MAJOR ARTICLE. CMV; donor; recipient; serological status; stem cell transplant.

Transcription

1 MAJOR ARTICLE Donor Cytomegalovirus Status Influences the Outcome of Allogeneic Stem Cell Transplant: A Study by the European Group for Blood and Marrow Transplantation er Ljungman, 1 Ronald Brand, 2 Jennifer Hoek, 2 Rafael de la Camara, 3 Catherine Cordonnier, 4 Hermann Einsele, 5 Jan Styczynski, 6 Katherine N. Ward, 7 and Simone Cesaro 8 ; for the Infectious Diseases Working arty of the European Group for Blood and Marrow Transplantation 1 Department of Hematology, Karolinska University Hospital, and Division of Hematology, Department of Medicine Huddinge, Karolinska Institutet, Stockholm, Sweden; 2 Department of Medical Statistics and Bioinformatics, Leiden University Medical Centre, The Netherlands; 3 Department of Hematology, Hospital de la rincesa, Madrid, Spain; 4 Sve d Hematologie, Hôpital Henri Mondor, Creteil, France; 5 Med. Klinik und oliklinik II, Universitätsklinikum Würzburg, Germany; 6 ediatric Hematology and Oncology, Collegium Medicum UMK, University Hospital, Bydgoszcz, oland; 7 Division of Infection and Immunity, University College London, United Kingdom; and 8 aediatric Haematology Oncology, oliclinico G.B. Rossi, Verona, Italy (See the Editorial Commentary by Snydman on pages ) Background. The use of a cytomegalovirus (CMV) seronegative donor for a CMV-seronegative allogeneic hematopoietic stem cell transplant (HSCT) recipient is generally accepted. However, the importance of donor serostatus in CMV-seropositive patients is controversial. Methods. A total of HSCT patients, seropositive and seronegative, were identified from the European Group for Blood and Marrow Transplantation database. Cox multivariate models were fitted to estimate the effect of donor CMV serological status on outcome. Results. Seronegative patients receiving seropositive unrelated-donor grafts had decreased overall (hazard ratio [HR], 1.13; 95% confidence interval [CI], ; <.0001) compared with seronegative donors, whereas no difference was seen in patients receiving HLA-matched sibling grafts. Seropositive patients receiving grafts from seropositive unrelated donors had improved overall (HR, 0.92; 95% CI,.86.98; <.01) compared with seronegative donors, if they had received myeloablative conditioning. This effect was absent when they received reduced-intensity conditioning. No effect was seen in patients grafted from HLA-identical sibling donors. The same association was found if the study was limited to patients receiving transplants from the year 2000 onward. Conclusions. We confirm the negative impact on overall if a CMV-seropositive unrelated donor is selected for a CMV-seronegative patient. For a CMV-seropositive patient, our data support selecting a CMVseropositive donor if the patient receives a myeloablative conditioning regimen. Keywords. CMV; donor; recipient; serological status; stem cell transplant. Received 10 November 2013; accepted 13 April 2014; electronically published 20 May Correspondence: er Ljungman, MD, hd. Department of Hematology, Karolinska University Hospital, SE Stockholm, Sweden (per.ljungman@ki.se). Clinical Infectious Diseases 2014;59(4): The Author ublished by Oxford University ress on behalf of the Infectious Diseases Society of America. All rights reserved. For ermissions, please journals.permissions@oup.com. DOI: /cid/ciu364 Cytomegalovirus (CMV) has for many years been an important cause of morbidity and mortality after allogeneic hematopoietic stem cell transplant (HSCT). During recent years, major advances have been reached regarding antiviral prophylactic strategies [1 3], and new and more sensitive diagnostic techniques have Donor CMV Status and Outcome After HSCT CID 2014:59 (15 August) 473

2 been developed, allowing better monitoring of patients and early intervention with antiviral therapy [4 6]. However, despite these advances, patients CMV serological status still has a strong influence on outcome [7]. It has been shown that the use of a CMV-seronegative donor for a CMV-seronegative patient reduces the risk of transplant-related mortality, especially that caused by infections [8]. The influence of the donor s serological status on outcome after transplant of a CMV-seropositive patient is more controversial [9 11]. In a previous study from the Infectious Diseases Working arty of the European Group for Blood and Marrow Transplantation (EBMT), we found that using a CMV-seropositive donor improved outcome for a CMVseropositive patient receiving an unrelated donor graft [12]. During the last decade, new transplant technologies have been introduced, such as improved