Original Article

Bone Marrow Transplantation (2015) 50, 1445–1452; doi:10.1038/bmt.2015.173; published online 27 July 2015

Transplant Toxicities

Predicting survival using clinical risk scores and non-HLA immunogenetics

Y Balavarca1,14, K Pearce2, J Norden2, M Collin2, G Jackson3, E Holler4, R Dressel5, H-J Kolb6, H Greinix7, G Socie8, A Toubert9, V Rocha10, E Gluckman10, I Hromadnikova11, P Sedlacek12, D Wolff4, U Holtick13, A Dickinson2,15 and H Bickeböller1,15

  1. 1Department of Genetic Epidemiology, University Medical Center, Göttingen, Germany
  2. 2Department of Haematological Sciences, Institute of Cellular Medicine, Medical School, Newcastle University, Newcastle upon Tyne, UK
  3. 3Northern Centre for Cancer Care, Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle upon Tyne, UK
  4. 4Department of Hematology and Oncology, University of Regensburg, Regensburg, Germany
  5. 5Department of Cellular and Molecular Immunology, University Medical Center, Göttingen, Germany
  6. 6Department of Haematology and Oncology, Klinikum Grosshadern, Medical Klinik III, Munich, Germany
  7. 7Department of Haematology, Division of Haematology, Medical University of Graz, Graz, Austria
  8. 8Department of Haematology, Immunology and Oncology, AP-HP, Saint Louis Hospital, Hematology Transplantation, Paris, France
  9. 9Departement d′Immunologie, Université Paris Diderot, INSERM UMRS-940, AP-HP, Paris, France
  10. 10Department of Bone Marrow Transplantation, EUROCORD, St Louis Hospital, Paris, France
  11. 11Department of Molecular Biology and Cell Pathology, Third Faculty of Medicine, Charles University, Prague, Czech Republic
  12. 12Department of Pediatric Hematology and Oncology, Second Faculty of Medicine, Charles University, Prague, Czech Republic
  13. 13Department I of Internal Medicine, University of Cologne, Cologne, Germany

Correspondence: Dr K Pearce, Department of Haematological Sciences, Institute of Cellular Medicine, William Leech Building, Medical School, Newcastle University, Framlington Place, Newcastle upon Tyne NE2 4HH, UK. E-mail: Kim.Pearce@ncl.ac.uk

14Current address: Department of Preventive Oncology, National Center for Tumour Diseases and German Cancer Research Center, Heidelberg, Germany.

15These authors contributed equally to this work.

Received 18 November 2014; Revised 16 June 2015; Accepted 18 June 2015
Advance online publication 27 July 2015



Previous studies of non-histocompatibility leukocyte antigen (HLA) gene single-nucleotide polymorphisms (SNPs) on subgroups of patients undergoing allogeneic haematopoietic stem cell transplantation (HSCT) revealed an association with transplant outcome. This study further evaluated the association of non-HLA polymorphisms with overall survival in a cohort of 762 HSCT patients using data on 26 polymorphisms in 16 non-HLA genes. When viewed in addition to an already established clinical risk score (EBMT-score), three polymorphisms: rs8177374 in the gene for MyD88-adapter-like (MAL; P=0.026), rs9340799 in the oestrogen receptor gene (ESR; P=0.003) and rs1800795 in interleukin-6 (IL-6; P=0.007) were found to be associated with reduced overall survival, whereas the haplo-genotype (ACC/ACC) in IL-10 was protective (P=0.02). The addition of these non-HLA polymorphisms in a Cox regression model alongside the EBMT-score improved discrimination between risk groups and increased the level of prediction compared with the EBMT-score alone (gain in prediction capability for EBMT-genetic-score 10.8%). Results also demonstrated how changes in clinical practice through time have altered the effects of non-HLA analysis. The study illustrates the significance of non-HLA genotyping prior to HSCT and the importance of further investigation into non-HLA gene polymorphisms in risk prediction.



Haematopoietic stem cell transplantation (HSCT) is the major curative therapy for disorders of the blood and immune system. However, the rate of survival in patients with HSCT has remained at 40–60% for the last two decades, owing to post-transplant complications including infection, GvHD and relapse.

Five relevant clinical factors influencing transplantation success in patients with haematological disorders, including CML, ALL and AML, have been identified by the European Group for Blood and Marrow Transplantation (EBMT). These risk factors (EBMT-factors) are patient age, sibling donor/matched unrelated donor (MUD), patient–donor gender combination, stage of disease and time from diagnosis to transplant. A clinical risk score (EMBT-score) utilising the EBMT-factors was proposed in order to aid the prediction and prevention of post-transplant complications.1, 2, 3, 4

Previous genetic association studies have suggested that, besides HLA genes, non-HLA genes may also have an important role in transplant outcome. To date, these studies with single-nucleotide polymorphisms (SNPs) have used small subsets of patients.5, 6, 7, 8, 9 Although genome-wide association studies have been performed,10 no SNP genotypes have been clearly identified so far that could be used to predict the outcome.

