Cytomegalovirus infection in recipients of related and unrelated donor bone marrow transplants: no evidence of increased incidence in patients receiving unrelated donor grafts


Dr Annabel Foot Royal Hospital for Sick Children, St Michael's Hill, Bristol BS2 8BJ.


Recent reports suggest an increased incidence of cytomegalovirus (CMV) infection in recipients of unrelated donor (UD) bone marrow transplantation (BMT).

In this study we have collated the incidence of CMV infection and disease in sequential UD (n = 119) and related donor (RD; n = 79) BMT performed in a single institution over a 7-year period. Low-risk patients (CMV seronegative recipient and donor) accounted for 51% of UD BMT (n = 61) and 62% of RD BMT (n = 49), with CMV excretion documented in one RD BMT only. The remaining high-risk patients received identical prophylaxis regimens with aciclovir and intravenous immunoglobulin (IVIG). Two groups consisting of 58 UD BMT (median age 9.0 years, range 0.7–45.3 years) and 30 RD BMT (median age 13.6 years, range 1.6–47.6 years) were analysed. CMV reactivation/re-infection was documented in 15 UD BMT (26%) and 10 RD BMT (33%) (P = 0.72), and CMV disease in four UD BMT (8%) and four RD BMT (13%) (P = 0.533). In this series the risk of CMV excretion and disease following UD BMT was similar to that following RD BMT.

Cytomegalovirus (CMV) disease continues to be one of the major infectious complications encountered following allogeneic bone marrow transplantation (BMT). Reactivation of latent virus in a seropositive recipient is the major predisposing factor. The reactivation rate has been reported to be as high as 80% in seropositive recipients, with CMV disease (most commonly manifested as interstitial pneumonitis (IP)) developing in up to a quarter of patients ( Rubie et al, 1993 ). Other reported risk factors include a seropositive donor, CMV-positive blood products, graft-versus-host disease (GVHD), age, and HLA mismatch ( Meyers et al, 1986 ; Wingard et al, 1990 ).

Since only 30% of the patients in Europe who would benefit from allogeneic BMT have an HLA-matched sibling, bone marrow donor registries are increasingly searched to identify matched and even mismatched unrelated donors (UD). UD BMT is now a feasible option in many cases where related donor (RD) BMT is not possible, although such transplants are associated with a higher rate of complications, including both acute and chronic graft-versus-host disease (GVHD) and a variety of viral infections ( Beatty et al, 1991 ; Marks et al, 1993 ).

Over the last decade there have been considerable advances in both the prevention and management of CMV disease. Improved conditioning regimens, more effective GVHD prophylaxis and blood screening techniques have been associated with a reduction in complications related to CMV ( Winston & Gale, 1991). Therapeutic intervention with antiviral agents such as ganciclovir and intravenous immunoglobulin (IVIG) have halved fatalities, and the move to both prophylactic ( Meyers et al, 1988 ; Goodrich et al, 1993 ; Prentice et al, 1994 ) and pre-emptive ( Schmidt et al, 1991 ; Goodrich et al, 1991 ) therapy has also significantly reduced the incidence of CMV disease. However, despite these advances, CMV disease has not been eradicated and in a significant proportion of patients the disease or the consequences of its treatment continue to be life-threatening. In particular there have been several recent reports detailing increased early reactivation and, more importantly, fatal pneumonitis in patients undergoing T-cell depletion of the marrow ( Przepiorka et al, 1994 ; Herenstein et al, 1995 ; Baciagalupo et al, 1995 ; Couriel et al, 1996 ). In addition, the specific group of UD transplants has also been reported to be associated with an increased risk of CMV infection and disease ( Takenaka et al, 1997 ; Marks et al, 1993 ; Ward et al, 1994 ).

We have therefore reviewed our experience in 198 consecutive UD and RD BMT performed at a single institution over a 7-year period. All patients received identical CMV prophylaxis regimens and T-cell depletion was performed in the majority of cases. We show in this report that there was no difference in rates of CMV excretion or disease.