human leukocyte antigen (HLA) typing allowing selection of better-matched unrelated donors and reduced-intensity conditioning (RIC). The aim of this analysis was to analyze the influence of donor CMV serological status on outcome after allogeneic HSCT in a large population of CMV-seropositive and CMV-seronegative patients. ATIENTS AND METHODS The EBMT has collected data on patients since the 1970s. For the purpose of this study, we included patients having undergone allogeneic HSCT between 1992 and 2008 and for whom CMV serological status was known for both patient and donor. The EBMT megafileusedforthisanalysisincluded patients who fulfilled these criteria. atients who had undergone acord blood transplant and transplant for plasma cell disorders or solid tumors were excluded from the analysis. After these exclusions, patients remained in the study population; were seropositive and seronegative. A total of seropositive patients had HLA-identical sibling donors, had unrelated donors, and 2248 had mismatched family donors. The corresponding numbers for seronegative patients were patients with HLA-identical sibling donors, 8801 with unrelated donors, and 1304 with mismatched family donors. atient characteristics of the study population are shown in Table 1. and RIC regimens were defined according to guidelines for reporting to the EBMT registry ( According to EBMT policy, patients give informed consent for data reporting to the EBMT registry. Statistical Analysis CMV serological status of the recipient was used as the main stratification variable; all analyses were done separately for seropositive and seronegative patients. The data analysis approach is the Cox proportional hazards framework. Instead of adjusting for the type of donor (HLA-identical siblings, mismatched family, or unrelated), the final models were stratified by donor type to allow for nonproportionality of the curves as well as a simpler interpretation of the coefficients, in which case interaction with donor type does not apply. The following factors were included in the analyses: calendar year of HSCT (continuous), patient age (continuous), diagnosis (acute leukemia, chronic leukemia, lymphoma, myelodysplastic syndrome or myeloproliferative disorder, and nonmalignant disease), stem cell source, country where the transplant was performed, patient and donor sex (as well as their interaction), conditioning intensity (according to the EBMT definitions), and use of T-cell depletion. We have previously shown that the CMV serological status of the donor and the recipient varied significantly between different parts of Europe [13]. The variable country was therefore divided as follows: the Netherlands plus Belgium, France, Spain, Italy, United Kingdom, the Nordic countries, Germany, and other countries. The T-cell depletion variable was categorized as no T-cell depletion, in vivo (use of anti-thymocyte globulin or alemtuzumab), or ex vivo (with or without addition of in vivo T-cell depletion). All conclusions are based on final models with patient age but without donor age due to the loss of power caused by a high percentage of missing donor age values. However, to exclude substantial bias due to donor age, we also fit the data to include this parameter so as to verify (taking into account the wider confidence intervals [CIs] due to the reduced sample size) our results. Cox models were constructed, stratified by donor type, from which the estimated effects were quantified as a hazard ratio (HR) to test the association between CMV positivity of the donor with overall, relapse incidence, relapse-free, and nonrelapse mortality. Each model contained interaction terms for donor positivity with T-cell depletion and conditioning intensity. Backward selection of these interaction terms retained only terms with a value <.10. In such a case, separate HRs are reported as well as the value of the interaction term. Although the HRs are all based on the Cox model framework, construction of curves was performed properly in the competing risk framework, leading to simultaneous estimates for nonrelapse mortality, death after relapse, alive with relapse, and alive without relapse (and thus of overall and relapse-free by combining some of these estimates). Univariate and bivariate curves are presented but no statistical tests were performed to compare the curves themselves in a model approach (eg, Fine and Gray). We also compared the crude overall of those who entered the analysis models and those who did not. We did apply the patient population selection criteria (eg, on year and diagnosis) but retained the patients with missing covariates and/or missing donor serology for a comparison (Supplementary Figures 1A and 1B). For the analysis of causes of death, a logistic regression approach was used, conditioning on death without relapse between transplant and 60 months follow-up, thus avoiding the 474 CID 2014:59 (15 August) Ljungman et al

