(GT)n Repeat Polymorphism in Heme Oxygenase-1 (HO-1) Correlates with Clinical Outcome after Myeloablative or Nonmyeloablative Allogeneic Hematopoietic Cell Transplantation

Allogeneic hematopoietic cell transplantation (HCT) is a treatment for various hematologic diseases where efficacy of treatment is in part based on the graft versus tumour (GVT) activity of cells in the transplant. The cytoprotective enzyme heme oxygenase-1 (HO-1) is a rate-limiting enzyme in heme degradation and it has been shown to exert anti-inflammatory functions. In humans a (GT)n repeat polymorphism regulates the expression of HO-1. We conducted fragment length analyses of the (GT)n repeat in the promotor region of the gene for HO-1 in DNA from donors and recipients receiving allogeneic myeloablative- (MA) (n = 110) or nonmyeloablative- (NMA-) (n = 250) HCT. Subsequently, we compared the length of the (GT)n repeat with clinical outcome after HCT. We demonstrated that transplants from a HO-1high donor after MA-conditioning (n = 13) is associated with higher relapse incidence at 3 years (p = 0.01, n = 110). In the NMA-conditioning setting transplantation of HO-1low donor cells into HO-1low recipients correlated significantly with decreased relapse related mortality (RRM) and longer progression free survival (PFS) (p = 0.03 and p = 0.008, respectively). Overall, our findings suggest that HO-1 may play a role for the induction of GVT effect after allogeneic HCT.


Introduction
Allogeneic hematopoietic cell transplantation (HCT) is an effective therapeutic option for patients with various hematologic malignancies. Current challenges include reducing graft-versus-host disease (GVHD) while maintaining or augmenting the beneficial graft-versustumour effect (GVT) [1]. Donor-derived type 1 T cells are known to be major players in GVT and GVHD which involve killing of tumour cells as well as healthy cells and tissues, respectively [1].
Heme oxygenase-1 (HO-1) is a rate limiting stress-responsive enzyme that degrades heme into biliverdin, carbon monoxide and free iron [2]. It is induced by various oxidative agents and exerts anti-inflammatory and cytoprotective effects hereby controlling inflammation and inducing tolerance [3]. To this end, dendritic cells expressing HO-1 retain an immature phenotype which promote differentiation of Tregs and inhibit effector T-cell responses [3]. Indeed, it could be speculated that the immune suppressive functions of HO-1 may influence the immune mechanisms underlying the effects of allogeneic HCT. It has been demonstrated, that a (GT) n dinucleotide repeat in the promoter region of the HO-1 encoding gene, HMOX1, regulates the expression of HO-1 [4]. Shorter (GT) n repeats leads to increased transcriptional activity and, thus, a higher expression of HO-1 upon oxidative stimuli [5].
The purpose of our study was to examine a possible association of the association between the HO-1 (GT) n repeat polymorphism and clinical outcomes after both myeloablative (MA) or nonmyeloablative (NMA) allogeneic HCT.