In this study, we assessed the association of candidate polymorphisms with overall survival using a large cohort of patients undergoing HSCT. The goal was to identify non-HLA SNPs with an impact on overall survival when viewed in addition to the already established EBMT-score. Prediction capability was also evaluated. Because changes in transplant protocols can affect transplant outcome, the study also took into account date of transplant.


Materials and methods


A total of 762 patients with malignant haematological diseases, having complete data on the EBMT-factors1, 3, 4 and with known date of death or last contact, were included in the study. These patients were transplanted between November 1983 and December 2005 at seven European transplant centres. The patients and donors gave informed consent to participate in the study in accordance with the Declaration of Helsinki and EBMT guidelines. The protocol was approved by the local research ethics committee at the coordinating centre (Newcastle upon Tyne, UK). Follow-up time was between 7 months and 20 years, with a median of 5–6 years. Patient and donor characteristics are presented in Table 1. Overall, death occurred in 399 patients (52%). Causes of death were relapse (41%), GvHD (18%), infection (18%), multiple organ failure (5%), acute respiratory-distress syndrome (4%), veno-occlusive disease (2%), interstitial pneumonitis (1%) and others (11%). The majority of the cohort after the year 2000 had high-resolution tissue typing for HLA Class I A,B,C and Class II DP,DQ and DR.

Candidate non-HLA polymorphisms

DNA was prepared from archived frozen peripheral blood mononuclear cells. Genotyping was outsourced to Kbioscience (http://www.kbioscience.co.uk) who used fluorescence-based competitive PCR technology and designed the assays for the SNPs based on the DNA sequence (50 bases) either side of the SNP. Genotypes were available for 743 patient–donor pairs on the following genes: CD14, CD91, C3, ESR1, GCR, HSP70-hom, IFNG, IL1RN, IL4, IL-6, IL-10, IL12B, IL13, LOX1, MAL, MDR1, NOD2, TNF, TNFRSF1B and VDR (Table 2). Candidate SNPs were selected according to findings on smaller patient cohorts by our coordinating centre in Newcastle (ESR1,11 IFNG,12 IL1RN,13 IL4,14 IL-6,15, 16 IL-10,17 IL13,18 TNF,19 TNFRSF1B20 and VDR21). SNPs were also selected according to findings by other groups in HSCT (MAL,22 MDR123, 24, 25 and NOD226) and according to previous disease association studies in autoimmune (CD14,27 GCR,28, 29 HSP70-hom30 and IL12B31) or inflammatory disease (and recently found to be deregulated in a rat model for GvHD:32 C3,33 LOX134 and CD9135).

Statistical analysis

Clinical differences in patients treated up to and after the year 2000 were assessed using Fisher’s exact test. This division in time was chosen as transplant protocols were changed at that time with the introduction of Imatinib and a subsequent increase in survival rate.2

Biallelic SNPs were considered under the additive, dominant and recessive modes of inheritance10 (Supplementary Section A). In the models, each SNP was used with the mode demonstrating the strongest association with survival.

The genes HSP70-hom, IL12B and MDR1 and GCR-haplotype were excluded from the analysis owing to missing genotypes (>33%). A total of 26 polymorphisms from 16 genes were available for analysis (Table 2).

A power of 80% in the cohort with 52% deaths was achieved for SNPs in an additive mode for sample sizes ngreater than or equal to300, MAF (minor allele frequency)>15%, and an expected hazard ratio (HR)greater than or equal to1.50. In a dominant mode, this power was achieved for ngreater than or equal to400. In a recessive mode, this was attained for ngreater than or equal to500, MAFgreater than or equal to25% and HR greater than or equal to2.00. A few SNPs with a MAF<10% (that is, IL-6 rs1800796, LOX1 rs11053646 and the three SNPs in NOD2) had a very low power at all settings; even so, we retained these SNPs for further analysis.

Association of risk score for overall survival

The EBMT risk score (EBMT-score) was derived by summation of the EBMT-factors.1, 3, 4 The EBMT-score was subsequently implemented in analyses on an ordinal scale (low-to-high risk 0–7). The additional effect of each individual polymorphism was evaluated using separate Cox regression models. The likelihood ratio test was applied to compare a Cox regression model including the EBMT-score and one polymorphism to a model including only the EBMT-score; a nominal P-value of 0.05 was used. Analyses were performed for the cohort as a whole, as well as for several subgroups.

Using step-wise selection, we built an additional Cox regression model to establish whether multiple non-HLA polymorphisms improved the model when added in with the EBMT-score (α=0.05 for variable entry and α=0.10 for variable removal36). All available polymorphisms were chosen as candidates in this procedure. HRs and 95% confidence intervals (95% CI) were reported.

