Kinetics of myeloid and lymphocyte recovery and infectious complications after unrelated umbilical cord blood versus HLA-matched unrelated donor allogeneic transplantation in adults


Mary J. Laughlin MD, Allogeneic Transplant Program, Case Western Reserve University, University Hospitals Research Institute, 11100 Euclid Ave, Wearn 433, Cleveland, OH 44106-5065, USA.


Sources for allogeneic stem cells for patients with haematological disorders lacking a histocompatible sibling donor include matched unrelated donor (MUD) and umbilical cord blood (UCB). A total of 51 patients with haematological disorders, treated with myeloablation and transplantation with either unrelated human leucocyte antigen (HLA) partially matched UCB (28 patients) or HLA-matched MUD grafts (23 patients) during 1997–2003, were evaluated for life-threatening infections, haematological reconstitution, graft versus host disease, relapse and event-free survival (EFS). The median duration of neutropenia after transplantation was longer (29 d vs. 14 d) in the UCB group. The probability of donor-derived neutrophil engraftment by day 42 was 0·86 [95% confidence interval (CI) 0·71–1·0] in UCB recipients versus 0·96 (95% CI 0·87–1·0) in MUD recipients surviving >28 d. Overall infection rates were higher in UCB recipients, particularly at the early time points (before day +50) after transplantation. Graft failure occurred in five UCB recipients and two MUD recipients and was associated with the occurrence of bacteraemia during neutropenia. The EFS at 3-year follow-up was 0·25 in UCB and 0·35 in MUD recipients. UCB transplantation in adults is associated with delayed neutrophil and lymphocyte recovery compared with MUD grafting, and higher rates of bacteraemia at early time points after transplantation.

Allogeneic stem cell transplantation following myeloablative conditioning is curative in a proportion of patients with life-threatening haematological disorders (Appelbaum, 1997; Lennard & Jackson, 2000). However, this treatment strategy is limited by the availability of human leucocyte antigen (HLA)-matched sibling donors. For those patients lacking an histocompatible sibling donor, alternative graft sources for haemopoietic stem cells include HLA-matched unrelated adult donors (Kernan et al, 1993; McGlave et al, 2000) HLA full-haplotype mismatch (Aversa et al, 1998) and HLA-mismatched unrelated donor umbilical cord blood (UCB) (Kurtzberg et al, 1996; Gluckman et al, 1997; Rubinstein et al, 1998; Laughlin et al, 2001).

Retrospective analyses reveal similar survival after infusion of unrelated UCB compared with matched unrelated donor (MUD) grafts in paediatric recipients, with few available data for adults (Barker et al, 2001; Rocha et al, 2001). However, there may be differences in UCB and MUD grafts, related to stem cell and lymphocyte populations, of importance in determining the kinetics of myeloid and lymphocyte recovery and graft versus host disease (GVHD) severity; factors which may contribute to morbidity and mortality at early and late time points after transplant.

Infection remains a significant cause of mortality and morbidity following allogeneic bone marrow or stem cell transplantation (Wingard, 1999). The incidence and risk factors for bacterial, fungal and viral infections have been shown to correlate with the kinetics of immune reconstitution, and serve as a guide for appropriate antimicrobial prophylaxis for allogeneic transplant patients (Ochs et al, 1995; Collin et al, 2001; Ninin et al, 2001).

We report a single institution retrospective comparative study, addressing kinetics of donor-derived myeloid and lymphocyte recovery, infectious complications, patient demographics, graft characteristics and transplant outcomes after unrelated allogeneic transplantation in adults consecutively treated during the time period 1997–2003, and infused with either UCB or MUD grafts following full myeloablative conditioning.

Patients and methods

Transplant protocols and patient eligibility

The Institutional Review Board of Case Western Reserve University/University Hospitals of Cleveland approved the clinical protocols for UCB and MUD transplantation. Informed consent was obtained from all patients. Patients were eligible for treatment within MUD clinical trials if allogeneic stem cell transplantation was determined to be clinically indicated and an HLA-identical related donor was not available. Patients were eligible for enrolment in UCB clinical trials if they had stable haematological disease and lacked an HLA-identical related or unrelated donor, or had unstable disease, lacked a related donor, and an HLA-matched unrelated bone marrow donor could not be identified within 6–8 weeks. All adult patients consecutively transplanted with UCB at this institution since April 1998 and all MUD adult recipients treated since January 1997 were included in this retrospective comparative analysis. Four MUD recipients undergoing transplant during the time period 1994–96 were excluded from analyses, because these early patients were not treated concurrently and received differing preparative regimens and antimicrobial prophylaxis.

