Impact of in vivo alemtuzumab dose before reduced intensity conditioning and HLA-identical sibling stem cell transplantation: pharmacokinetics, GVHD, and immune reconstitution

Reduced intensity conditioning before allogeneic hematopoietic stem cell transplantation has considerably broadened the applicability of this approach as therapy for patients with hematologic malignancy. This treatment strategy relies on the graft-versus-tumor (GVT) effect to eradicate persistent cancer. A critical barrier to the success of this approach is the risk of graft-versus-host disease (GVHD), which can affect between 30% and 70% of patients according to the protocol used and population treated.¹  The development of GVHD contributes to a higher risk of nonrelapse mortality (NRM) either directly or indirectly, as a result of infection occurring in persons who require intense immunosuppression. Efforts to reduce the risk of GVHD have included incorporation of alemtuzumab into reduced intensity conditioning regimens.^2,3  Alemtuzumab is a humanized immunoglobulin G₁ anti-CD52 monoclonal antibody that has broad lymphocyte-depleting properties.⁴  When used in vivo before graft infusion, it is highly effective in preventing GVHD with reported incidences of 2%-21% for acute grade II-IV GVHD and 0%-13% for chronic extensive GVHD.^2,3,5-8  This lack of GVHD translates into significant reductions in treatment-related deaths, a benefit that is particularly evident in the unrelated donor setting.²  Furthermore, this approach permits the use of subsequent cellular therapies, such as donor leukocyte infusions (DLIs)⁹  or antigen-specific T cells, otherwise contraindicated in the presence of early GVHD.

We have previously reported the use of alemtuzumab at a fixed total dose of 100 mg divided over 5 days (days −8 to −4) in the context of a fludarabine and melphalan–conditioning protocol.³  With the use of this dose regimen, the terminal half-life of the antibody is 8 days, and the antibody is still detectable at levels sufficient to induce antibody-mediated cellular cytotoxicity (> 100 ng/mL ¹⁰ ) 2 months after transplantation.¹¹  As a consequence, immune reconstitution is markedly delayed and patients have frequent infections.^12-14  Slow kinetics of immune reconstitution after reduced intensity conditioning may also lead to high rates of mixed chimerism and may negatively affect the overall integrity of the GVT response. Delayed adoptive transfer of donor T cells is efficient at correcting mixed chimerism and in certain disorders at mediating GVT effects.⁹  However, the delay in immune recovery may be a significant disadvantage in patients with advanced disease or rapid tumor growth when GVT responses are required early after transplantation.

We reasoned that reductions in the dose of alemtuzumab at the time of transplantation might improve immune reconstitution without increasing the risk of GVHD. We therefore conducted a prospective, multicenter trial to determine the feasibility of alemtuzumab dose deescalation in adult patients undergoing human leukocyte antigen (HLA)–identical sibling transplantation. The results show the feasibility of safely reducing the alemtuzumab dose to 30 mg. However, further dose reductions to 20 mg are associated with a greater risk of severe GVHD and unfavorable pharmacokinetic properties, particularly in patients with persistent CD52⁺ disease at the time of transplantation.

Patients were recruited from September 2003 to February 2007 at 10 United Kingdom transplantation centers. The protocol was approved by the United Kingdom Multicenter Research Ethics Committee and by individual center institutional review boards. All patients and donors gave written informed consent in accordance with the Declaration of Helsinki. Eligible patients were those with hematologic malignancies who because of age, prior therapy or diagnosis, would not normally be offered allogeneic transplantation with ablative conditioning but were suitable for a reduced intensity hematopoietic stem cell transplantation. Other inclusion criteria were age 18-65 years, life expectancy > 3 months, and possession of ≥ 1 HLA identical sibling. Exclusion criteria were creatinine clearance < 40 mL/minute, left ventricular ejection fraction < 40%, bilirubin > 3 times above normal limit, pregnancy or lactating women, and serious comorbidity that would limit lifespan or ability to tolerate chemotherapy. Patient pretransplantation characteristics are detailed in Table 1.