Patients and controls
The study population consisted of patients with malignant hematologic diseases that received related or unrelated donor HCT after MA conditioning (n = 110) or NMA conditioning (n = 250) ( Table 1) between March 2000 and July 2007 at the bone marrow transplantation unit at Rigshospitalet, Copenhagen, Denmark. Transplant regimens and supportive care have been described elsewhere [6]. Acute GVHD and chronic GVHD were graded according to standard criteria [7]. Samples from related and unrelated donors were obtained from leukapheresis products whereas peripheral blood samples were obtained from healthy donors. DNA was extracted as previously reported using the Promega Maxwell 16 blood DNA kit (Promega Corporation, Madison, WI) [8]. Written informed consent was obtained from all recipients and donors and the study and consent procedure was approved by the Scientific Ethics Committee for The Capital Region of Denmark. Our healthy cohort included DNA from 74 individuals from the Danish general population [9].
Fragment length analyses of the (GT) n repeat in the promoter region of the HMOX1 gene To determine the number of (GT) n repeats in the promoter region, polymerase chain reaction (PCR) was carried out to amplify the (GT) n repeat region using a forward primer labeled with a dye (FAM). Forward primer: 5'-FAM-AGAGCCTGCAGCTTCTCAG-3', reverse primer: 5-AAACAAAGTCTGGCCATAGGAC-3' (TAG Copenhagen, Denmark). Amplifications were carried out in a total volume of 15 μl containing 1x PCR buffer (50 mM KCl, 20 mM Tris pH 8.4, 2.0 mM MgCl 2 , 0.2 mM cresol red, 12% sucrose, 0.005% (wt/v) BSA (Boehringer-Mannheim, Germany), 5 pmol of each primer, 40 mM dNTPs (Pharmacia LKB, Sweden) and 1.25 U of AmpliTaq polymerase (Perkin Elmer Cetus Corporation, Emeryville, CA, USA). To minimize non-specific amplification, we carried out "touch down" PCR where annealing temperature is higher than target optimum in early PCR cycles followed by a decrease by 0.5˚C in every following cycle. Genotyping was performed by analyses of PCR products from amplification of the (GT) n repeat region in a 3500 Genetic Analyzer (Applied Biosystems). 1 μL of PCR product was added to 20 μL formamide (95%) and 0.3 μL GeneS-can™-600 LIZ 1 SIZE standard (Applied Biosystems). Fragment length examination was performed using the GeneMapper software version 4.1 (Applied Biosystems).

Statistical analyses
For our statistical analyses differences in distribution of number of repeats between patients and donors were tested using a Kolomogorov-Smirnov test. Comparisons of groups of patients and discrete variables were done using Fisher's Exact test and comparisons of groups of patients and continuous variables were performed using the Mann-Whitney test. Overall survival (OS) and progression-free survival (PFS) was estimated by the Kaplan-Meier method. Cumulative incidences of relapse/progression, relapse related mortality (RRM), acute and chronic GVHD were estimated by standard methods for competing risks [10]. A Cox proportional hazard regression model was used for calculating p-values and hazard ratios. Significant associations between genotype and outcome were further investigated in multivariable Cox regression models, restricted to include patient age, sex, donor type (related versus unrelated) and relapse risk (i.e Kahl score for NMA cohort [11] and EBMT for MA cohort [12]) to calculate adjusted hazard ratios and p-values. This was not done on subset analyses on subsets of data due to the low number of cases. In cases of very few events a permutation test was used for calculating differences. All tests were two-sided and p-values below 0.05 were considered significant. IBM SPSS Statistics version 19.0.0 (2010, IBM, Somer, NY, USA) or R Statistical Software version 3.2.2 (2015, R Foundation, Vienna, Austria) were used for all calculations. R package "crr" was used for cumulative incidence calculations and function "permlogrank" in the "clinfunc" package was used for permutation tests.