Graft selection

Preliminary searches of the National Marrow Donor Program and UCB banks were performed using the patient's HLA phenotype, based on serologic typing prior to year 2000 for class I HLA-A and HLA-B antigens and low-resolution DNA typing for class II HLA alleles. High resolution molecular typing for HLA-DRB1 alleles was performed as confirmatory typing. Subsequent to year 2000, DNA typing was routinely performed for class I HLA-A, HLA-B, HLA-C alleles. The MUD grafts were matched at 6/6 HLA loci (HLA-A, HLA-B and DRB1), containing a minimum 20 × 107 mononuclear cells/kg recipient body weight. Five patients in the MUD group received peripheral blood stem cells, while the other 18 patients received bone marrow stem cells. Preferred UCB units were those that matched at least 3/6 HLA loci and contained a minimum prefreeze cell dose of 1 × 107 nucleated cells/kg recipient body weight. UCB units were not depleted of T lymphocytes. Some units were reduced in volume and depleted of red cells with hetastarch before freezing, as previously described (Rubinstein et al, 1995).

Preparative regimens and transplant supportive care

Disease status for leukaemia was categorized according to the International Bone Marrow Transplant Registry (IBMTR) criteria (Sobocinski et al, 1994). Conditioning regimens for MUD recipients diagnosed with acute myeloid leukaemia (AML) and chronic myeloid leukaemia (CML) included, cyclophosphamide (Cy) 120 mg/kg with either total body irradiation (TBI) 1200 cGy (Clift et al, 1994) or busulphan (Bu) 16 mg/kg (or adjusted-dose AUC) (Copelan et al, 1993). Patients with acute lymphoblastic leukaemia (ALL) received TBI 1320 cGy and etoposide 60 mg/kg (Blume et al, 1993). All UCB patients received either TBI/Cy or Bu/Cy conditioning and antithymocyte globulin (ATGAM; Pharmacia-Upjohn, Kalamazoo, MI, USA) 90 mg/kg over 3 d (day −3 to −1) prior to UCB infusion (Laughlin et al, 2001). Supportive care included the administration of recombinant granulocyte colony-stimulating factor (Filgrastim; Amgen, Thousand Oaks, CA, USA) subcutaneous daily, starting at day of graft infusion (day 0) for UCB patients (10 μg/kg/d) and day +7, after completion of methotrexate (MTX) administration in MUD recipients (5 μg/kg/d), and was continued daily until full donor-derived neutrophil recovery was achieved.

GVHD and antimicrobial prophylaxis

Prophylactic regimens for GVHD included cyclosporine (CSA) (day −2 to +180) combined with methylprednisolone (MP) at a dose of 1 mg/kg/d (day +5 to +13) in the UCB group, or short-course MTX (10 mg/m2 day +1, +3 and +6) in the MUD group. Acute GVHD (aGVHD) was scored weekly using standardized guidelines (Przepiorka et al, 1995). Patients surviving more than 100 d after transplantation were evaluated and graded for chronic GVHD (cGVHD) according to standard criteria (Flowers et al, 1999). Acute GVHD (grade II or higher) was treated with MP, 2–10 mg/kg/d (i.v. or p.o.) in three divided doses for 7–14 d, then tapered by 0·5 mg/kg/d every 5 d. An alternative regimen of MP 500 mg/m2/dose (12 h × 2 doses), repeated every 48–72 h for up to four courses was also used. Corticosteroid-resistant aGVHD and cGVHD were treated with second-line immunosuppressive agents according to institutional protocols.

Prophylaxis for Pneumocystis carinii pneumonia, consisting of trimethoprim-sulphamethoxazole, dapsone or inhaled pentamidine was administered to all patients after transplant. Patients with serologic evidence of prior exposure to cytomegalovirus (CMV IgG+) or those who received grafts from CMV serologic positive donors received ganciclovir prophylaxis, starting after myeloid engraftment and continued until day +100 (Leather & Wingard, 2001). When both the patient and donor were CMV seronegative, no specific CMV prophylaxis was administered, other than the use of CMV seronegative or leucocyte depleted blood products. Prior to January 2001, MUD and UCB patients received immunoprophylaxis with 5–10% intravenous immunoglobulin (IVIG) weekly (100–500 mg/kg/week), beginning the week before marrow infusion and continuing until day +100 (Sokos et al, 2002). This protocol was amended in January 2001, and supplemental doses of IVIG (250 mg/kg) were administered only if total serum IgG level was <0·4 g/l.

To minimize the risk of invasive fungal infection (IFI), all patients received oral nystatin prophylaxis. Fluconazole was used as antifungal prophylaxis in the MUD recipients. In the UCB group, those patients receiving aGVHD treatment with MP at a dose of ≥1 mg/kg/d received either oral itraconazole, amphotericin  B (1·0 mg/kg/dose i.v., three times/week), or liposomal amphotericin  B (Ambisome; Fugisawa, Deerfield, IL, USA) 1 mg/kg/dose i.v., three times/week in patients experiencing amphotericin  B intolerance or renal toxicity. Twelve patients in the UCB group required high dose steroid to treat aGVHD, and these patients received prophylactic antifungal as described above.

MUD and UCB recipients did not receive prophylactic systemic antibiotics. For both transplant groups, febrile neutropenic patients were initiated on an empiric regimen of broad-spectrum antibiotics according to published guidelines (Hughes et al, 2002).

