyses were performed to identify other significant independent predictors. Variables that exhibited univariable significance of p<0.20 were considered for multivariable modeling and a step-wise elimination method was then used. Results are expressed as hazard ratios (HR) in univariable analyses and adjusted hazard ratios (aHR) in multivariable models with accompanying 95% confidence intervals (CI). Correlations between graft T cell content and clinical variables were assessed using Pearson’s test. A two-sided p value of less than 0.05 was considered to be statistically significant. Data were analyzed using STATA v14.0 (STATA, College Station, TX).

RESULTS

Patients

Between April 2007 and June 2014, 53 consecutive patients underwent a first alloSCT for MDS with fludarabine-busulfan and a PBSC graft from HLA-matched related donors (n=19), 8/8 allele-matched unrelated donors (n=23) or single allele mismatched unrelated donors (n=11). Patient, disease and transplantation characteristics are summarized in Table 1. The median age of patients at the time of transplant was 63 years (range 50–72). The median time from diagnosis to transplant was 11.7 months (range 4.5–135.8) and was longer for patients receiving allografts from unrelated donors compared to related donors (14.9 months versus 9.2 months). However, this difference did not reach statistical significance (p=0.0504).

TABLE 1.

BASELINE CHARACTERISTICS OF PATIENTS AND DONORS

	Unrelated donor (n=34)	Related donor (n=19)	p value
Median recipient age at transplant – years (range)	64 (50–72)	60 (52–70)	0.17
Median donor age at transplant – years (range)	32 (19–53)	60 (42–72)	<0.0001
Median time from diagnosis to transplant – months (range)	14.9 (5.0–135.8)	9.2 (4.5–98.0)	0.0504
Patient gender			0.56
Male – no. (%)	22 (65)	14 (74)
Female – no. (%)	12 (35)	5 (26)
Donor Gender			1.00
Male – no. (%)	19 (56)	10 (53)
Female – no. (%)	15 (44)	9 (47)
Patient CMV status			0.39
Negative – no. (%)	18 (53)	13 (68)
Positive – no. (%)	16 (47)	6 (32)
Donor CMV status			0.56
Negative – no. (%)	23 (68)	11 (58)
Positive - no. (%)	11 (32)	8 (42)
ABO compatibility			0.57
Compatible – no. (%)	16 (47)	7 (37)
Incompatible – no. (%)	18 (53)	12 (63)
HLA compatibility			0.004
Matched – no. (%)	23 (68)	19 (100)
Mismatched – no. (%)	11 (32)	0
Cytogenetics at diagnosis			0.85
Good1 – no. (%)	9 (26)	6 (32)
Intermediate1 – no. (%)	5 (15)	3 (16)
Poor1 – no. (%)	12 (35)	9 (47)
Unknown – no. (%)	8 (24)	1 (5)
IPSS score at diagnosis			1.00
Low/Int-1 – no. (%)	12 (35)	8 (42)
Int-2/High – no. (%)	15 (44)	9 (47)
Unknown – no. (%)	7 (21)	2 (11)
WHO classification at diagnosis			0.58
RAEB1/2 – no. (%)	14 (41)	10 (53)
Other – no. (%)	18 (53)	9 (47)
Unknown – no. (%)	2 (6)	0
WHO classification at transplant			0.72
RAEB1/2 – no. (%)	6 (18)	4 (21)
Other – no. (%)	26 (76)	13 (68)
Unknown – no. (%)	2 (6)	2 (11)
GVHD prophylaxis			0.76
CSA2 + MTX3 or MMF4	5 (15)	4 (21)
TAC5 + MTX3 or MMF4	23 (68)	13 (68)
TAC5 + MTX3 + MVC6	6 (18)	2 (11)

¹Good cytogenetics: Normal, -Y alone, del (5q) alone or del (20q) alone; Poor cytogenetics: complex karyotype (≥3 abnormalities) or abnormal chromosome 7; Intermediate cytogenetics: all others

²Cyclosporine

³Methotrexate

⁴Mycophenolate Mofetil

⁵Tacrolimus

⁶Maraviroc

There were no differences in recipient age, sex, CMV status, cytogenetics at diagnosis, IPSS score at diagnosis and WHO classification at diagnosis or at transplant between the groups (Table 1). The median duration of follow up for our study cohort was 34.4 months.

Donors

The median donor age at the time of transplantation was 42 years (range 19–72). Unrelated donors were significantly younger than related donors (median age: 32 years versus 60 years, p<0.0001, Table 1). There were no significant differences in sex, CMV status or ABO compatibility between related and unrelated donors. HLA mismatches occurred in approximately one third of unrelated donors and in none of the related donors (p=0.004).