Conditioning consisted of fludarabine 30 mg/m²/day from day −7 to day −3, melphalan 140 mg/m² on day −2, and alemtuzumab. The dose of alemtuzumab was reduced stepwise in 4 groups: 60 mg (2 × 30 mg given intravenously on days −2 and −1), 40 mg (2 × 20 mg given intravenously on days −2 and −1), 30 mg (1 × 30 mg given intravenously on day −1), and 20 mg (1 × 20 mg given on day −1). Each cohort was planned to be of 25 patients. With the use of a type 1 error of 0.05, a group size of 25 would provide a 0.8 probability of detecting an increase in the proportion of patients with acute grade II-IV GVHD from 10% (the expected level with the use of in vivo alemtuzumab) to a clinically significant proportion of 30% (the lower limit of the expected level after HLA-identical sibling transplantations in the absence of alemtuzumab). Recruitment to subsequent cohorts required the Data and Safety Monitoring Committee to review data from the previous cohorts and invoke stopping rules if required. Further accrual within the current cohort was permissible until this had been completed. All patients were given intravenous methylprednisolone 2 mg/kg intravenously at the time of the first alemtuzumab infusion. According to our standard protocol, patients were given ciclosporin (3 mg/kg/day intravenously from day −1), which was tapered from 3 months after transplantation in the absence of GVHD. The preferred donor stem cell source was granulocyte-colony stimulating factor–mobilized and collected from peripheral blood. Bone marrow was permissible when donors indicated a preference.

All patients received prophylaxis against Pneumocystis jiroveci and acyclovir prophylaxis against varicella zoster virus reactivation. Surveillance for cytomegalovirus (CMV) infection was performed by weekly polymerase chain reaction or antigenemia testing, and preemptive treatment was administered according to institutional guidelines. Peripheral blood was assayed for chimerism (either on unseparated mononuclear cells, or in T-cell, B-cell, and myeloid lineages) by polymerase chain reaction analysis of informative minisatellite regions. Disease status after transplantation was monitored regularly from 3 months, and patients were eligible for treatment with DLI in the presence of mixed chimerism or residual disease from 6 months. DLI were contraindicated in the presence of > grade I GVHD, and the standard starting dose was 1 × 10⁶/kg, with further DLI given if needed at 3 monthly intervals, at 3-fold escalating doses. DLI could also be administered at any time in the presence of clinically progressive disease, the starting dose varied according to time from transplantation and disease.

Alemtuzumab levels were determined by enzyme-linked immunoabsorbent assay. CD52 fusion protein (obtained from Prof Herman Waldmann, University of Oxford¹⁵ ) was absorbed onto a microtiter plate at a concentration of 2.5 μg/mL, 100 μL/well, and the plates were blocked with PBS containing 2% (wt/vol) bovine serum albumin (BSA). Test samples were applied and incubated for 1 hour at 25°C. The plates were washed, and bound alemtuzumab was detected with a monoclonal peroxidase-labeled mouse anti–human immunoglobulin G₁ followed by tetramethylbenzidine substrate. After stopping the reaction with 5% (vol/vol) HCl, the absorbance was measured at 450 nm with background correction at 540 nm. The concentration of test and quality control samples was determined by interpolation on a standard curve with the use of a 4-parameter logistic model. The detection limit was 37.5 ng/mL.

Mixed-effects modeling was performed in NONMEM VI 2.0 (ICON Development Solutions). NONMEM run was called by PsN 3.1.¹⁶  A 2-compartment model with nonlinear elimination was first fitted with the first-order method by calling subroutine ADVAN 6. A 2-compartment model with linear elimination was fitted with first-order conditional estimation by calling subroutine ADVAN 3. Intersubject variability was evaluated for all parameters except for the intercompartmental clearance in ADVAN 6. Random effects were treated as log-normal distribution. The residual error was modeled as a combined constant coefficient and additive error. For the linear elimination model (ADVAN 3), the dose was used as a covariate and entered the model in a power function and screened for all parameters. The final model was selected as the one with the lowest objective function value. A visual predictive check was performed to verify the adequacy of the final model.

The primary end points of the study were the cumulative incidence of acute and chronic GVHD, NRM, the incidence of full donor chimerism, and infection. Pharmacokinetic studies were performed in a subset of patients. Secondary end points included overall survival (OS) and progression-free survival (PFS).

Analysis was performed on data accrued to April 30, 2008. OS and PFS were estimated according to the Kaplan-Meier method. The log-rank test was used to compare survival curves for different subgroups. Time-to-event outcomes with competing risks (NRM and GVHD) were estimated by cumulative-incidence analyses. NRM was determined from the date of transplantation until death from causes other than relapse. Relapse was considered a competing event for NRM analyses. For GVHD analyses, death or relapse were considered competing events, and patients receiving DLI were censored at the time of DLI. Acute and chronic GVHD were determined and scored according to criteria of the European Group for Blood and Bone Marrow Transplantation. Outcomes were compared with the Gray method. For continuous variables, comparisons across multiple groups were made with the 1-way analysis of variance and comparisons between 2 groups by the Student t test. For proportions, comparisons were made by the Fisher exact test. P values ≤ .05 were considered significant.