Study definitions of infection

We used standard criteria to define fever, blood stream infections and other infectious disease syndromes (Engels et al, 1999; Ketterer et al, 1999; Hughes et al, 2002). Infections were classified as microbiologically documented when pathogenic microorganisms were recovered from affected sites, and clinically documented when pathogenic microorganisms were not recovered. The diagnosis and management of catheter-related infections were adapted from the Infectious Diseases Society of America guidelines (Mermel et al, 2001). Early infections were defined as those occurring within the first 50 d after graft infusion (Ochs et al, 1995).

Cytomegalovirus disease was defined as symptoms from the affected organ, combined with detection of CMV by histology, isolation, immunohistochemistry or typical retinitis changes (Zaia & Forman, 1995; Takenaka et al, 1997; Ljungman et al, 2002). Definitive invasive adenovirus disease was defined by a compatible clinical presentation, positive viral culture, and/or the presence of characteristic intranuclear viral inclusion (Howard et al, 1999).

Criteria for definitive aspergillosis according to definitions by the Mycoses Study Group included histological diagnosis with or without positive culture, or a positive culture from tissue obtained by an invasive procedure such as transbronchial biopsy or percutaneous needle aspiration (Jantunen et al, 1997; Yeghen et al, 2000; Ascioglu et al, 2002).

Statistical analysis

The chi-square or Fisher's exact test was used to examine the association between categorical factors. Wilcoxon's rank sum test was used to examine the difference of age, cell dose infused, CD34+ and duration of stay in hospital between the two groups. Neutrophil and platelet recoveries were analysed separately, and defined as the first of three consecutive days of an absolute neutrophil count (ANC) ≥ 0·5 × 109/l, and the first of seven consecutive days of a transfusion-independent platelet count >20 × 109/l respectively. Primary graft failure was defined as the failure of donor-derived ANC ≥ 0·5 × 109/l in patients who survived for more than 28 d after transplantation. Event-free survival (EFS) was measured from the date of transplantation to date of death or relapse and censored at the date of last follow-up for survivors without relapse. Data analyses were performed on follow-up for all patients through July 2003. Observations of day to attained ANC ≥ 0·5 × 109/l were censored at: death, relapse, or +42 d, whichever came first. Survival distribution was estimated using Kaplan–Meier methods (Kaplan & Meier, 1958) and the difference between groups was tested using log-rank or Wilcoxon's test. Absolute lymphocyte counts (ALC) (×109/l) were calculated (WBC×% lymphocytes on differential) at specified time points (days 30, 60, 90, 180, 270 and 365) after transplant and Wilcoxon's rank sum test was used to examine differences between MUD and UCB study groups. The mean value for ALC measurements (×109/l) at all time points during the first year after transplantation in each MUD and UCB patient was calculated and divided into two groups (≤ and >0·3 × 109/l divided at the median) and compared with survival using log-rank test.

To account for multiple infections for each patient and differing lengths of follow-up after MUD and UCB transplantation, density incidence (number of infections per 100 patient days at risk) was calculated by dividing the total episodes of infection in all recipients of MUD or UCB by the number of days at risk (survival after transplantation) and multiplied by 100, and infection rates were compared in the two study groups using the likelihood ratio test. The number of infections was treated as a Poisson random variable. Generalized linear models were used with log link and the log of the number of days at risk as offset (fixed predictor) (McCullagh & Nelder, 1989). For infection rates during neutropenia, days at risk were calculated as day to attained ANC ≥ 0·5 × 109/l or day +42 or day of death if patients died before day +42 after transplantation. For infection rates after 50 d, days at risk were days of survival minus 50. For early infection rates (first 50 d), the days at risk were 50 d or days of survival if patient died prior to day 50 (Ochs et al, 1995). For total infection rate, days at risk were days of survival after transplantation. The likelihood ratio test was used to compare ALC with infection rates in all study patients. All tests were two-sided and P ≤ 0·05 was considered statistically significant.


Patient and graft characteristics

Patients receiving UCB or MUD grafts were similar in terms of age, gender and race (Table I). UCB patients differed from MUD recipients, with a higher proportion of patients scored as intermediate high risk by IBMTR risk assessment, although this did not attain statistical significance (P = 0·46). In the UCB group, 22 of 28 patients had high/intermediate-risk malignancies including eight patients with ALL > first complete remission (CR-1) status, seven patients with CML in accelerated phase or prior blast crisis, according to the standard definition (Sokal et al, 1988), six patients with AML > CR-1 status, and one patient with refractory non-Hodgkin's lymphoma. Sixteen of the 23 patients in the MUD group were considered high/intermediate risk including seven AML, five CML, two ALL and two patients with non-Hodgkin's lymphoma, who failed prior autologous transplantation. The time from diagnosis to allogeneic transplantation was longer in UCB (21·7 months; range 2–146) compared with the MUD group (8·2 months; range 2–87) (P = 0·001). The median dose of CD34+ cells infused was lower in the UCB group (1·7 × 105 CD34+/kg; range 0·2–9·5) than the MUD group (4·4 × 105 CD34+/kg; range 1·7–10·2) (P = 0·001). The median dose of the total nucleated cells infused was also significantly lower (2·1 × 107 cell/kg; range 0·9–4·5) in UCB compared with MUD patients (40·5 × 107/kg; range 20–100) (P = 0·001).