Relapse

The 1 year cumulative incidence of relapse in unrelated donors and related donors was 38% (95% CI: 22–54%) and 58% (95% CI: 32–77%), respectively. The 2 year cumulative incidence of relapse in unrelated donors and related donors was 41% (95% confidence interval [CI]: 24–57%) and 63% (95% CI: 36–81%), respectively. After adjusting for recipient age, patients who received allografts from unrelated donors had a decreased risk of relapse (adjusted hazard ratio [aHR]=0.35, 95% CI [0.15–0.79], p=0.012, Fig 1A). Similarly, patients who received allografts from unrelated donors had improved RFS after adjusting for differences in the time from diagnosis to transplant (aHR=0.47, 95% CI [0.25–0.88], p=0.018, Fig 1B). The 1 year RFS for unrelated and related donors was 41% (95% CI: 28–62%) and 16% (95% CI: 5.6–45%), respectively. The 2 year RFS for unrelated and related donors was 27% (95% CI: 15–46%) and 11% (95% CI: 2.8–39%), respectively. Because 32% of unrelated donors were single allele mismatched unrelated donors, we performed a subset analysis and found that HLA matched unrelated donors were associated with a decreased risk of relapse (HR=0.26, 95% CI [0.10–0.72], p=0.009, Table 2) and improved RFS (HR=0.38, 95% CI [0.19–0.78], p=0.008, Table 2) as compared to HLA matched related donors. However, patients who received allografts from single allele HLA mismatched unrelated donors had a similar risk of relapse (p=0.80) and RFS (p=0.30) as compared to patients who received allografts from matched related donors (Table 2). In summary, unrelated donors were associated with a reduced risk of relapse and improved relapse-free survival. The improved outcomes in unrelated donors appeared to be driven by HLA matched unrelated donors while HLA mismatched unrelated donors and related donors were associated with similarly poor outcomes.

FIGURE 1. Unrelated donors are associated with a decreased risk of relapse and improved relapse free survival (RFS) but not associated with a higher risk of non-relapse mortality (NRM) as compared to related donors. Patients allografted from unrelated and related donors had similar overall survival (OS).

A cumulative incidence plot of relapse (A), Kaplan-Meier plot of RFS (B), cumulative incidence plot of NRM (C) and Kaplan-Meier plot of OS (D) are shown, stratified by donor type. Unrelated donors were associated with a decreased risk of relapse (p=0.012) and improved RFS (p=0.018) but were not associated with an increased risk of NRM (p=0.33) as compared to related donors. Patients allografted from unrelated and related donors had similar OS (p=0.54).

TABLE 2.

IMPACT OF DONOR TYPE ON RELAPSE, SURVIVAL AND GRAFT VERSUS HOST DISEASE OUTCOMES

	Hazard Ratio	95% CI	p value
Relapse
Matched related donor	1
Matched unrelated donor	0.26	0.10–0.72	0.009
Mismatched unrelated donor	0.90	0.39–2.05	0.80
Relapse Free Survival
Matched related donor	1
Matched unrelated donor	0.38	0.19–0.78	0.008
Mismatched unrelated donor	0.66	0.30–1.44	0.30
Non Relapse Mortality
Matched related donor	1
Matched unrelated donor	1.81	0.62–5.29	0.28
Mismatched unrelated donor	0.61	0.12–3.16	0.56
Overall Survival
Matched related donor	1
Matched unrelated donor	0.70	0.33–1.48	0.35
Mismatched unrelated donor	0.98	0.42–2.27	0.96
Grade 2–4 aGVHD
Matched related donor	1
Matched unrelated donor	2.22	0.93–5.31	0.074
Mismatched unrelated donor	4.64	1.74–12.41	0.002
Grade 3–4 aGVHD
Matched related donor	1
Matched unrelated donor	1.53	0.54–4.35	0.43
Mismatched unrelated donor	1.33	0.34–5.26	0.68
Chronic GVHD
Matched related donor	1
Matched unrelated donor	2.61	0.60–11.32	0.20

T Cell Engraftment

We hypothesized that the improvement in RFS and reduction in the risk of relapse observed with unrelated donors was associated with more rapid donor T cell engraftment. Median donor T cell chimerism levels measured at day 60 were 77% (interquartile range [IQR]: 67–88%) and 54% (IQR: 28–58%) for unrelated donors and related donors, respectively (p=0.003). Similarly, median donor T cell chimerism levels measured at day 100 were 80% (IQR: 55–89%) and 60% (IQR: 31–80%) for unrelated and related donors, respectively (p=0.03).