A new risk score (EBMT-genetic-score) was developed using this last Cox regression model. The score was derived by summing up the clinical EBMT-score and the genetic score points given for the polymorphisms. The latter were obtained by dividing the respective regression coefficients by the coefficient of the EBMT-score and rounding to the nearest integer37 (Supplementary Section B).

Assessment of prediction

Two statistical approaches were used to assess the risk score-prediction capability: the concordance index (C-index)3, 4, 38, 39, 40 and the r2 measure of the gain in prediction.41 The C-index measures the agreement between risk score and observed survival time. A higher risk score should correspond to a shorter observed survival time. A C-index=0.5 implies that a risk score has no predictive discrimination, whereas C-index=1 implies maximum predictive discrimination. The U-statistic40 was also used to test whether the EBMT-genetic-score was better than the EBMT-score as regards agreement with observed survival time. In addition, r2 was used to quantify the improvement in prediction41 of the new EBMT-genetic-score over the EBMT-score. This measure is a combination of the 0.632 bootstrap estimate of prediction error42 and the explained variation using Schoenfeld residuals43, 44 (see Supplementary Section C). Larger values of r2 (>0%) mean that the EBMT-genetic-score correctly predicts outcome more often than the EBMT-score. r2=0% means that both scores have equivalent predictive ability and r2=100% means that the EBMT-genetic-score has perfect predictive ability (that is, the predicted and the actual outcomes always agree).



Clinical characteristics

Sixty percent of the patients underwent HSCT after the year 2000. There was evidence that, after 2000, transplants involved more patients and donors over 40 years of age, more HLA-MUDs, more lymphoma, more donor cells from peripheral blood, more later-stage disease, more T-cell depletion and more reduced-intensity conditioning (Table 1).

There was no significant centre effect on overall survival (likelihood ratio test, P-value=0.49).

Association of the EBMT risk score and single polymorphisms with overall survival

The EBMT-score1, 3, 4 was significantly associated with overall survival (HR=1.16, 95% CI=1.09–1.24, P<0.001).

The top ten candidate polymorphisms associated with overall survival in the whole cohort, while controlling for EBMT-score, are given in Table 3. The IL-10 haplotype in donors demonstrated the lowest P-value and the presence of ACC/ACC was protective (HR=0.48, 95% CI=0.29–0.80, likelihood ratio test P-value=0.002). This haplotype was also the only polymorphism significantly associated with survival in all subgroups (Supplementary Table 1). Nine of the ten polymorphisms were also highest ranked amongst patients without T-cell depletion and with myeloablative conditioning regimens. IL-10 haplotype, IL-10 rs1800896(G), IL4 rs2243250(T) and IL-6 rs1800797(A) in donors, ESR1 rs2234693(C) and GCR rs33388(T) in patients were observed in over 50% of the assessed clinical subgroups.

Association and prediction of multiple polymorphisms with overall survival

After step-wise selection, the final model contained the EBMT-score and four selected polymorphisms (n=419 owing to missing genotypes). The presence of haplo-genotype ACC/ACC of IL-10 in donors was protective against patient death (HR=0.49, 95% CI=0.26–0.89, P-value=0.020). The risk of death increased with an increased number of T alleles in MAL rs8177374 in patients (additive, HR=1.34, 95% CI=1.04–1.74, P-value=0.026), with the presence of allele G in ESR1 rs9340799 in patients (dominant, HR=1.52, 95% CI=1.15–2.01, P-value=0.003), and with the increased number of C alleles in IL-6 rs1800795 in donors (additive, HR=1.29, 95% CI=1.07–1.55, P-value=0.007; Table 4, Supplementary Figure 1).

Three out of four of these polymorphisms were significantly associated with death due to relapse. In addition, MAL rs8177374 was associated with death due to GvHD and IL-6 rs1800795 was associated with death due to infection (Supplementary Table 2).

When the multivariate Cox regression model (Table 4) was compared with a model containing the EBMT-score alone, the estimated r2 for gain in prediction indicated a 5.1% gain in prediction ability by adding the four polymorphisms (separately) to the EBMT-score.

Comparing EBMT-genetic-score and EBMT-score

The new EBMT-genetic-score, derived through summing the individual risk score values (Supplementary Section B), ranged from 1 to 15. The scores were grouped into five distinct categories according to the observed groupings of the Kaplan–Meier survival curves (results not shown). The Cox regression model, with five ordered categories of the EBMT-genetic-score, revealed increasing HRs with increasing EBMT-genetic-score (n=419, Table 5). The risk values of the EBMT-genetic-score displayed a clearer separation of the survival curves when compared with values of the EBMT-score (Supplementary Figure 2).