Table I.  Patient and graft characteristics.
CharacteristicsUCB group (n = 28)MUD group (n = 23)P-value
  1. UCB, umbilical cord blood; MUD, matched unrelated donor; CMV, cytomegalovirus; HSV, herpes simplex virus; VZV, herpes zoster virus.

  2. *International Bone Marrow Transplant Registry criteria.

Sex (M/F)13/1515/80·18
Age (years), mean (range)36·4 (19–54)35·8 (18–65)0·73
Race (%)
 Caucasian24 (86)21 (91)0·68
 Black 4 (14) 2 (9) 
Preparative regimen (%)
 Chemotherapy + radiation21 (75) 8 (35)0·004
 Chemotherapy alone 7 (25)15 (65) 
Risk (intermediate/advanced/low)*22/616/70·46
Time [diagnosis–transplant; months (range)]21·7 (2–146) 8·2 (2–87)0·001
Recipient CMV serology (+/−)16/1213/100·96
Donor CMV serology (+/−) 0/2810/130·001
CMV status (recipient/donor) (%)
 Positive/positive 0 4 (17) 
 Positive/negative16 (57) 6 (26)0·001
 Negative/negative12 (43) 7 (30) 
 Negative/positive 0 6 (26) 
Recipient HSV serology (+/−)19/916/70·9
Recipient VZV serology (+/−)16/1215/80·7
Cells infused, mean (range)
 Infused nucleated cells × 107/kg 2·1 (0·9–4·5)40·5 (20–100)0·001
 Infused CD34+ × 105/kg 1·7 (0·2–9·5) 4·4 (1·7–10·2)0·001

Myeloid and lymphocyte engraftment

In the UCB group, the estimated probability of donor-derived neutrophil recovery during the first 42 d after transplantation was 0·86 [95% confidence interval (CI) 0·81–1·0], compared with 0·96 (95% CI 0·9–1·0) (P = 0·001) in the MUD group. Chimaerism analyses of peripheral blood mononuclear cells was performed in all patients and full donor chimaerism (>98% donor) was noted in all patients who survived >28 d and engrafted. The median time to an ANC ≥ 0·5 × 109/l was prolonged in the UCB group: 29 d (range 13–42 d) compared with 14 d (range 11–20 d) for MUD transplant patients (P = 0·001) (Fig 1). There were two early deaths in the UCB group prior to day 28, caused by multiple organ failure and sepsis. Of the 26 patients in the UCB group who survived for 28 d or more, five patients experienced primary graft failure while only two patients in the MUD group failed to engraft. These MUD patients received a second transplantation with successful donor-derived myeloid recovery. The five patients in the UCB with graft failure died between days 28 and 65 because of multiple organ failure, and two of these patients had signs of overwhelming infection/sepsis at the time of death. The median time to platelet engraftment was 59 d (range 37–104 d) in the UCB recipients, and 34 d (range 12–65 d) in the MUD recipients. No secondary graft failures were observed in any study patient.

Figure 1.

Neutrophil recovery after umbilical cord blood (UCB) versus matched unrelated donor (MUD) transplant. Kaplan–Meier curve of neutrophil engraftment as defined by the number of days required for donor-derived absolute neutrophil count (ANC) to reach 0·5 × 109/l, stratified by graft source.

Figure 2 outlines the kinetics of the ALC recovery during the first year after transplantation. Donor-derived lymphocyte recovery was slower in UCB patients and remained below that measured in MUD recipients at early time points (days 30 and 60; P = 0·018, 0·126 respectively). However, lymphocyte recovery in UCB patients surpassed that of MUD recipients from day +60 to +365. The ALC remained below normal range in MUD recipients throughout the first year, but was noted to normalize by day +200 in UCB recipients.

Figure 2.

Profile of absolute lymphocyte counts (ALC): matched unrelated donor (MUD) versus umbilical cord blood (UCB). The area between the horizontal lines represents the normal values.

The mean ALC correlated highly with both the overall infection rate and with bacterial infections (P < 0·001 and P = 0·001 respectively), occurring during the first year after transplantation. Each decrease of the mean ALC by 1 × 109/l was associated with an 8·6-fold increase in the patient's infection rate (95% CI 4·8–15·3).

Overall infections and bacterial infections

All study patients developed at least one febrile episode during their initial hospitalization for transplantation. The total number of microbiologically documented infections was 159 episodes (84 in the UCB and 75 in the MUD group). The mean duration of fever was longer in the UCB group: 7·17 d (range 1–30 d) compared with 2·3 d (range 1–12 d) in the MUD group. The aetiology of the fever was identified in 46% of the cases in the UCB group and in 41% in the MUD group.