A total of 106 patients were recruited to the trial. Median follow-up was 2.6 years, and minimum follow-up was 14 months. The mean age was 49 years. Patient pretransplantation characteristics were similar in each cohort, although there were more patients with advanced disease (defined as acute leukemia > CR1, chronic myeloid leukemia > first chronic phase) or prior autologous transplantation in the 60-mg cohort (Table 1).

The median time for neutrophil engraftment overall was 13 days (> 0.5 × 10⁹/L, interquartile range [IQR], 11-15 × 10⁹/L), and the median time for platelet engraftment overall was 14 days (> 50 × 10⁹/L, IQR, 12-22 × 10⁹/L). No differences were observed in engraftment kinetics between the groups (data not shown). A single patient in the 60-mg group did not engraft and went on to receive a successful donor CD34-selected top-up without further conditioning.

The numbers of patients in each dose cohort developing acute pattern grade II-IV or chronic extensive GVHD before DLI are shown in Tables 2, 3, and 4. As shown in Table 3, there was a significantly higher risk of acute grade II-IV GVHD (hazard ratio [HR], 16.6; 95% CI, 1.8-150), chronic extensive GVHD (HR, 4.1; 95% CI, 1.4-12.4) and any severe GVHD (acute grade II-IV or chronic extensive; HR, 6.7; 95% CI, 2.5-18.3) in the 20-mg group compared with the > 20-mg cohort. In contrast, HR comparisons for GVHD between the 30-mg and > 30-mg group showed no significant differences (Table 4). Competing risk estimates for 1-year cumulative incidences of severe GVHD in the 20-mg group were 57% (95% CI, 28%-78%) versus 11% (95% CI, 4%-21%) in the > 20-mg group (Figure 1A; P < .001). In contrast, the 1-year cumulative incidences of severe GVHD were similar in the 30-mg group compared with the > 30-mg groups (6%, 95% CI, 0%-25% vs 14%, 95% CI, 5%-27%; Figure 1C; P = NS). In total, 5 patients died of complications relating to GVHD (n = 1 in the 60-mg group; n = 1 in the 40-mg group, and n = 3 in the 20-mg group).

Forty-three patients received DLI for mixed chimerism (n = 30) or persistent/relapsed disease (n = 13). The number of patients developing either acute pattern grade III-IV or chronic extensive GVHD after DLI was 7 in total (16%), and no significant differences were observed between the groups (60 mg, 3 of 13; 40 mg, 0 of 9; 30 mg, 1 of 11; and 20 mg, 3 of 10).

Competing risk estimates for cumulative incidences of day 100, 1-year, and 2-year NRM are shown in Tables 2 and 4. The major causes of death were infection (n = 9) or infection complicating GVHD (n = 5). Other causes of death included respiratory failure (n = 2), ciclosporin toxicity (n = 1), intracranial hemorrhage (n = 1), and secondary cancer (n = 1). Overall, day 100, 1-year, and 2-year estimates of NRM were 5% (95% CI, 2%-10%), 12% (95% CI, 7%-19%), and 17% (95% CI, 10%-24%), respectively. As shown in Tables 2 and 4 and Figure 1B and D, no differences were evident in the cumulative incidences of NRM between the individual groups (Table 2) or in comparisons between the 20-mg and > 20-mg (Table 3; Figure 1B) or 30-mg and > 30-mg groups (Table 4; Figure 1D).

Overall, Kaplan-Meier estimates of 2-year OS and PFS were 72% (95% CI, 63%-80%) and 59% (95% CI, 48%-68%), respectively. As shown in Tables 2 and 4, no differences were evident between the groups.

Chimerism data were available for 87 of 99 evaluable patients (Table 2-4). Overall, initial chimerism at 3 months was full donor in 34 patients (39%), mixed in 52 patients, and autologous in 1 patient. No differences were evident between the groups. On withdrawal of immunosuppression or subsequent DLI, the proportion of patients attaining full donor chimerism at 2 years increased to 75% overall, again with no differences between the individual groups.