Over a 7-year period from September 1987 to September 1994, 198 allogeneic BMT (119 UD BMT, 79 RD BMT) were performed in both adult and paediatric populations at our institution. CMV status of both recipient and donor was established pre-transplant by means of serological assays (complement fixation and latex agglutination tests). Of the total, 61 UD (51%) and 49 RD (62%) BMTs were considered to be at minimal risk for development of CMV infection (CMV seronegative recipient and donor). This relatively high proportion of low-risk patients results partly from the fact that these cohorts are predominantly children, but also partly reflects the Unit's policy to administer CMV-negative blood products to all potential BMT recipients and to undergo an extensive search for CMV seronegative donors wherever possible. All patients were given CMV-negative red cell and platelet transfusions whenever indicated throughout the transplant period.

The remaining 88 allogeneic BMT were performed in 84 patients (four patients receiving a second transplant) and consisted of 58 UD and 30 RD transplants. Two recipients also received infusions of donor lymphocytes as therapy for relapse. All cases were analysed for development of CMV excretion and disease, with a follow-up of at least 12 months (or death if sooner). Patient characteristics of this high-risk group (CMV seropositive recipient and/or donor) are shown in 1 Table I.

Table 1. Table I. Characteristics of unrelated donor and related donor bone marrow transplants at risk of CMV reactivation.Thumbnail image of
  • a

    ALL, acute lymphoblastic leukaemia; AML, acute myeloid leukaemia; CML, chronic myeloid leukaemia; SAA, severe aplastic anaemia.* Includes 6/30 phenotypically matched parental BMTs.† Excludes deaths <30 d and rejections.‡ Excludes deaths <100 d and rejections.

  • The selection of unrelated donors was made from either the British Bone Marrow Registry, the Anthony Nolan Bone Marrow Trust, or in two cases from international registries. All prospective donor/recipient pairs were typed by serology for HLA-A,-B and by restriction fragment length polymorphism (RFLP) for HLA-DR,-DQ. Mismatches were recorded as detected at HLA class I, HLA class II or both. Related donors were matched at all loci tested and included six parental donors.

    Pretransplant conditioning regimens consisted of cyclophosphamide 120 mg/kg and total body irradiation (TBI) 1440 cGy in eight fractions in 82% of procedures (48 UD, 24 RD), busulphan 16 mg/kg plus cyclophosphamide 200 mg/kg to nine patients (six UD, three RD), and in the remainder conditioning was either total nodal irradiation (TNI) and cyclophosphamide (two UD), reduced dosage TBI/TNI and cyclophosphamide for Fanconi's anaemia (two UD, one RD), cyclophosphamide alone (one RD), and in a single case TBI, cyclophosphamide and cytosine arabinoside (one RD). Day 0 was defined as the day of bone marrow infusion.

    T-cell depletion (TCD), both in vivo and ex vivo, was performed in most cases. All but one of the UD BMTs received in vivo CAMPATH-1G (day −9 to day −5) and ex vivo T-cell-depleted grafts (CAMPATH-1M plus complement, or CAMPATH-1G alone added to the bag) ( Hale & Waldmann, 1994, 1996); the remaining case was T replete. The RD BMTs were a more heterogenous group. In 15 cases management was as described above with in vivo and ex vivo T-cell depletion. In a further six cases an additional T-cell addback was given (0.1–1.0 × 106 CD3 cells/kg of recipient weight), two cases had ex vivo T depletion with CAMPATH-1M alone, and the remaining cases were all T replete. Post-graft GVHD prophylaxis was with cyclosporin A alone (43 UD, 17 RD) or with methotrexate (nine UD, four RD); no prophylaxis was given to the remainder.

    Prophylaxis against CMV was standard in all high-risk patients. Aciclovir 500 mg/m2 i.v. was given three times daily from day −4 to day 30, followed by oral aciclovir 200 mg five times daily to at least day 100, and continued to 6 months in patients requiring significant immunosuppression for GVHD. Intravenous immunoglobulin (IVIG) 200 mg/kg was commenced immediately pre-BMT on day −1, with a second dose on day 13 and thereafter every 3 weeks to day 100. All BMT recipients received CMV-negative blood products.