Kaplan–Meier curves for the EBMT-genetic-score were plotted to establish if the EBMT-genetic-score is appropriate for the clinical subgroups: disease diagnosis, sibling donor/MUD, T-cell depletion, conditioning regimen and year of transplantation. The plots consistently demonstrated that higher risk scores corresponded to lower survival (Supplementary Figure 3).

For the whole cohort, the higher EBMT-genetic-score corresponded to shorter observed survival times compared with the EBMT-score (U-statistic P-value<0.001, Table 6). This also proved to be the case in subgroups of patients transplanted before and after 2000. Estimation of the gain in prediction ability indicated that there was benefit in utilising the single EBMT-genetic-score (r2=10.8%, Table 6) over the previous model containing the EBMT-score and four separate polymorphisms (r2=5.1%).

Kaplan–Meier survival curves for EBMT-score and EBMT-genetic-score before and after 2000 (n=419) appear in Figures 1a–d. Compared with the EBMT-score, the EBMT-genetic-score better discriminates the survival curves and a higher score consistently corresponds to a lower survival probability. It was also apparent that, when using the EBMT-genetic-score, those treated after 2000 had improved chances of survival in comparison with those treated before 2000 (Figures 1c and d). For those with the lowest scores (1–6), 3-year survival was 85 and 95% for patients treated before and after 2000, respectively; for those with the highest scores (13–15), 1-year survival was 15 and 42% for patients treated before and after 2000, respectively.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Kaplan–Meier curves of patients undergoing HSCT up to and after the year 2000 (n=419) for 5 years follow-up (a) EBMT-scores before 2000, (b) EBMT-scores after 2000, (c) EBMT-genetic-scores before 2000 and (d) EBMT-genetic-scores after 2000.

Full figure and legend (129K)



The aim of this work was to study the effect of non-HLA polymorphisms on overall survival of HSCT patients in addition to the EBMT-score. In our study, the IL-10 promoter haplo-genotype ACC/ACC in donors was one of the most important polymorphisms associated with improved overall survival. IL-10 is an important cytokine in the regulation of the immune response. However, it can have a stimulatory effect on B cells, increasing MHC class II expression and antibody production. IL-10 haplotypes have been shown to correlate with IL-10 protein production,45 with the GCC haplotype being associated with the highest IL-10 production.

The ATA/ACC genotype has been identified as a protective factor for overall survival in CML patients with sibling donors.9 SNP rs1800872(A) and haplotype ACC in patients have been shown to demonstrate a strong association with severe acute GvHD (aGvHD III–IV) in patients with matched related donors.7, 10 In our whole cohort, the IL-10 promoter haplo-genotype ACC/ACC in donors proved to be significantly associated with increased overall survival, whereas SNP rs1800872(A) in patients revealed only borderline association. The ACC/ACC genotype in the donor is associated with intermediate production of IL-10 and the ACC haplotype has been shown to be protective in aspergillosis.46 The AA genotype of IL-10 rs1800872 in the patient is associated with a decreased risk of aGvHD47 and an increased risk of non-relapse mortality; this was confirmed in follow-up genome-wide association studies.10, 48 In addition, the presence of the GG genotype of IL-10 rs1800896 in the patient was found to be associated with the risk of chronic GvHD.49

Furthermore, we discovered that MAL rs8177374(T) in patients is associated with reduced overall survival whereas parallel smaller studies have reported that the presence of the T allele in donors resulted in less chronic GvHD and a reduction in transplant-related mortality. In addition, the presence of the T allele in patients was associated with an increased risk of relapse.50 Our study revealed a strong association between the presence of the T allele and relapse and an increased risk of death in patients who relapsed. Moreover, the T allele was strongly associated with GvHD. The T allele is regarded in the literature as the inflammatory allele, and T-heterozygous individuals have increased protection from infection. Interestingly, another study revealed that patients transplanted from donors with the T allele have a lower incidence of fungal infections, aGvHD and improved overall survival.22 MAL protein was originally identified in intermediate and late stages of T-lymphocyte differentiation51 and the MAL mRNA expression was also found to be related with differentiation in urothelial cells, neuronal cells52, 53 and oesophageal epithelium.54 The MAL-A variant containing all four exons is abundantly expressed in peripheral blood lymphocytes55 and positively expressed in the gastrointestinal tract, respiratory tract and haematopoietic system. MAL is important in the innate immune response, it is involved in Toll-like receptor signalling,56 so it could be important in the development of aGvHD in the recipient.

Our study also indicated that ESR1 rs9340799(G) in patients is associated with reduced overall survival. The G allele has been previously reported to be associated with reduced overall survival and increased risk of aGvHD in patients with HLA-matched siblings.11 ESR1 is thought to inhibit IL-6 production.57

Furthermore, we found IL-6 rs1800795(C) in donors to be associated with reduced overall survival. The G allele in patients and/or donors has been reported by genome-wide association studies as a risk factor for both aGvHD and chronic GvHD in patients with HLA-matched related10, 21 and unrelated10 donors. In our study, there was evidence of an increased risk of aGvHD in patients transplanted from donors with the C allele present. Among these patients, those transplanted before 2000 had evidence of reduced survival compared with those transplanted later (Supplementary Figure 4a), possibly as a result of a higher rate of standard myeloablative conditioning before 2000.