3 Table 1. atient Characteristics CMV-Negative atients CMV-ositive atients Characteristic HLA-Identical Sibling Donor (n = ) Mismatched Family Donor (n = 1304) Unrelated Donor (n = 8801) HLA-Identical Sibling Donor (n = ) Mismatched Family Donor (n = 2248) Unrelated Donor (n = ) Age, y, mean (SD) 33.6 (17.4) 20.8 (18.3) 31.1 (18.6) 36.9 (17.0) 27.1 (18.6) 37.6 (18.6) Donor Seronegative 7249 (71.9%) 778 (59.7%) 6309 (71.1%) 4179 (25.0%) 530 (23.6%) 5255 (50.6%) Seropositive 2839 (28.1%) 526 (40.3%) 2492 (28.3%) (75.0%) 1718 (76.4%) 5136 (49.4%) Stem cell source Bone marrow 5372 (53.3%) 601 (46.1%) 4474 (50.8%) 7545 (45.2%) 814 (36.2%) 4334 (41.7%) eripheral blood 4716 (46.7%) 703 (53.9%) 4327 (49.2%) 9165 (54.8%) 1434 (63.8%) 6057 (58.3%) Diagnosis Acute leukemia 5103 (50.6%) 667 (51.1%) 4503 (51.2%) 8890 (53.2%) 1325 (58.9%) 5568 (53.6%) Chronic 1565 (15.5%) 122 (9.3%) 1425 (16.2%) 2425 (14.5%) 209 (9.3%) 1495 (14.4%) leukemia Lymphoma 1417 (14.0%) 95 (7.3%) 943 (10.7%) 1865 (11.1%) 172 (7.6%) 1121 (10.8%) MDS/MN 1031 (10.2%) 107 (8.2%) 1153 (13.1%) 1681 (10.1%) 185 (8.2%) 1456 (14.0%) Nonmalignant 972 (9.6%) 313 (24.0%) 777 (8.8%) 1849 (11.1%) 357 (15.9%) 751 (7.2%) disorders regimen 7903 (78.3%) 1151 (88.3%) 6567 (74.6%) (75.5%) 1914 (85.1%) 7101 (68.3%) Reduced 2185 (21.7%) 153 (11.7%) 2234 (25.4%) 4088 (24.5%) 334 (14.9%) 3290 (31.7%) intensity T-cell depletion None 7213 (71.5%) 426 (32.7%) 2745 (31.2%) (74.2%) 687 (30.6%) 3151 (30.3%) In vivo 1699 (16.8%) 257 (19.7%) 4839 (55.0%) 2799 (16.7%) 498 (22.1%) 6330 (60.9%) Ex vivo (with or without in vivo) 1176 (11.7%) 621 (47.6%) 1217 (13.8%) 1513 (9.1%) 1063 (47.3%) 910 (8.8%) Data are presented as No. (%) unless otherwise specified. Abbreviations: CMV, cytomegalovirus; HLA, human leukocyte antigen; MDS, myelodysplastic syndrome; MN, myeloproliferative neoplasm; SD, standard deviation. notion of censoring. A hierarchical definition of cause of death was applied: any death preceded by a relapse was classified as relapse ; if not a relapse but a secondary malignancy was coded, the cause was classified as such; if neither happened but the death was explicitly coded as being caused by graftvs-host disease, that cause was taken; if none of these were observed and an infection was explicitly mentioned as the cause of death, then the cause was classified as infection. If none of these applied but the cause of death was coded as rejection/ poor graft function, veno-occlusive disease, or hemorrhage, we used organ failure as the cause of death; finally, all unclassified deaths were coded as other cause. Association between CMV donor positivity and cause of death was assessed by a simple χ 2 analysis in a 2 6 table. To study death due to infection in detail, the subset of patients was selected, who were all classified as cause of death = infection. Conditional on having died due to infection, the following analyses were performed to assess possible associations between CMV donor positivity and the type of infection: bacterial, viral, or fungal. For each patient, we used 2 approaches: (1) an infection (bacterium or virus or fungus) given as the only cause of death vs all other deaths due to infection (possibly including the same pathogen); (2) the pathogen type under investigation given as a cause of death either as single infection or as a part of a multipathogen infection vs death due to infection but not specifying the pathogen(s). Both approaches lead to the quantification of the association as an odds ratio in a 2 2 table (likelihood ratio test). RESULTS CMV-Seronegative atients HLA-Identical Sibling Donors and Mismatched Family Donors The estimated 5-year (Kaplan-Meier) after HLAidentical sibling HSCT was 54% for patients with CMVseropositive donors and 59% for patients with CMV-seronegative Donor CMV Status and Outcome After HSCT CID 2014:59 (15 August) 475