Bloodstream infections were the most common, accounting for 45% (69/159) of all documented infections. Gram-positive organisms accounted for 84% (58/69) of the bacteraemia cases and Gram-negative organisms for less than one-fifth of the cases (Table II). Overall, coagulase-negative staphylococci were the most commonly isolated organisms, followed by Enterococcus spp., Streptococcus spp. and Pseudomonas aeruginosa.

Table II.  Microbiologically documented infections.
InfectionUCB group (n = 28)MUD group (n = 23)
  1. UCB, umbilical cord blood; MUD, matched unrelated donor; CMV, cytomegalovirus; HSV, herpes simplex virus I, II; IFI, invasive fungal infections.

Total episodes8475
 Gram positive40/4718/22
 Gram negative7/474/22
C. difficile colitis64
Bacterial pneumonia43
Urinary tract infection24
Viral infections1330
Fungal infections910

Fungal infections

Excluding thrush and superficial fungal infections, the total number of documented fungal infections was 19, of which 12 were IFIs. There was no significant difference in the incidence of IFI comparing UCB versus MUD patients (P = 0·39). There were nine episodes of fungal infections in the UCB group; three occurred early (prior to day 50) and six late, with a median time of late infection 376 d after transplantation (Table II). Of the nine documented fungal infections in the UCB group, seven episodes were invasive, including a brain abscess caused by Cryptococcus neoformans, two episodes of invasive pulmonary aspergillosis, three episodes of disseminated fungal infection caused by Scedosporium apiospermum, Aspergillus niger and Candida tropicalis, and one episode of fungaemia caused by Candida albicans.

In the MUD group, 10 episodes of fungal infections were documented, three occurred during the first 50 d and seven were late infections, with a median 367 d after transplantation. Five of the 10 episodes were considered invasive, one case of fungaemia caused by Candida glabrata, a second case of endophthalmitis caused by Aspergillus fumigatus, and three additional episodes of invasive pulmonary fungal pneumonia with Scedosporium prolificans, A. niger and C. glabrata as the causative organisms (Table II).

Viral infections

In the UCB group, CMV was recovered in only three patients encompassing six episodes (five cases of viraemia and one case of CMV pneumonitis); all occurring between day +63 and +257 after transplantation (Table II). One of these patients was CMV seronegative and two were seropositive before transplantation.

In the MUD group, CMV was recovered in nine patients (17 episodes). Three were two episodes of CMV pneumonitis, two had histologic-confirmed CMV colitis and one episode of CMV retinitis, while the others patients had asymptomatic viraemia (Table II). Seven of the above-mentioned patients were CMV seropositive and two seronegative before transplantation. The median day to CMV reactivation in UCB was +159 and was later in the MUD group, +245 d after transplant.

Herpes simplex virus (HSV)-related infections were documented in 15 patients (Table II); the clinical syndromes were mostly stomatitis, oesophagitis or genital ulcers. Seven of the HSV-related disease developed early in the course of transplantation with a median time of 19 d, the other seven developed late, with a median time of 209 d.

Haemorrhagic cystitis caused by adenovirus was diagnosed in a 19-year-old UCB recipient, who presented with recurrent haemorrhagic cystitis. The clinical presentation, positive viral culture and the presence of characteristic intranuclear viral inclusion fulfilled the definition of invasive adenovirus disease.

JC papovavirus genome was detected in the cerebrospinal fluid of one patient in the MUD group who presented with bilateral upper extremities weakness, incoordination, progressive aphasia, apraxia and unsteady gait, 3 years after transplantation. Serial cerebral magnetic resonance imaging scans showed progressive worsening of white matter changes. Histological examination of stereotactic biopsies showed multiple foci of demyelination in the white matter.

Early versus late infections after transplant

Table III outlines the incidence of early (<50 d) and late (beyond day 50) infections in all study patients. There was no significant difference in infection rates comparing UCB versus MUD patients at later time points (after 50 d) after transplant. However, rates of bacterial infections, particularly Gram-positive organisms after transplant were higher in UCB recipients, at early time points (prior to day +50). The bacterial infection rate prior to day +50 in UCB recipients was 3·13 per 100 patient days compared with 1·25 in MUD recipients (P = 0·003). The incidence of early infection was associated with the total nucleated cells in the UCB graft (P = 0·008), the total infused CD34+ dose (P = 0·003), and the total graft colony-forming units infused (P = 0·008). In the MUD patients, the incidence of early (or late) infections was not associated with either the total nucleated cells infused (P = 0·59) or the CD34+ cell dose infused (P = 0·55). In both groups, patients experiencing bacterial infections while neutropenic demonstrated higher rates of graft failure (P = 0·04) compared with patients not experiencing bacterial infections while neutropenic. There was no association of graft failure with fungal or viral infections (P = 0·13 and 0·46 respectively).