The G allele of this SNP is reported to correlate with higher serum IL-6 levels in systemic-onset juvenile chronic arthritis.58 However, the CC genotype has been associated with higher levels of IL-6 in polymyalgia rheumatica.59 It could therefore be that increased levels of IL-6 in our cohort exacerbated the inflammatory milieu, leading to increased transplant-related complications such as infection and poorer survival.

We have also demonstrated that polymorphisms modelled via Cox regression, either in a joint model with the EBMT-score or combined with the latter as a single EBMT-genetic-score factor, contribute to a better discrimination of the risk groups (C-index) and increase the prediction of survival (r2) compared with the EBMT-score alone. Changes in HSCT clinical protocols during 2000 greatly improved patient survival. The consideration of an EBMT-genetic-score highlighted an improvement in survival, especially for those at higher risk of death. More transplants from MUDs were performed after 2000 and it could be argued that the improvement in survival of patients with high-risk non-HLA genotypes is due to the improved quality of HLA matching after 2000. However, no difference in survival was evident between siblings or MUD patients transplanted after 2000 (Supplementary Figure 4b).

In addition, a recent review has revealed that other pre-transplant clinical factors (for example, CMV status and Karnofsky performance score) also have a role in survival and could be used alongside the EBMT-score.4

In conclusion, we hypothesise that implementing risk scores for pre-transplant risk assessment from clinical and genetic factors enhances the prediction of overall survival for patients undergoing HSCT. The potential of considering non-HLA polymorphisms in pre-transplant risk assessment is evident with the promising results for polymorphisms in genes IL-10, MAL, ESR1 and IL-6. Further investigations into pre-transplant risk assessment could also include other potential predictors such as mRNA and microRNA expression.60


Conflict of interest

The authors declare no conflict of interest.