    CMV surveillance methods were similar in both cohorts but expanded during the time course covered. Initial screening procedures included weekly urine samples and throat swabs throughout the transplant period to day 100. Since May 1991 additional routine bronchoalveolar lavage (BAL) was performed at days 35, 63 and 91. In the event of development of symptoms suggestive of CMV disease, all specimens were repeated and a BAL performed wherever possible. Laboratory methods included the rapid identification of CMV in cell culture using monoclonal antibodies in the direct early antigen fluorescent foci method (DEAFF) and conventional cell culture techniques ( Griffiths et al, 1984 ).

    Treatment of suspected or proven CMV disease included ganciclovir therapy (5 mg/kg twice daily for 14 d, then reduction to once daily Monday–Friday until day 100 if repeat BAL specimens were CMV negative), together with immunoglobulin (200 mg/kg twice weekly for 14 d, then once weekly to day 100). In addition, since May 1991 all asymptomatic BAL-positive patients have been treated pre-emptively with a similar regimen.

    For the purpose of this analysis, CMV excretion was defined as isolation of CMV from any specimen, whereas CMV disease implied isolation of CMV in association with clinical changes.

    Statistical analysis

    Initial exploratory analysis was performed using the Chi-square test or Fisher exact test (for small expected frequencies) to compare proportions with CMV infection or disease between subgroups defined by the following variables: donor (UD/RD), sex, age ≷10 years, TBI, acute GVHD (0, ≷ II), ±TCD, recipient/donor CMV status, year of BMT (≷1991). For the CMV infection rates we were able to calculate Kaplan-Meier estimates of cumulative proportions remaining infection free, censoring at 12 months. Numbers were too small to do this for CMV disease- free survival. The log-rank test was used to compare differences between subgroups. A Cox proportional hazards regression model was used to compare the risk of developing CMV infection in UD and RD BMT, making adjustment for other significant variables.


    Of all the 110 CMV low-risk transplants performed (CMV seronegative recipient and donor), CMV excretion (viruria only) was documented in a single RD BMT who did not receive specific antiviral therapy.

    Of the remaining 88 high-risk procedures, CMV excretion was documented in 14/58 UD and 9/30 RD BMT, with a further case in each group diagnosed at post-mortem (incidence of CMV 26% and 33% respectively) ( 2Tables IIa and 2 IIb). Early deaths (<30 d) occurred in five cases (five UD BMT), with evidence of CMV excretion in only one UD BMT. Exclusion of these patients did not alter the incidence of CMV reactivation (14/53 UD BMT, 26%). When only CMV seropositive recipients were considered, CMV excretion occurred in 12/27 UD (44%) and 9/21 RD BMT (43%) ( 3 Table III).

    Table 2. Table II. Characteristics of UD BMT recipients excreting CMV.Thumbnail image of
  • a

    BAL, bronchoalveolar lavage; IP, interstitial pneumonitis; NPA, nasopharyngeal aspirate; PM, postmortem; VOD, veno-occlusive disease.* Routine surveillance bronchoalveolar lavage.† Pre-emptive ganciclovir given.

  • Table 3. Table III. The incidence of CMV excretion and disease in related and unrelated donor bone marrow transplantation according to recipient and donor CMV status.Thumbnail image of

    CMV disease developed in four UD BMT (7%) and four RD BMT (13%). When only CMV seropositive recipients were considered, CMV disease occurred in 1/27 UD BMT (4%) and 3/21 RD BMT (14%). There was no evidence of CMV disease at postmortem in the patient who died at day 25 who had previously been found to have viruria (UD, UPN 132; 2 Table IIa). Pneumonitis was evident in all eight of the symptomatic recipients with CMV disease, with two having additional sites of disease ( 2Tables IIa and 2 IIb). In all cases of CMV disease, therapeutic manoeuvres included ganciclovir 10 mg/kg/d and immunoglobulin 200 mg/kg twice weekly. In seven cases pneumonitis was fatal despite therapy, and the final patient (RD, UPN 163) recovered but died from complications related to treatment.