4 Table 2. Impact of Using a Cytomegalovirus (CMV) Seropositive Donor for Survival, Transplant-Related Mortality, Nonrelapse Mortality, and Relapse in CMV-Seronegative and CMV- Seropositive atients Receiving Sibling or Unrelated Donor Transplants Adjusted for Covariates Sibling Donor Unrelated Donor Impact HR 95% CI Value HR 95% CI Value Figure 1. Kaplan-Meier estimates of overall in cytomegalovirus (CMV) seronegative patients undergoing hematopoietic stem cell transplant with CMV-seropositive (green) or seronegative (blue) donors. A, Human leukocyte antigen identical sibling donors. B, Unrelated donors. Abbreviation: HSCT, hematopoietic stem cell transplant. donors (Figure 1A). In a multivariate Cox regression model, donor CMV serological status had a borderline significant effect on overall after HLA-identical sibling HSCT (HR, 1.07 [95% CI, ]; =.06) after adjusting for calendar year of HSCT, donor sex, patient age, diagnosis, stem cell source, country where the transplant was performed, patient sex, conditioning intensity, and use of T-cell depletion (in vivo or ex vivo) (Table 2). The inclusion of donor age in the model reduced the HR slightly (data not shown). There was no significant effect of the donor CMV serological status on relapse-free, nonrelapse mortality, or relapse incidence (Table 2). There was no effect on outcome of CMV serological status in patients transplanted from mismatched family donors. The estimated 5-year (Kaplan-Meier) was 44% for patients with CMV-seropositive donors and 47% for patients with CMV-seronegative donors (data not shown). CMV-seronegative patients Overall <.0001 Relapse-free Nonrelapse mortality Relapse incidence CMV-seropositive patients Overall Relapse-free Nonrelapse mortality Relapse incidence Abbreviations: CI, confidence interval; CMV, cytomegalovirus; HR, hazard ratio. Unrelated Donors The estimated 5-year (Kaplan-Meier) was 45% for patients with CMV-seropositive donors and 50% for patients with CMV-seronegative donors (Figure 1B). CMV-seronegative patients receiving grafts from CMV-seropositive donors had in a multivariate Cox regression model significantly lower overall (HR, 1.13 [95% CI, ]; <.0001) compared with CMV-seronegative patients receiving grafts from CMVseronegative donors (Table 2). This difference in was also seen with the inclusion of donor age in the model (data not shown). There was also a decreased relapse-free (HR, 1.10 [95% CI, ]; <.003) and an increased nonrelapse mortality (HR, 1.14 [95% CI, ]; <.01) in patients receiving grafts from CMV-seropositive donors compared with patients receiving grafts from CMV-seronegative donors. The same effect was seen also with donor age in the model, although the HR was somewhat reduced. Furthermore, when we restricted the sample to patients transplanted from the year 2000 onward (n = ) to get a sample more consistent with today s practice, the same negative effects of using a CMVseropositive unrelated donor were consistently observed. Causes of Death A total of 2697 of 5857 (46.0%) patients died in the CMVseropositive donor group and 5955 of (41.5%) died in 476 CID 2014:59 (15 August) Ljungman et al

5 the CMV-seronegative donor group. Of the 2697 patients with a seropositive donor, 1524 (57%) patients died of transplantrelated causes and 1173 (43%) from the underlying disease in the CMV-seropositive donor group. The corresponding numbers in the CMV-seronegative donor group were 3243 (54%) of transplant-related causes and 2712 (46%) of the underlying disease. The distribution of causes of nonrelapse mortality between patients with CMV-seropositive or CMV-seronegative donors was similar (data not shown). Although more patients died from viral infections in the CMV-seropositive donor group, this was not significant (5.8% vs 4.9%; =.21). CMV-Seropositive atients HLA-Identical Sibling or Mismatched Family Donors DonorCMVserologicalstatushadnosignificant effect on overall, relapse-free, nonrelapse mortality, or Figure 2. Kaplan-Meier estimates of overall in cytomegalovirus (CMV) seropositive patients undergoing hematopoietic stem cell transplant with CMV-seropositive (green) or seronegative (blue) unrelated donors. A, conditioning. B, Reduced-intensity conditioning. Abbreviation: HSCT, hematopoietic stem cell transplant. relapse incidence (Table 2). The estimated 5-year overall (Kaplan-Meier) after HLA-identical sibling HSCT was 52% both for patients with CMV-seropositive donors and those with CMV-seronegative donors. The corresponding Kaplan- Meier estimates were 39% and 34% for patients with HLA-mismatched family donors (not significant in univariate or multivariate analysis). There was no difference when donor age was included in the models (data not shown). Unrelated Donors The estimated 5-year (Kaplan-Meier) after unrelated donor HSCT was 43% for patients with CMV-seropositive donors and 41% for patients with CMV-seronegative donors. When simply adjusting for conditioning intensity in the Cox model (ie, averaging the effects over both intensity categories), CMV-seropositive patients receiving grafts from CMVseropositive donors had a nonsignificant (but beneficial) effect on overall (HR, 0.96 [95% CI, ]; =.11; Table 2) and a significantly beneficial effect on nonrelapse mortality (HR, 0.94 [95% CI, ]; =.05) compared with CMVseropositive patients receiving grafts from CMV-seronegative donors. The HR for overall decreased slightly more after addition of donorage. Ex vivo T-cell depletion had a negative impact on overall ( <.01) and relapse-free ( <.01). We found, however, a significant interaction between donor CMV serological status and conditioning intensity showing that merely adjusting for such intensity is an inappropriate and biased assessment of the CMV effect (which then depends on the actual relative contribution of RIC to the total number of transplants). We therefore estimated the effect separately in patients having received myeloablative or RIC regimens. The 5-yearestimatewas44%forpatientswithCMV- Table 3. Impact of Using a Cytomegalovirus (CMV) Seropositive Donor for Survival, Transplant-Related Mortality, and Relapse in CMV-Seropositive atients Receiving Unrelated Donor Transplants, Split by Regimen Intensity, Adjusted for Covariates Impact Overall Relapse-free Nonrelapse mortality Relapse incidence HR 95% CI Value HR 95% CI Value Abbreviations: CI, confidence interval; HR, hazard ratio. Donor CMV Status and Outcome After HSCT CID 2014:59 (15 August) 477

6 Table 4. Univariate and Multivariate Analysis of Risk Factors for Overall Survival and Nonrelapse Mortality in Cytomegalovirus- Seropositive atients Receiving Grafts From Unrelated Donors Univariate Analysis Multivariate Analysis Factor HR 95% CI Value HR 95% CI Value HR 95% CI Value HR 95% CI Value Overall CMV donor status: seropositive Year of HSCT < < < <.0001 atient age < < < <.0001 Donor sex: male atient sex: male Diagnosis Acute leukemia 1.00 < < < Chronic leukemia Lymphoma MDS/MN Nonmalignant Stem cell source: peripheral blood stem cells Country Other 1.00 < < Netherlands/ Belgium France Spain Italy United Kingdom Nordic countries Germany T-cell depletion None 1.00 < In vivo Ex vivo Nonrelapse mortality CMV donor status: < seropositive Year of HSCT < < <.0001 atient age < < Donor sex: male atient sex: male Diagnosis Acute leukemia < < Chronic leukemia Lymphoma MDS/MN Non-malignant Stem cell source: peripheral blood stem cells CID 2014:59 (15 August) Ljungman et al

7 Table 4 continued. Univariate Analysis Multivariate Analysis Factor HR 95% CI Value HR 95% CI Value HR 95% CI Value HR 95% CI Value Country Other < < Netherlands/ Belgium France Spain Italy United Kingdom Nordic countries Germany T-cell depletion None < In vivo Ex vivo Abbreviations: CI, confidence interval; CMV, cytomegalovirus; HR, hazard ratio; HSCT, hematopoietic stem cell transplant; MDS/MN, myelodysplastic syndrome/ myeloproliferative neoplasm. seropositive donors and 41% for patients with CMV-seronegative donors who had undergone myeloablative conditioning (Figure 2A). The corresponding estimates in patients who had undergone HSCT with RIC were 40% and 39%, respectively (Figure 2B). In multivariate modeling, CMVseropositive patients with seropositive donors who received myeloablative conditioning had significantly improved overall (HR, 0.91 [95% CI,.86.97]; <.01) and a decreased nonrelapse mortality (HR, 0.88 [95% CI,.81.95]; <.01). A summary of the effects is shown in Table 3, and the univariate and multivariate HR estimates for overall and nonrelapse mortality are shown in Table 4. A positive effect on overall was also seen when the analysis was restricted to patients with myeloablative conditioning transplanted from the year 2000 onward (n = ; HR, 0.91 [95% CI,.85.98]; =.02). In contrast, no significant effect of donor CMV status was seen in patients receiving RIC regimens, although the HRs were between 1.05 and Causes of Death A total of 9024 of (46.6%) patients died in the CMVseropositive donor group, and 492 of 9964 (49.4%) died in the CMV-seronegative donor group. In the CMV-seropositive donor group, 5109 (57%) patients died of transplant-related causes; in the CMV-seropositive donor group, 3915 (43%) died from the underlying disease. The corresponding numbers in the CMV-seronegative donor group were 2915 (59%) of transplantrelated causes and 2011 (41%) of the underlying disease. There were differences in the causes of death between patients with CMV-seropositive or CMV-seronegative donors. atients who received grafts from CMV-seronegative donors were more likely to die from viral infection given as the only cause of death (6.2% vs 4.7%; =.005) or where a viral pathogen was included as a part of a mixed infection with different organisms (8.3% vs 6.4%; <.01). This increase in risk for death from viral infections was found in all 3 