  1. Gratwohl A, Hermans J, Goldman JM, Arcese W, Carreras E, Devergie A et al. Risk assessment for patients with chronic myeloid leukaemia before allogeneic blood or marrow transplantation. Lancet 1998; 352: 1087–1092. | Article | PubMed | ISI | CAS |
  2. Gratwohl A, Brand R, Apperley J, Crawley C, Ruutu T, Corradini P et al. Allogeneic hematopoietic stem cell transplantation for chronic myeloid leukemia in Europe 2006: transplant activity, long-term data and current results. An analysis by the Chronic Leukemia Working Party of the European Group for Blood and Marrow Transplantation (EBMT). Haematologica 2006; 91: 513–521. | PubMed | ISI |
  3. Gratwohl A, Stern M, Brand R, Apperley J, Baldomero H, de Witte T et al. Risk score for outcome after allogeneic hematopoietic stem cell transplantation: a retrospective analysis. Cancer 2009; 115: 4715–4726. | Article | PubMed | ISI |
  4. Gratwohl A. The EBMT risk score. Bone Marrow Transplant 2012; 47: 749–756. | Article | PubMed | ISI | CAS |
  5. Mullally A, Ritz J. Beyond HLA: the significance of genomic variation for allogeneic hematopoietic stem cell transplantation. Blood 2007; 109: 1355–1362. | Article | PubMed | ISI | CAS |
  6. Ambruzova Z, Mrazek F, Raida L, Jindra P, Vidan-Jeras B, Faber E et al. Association of IL6 and CCL2 gene polymorphisms with the outcome of allogeneic haematopoietic stem cell transplantation. Bone Marrow Transplant 2009; 44: 227–235. | Article | PubMed | ISI |
  7. Dickinson AM, Middleton PG, Rocha V, Gluckman E, Holler E. Genetic polymorphisms predicting the outcome of bone marrow transplants. Br J Haematol 2004; 127: 479–490. | Article | PubMed | ISI | CAS |
  8. Dickinson A. Non-HLA genetics and predicting outcome in HSCT. Int J Immunogenet 2008; 35: 375–380. | Article | PubMed | ISI | CAS |
  9. Dickinson AM, Pearce KF, Norden J, O’Brien SG, Holler E, Bickeböller H et al. Impact of genomic risk factors on outcome after hematopoietic stem cell transplantation for patients with chronic myeloid leukemia. Haematologica 2010; 95: 922–927. | Article | PubMed | ISI |
  10. Chien JW, Zhang XC, Fan W, Wang H, Zhao LP, Martin PJ et al. Evaluation of published single nucleotide polymorphisms associated with acute GVHD. Blood 2012; 119: 5311–5319. | Article | PubMed | ISI | CAS |
  11. Middleton PG, Norden J, Cullup H, Cavet J, Jackson GH, Taylor PR et al. Oestrogen receptor alpha gene polymorphism associates with occurrence of graft-versus-host disease and reduced survival in HLA-matched sib-allo BMT. Bone Marrow Transplant 2003; 32: 41–47. | Article | PubMed |
  12. Pravica V, Asderakis A, Perrey C, Hajeer A, Sinnott PJ, Hutchinson IV. In vitro production of IFN-gamma correlates with CA repeat polymorphism in the human IFN-gamma gene. Eur J Immunogenet 1999; 26: 1–3. | Article | PubMed | ISI | CAS |
  13. Hurme M, Santtila S. IL-1 receptor antagonist (IL-1Ra) plasma levels are co-ordinately regulated by both IL-1Ra and IL-1beta genes. Eur J Immunol 1998; 28: 2598–2602. | Article | PubMed | ISI | CAS |
  14. Walley AJ, Cookson WO. Investigation of an interleukin-4 promoter polymorphism for associations with asthma and atopy. J Med Genet 1996; 33: 689–692. | Article | PubMed | ISI |
  15. Cavet J, Dickinson AM, Norden J, Taylor PR, Jackson GH, Middleton PG. Interferon-gamma and interleukin-6 gene polymorphisms associate with graft-versus-host disease in HLA-matched sibling bone marrow transplantation. Blood 2001; 98: 1594–1600. | Article | PubMed | ISI | CAS |
  16. Leland JY, KyungAh I, Borg B, Yang H, Liang TJ. Interleukin-6 (IL-6) haplotypes and the response to therapy of chronic hepatitis C virus infection. Genes Immun 2009; 10: 365–372. | Article | PubMed |
  17. Morse HR, Olomolaiye OO, Wood NA, Keen LJ, Bidwell JL. Induced heteroduplex genotyping of TNFα, IL-1β, IL-6 and IL-10 polymorphisms associated with transcriptional regulation. Cytokine 1999; 11: 789–795. | Article | PubMed | CAS |
  18. Graves PE, Kabesch M, Halonen M, Holberg CJ, Baldini M, Fritzsch C et al. A cluster of seven tightly linked polymorphisms in the IL-13 gene is associated with total serum IgE levels in three populations of white children. J Allergy Clin Immunol 2000; 105: 506–513. | Article | PubMed | ISI | CAS |
  19. Imboden M, Nieters A, Bircher AJ, Brutsche M, Becker N, Wjst M et al. Cytokine gene polymorphisms and atopic disease in two European cohorts. (ECRHS-Basel and SAPALDIA). Clin Mol Allergy 2006; 4: 9. | Article | PubMed |
  20. Stark GL, Dickinson AM, Jackson GH, Taylor PR, Proctor SJ, Middleton PG. Tumour necrosis factor receptor type II 196 M/R genotype correlates with circulating soluble receptor levels in normal subjects and graft-versus-host disease after sibling allogeneic bone marrow transplantation. Transplantation 2003; 76: 1742–1749. | Article | PubMed | ISI | CAS |