Table III.  Infection rates after transplantation.
InfectionMUD*UCB*Rate ratio for MUD/UCB (95% CI)P-value
  1. MUD, matched unrelated donor; UCB, umbilical cord blood; CI, confidence interval; CMV, cytomegalovirus.

  2. * Number of infections/100 patient days, i.e. number of infections in all recipients of MUD or UCB divided by the number of days at risk and multiplied by 100.

  3. † The total number of days at risk was 14 454 for recipients of MUD and 10 695 for recipients of UCB.

  4. ‡ During first 50 d of transplantation, the number of days at risk was 1117 for recipients of MUD and 1247 for recipients of UCB.

  5. § After first 50 d of transplantation, the number of days at risk was 13 337 for recipients of MUD and 9448 for recipients of UCB.

Total infections†
 Bacterial0·250·540·46 (0·3–0·7)0·02
 Fungal0·060·080·66 (0·2–1·6)0·38
 Viral0·190·141·33 (0·7–2·6)0·37
 CMV0·090·061·6 (0·6–4·2)0·33
 Total infections0·490·770·64 (0·4–0·8)0·06
Infection rates during first 50 d after transplantation‡
 Bacterial1·253·130·44 (0·2–0·74)0·03
 Fungal0·180·320·56 (0·1–3·05)0·50
 Viral0·540·163·35 (0·6–16·6)0·13
 Total infections1·973·610·55 (0·3–0·91)0·02
Infection rates beyond 50 d after transplantation§
 Bacterial0·160·20·82 (0·4–1·5)0·52
 Fungal0·050·060·83 (0·2–2·7)0·78
 Viral0·160·141·14 (0·5–2·2)0·72
 Total infections0·370·390·94 (0·6–1·4)0·67

Survival and transplant outcomes

As of 1 July 2003, the median time of follow-up was 39 months (range 27–52 months) in the UCB group and 38 months (range 25–59 months) in the MUD group. EFS for UCB patients at day +100 was low: 0·34 (95% CI 0·25–0·61) and at the 3-year follow-up was 0·25 (95% CI 0·09–0·35) (Fig 3). EFS for MUD patients at day +100 was 0·78 (95% CI 0·6–0·95) and was noted to drop to 0·35 at the 3-year follow-up (95% CI 0·15–0·54). EFS for MUD patients at the 3-year follow-up was higher when compared with UCB patients (P = 0·014). Of note, the EFS of all study patients with lower ALC (<0·3 × 109/l) during the first year after transplant was significantly lower than that of patients with higher ALC (≥0·3 × 109/l) (P = 0·001).

Figure 3.

Event-free survival. Event times analysed as of January 2003: event-free survival stratified by graft source comparing matched unrelated donor (MUD) and umbilical cord blood (UCB) transplant recipients.

Among the 21 patients in the UCB group who survived >28 d and engrafted, seven had severe (grades III/IV) aGVHD (Table IV). The actuarial probability of severe aGVHD (grades III/IV) in this group by day +100 was 0·33 (95% CI 0·16–0·50). In the MUD group, 22 of the 23 patients had a successful myeloid recovery and survived for >28 d after transplantation, Severe aGVHD was documented in six of these 22 patients. The actuarial probability of severe aGVHD in this group was 0·27 (95% CI 0·12–0·43). There was no statistical difference in the probability of severe aGVHD comparing UCB and MUD groups (P = 0·66). Moreover, there was no association between mean ALC and incidence of acute GVHD III/IV (P = 0·616).

Table IV.  Transplant outcomes.
EventUCB (n = 28)MUD (n = 23)P-value
  1. UCB, umbilical cord blood; MUD, matched unrelated donor; GVHD, graft versus host disease; ICU, intensive care unit.

  2. * During the immediate post-transplantation.

  3. Admission during the immediate post-transplant period.

Neutrophil recovery (d), median (range)29 (13–42)14 (11–42)<0·001
Primary graft failure520·436
Duration of stay in hospital (d), mean (range)*55 (19–111)40 (19–83)0·008
Death related to infection750·785
Acute GVHD (grades II–IV)17120·104
Acute GVHD (grades III and IV)760·665
Chronic GVHD7160·206
Admission to the ICU1460·082

In the UCB group, of the 12 patients surviving >100 d, five patients developed extensive cGVHD, and four developed limited cGVHD. In the MUD group, of the 19 patients surviving >100 d, nine patients developed extensive and seven patients developed limited cGVHD. The actuarial probabilities of extensive cGVHD were 0·42 and 0·47, respectively, in UCB and MUD recipients (P = 0·65). A higher incidence of relapse of underlying malignancy was noted in the MUD group compared with the UCB group (P = 0·02). Seven patients relapsed after MUD transplant: four patients within the first year (three patients with AML and one patient with chronic lymphocytic leukaemia); and three patients before the end of the third year (two patients with ALL and one with AML). Relapse incidence in the UCB group was low, with only one patient diagnosed with Philadelphia chromosome positive ALL who relapsed nine months after transplantation.