Non-Relapse Mortality and Survival

The 1 year cumulative incidence of NRM in unrelated donors and related donors was 21% (95% CI: 9–36%) and 26% (95% CI: 9–48%), respectively. The 2 year cumulative incidence of NRM in unrelated donors and related donors was 32% (95% CI: 17–48%) and 26% (95% CI: 9–48%), respectively. After adjusting for recipient age, patients receiving allografts from unrelated donors did not have a significantly higher rate of NRM (aHR=1.74, 95% CI [0.58–5.21], p=0.33, Fig 1C) when compared to patients receiving allografts from related donors. The causes of NRM stratified by donor type are summarized in the Online Data Supplement. Similarly, after adjusting for recipient age, there was no difference in OS between patients who had an unrelated donor versus those who had a related donor (aHR=0.81, 95% CI [0.41–1.60], p=0.54, Fig 1D). The 1 year OS for unrelated donors and related donors was 52% (95% CI: 38–72%) and 42% (95% CI: 25–71%), respectively. The 2 year OS for unrelated donors and related donors was 34% (95% CI: 21–54%) and 26% (95% CI: 12–56%), respectively. Likewise, a subset analysis of patients who received allografts from HLA matched unrelated donors and single allele HLA mismatched unrelated donors did not show a significant difference in the risk of NRM or OS as compared to patients who received allografts from matched related donors (Table 2).

Graft versus Host Disease

The 6 month cumulative incidence of acute grade 2–4 GVHD for matched related donors, matched unrelated donors and single HLA mismatched unrelated donors was 30% (95% CI: 10–53%), 61% (95% CI: 37–78%) and 91% (95% CI: 29–99%), respectively. Patients who had single HLA mismatched unrelated donors had a significantly higher risk of developing acute grade 2–4 GVHD (HR=4.64, 95% CI [1.74–12.41], p=0.002, Table 2, Fig 2) as compared to patients who had HLA matched related donors. In contrast, the increased hazard for acute grade 2–4 GVHD for patients who received an allograft from an HLA matched unrelated donor was more modest and did not reach statistical significance (HR=2.22, 95% CI [0.93–5.31], p=0.074, Table 2, Fig 3). The 6 month cumulative incidence of acute grade 3–4 GVHD for matched related donors, matched unrelated donors and single HLA mismatched unrelated donors was 24% (95% CI: 7–47%), 30% (95% CI: 13–50%) and 36% (95% CI: 10–64%), respectively. Donor type did not have a significant impact on the risk of grade 3–4 acute GVHD (Table 2). The 1 year cumulative incidence of chronic GVHD for matched related donors and matched unrelated donors was 8.5% (95% CI: 0.35–34%) and 14% (95% CI: 3.2–32%), respectively. Patients who received allografts from matched unrelated donors were not at a significantly higher risk of developing chronic GVHD (HR=2.61, 95% CI [0.60–11.32], p=0.20, Table 2) as compared to patients who had matched related donors. The hazard ratio comparing the risk of chronic GVHD between patients who received allografts from single HLA mismatched unrelated donors to patients who received allografts from matched related donors could not be computed because none of the patients who received allografts from single HLA mismatched unrelated donors developed chronic GVHD during the follow up period.

FIGURE 2. Mismatched unrelated donors are associated with a higher risk of acute grade 2–4 graft versus host disease (GVHD).

A cumulative incidence plot of acute grade 2–4 GVHD is shown, stratified by donor type. Mismatched unrelated donors were associated with an increased risk of acute grade 2–4 GVHD (p=0.002) when compared with matched related donors. In contrast, matched unrelated donors were not associated with an increased risk of acute grade 2–4 GVHD (p=0.074) when compared with matched related donors.

Allograft cell content

We then hypothesized that the benefit in disease control for young unrelated donors was related to differences in T cell graft content. We therefore analyzed the impact of donor age and relatedness on graft CD4, CD8 and CD34 cell doses and found that donor age correlated inversely with CD8 cell doses (r=−0.42, p = 0.0024) but did not have a significant association with CD34 (r=−0.19, p=0.16) or CD4 cell doses (r=−0.093, p=0.52).

Unrelated and related donors provided a mean CD8 cell dose of 0.67 × 10⁸/kg (95% CI: 0.53–0.82) and 0.40 × 10⁸/kg (95% CI: 0.25–0.55), respectively (p=0.014). In contrast, there were no significant differences in mean CD34 (p=0.22) and CD4 cell doses (p=0.14) between allografts from unrelated and related donors. In summary, younger donor age and unrelated donors were associated with higher CD8 cell doses.