Results and Discussion
First, we used fragment analyses to determine the number of (GT) n repeats among 339 recipients and 360 donors as well as 74 healthy donors. Alleles were classified as either short with 26 (GT) n repeats (S) or long with >26 (GT) n repeats (L) [13]. The (GT) n repeat polymorphism adhered to the Hardy-Weinberg equilibrium in all 3 cohorts (p = 0.55, p = 0.56, p = 0.75). The number of (GT) n repeats ranged from 21 to 39 for recipients and 19 to 38 for donors (S1 Fig). In agreement with previous reports distribution of the alleles was bimodal with the most common alleles consisting of 23 and 30 (GT) n repeats [13]. The distribution of alleles was similar between recipients and donors (p = 0.8) and did not differ from that of healthy donors (p = 0.9 and 0.4, respectively). Next, we investigated the association between HO-1 genotype in recipients and donors and clinical outcome after allogeneic MA-HCT. Recipients of grafts from donors with genotype S/ S (HO-1 high ) had a significantly higher relapse/progression incidence (time to progression) at 3 years (38%, n = 13) than patients receiving grafts from donors with genotypes S/L or L/L (16%, n = 97) (adjusted hazard ratio (HR) 3.4 (95% confidence interval (CI) 1.3-9.1), p = 0.01, Fig 1A, Table 2), which translated into lower PFS (S/S, 39%; S/L or L/L, 54%; adjusted HR 2.2 (95% CI 1.0-4.46), p = 0.04) and a trend towards increased RRM (S/S, 38%; S/L or L/L, 16%; adjusted HR 2.8 (95% CI 1.0-7.9), p = 0.05) at 3 years (Table 2).
In the NMA setting no association between outcome and genotype where observed, when statistical analyses were isolated to host genotype only (Table 2). However, HO-1 low genotype (L/L) recipients of HO-1 low donors experienced significantly lower RRM (L/L donor into L/L recipient, 18%, n = 29; L/L donor into S/L or S/S recipient, 31%, n = 25; HR 0.3 (95% CI 0.1-0.9), p = 0.03, Fig 1B) and increased PFS (L/L donor into L/L recipient, 64%, n = 29; L/L donor into S/L or S/S recipient, 29%, n = 25; HR 0.4 (95% CI 0.2-0.8), p = 0.008, Fig 1C) at 3 years. Furthermore, we observed a trend towards higher incidence of grade III-IV acute GVHD (L/L, We did not observe the same effect of HO-1 genotype in the two cohorts. However, the results were complementary and in support of previously published animal data indicating that HO-1 expression favors induction of regulatory T cells (Tregs) and inhibits proliferation of effector T cells [14], thus mitigating GVT effects. In the MA cohort donorswith an antiinflammatory phenotype(HO-1 high (S/S) genotype) were associated with increased relapse related outcome measures, implying that high levels of the anti-inflammatory HO-1 may suppress GVT effects. In the NMA cohort transplantation using donors with the pro-inflammatory HO-1 low (L/L) genotype in equally phenotypically pro-inflammatory HO-1 low (L/L) recipients was associated with less relapse and superior PFS, implying augmented GVT effects with less anti-inflammatory drive. The reason for only being able to observe significant association between complementary endpoints, and not the same endpoints, across the two cohorts, may be due to the relatively small cohort sizes and heterogeneity. Due to the same reason the current observations could not be reproduced in joint analyses of the cohorts.
Only 1 study has previously been published investigating the role of HO-1's genetic variation after allogeneic HCT. In a cohort of 92 patients transplanted with in vivo T-cell depleted grafts from matched related donors MA-HCT, Gerbitz et al. observed that donors with high HO-1 expression (based on (GT) n repeat genotype) and thus increased anti-inflammatory drive, were associated with inferior survival due to grade III-IV acute GVHD!3 [15]. Gerbitz et al. hypothesize that expression of HO-1 above a certain threshold may lead to an excessive oxidative burst that induces tissue damage [16]. In contrast, our study implied that high expression of HO-1 suppresses the GVT effects. The apparently inconsistent results observed in the current study and often across genetic association studies may be due to several factors that make direct comparisons difficult. The generally small cohorts in transplant studies increase the risk of type I errors as the effect of single genetic variants usually is modest. Furthermore heterogeneity between patient populations, with differences in treatment regimens, diagnoses, racial admixture and other yet unknown risk factors, as well as use of different laboratory methods and analyses platforms all contribute to an unclear picture.
In conclusion, data from the current study suggest that donor HO-1 genotype may be associated with the GVT effect in both the MA and NMA setting, hereby implying a role for HO-1 in alloreactivity. To validate the significance of these findings further experimental and clinical studies are needed to explain the role of HO-1 in alloreactivity and assess the impact on outcome in a prospective manner with consequences in a clinical setting.

S1 Fig. Frequency of number of HO-1 (GT) n repeats.
We determined the number of (GT) n repeats among 74 normal donors (normal cohort), 360 donors (donor cohort) and 339 recipients (recipient cohort). As every individual carries two alleles we analysed 148 normal donor, 720 donor and 678 recipient alleles. (TIF)