The median number of infections per patient was 2 (IQR, 1-4), and the number did not differ between the groups (Tables 2,Table 3–4). No differences were observed in the number of patients dying as a result of infection (data not shown). The proportion of CMV seropositive patients developing CMV reactivation was also similar in each group, and there were no differences in the incidence of CMV disease (Tables 2,Table 3–4).

Although no differences were evident in the risk of infections, we also examined the recovery of lymphocytes at 1, 3, 6, and 12 months after transplantation. In this evaluation, patients receiving protocol DLI for mixed chimerism within the first 12 months were included (60 mg: n = 7, median DLI dose, 4 × 10⁶ CD3⁺ cells/kg of recipient weight; 40 mg: n = 5, median dose, 3 × 10⁶ CD3⁺/kg; 30 mg, n = 7, median dose, 4 × 10⁶ CD3⁺/kg; and 20 mg, n = 9, median dose, 1 × 10⁶ CD3⁺/kg), whereas patients with relapsed or progressive disease were excluded. No differences in the timing of DLI initiation were observed between the groups (60 mg: median, 229 days, range, 183-245 days; 40 mg: median, 239 days, range, 223-301 days; 30 mg: median, 234 days, range, 205-317 days; and 20 mg: median, 231 days, range, 173-313 days). As shown in Table 4, absolute lymphocyte numbers at 12 months in evaluable patients were greater in the 30-mg compared with the > 30-mg cohort, with more than 75% of patients in the former group having counts greater than the lower limit of the normal range (number of evaluable patients 30 mg, n = 20 and > 30 mg, n = 30). In contrast, in the > 30-mg group, lymphocyte recovery was relatively delayed with > 50% of patients having subnormal lymphocyte counts.

At a number of centers, blood samples were taken before transplantation, during conditioning, and up to 28 days after transplantation (day −3, 0, 7, and 28; n = 5-6 each group; Figure 2A). After completion of the trial, we determined alemtuzumab concentrations by enzyme-linked immunoabsorbent assay (lower limit of detection, 37.5 ng/nL). Although wide interindividual variation was observed within each group, there were significant differences in peak concentrations measured in the 60-mg group compared with the lower dose groups (Figure 2B). Peak serum concentrations in the 40-mg cohort overlapped considerably with the 20- and 30-mg cohorts. By day 28, alemtuzumab levels had fallen in all groups, but especially in patients receiving 20- and 30-mg doses, whereby 2 of 5 patients in the 30-mg group and 3 of 5 patients in the 20-mg group had levels < 100 ng/mL (Figure 2B). Pharmacokinetic data were analyzed with NONMEM mixed-effects modeling, and choice between individual models was made by data adequacy and a goodness-of-fit evaluation. In this sample, a linear 2-compartment model was most suitable for describing alemtuzumab concentrations, and we used this to estimate clearance, central volume of distribution (V₁, representing unbound antibody in serum), peripheral volume of distribution (antibody outside the circulation), and intercompartmental clearance in each group. As shown in Figure 2B through D, reductions in the dose of alemtuzumab were associated with progressive increases in peripheral distribution but reciprocal decreases in central distribution, reflecting incomplete saturation of antibody binding sites with reduced doses. These compartmental differences were reflected in greater clearance at lower alemtuzumab doses (Figure 2B-D).

One implication of this finding was that patients with CD52-expressing tumors at the time of transplantation might show more rapid clearance of antibody, especially at low dose. To test this, we categorized 49 patients, when peak alemtuzumab levels were available, according to status at transplantation and predicted tumor expression of CD52.¹⁷  As shown in Figure 3A, peak alemtuzumab levels were broadly similar in patients with and without CD52-expressing tumors at the time of transplantation in the 30-, 40-, and 60-mg groups. However, in the 20-mg cohort, the presence of CD52⁺ tumor at the time of transplantation was associated with significantly lower peak concentrations. In Figure 3B, examples of pharmacokinetic profiles of alemtuzumab elimination are shown for 2 patients with chronic lymphocytic leukemia (CLL) in partial remission treated within the 60-mg and 20-mg groups and compared with patients without CD52-expressing tumor. The number of patients with tumors predicted to express CD52 at the time of transplantation was not significantly different between the groups (60 mg, n = 6; 40 mg, n = 7; 30 mg, n = 9; and 20 mg, n = 11). However, 7 of 33 evaluable patients (21%) with CD52-expressing tumors present at the time of transplantation developed grade II-IV acute GVHD compared with 1 of 72 patients (1.4%) without CD52⁺ tumor (P = .001). The risk of chronic extensive GVHD did not differ significantly between evaluable patients with or without CD52⁺ tumor at the time of transplantation (5 of 32 versus 8 of 67, respectively). In the 20-mg group, all 4 patients with grade II-IV acute GVHD had a CD52-expressing tumor present at the time of transplantation, as did 4 of 6 patients developing chronic extensive GVHD.