During this 5-year single institution experience, the rate of bacterial infections during the early post-transplantation period (<50 d) was higher in UCB patients compared with MUD transplant recipients, while overall rates of infections at later time points (beyond 50 d) were similar in the two groups. One important contributing factor to this higher incidence of early bacterial infections in the UCB patients may be related to the prolonged duration of neutropenia and lymphopenia after infusion of smaller numbers of total graft nucleated cells and CD34+ cells. A second contributing factor to these higher rates of early bacterial infections may be related to patient selection. The time interval between diagnosis and transplantation was much longer in the UCB recipients and a higher proportion of these patients were considered intermediate high risk haematological malignancy using IBMTR criteria.

Early infections were predominantly bacterial in origin and caused primarily by Gram-positive organisms. Walter and Bowden (1995) noted a shift in the microbiology of bloodstream infections after BMT. Gram-positive organisms, previously in the minority, were now responsible for 60% of bacteraemias (Walter & Bowden, 1995). Possible reasons for this shift included the increased use of indwelling catheters, mucositis associated with intensive chemotherapy, and use of empiric antibiotics, such as fluoroquinolones, thereby reducing the number of Gram-negative bacteraemias (Walter & Bowden, 1995; Collin et al, 2001).

Our data point to a potential association between bacteraemia and primary graft failure; however, no such association was observed in the setting of fungal or viral bloodstream infection. In both UCB and MUD recipients, those patients experiencing bacterial infections whilst neutropenic demonstrated a higher rate of graft failure (P = 0·04). This correlation has not been previously reported. Whether there is a causal relationship between bacteraemia and failure of donor engraftment is not clear and remains a subject for future investigation. Based on these data of high incidence of bacteraemia in UCB recipients, their profound and prolonged duration of neutropenia and association with primary graft failure, consideration may be given for the clinician to initiate broad-spectrum antibiotics at the onset of neutropenia rather than waiting for fever and probable bacteraemia.

Serious fungal infections constituted 12% of the total microbiologically documented infections, with similar rates in the two groups, however patients in the MUD group received fluconazole prophylaxis, whereas UCB patients receiving high dose steroids for GVHD treatment received amphotericin  B, including liposomal amphotericin or itraconazole prophylaxis. Broader antifungal coverage was used in the UCB group because of the prolonged and profound neutropenia after transplantation, and anticipated increased risk of IFIs. In early reports, the incidence of invasive aspergillosis and other fungal infections after bone marrow transplantation (BMT) was <10% with a mortality rate of 94% (McWhinney et al, 1993; Morrison et al, 1994). Baddley et al (2001) documented an increased incidence of invasive aspergillosis in stem cell transplant recipients, rising from 7·3% in 1992 to 16·9% in 1998. This was felt to be related to prolonged neutropenia, cytotoxic chemotherapy, recent or concurrent therapy with broad-spectrum antibacterial agents, glucocorticoid treatment and disease relapse (Morrison et al, 1994; Yeghen et al, 2000; Marr et al, 2002).

In this study, CMV was recovered from only three patients (six episodes) in the UCB group and nine patients in the MUD group (17 episodes). In a study of 72 patients who received allogeneic BMT, the incidence of CMV antigenaemia was 57% during the first 18 weeks after transplantation, and the incidence of CMV-associated disease was 28% (Takenaka et al, 1997). Our observations of lower rates of CMV infection may be related to ganciclovir prophylaxis and, in the UCB group, the fact that all UCB grafts infused were CMV negative, because of the extremely low rates of maternal CMV transplacental transfer. In a study of 27 patients who received UCB grafts, only one patient developed CMV-related disease enteritis (Saavedra et al, 2002). In a more recent study of 28 patients who received UCB, the incidence of CMV reactivation/disease was higher than that in our study (79% vs. 24%); however, one of our patients developed CMV-disease versus none in their group (Tomonari et al, 2003). This difference could be explained by the difference in CMV serostatus; in our group, 57% were CMV positive, while 86% were seropositive in the study by Tomonari et al (2003). Furthermore, the strategies used for controlling CMV in these two studies differed; our patients received prophylactic ganciclovir for the first 100 d after transplantation, while Tomonari et al (2003) used a pre-emptive approach. Finally, the techniques for detecting CMV reactivation/diseases also differed, in our study we used a polymerase chain reaction and shell vial techniques versus CMV antigenaemia (Tomonari et al, 2003).

Progressive multifocal leucoencephalopathy is an uncommon demyelinating disease of the central nervous system, caused by reactivation of JC papovavirus, during periods of profound immunosuppression (Weber et al, 2001; Dworkin, 2002; Osorio et al, 2002). Immunosuppressive-associated leucoencephalopathy is a significant complication of CSA or tacrolimus therapy. This MUD recipient developed extensive steroid-resistant cGVHD, was treated with prolonged courses of steroids, CSA and Psoralen plus ultraviolet A light therapy.