donor groups, although it was significantly different only in patients receiving unrelated donor grafts (data not shown). There was no difference in the rates of patients dying from bacterial or fungal infections (data not shown). DISCUSSION CMV is one of the major causes of transplant-related mortality in patients undergoing allogeneic HSCT [14 16]. One of the key factors for outcome has been patient CMV serological status, especially in those receiving unrelated-donor transplants [7, 9, 17 21]. CMV-seropositive patients have a poorer outcome than CMV-seronegative patients despite improvement in preventive strategies against CMV disease such as antiviral prophylaxis and preemptive therapy [7, 19]. The effect of donor serostatus in CMV-seropositive patients remains a controversial issue. Donor CMV Status and Outcome After HSCT CID 2014:59 (15 August) 479

8 CMV can be transmitted from a seropositive donor to a seronegative patient and therefore the importance of selecting a CMV-seronegative donor for a CMV-seronegative patient has been recognized for a long time [22] and was confirmed in a recent study from the EBMT analyzing outcome of patients with acute leukemia [21], but the impact of different donor categories was not analyzed. In our study, we were only able to verify this negative impact on overall in patients receiving grafts from unrelated donors. This was also true when the analysis was restricted to patients transplanted from the year 2000 onward. In sibling donors, there were similar tendencies, but donor CMV serological status was not an independent risk factor for outcome. This is of practical importance as the option to select a matched sibling donor on CMV serological status rarely exists. It should be noted that there was a negative impact of donor age in the mismatched family donor subgroup; because haploidentical family donors are being increasingly used [23, 24], this might be important when selecting the best donors when different options exist. In our study, there was an increase in the proportion of patients dying from transplant-related causes if the donor was CMV seropositive vs CMV seronegative. This might have been explained by the observation of Nicholsetalthatthemortalityfromfungalandbacterial infections was increased in this donor/recipient combination [8]. However, we were unable to show that these particular causes of death were more common in CMV-seronegative patients receiving CMV-seropositive donor grafts. However, it should be recognized that a retrospective analysis of the causes of death reported from many centers should be interpreted with caution. Although the importance of the negative effect of a CMVmismatched donor for CMV-seronegative patients has been generally accepted, the influence of a similar mismatch in a CMV-seropositive patient has remained controversial. Kollman et al, in a large study from the National Marrow Donor rogram, reported that increasing bone marrow donor age was a significant risk factor for decreased, but no effect of CMV donor serological status could be found [11]. In our previous study, we detected a strong influence of donor CMV seropositivity on outcome in unrelated donor transplants [12]. atients receiving grafts from CMV-seropositive unrelated donors had improved, event-free, and reduced nonrelapse mortality. These effects were also abrogated by T-cell depletion. Since these 2 studies were published, several observations have been made regarding the negative effect of CMV-seronegative donors for CMV-seropositive patients on directly CMV-associated outcomes; in particular, delayed CMV-specific immune reconstitution [25 29], repeated CMV reactivations [26, 30], requirement of additional courses of antiviral therapy [29], late CMV recurrence [31], and development of CMV disease [26, 32, 33]. In the current analysis, we show that using a CMV-seronegative unrelated donor for a CMV-seropositive patient did not have an effect on overall (HR, 0.96 [95% CI, ]; =.11). However, there was a strong interaction between the CMV serological status of the donor and conditioning intensity. When accounting for this interaction by estimating the CMV donor effect of type of conditioning intensity, we found something new; namely, that there was a protective effect of donor seropositivity resulting in improved overall in patients having received myeloablative conditioning, whereas there was no such effect in patients receiving RIC (the discrepancy between the 2 effects being significant in itself). The effect is most likely mediated through retention of CMV-specific recipient T-cell function in patients receiving RIC. This fits well with previous