DISCUSSION

Our study found that patients who received allografts from unrelated donors had a lower risk of relapse and improved relapse-free survival as compared to patients who received allografts from related donors. We also showed that the reduction in the risk of relapse and improved RFS in unrelated donors was driven largely by patients who received allografts from HLA matched unrelated donors. Patients who had single allele mismatched unrelated donors had a significantly higher risk of developing acute grade 2–4 GVHD as compared to patients who had HLA matched related donors. Despite the higher risk of acute grade 2–4 GVHD in unrelated donors, the risk of non-relapse mortality was similar across donor types. We also found that unrelated donors were significantly younger than related donors, provided allografts with higher CD8 cell doses and were associated with more rapid T cell engraftment.

There are several possible biologic reasons for improved outcomes with younger unrelated donors. Advanced donor age has been shown to result in loss of stem cell repopulation potential [14], impaired homing ability [15] and an overall loss of function [16]. The success of RI alloSCT strongly depends on T cell mediated graft-versus-tumor effect for eradication of malignant cells, suggesting the importance of allograft T cell content. In our study, we found that allografts from unrelated donors had a higher mean CD8 T cell content, which was related to their younger age. We therefore hypothesize that age related differences in graft T cell content could, amongst other factors, lead to reduced relapse risk and improved survival possibly through more rapid T cell engraftment and enhanced graft versus tumor response [10]. Our study did not show an improvement in overall survival likely due to our relatively small sample size or because some MDS patients who relapse after transplant still benefit from subsequent salvage therapies such as hypomethylating agents, donor lymphocyte infusions or second transplants.

Several earlier studies reported an association between younger donor age and improved overall and disease-free survival [5–8]. However, these studies included heterogeneous cohorts in terms of diseases, conditioning regimens, graft sources and the use of T cell depleting antibodies, making it difficult to apply the results clinically to MDS patients undergoing RI alloSCT. To overcome these limitations, our study focused on a uniform patient population that was allografted following a standard and commonly used RI conditioning regimen for MDS.

Interestingly, a Center for International Blood and Marrow Transplant Research (CIBMTR) analysis showed worse OS after an alloSCT from younger matched unrelated donors compared with older sibling donors in patients with good performance scores and comparable outcomes between donor types for patients with poor performance scores [9]. However, only 12% of patients in this study were transplanted for MDS and 50% of patients in this study received myeloablative regimens. In contrast, another CIBMTR study analyzing outcomes specifically in patients with MDS reported similar disease free survival (DFS) and OS rates in patients allografted from matched related donors and 8/8 matched unrelated donors while patients allograted from single allele mismatched unrelated donors had poorer DFS and survival rates [17]. However, only 28% of patients in this study received RI conditioning regimens. In addition, there was considerable heterogeneity in stem cell source as well as the use of T cell depleting antibodies, particularly in unrelated donor transplants, making it difficult for the results to be applied clinically.

Our study has a few limitations. First, the retrospective nature of our study meant that donor type was not randomly assigned. Because donor availability is a practical consideration in organizing an unrelated donor transplant, treating physicians may have been more inclined to opt for matched related donors in patients with more advanced disease to avoid a prolonged search for an unrelated donor. This may have led to a selection bias resulting in patients with more advanced disease represented to a greater degree in the group that received allografts from related donors. However, aside from donor age, we did not observe any statistically significant differences in baseline characteristics between the two groups, making it unlikely that this effect led to a significant degree of bias in our study. In addition, because our analysis was retrospective and most of our patients were diagnosed prior to the ubiquitous use of next generation sequencing techniques in clinical practice, we were unable to analyze the impact of specific gene mutations such as TP53, EZH2, ETV6, RUNX1 and ASXL1 that have been independently associated with poor prognosis in MDS [18]. Second, we examined a single common reduced intensity conditioning regimen (fludarabine-busulfan) in a single center which may limit the generalizability of our results. Whether young unrelated donors are associated with improved relapse free survival with other conditioning regimens remains unknown. Third, our relatively small sample size limited our ability to perform adequately powered subset analyses or demonstrate an association between graft CD8 cell dose with more rapid engraftment and/or improved outcomes.

In conclusion, our data suggests that patients with MDS undergoing RI alloSCT who receive allografts from unrelated donors have a lower risk of relapse and improved relapse free survival without an accompanying increase in non relapse mortality as compared to patients who are allografted from matched related donors. Unrelated donors were significantly younger than related donors, provided allografts with higher CD8 cell doses and were associated with more rapid donor T cell engraftment. These findings should be confirmed in a prospective study.