In general the kinetics of immune reconstitution after UCB in our study were similar to that recently reported by Inoue et al (2003). Full analysis of lymphocyte immune reconstitution is beyond the scope of this study but would be an important focus of future analyses. Despite apparently normal myeloid recovery in UCB and MUD recipients, about 61% of all patients who survived for 50 d or more after transplantation developed one or more infections, suggesting continued impaired immune reconstitution. Although the myeloid recovery after MUD graft infusion was more rapid than after UCB, the rate of late infectious complications (beyond day +50) was similar in the two groups (62% and 61% respectively). One possible explanation for the delay in lymphoid recovery in the UCB group is the incorporation of ATG in the conditioning regimen. Studies have shown that ATG could be detected in sera at least 2 months post-transplant with 25% of the initial peak concentration on day 60, which correlated with the severely impaired T-cell recovery (Duval et al, 2002). In this study, we observed a significant improvement in the ALC count in UCB recipients beyond day +60 after transplantation (Fig 2).

We observed that slow ALC recovery in unrelated transplant recipients correlated with a significant increase in the rate of overall infection and bacterial infections, occurring throughout the first year. Although prior studies have reported correlation between faster lymphocyte recovery and EFS, relapse, and overall survival after transplantation (Pavletic et al, 1998; Kumar et al, 2001), the correlation with overall infection rates has not previously been reported.

Our study is limited by a relatively small sample size, and heterogeneity of the patients analysed. However, the benefit of this study lies in the information on specific pathogens and kinetics of infections outlined in this single institution retrospective analysis, data that is not generally available in large multi-institutional registry databases. Further benefit is the fact that these study patients were drawn from a single institution, thereby eliminating the potential ‘centre effect’ observed in larger patient cohort studies derived from cooperative group registry data. Nevertheless, as this study is retrospective and treatments were not randomly assigned, no firm conclusions can be drawn that confounding factors, including use of growth factors or antimicrobial agents, changed the risk or the course of infectious complications after transplant.

Infection-related mortality in this and other studies is difficult to determine, because majority of the patients died with multiple organ failure. Seven of our patients in the UCB group (35% of deaths) and six in the MUD group (33% of deaths) had serious infections at the time of death. In a study of 462 patients who underwent allogeneic transplantation from unrelated donors, infections were considered as the primary or secondary cause of death in 36% of patients (Kernan et al, 1993). In another study of 562 recipients of UCB transplants from unrelated donors, infection was reported to contribute to the death of 47% of the patients (Rubinstein et al, 1998).

In this study, EFS at 3 years was 25% in adults treated with full myeloablation and single unit unrelated UCB grafting, similar to prior reports (Barker et al, 2001; Laughlin et al, 2001). This is in contrast to a higher survival rate reported in a recent study, 7/8 of UCB recipients were alive and disease free at 1 year after transplantation (Ooi et al, 2002). These differences in survival comparing North America versus Japanese outcomes may be attributable to a more homogenous HLA genotype and shorter stature in Japanese reports rendering improved graft HLA matching and cell dose for the recipient. Approaches to optimize outcomes in UCB recipients are being explored, including multi-unit UCB transplantation and UCB ex vivo expansion (Shpall et al, 2002; Barker & Wagner, 2003; Jaroscak et al, 2003).

In summary, the major findings in this study are that unrelated UCB in adults is marked by a high incidence of early Gram-positive bacterial infections during the first 50 d after transplant, which correlates with the duration of neutropenia. Patients undergoing unrelated UCB or MUD allogeneic transplantation who develop bacteraemia prior to myeloid recovery appear to be more likely to experience primary graft failure. Lymphocyte recovery was delayed during the first 2 months in UCB patients but, importantly, normalized by day +200. Lymphocyte recovery remained below normal levels in MUD recipients throughout the first year after transplantation. We noted in our analyses that lymphocyte recovery correlated with overall infection rates in all unrelated allogeneic transplant patients. It is anticipated that this retrospective study of the kinetics of myeloid and lymphocyte engraftment and infectious complications after unrelated UCB transplantation may serve as a basis for future prospective studies to determine appropriate prophylactic and therapeutic antimicrobial regimens.


The authors thank the clinical staff of the University Hospitals of Cleveland Blood and Marrow Transplant Program for their compassion and hard work in the care of their patients; and specifically to the laboratory staff of this transplant program laboratory, for characterizing and processing the infused UCB units and MUD grafts. The authors also thank Pablo Rubinstein MD, Cladd Stevens MD, and Carmelita Carrier PhD, of the Lindsley F. Kimball Research Institute, New York Blood Center, Placental Blood Program, and the staff of the Milano Cord Blood Bank, and Knochenmarkspenderzentrale (Eurocord-Germany, Dusseldorf), for co-ordinating UCB search processing, and careful analyses of potential UCB grafts. This work was supported by the Phyllis and Samuel Katz Foundation (M.J.L.), Elsa U. Pardee Foundation (M.J.L.), and Leukemia and Lymphoma Society of America (grant no. 2009–99, M.J.L.). M. J. Laughlin MD is a Leukemia and Lymphoma Society of America Scholar in Clinical Research.