observations that RIC is associated with better early immune reconstitution and a lower risk for CMV infection early after HSCT [34] and a lower risk for high-grade CMV infection [35]. In contrast to our previous study, we were unable to see an effect of in vivo or in vitro T-cell depletion on the impact of CMV serological status on overall or nonrelapse mortality in patients receiving either myeloablative or reduced-intensity conditioning. In addition, CMV-seropositive patients receiving grafts from CMV-seronegative donors were more likely to die from viral infections. Although the EBMT registry does not make provision for distinction between different viral infections as the cause of death, the finding suggests that the presence of CMV-specific T cells is the key factor mediating the protective effect of donor serological status on overall. Supplementary Data Supplementary materials are available at Clinical Infectious Diseases online ( Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Notes Acknowledgments. We are grateful to Anja van Biezen at the European Group for Blood and Marrow Transplantation (EBMT) Leiden office for helping with the data selection from the EBMT database. We are also grateful to all centers providing data to the EBMT registry (the main participating centers are shown in the Supplementary Appendix. Author contributions.. L. designed the study, interpreted the data, and wrote the manuscript; R. B. designed the study, performed the statistical analysis, interpreted the data, and reviewed and approved the manuscript; J. H. worked with the database and reviewed and approved the manuscript; R. C., C. C., H. E., J. S., K. N. W., and S. C. interpreted the data and reviewed and approved the manuscript. Financial support. The work was supported by the Stockholm County Council and the Karolinska Institutet (funds to. L.). otential conflicts of interest. All authors: No reported conflicts. 480 CID 2014:59 (15 August) Ljungman et al

9 All authors have submitted the ICMJE Form for Disclosure of otential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed. References 1. Goodrich J, Bowden R, Fisher L, Keller C, Schoch G, Meyers J. Ganciclovir prophylaxis to prevent cytomegalovirus disease after allogeneic marrow transplant. Ann Intern Med 1993; 118: rentice HG, Gluckman E, owles RL, et al. Impact of long-term acyclovir on cytomegalovirus infection and after allogeneic bone marrow transplantation. European Acyclovir for CMV rophylaxis Study Group. Lancet 1994; 343: Winston DJ, Ho WG, Bartoni K, et al. Ganciclovir prophylaxis of cytomegalovirus infection and disease in allogeneic bone marrow transplant recipients. Results of a placebo- controlled, double-blind trial. Ann Intern Med 1993; 118: Einsele H, Ehninger G, Hebart H, et al. olymerase chain reaction monitoring reduces the incidence of cytomegalovirus disease and the duration and side effects of antiviral therapy after bone marrow transplantation. Blood 1995; 86: Gor D, Sabin C, rentice HG, Vyas N, Man S, Griffiths D. Longitudinal fluctuations in cytomegalovirus load in bone marrow transplant patients: relationship between peak virus load, donor/recipient serostatus acute GVHD and CMV disease. Bone Marrow Transplant 1998; 21: Boeckh M, Bowden RA, Goodrich JM, ettinger M, Meyers JD. Cytomegalovirus antigen detection in peripheral blood leukocytes after allogeneic marrow transplantation. Blood 1992; 80: Broers AE, van Der Holt R, van Esser JW, et al. Increased transplantrelated morbidity and mortality in CMV- seropositive patients despite highly effective prevention of CMV disease after allogeneic T-celldepleted stem cell transplantation. Blood 2000; 95: Nichols WG, Corey L, Gooley T, Davis C, Boeckh M. High risk of death due to bacterial and fungal infection among cytomegalovirus (CMV)- seronegative recipients of stem cell transplants from seropositive donors: evidence for indirect effects of primary CMV infection. J Infect Dis 2002; 185: Meyers JD, Flournoy N, Thomas ED. Risk factors for cytomegalovirus infection after human marrow transplantation. J Infect Dis 1986; 153: Grob J, Grundy JE, rentice HG, et al. Immune donors can protect marrow-transplant recipients from severe cytomegalovirus infections. Lancet 1987; 1: Kollman C, Howe CW, Anasetti C, et al. Donor characteristics as risk factors in recipients after transplantation of bone marrow from unrelated donors: the effect of donor age. Blood 2001; 98: Ljungman, Einsele H, Frassoni F, Niederwieser D, Cordonnier C. Donor CMV serological status influences the outcome of CMV seropositive recipients after unrelated donor stem cell transplantation: an EBMT Megafile