    Thirteen recipients (nine UD BMT, four RD BMT) received successful pre-emptive therapy with ganciclovir and immunoglobulin for their CMV reactivation. Nine of these patients are still alive, and the remainder died subsequently of unrelated causes (two disease relapses, one GVHD, one sepsis). The introduction of routine BAL and pre-emptive therapy in May 1991 has resulted in a similar reduction in CMV disease in both groups, when pre-routine BAL/pre-emptive therapy data (CMV excretion 22% UD, 33% RD; CMV disease 22% UD, 17% RD) was compared with post-routine BAL/pre-emptive therapy data (CMV excretion 27% UD, 33% RD; CMV disease 4% UD, 11% RD). The four cases of CMV disease since 1991 have occurred in patients who have developed CMV infection before day 35 (n = 1), or following previous negative BALs (n = 3) at 2, 3 and 8 months post-BMT. The incidence of grade II–IV acute GVHD in evaluable patients (excluding early deaths <30 d and episodes of graft failure) was 29% (UD) and 34% (RD). The incidence of chronic GVHD (excluding deaths <100 d and episodes of graft failure) was 22% (UD) and 17% (RD) ( 1 Table I).

    When all subjects were considered, Kaplan-Meier estimates of percentage infection-free showed a significant advantage for females over males (estimate at 12 months 88.6% and 57.5% respectively; logrank test P = 0.004) and recipient/donor CMV status −/+ over +/+ and +/− (estimate at 12 months 87.1%, 57.4% and 52.1% respectively; logrank test P = 0.003). However, in those who had evidence of CMV infection, recipient/donor CMV status −/+ were more prone to go on to develop CMV disease (Fisher exact test P = 0.011). No other significant differences were found, in particular the risk of CMV infection (logrank test P = 0.720) or disease (Fisher exact test P = 0.533) did not differ significantly between UD and RD BMT. Cox regression failed to show a significant difference between the UD and RD BMT groups in development of CMV infection when adjusted for gender and recipient/donor CMV status.


    Although CMV has been a major infectious cause of death in bone marrow transplantation, there have been considerable advances both in the prevention and treatment of this condition. Both the risk of developing CMV disease and the toxicity of any therapy administered should be considered when developing a strategy with a view to improving outcome ( Prentice & Kho, 1997). The use of CMV seronegative blood products in low-risk CMV seronegative patients has been shown to remove virtually all risk of infection ( Rubie et al, 1993 ) and we have confirmed this in our series. In high-risk patients not given any form of prophylaxis, excretion rates can rise to 80% ( Meyers et al, 1986 ; Rubie et al, 1993 ). For these high-risk patients a number of prophylactic measures have been employed, including IVIG and antiviral therapy, often in combination. Intravenous immunoglobulin has been shown to decrease the incidence of CMV disease and this appears to be correlated with a reduction in GVHD ( Winston et al, 1987 ; Sullivan et al, 1990 ). The role of antivirals has been examined in several studies. Prophylactic aciclovir does reduce excretion and disease rates to a limited extent, probably reflecting an effect of delay in time to infection, and prolonged courses are associated with a significant overall survival advantage ( Meyers et al, 1988 ; Prentice et al, 1994 , 1997). On the other hand, although CMV seropositive recipients given prophylactic ganciclovir would appear to have virtually no risk of CMV excretion and disease compared with controls, overall mortality was found to be similar in both groups, and was presumed to reflect the increased risk of neutropenia and subsequent infection in the ganciclovir-treated group. In this unit we chose to adopt an aciclovir prophylaxis regimen (intravenous followed by oral) in all patients throughout the study period.