  21. Middleton PG, Cullup H, Dickinson AM, Norden J, Jackson GH, Taylor PRA et al. Vitamin D receptor gene polymorphism associates with graft-versus-host disease and survival in HLA-Matched sibling allogeneic bone marrow transplantation. Bone Marrow Transplant 2002; 30: 223–228. | Article | PubMed | ISI | CAS |
  22. Rocha V, Porcher R, Kabbara N, Peffault de Latour R et al. Mutant TIRAP gene polymorphism is associated with outcomes in HLA genoidentical bone marrow transplants recipients with leukemia. Blood ASH Ann Meet Abstracts 2007 abstract 327.
  23. Hoffmeyer S, Burk O, von Richter O, Arnold HP, Brockmoeller J, Johne A et al. Functional polymorphisms of the human multidrug-resistance gene: multiple sequence variations and correlation of one allele with P-glycoprotein expression and activity in vivo. Proc Natl Acad Sci USA 2000; 97: 3473–3478. | Article | PubMed | CAS |
  24. Kim RB, Leake BF, Choo EF, Dresser GK, Kubba SV, Schwarz UI et al. Identification of functionally variant MDR1 alleles among European Americans and African Americans. Clin Pharmacol Ther 2001; 70: 189–199. | Article | PubMed | ISI | CAS |
  25. Kim DH, Park JY, Sohn SK, Lee NY, Suh JS, Lee KB. The association between multidrug resistance-1 gene polymorphisms and outcomes of allogeneic HLA-identical stem cell transplantation. Haematologica 2006; 91: 848–851. | PubMed |
  26. Holler E, Rogler G, Herfarth H, Brenmoehl J, Wild PJ, Hahn J et al. Both donor and recipient NOD2/CARD15 mutations associate with transplant-related mortality and GvHD following allogeneic stem cell transplantation. Blood 2004; 104: 889–894. | Article | PubMed | ISI | CAS |
  27. Klein W, Tromm A, Griga T, Fricke H, Folwaczny C, Hocke M et al. A polymorphism in the CD14 gene is associated with Crohn disease. Scand J Gastroenterol 2002; 37: 189–191. | Article | PubMed | ISI | CAS |
  28. Stevens A, Ray DW, Zeggini E, John S, Richards HL, Griffiths CE et al. Glucocorticoid sensitivity is determined by a specific glucocorticoid receptor haplotype. J Clin Endocrinol Metab 2004; 89: 892–897. | Article | PubMed | ISI | CAS |
  29. Derijk RH, Schaaf MJ, Turner G, Datson NA, Vreugdenhil E, Cidlowski J et al. A human glucocorticoid receptor gene variant that increases the stability of the glucocorticoid receptor beta-isoform mRNA is associated with rheumatoid arthritis. J Rheumatol 2001; 28: 2383–2388. | PubMed | ISI | CAS |
  30. Bogunia-Kubik K, Lange A. HSP70-hom gene polymorphism in allogeneic hematopoietic stem-cell transplant rescipients correlates with the development of acute graft-versus-host disease. Transplantation 2005; 79: 815–820. | Article | PubMed |
  31. Cargill M, Schrodi SJ, Chang M, Garcia VE, Brandon R, Callis KP et al. A large-scale genetic association study confirms IL12B and leads to the identification of IL23R as psoriasis-risk genes. Am J Hum Genet 2007; 80: 273–290. | Article | PubMed | ISI | CAS |
  32. Novota P, Zinocker S, Norden J, Wang WN, Sviland L, Opitz L et al. Expression profiling of major histocompatibility and natural killer complex genes reveals candidates for controlling risk of graft versus host. PLoS One 2011; 6: e16582. | Article | PubMed |
  33. Park KH, Fridaley BL, Ryu E, Tosakulwong N, Edwards AO. Complement component 3 (C3) haplotypes and risk of age-related macular degeneration. Invest Ophthalmol Vis Sci 2009; 50: 3386–3393. | Article | PubMed |
  34. Wang L, Yanuck D, Beecham A, Gardener H, Slifer S, Blanton SH et al. A candidate gene study revealed sex-specific association between the OLR1 gene and carotid plaque. Stroke 2011; 42: 588–592. | Article | PubMed |
  35. Chalmers KA, Barker R, Passmore PA, Panza F, Seripa D, Solfrizzi V et al. LRP-1 variation is not associated with risk of Alzheimer’s disease. Int J Mol Epidemiol Genet 2010; 1: 104–113. | PubMed |
  36. Labopin M, Iacobelli S. Statistical guidelines for EBMT. The European Group for Blood and Marrow Transplantation. Available at https://portal.ebmt.org/sites/clint2/clint/Documents/StatGuidelines_oct2003.pdf accessed on 12 August 2011.
  37. Parmar M, Machin D. Survival Analysis. A practical approach. J Wiley & Sons: Chichester, UK, 1995.
  38. Harrell FE Jr., Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating the yield of medical tests. JAMA 1982; 247: 2543–2546. | Article | PubMed | ISI |
  39. Harrell FE Jr., Lee KL, Califf RM, Pryor DB, Rosati RA. Regression modelling strategies for improved prognostic prediction. Stat Med 1984; 3: 143–152. | Article | PubMed | ISI |
  40. Harrell F and with contributions from many other users. Hmisc: Harrell miscellaneous; rcorrp.cens, method=2. R package version 3.5-2; 2008.
  41. Balavarca Y. Assessing prediction error of genetic variants in Cox regression models, Doctoral thesis, available at http://webdoc.sub.gwdg.de/diss/2012/balavarca/ Accessed April 30, 2012.
  42. Efron B. Estimating the error rate of a prediction rule: improvement on cross-validation. J Am Stat Assoc 1983; 78: 316–331. | Article |