analysis. Blood 2003; 102: Ljungman, Brand R. Factors influencing cytomegalovirus seropositivity in stem cell transplant patients and donors. Haematologica 2007; 92: Ljungman, Aschan J, Lewensohn-Fuchs I, et al. Results of different strategies for reducing cytomegalovirus-associated mortality in allogeneic stem cell transplant recipients. Transplantation 1998; 66: Meyers J, Flournoy N, Thomas E. Nonbacterial pneumonia after allogeneic marrow transplantation: review of ten years experience. Rev Infect Dis 1982; 4: Zaia JA, Forman SJ. Cytomegalovirus infection in the bone marrow transplant recipient. Infect Dis Clin North Am 1995; 9: Bacigalupo A, Tedone E, Sanna MA, et al. CMV infections following allogeneic BMT: risk factors, early treatment and correlation with transplant related mortality. Haematologica 1992; 77: Takenaka K, Gondo H, Tanimoto K, et al. Increased incidence of cytomegalovirus (CMV) infection and CMV- associated disease after allogeneic bone marrow transplantation from unrelated donors. The Fukuoka Bone Marrow Transplantation Group. Bone Marrow Transplant 1997; 19: Craddock C, Szydlo RM, Dazzi F, et al. Cytomegalovirus seropositivity adversely influences outcome after T- depleted unrelated donor transplant in patients with chronic myeloid leukaemia: the case for tailored graft-versus-host disease prophylaxis. Br J Haematol 2001; 112: Ljungman, Hakki M, Boeckh M. Cytomegalovirus in hematopoietic stem cell transplant recipients. Hematol Oncol Clin North Am 2011; 25: Schmidt-Hieber M, Labopin M, Beelen D, et al. CMV serostatus has still an important prognostic impact in de novo acute leukemia patients after allogeneic stem cell transplantation: a report from the Acute Leukemia Working arty of EBMT. Blood 2013; 122: Bowden RA, Sayers M, Flournoy N, et al. Cytomegalovirus immune globulin and seronegative blood products to prevent primary cytomegalovirus infection after marrow transplantation. N Engl J Med 1986; 314: Raiola AM, Dominietto A, Ghiso A, et al. Unmanipulated haploidentical bone marrow transplantation and posttransplantation cyclophosphamide for hematologic malignancies after myeloablative conditioning. Biol Blood Marrow Transplant 2013; 19: Luznik L, O Donnell V, Symons HJ, et al. HLA-haploidentical bone marrow transplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biol Blood Marrow Transplant 2008; 14: Avetisyan G, Aschan J, Hagglund H, Ringden O, Ljungman. Evaluation of intervention strategy based on CMV-specific immune responses after allogeneic SCT. Bone Marrow Transplant 2007; 40: Ganepola S, Gentilini C, Hilbers U, et al. atients at high risk for CMV infection and disease show delayed CD8+ T-cell immune recovery after allogeneic stem cell transplantation. Bone Marrow Transplant 2007; 39: Lilleri D, Fornara C, Chiesa A, Caldera D, Alessandrino E, Gerna G. Human cytomegalovirus-specific CD4+ and CD8+ T-cell reconstitution in adult allogeneic hematopoietic stem cell transplant recipients and immune control of viral infection. Haematologica 2008; 93: Moins-Teisserenc H, Busson M, Scieux C, et al. atterns of cytomegalovirus reactivation are associated with distinct evolutive profiles of immune reconstitution after allogeneic hematopoietic stem cell transplantation. J Infect Dis 2008; 198: Zhou W, Longmate J, Lacey SF, et al. Impact of donor CMV-status on viral infection and reconstitution of multi-function CMV-specific T-cells in CMV-positive transplant recipients. Blood 2009; 113: Lin TS, Zahrieh D, Weller E, Alyea E, Antin JH, Soiffer RJ. Risk factors for cytomegalovirus reactivation after CD6+ T-cell-depleted allogeneic bone marrow transplantation. Transplantation 2002; 74: Ozdemir E, Saliba R, Champlin R, et al. Risk factors associated with late cytomegalovirus reactivation after allogeneic stem cell transplantation for hematological malignancies. Bone Marrow Transplant 2007; 40: Ljungman, erez-bercoff L, Jonsson J, et al. Risk factors for the development of cytomegalovirus disease after allogeneic stem cell transplantation. Haematologica 2006; 91: Marty FM, Bryar J, Browne SK, et al. Sirolimus-based graft-versus-host disease prophylaxis protects against cytomegalovirus reactivation after allogeneic hematopoietic stem cell transplantation: a cohort analysis. Blood 2007; 110: Maris M, Boeckh M, Storer B, et al. Immunologic recovery after hematopoietic cell transplantation with nonmyeloablative conditioning. Exp Hematol 2003; 31: Nakamae H, Kirby KA, Sandmaier BM, et al. Effect of conditioning regimen intensity on CMV infection in allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant 2009; 15: Donor CMV Status and Outcome After HSCT CID 2014:59 (15 August) 481