    Another approach that has been explored increasingly with the development of CMV diagnostic techniques is that of pre-emptive therapy given to recipients in whom CMV is detected systemically while at an asymptomatic stage. Ganciclovir given to patients with positive surveillance cultures and in particular to patients with viraemia or BAL positivity ( Schmidt et al, 1991 ; Goodrich et al, 1991 ) has resulted in decreased CMV disease and improved overall survival. As a result of these studies, in 1991 we adopted the policy of routine surveillance BAL and pre-emptive therapy in positive cases. The results here confirm a reduction in CMV disease (with no overall change in CMV excretion rates) in both our UD and RD groups, although the numbers were too small pre-1991 to attain statistical significance. At the time of this study, CMV PCR techniques were not established in our unit ( Einsele et al, 1991 ), although they have subsequently been included in the routine CMV surveillance programme. Although the results of this study may have differed if this diagnostic test had been employed at the time, the similar approach in both cohorts allows justifiable comparison between the two groups.

    Large national registries of volunteer unrelated donors are now increasingly utilized to expand the potential donor pool, such that a matched or partially mismatched donor can be identified for the majority of patients in whom BMT is indicated. Several studies have now reported concern over a higher incidence of complications from CMV in the unrelated donor BMT setting. In a study comparing the complications of RD and UD BMT for adults with chronic myeloid leukaemia ( Marks et al, 1993 ) there was a significantly increased incidence of CMV disease (37% UD as compared with 12% RD when only patients at high risk for CMV were considered, P = 0.03). In a further later report ( Ward et al, 1994 ) this group confirmed these findings, describing CMV excretion rates of 87% in UD BMT recipients as compared with 44% in RD BMT, with CMV disease occurring in 60% (27% proving fatal despite early treatment). Takenaka et al (1997 ) reported similarly high incidences of both CMV infection (87% v 53%) and disease (73% v 14%) in HLA-matched UD and RD BMTs. In our study the overall incidence of CMV excretion (26% v 33%) and disease (8% v 13%) was similar in both groups, and indeed only equivalent to that observed in the RD groups described above. From the results of this large series we would therefore debate whether UD BMTs are at increased risk from CMV disease.

    Patients receiving T-cell-depleted marrows have been reported to be at increased risk of early CMV infection and disease ( Przepiorka et al, 1994 ; Hertenstein et al, 1995 ; Baciagalupo et al, 1995 ; Couriel et al, 1996 ). The patients receiving UD transplants in our series underwent more T-cell depletion (TCD) than the RD transplants. The approach to TCD might be an important factor in this respect, although similar methods were employed both here and in the majority of previously reported series.

    Graft-versus-host disease is also an obvious risk factor to consider. UD BMTs have been shown to be associated with an increase in transplant-related complications which includes a higher incidence of both acute and chronic GVHD as compared to RD BMT ( Beatty et al, 1991 ; Ash et al, 1990 ). Our GVHD prophylaxis has been very effective as in this group of patients the incidence of both acute grade II–IV (29%) and chronic GVHD (22%) has been low, and in particular we have documented similar rates in both our RD and UD cohorts. In view of the relationship between GVHD and CMV disease, this could well have a bearing on the low overall rate of infection. Indeed, Marks et al (1993 ) noted a much higher incidence of both acute grade II–IV (52%) and chronic GVHD (77%) in recipients of UD grafts compared with ours. Moreover, our UD group contained a large proportion of mismatches (43%) where a higher rate of complications might be expected. Mismatched grafts could not be correlated with either CMV excretion or disease, a similar rate being observed in patients who received either matched or mismatched marrow.

    Older age has also been associated with an increased risk of CMV infection and therefore should be considered in this series where the majority of patients were children. However, we found no evidence that age had a significant effect on the development of CMV infection or disease in this series.

    Finally, a higher dose of TBI has also been proposed as a contributing factor to the development of CMV. Our UD BMT patients received a higher TBI dose (14.4 Gy) and had less disease than previously described ( Marks et al, 1993 ).

    The overall CMV infection rate reported here compared favourably with other recent studies ( Prentice & Kho, 1997), and confirmed the trend towards a reduced rate overall. In our experience, recipients of either matched or mismatched UD bone marrow do not experience a higher incidence of com-plications from CMV than their related counterparts and this in itself should not be regarded as a contraindication to BMT.