  43. Schoenfeld D. Partial residuals for the proportional hazards regression model. Biometrika 1982; 69: 239–241. | Article | ISI |
  44. O’Quigley J, Xu R. Explained variation in Cox regression. Handbook of Statistics in Clinical Oncology. Marcel Dekker: NewYork, NY, USA, 2001, pp 397–410.
  45. Turner DM, Williams DM, Sankaran D, Lazarus M, Sinnot PJ, Hutchinson IV. An investigation of polymorphism in the interleukin-10 gene promoter. Eur J Immunogenet 1997; 24: 1–8. | Article | PubMed | ISI | CAS |
  46. Seo KW, Kim DH, Sohn SK, Lee NY, Chang HH, Kim SW et al. Protective role of interleukin-10 promoter gene polymorphisms in the pathogenesis of invasive pulmonary aspergillosis after allogeneic stem cell transplantation. Bone Marrow Transplant 2005; 36: 1089–1095. | Article | PubMed | ISI |
  47. Lin MT, Storer B, Martin PJ, Tseng LH, Gooley T, Chen PJ et al. Relation of an interleukin-10 promoter polymorphism to graft-versus-host disease and survival after hematopoietic-cell transplantation. New Engl J Med 2003; 349: 2201–2210. | Article | PubMed |
  48. Hansen JA, Chien JW, Warren EH, Zhao LP, Martin PJ. Defining genetic risk for GVHD and mortality following allogeneic hematopoetic stem cell transplantation. Curr Opin Hematol 2010; 17: 483–492. | Article | PubMed |
  49. Rocha V, Franco RF, Porcher R, Bittencourt H, Silva WA Jr., Latouche A et al. Host defense and inflammatory gene polymorphisms are associated with outcomes after HLA-identical sibling bone marrow transplantation. Blood 2002; 100: 3908–3918. | Article | PubMed | ISI | CAS |
  50. Holtick U, Norden J, Jackson GJ, Holler E, Hromadnikova I, Sedlacek P et al. The TLR adapter Mal/TIRAP protein variant leu 180 influences graft-versus-leukaemia and graft-versus-host reactions after allogeneic haematopoietic stem cell transplantation. 34th Annual Meeting of the European Group for Blood and Marrow Transplantation 2008; 41(Suppl.1): S217.
  51. Alonso MA, Weissman SM. cDNA cloning of MAL, a hydrophobic protein associated with human T-cell differentiation. Proc Natl Acad Sci USA 1987; 84: 1997–2001. | Article | PubMed |
  52. Liebert M, Hubbel A, Chung M, Wedemeyer G, Lomax MI, Hegeman A et al. Expression of mal is associated with urothelial differentiation in vitro: identification by differential display reverse-transcriptase polymerase chain reaction. Differentiation 1997; 61: 177–185. | Article | PubMed |
  53. Wakeman JA, Heath PR, Pearson C, Andrews PW. MAL mRNA is induced during the differentiation on human embryonal carcinoma cells into neurons and is also localised within specific regions of the human brain. Differentiation 1997; 62: 97–105. | Article | PubMed |
  54. Mimori K, Nishida K, Nakamura Y, Ieta K, Yoshikawa Y, Sasaki A et al. Loss of MAL expression in precancerous lesions of the esophagus. Ann Surg Oncol 2007; 14: 1670–1677. | Article | PubMed |
  55. Rancano C, Rubio T, Correas I, Alonso M. Genomic structure and subcellular localisation of MAL, a human T-cell-specific proteolipid protein. J Biol Chem 1994; 269: 8159–8164. | PubMed | CAS |
  56. Akira S, Takeda K. Toll-like receptor signalling. Nat Rev Immunol 2004; 4: 499–511. | Article | PubMed | ISI | CAS |
  57. Girasole G, Jilka RL, Passeri G, Boswell S, Boder G, Williams DC et al. 17 beta-estradiol inhibits interleukin-6 production by bone marrow-derived stromal cells and osteoblasts in vitro: a potential mechanism for the antiosteoporotic effect of estrogens. J Clin Invest 1992; 89: 883–891. | Article | PubMed | ISI | CAS |
  58. Fishman D, Faulds G, Jeffery R, Mohamed-Ali V, Yudkin JS, Humphries S et al. The effect of novel polymorphisms in the interleukin-6 (IL-6) gene on IL-6 transcription and plasma IL-6 levels, and an association with systemic-onset juvenile chronic arthritis. J Clin invest 1998; 102: 1369–1376. | Article | PubMed | ISI | CAS |
  59. Boiardi L, Casali B, Farnetti E, Pipitone N, Nicoli D, Cantini F et al. Relationship between interleukin 6 promoter polymorphism at position -174, IL-6 serum levels, and the risk of relapse/recurrence in polymyalgia rheumatica. J Rheumatol 2006; 33: 703–708. | PubMed |
  60. Xiao B, Wang Y, Li W, Baker M, Guo J, Corbet K et al. Plasma microRNA signature as a noninvasive biomarker for acute graft-versus-host disease. Blood 2013; 122: 3365–3375. | Article | PubMed | ISI | CAS |


This work was supported by the Marie Curie Research Training Network (MCRTN) grant CT-2004-512253: TRANSNET (European Commission), grant LSHB-CT-2007-037703: Stemdiagnostics (European Commission), the German Research Foundation (DFG), grant GRK 1034, the Marie Curie Initial Training Network (MCITN) grant 315963: CELLEUROPE, and the Deutsche José Carreras Leukämie-Stiftung. We thank Andrew Entwistle for his support in reviewing and proofreading the manuscript.

Supplementary Information accompanies this paper on Bone Marrow Transplantation website