As shown in Figure 3A, substantial variation was observed in peak concentrations within each dose group and significant overlap between the 3 lower dose groups. To test whether peak concentrations would be predictive of outcome, we performed a post hoc analysis of outcomes for patients with peak levels below, or at and above the median value (4177 ng/mL). No differences between baseline variables were present between the 2 groups (data not shown). With the use of this approach, no differences in GVHD incidence (2-year cumulative incidence severe GVHD, 24%; 95% CI, 7%-47% below median vs 15%; 95% CI, 4%-34% at or above median) or NRM (2-year cumulative incidence NRM, 13%; 95% CI, 29% below median vs 9%; 95% CI, 2%-25% at or above median) were evident. Further analyses, on the basis of other categorizations (eg, highest vs lowest quartile) were not possible because of the small number of patients and events.

In this multicenter, prospective trial, we have demonstrated the feasibility of substantial reductions in the dose of in vivo alemtuzumab in the setting of fludarabine-melphalan conditioning for HLA-identical sibling transplantations. Dose reductions to 30 mg, given at day −1, were compatible with a low risk of GVHD, no increase in NRM, and improved lymphocyte recovery at 1 year. In contrast, further dose reduction to 20 mg led to higher rates of severe GVHD, and peak antibody levels were more likely to be influenced by the presence of CD52⁺ tumor at the time of transplantation.

One recent observational study has also reported the potential feasibility of reducing in vivo alemtuzumab dosing before reduced intensity allogeneic stem cell transplantation. Bertz et al¹⁸  used a similar time schedule to our study for alemtuzumab dose deescalation (day −2 and −1) in their reported experience of unrelated and related donor transplantations with the use of a variety of fludarabine-based protocols. In their report, reducing the dose to 10-20 mg on day −1 was associated with cumulative incidences of acute grade II-IV GVHD of 10%-16% after sibling and 28%-34% after unrelated donor transplantations. The incidences of chronic extensive GVHD in these groups were 14%-17% overall. They concluded that the optimal alemtuzumab dose was between 10 and 20 mg on day −1, irrespective of the donor type. This is a lower dose than our results would suggest is feasible if the risk of severe GVHD is to stay relatively low, with our optimal dose being 30 mg on day −1. One potential caveat to their findings is that most of the patients in their series were treated for myeloid malignancies that lack expression of CD52, and only 4% of patients in the lowest dose cohort had a lymphoproliferative disorder. Thus, it is possible that rates of alemtuzumab clearance differed substantially from our trial in which a greater proportion of trial participants had CD52-expressing indolent lymphoproliferative disorders. Accordingly, we would recommend that substantial alemtuzumab dose reductions in unrelated donor patients should preferably take place in the context of a prospective study in which issues such as the effect of diagnosis or the degree of mismatch can be addressed in more detail.

A linear, 2-compartment model was adequate to describe the pharmacokinetic data in this study, similar to those reported for other monoclonal antibodies.^19-21  At the highest dose, V₁ values (central distribution) corresponded to those reported for other antibodies that are largely unbound in the serum. With reducing doses, falls in V₁ were associated with increases in clearance and in peripheral distribution, suggesting incomplete target saturation. Recently, a more complex, nonlinear elimination model has been described in patients with B-cell CLL treated with alemtuzumab according to a multiple-dose schedule.²²  In this context, rates of elimination varied according to antibody dose and to patient lymphocyte numbers that changed with time.²²  Thus, there was an inverse correlation between lymphocyte counts and serum alemtuzumab levels, reflecting a higher number of CD52 binding sites in patients with significant tumor burdens. Consistent with this, we found that the risk of severe acute GVHD was associated with the presence of CD52-expressing tumor at the time of transplantation and reduced peak antibody concentrations when alemtuzumab was given at the lowest 20-mg dose. At doses of ≥30 mg, peak drug concentrations were less affected, suggesting that this need not be a consideration except perhaps in patients with very bulky disease. More detailed pharmacokinetic studies at this dose level will be required to determine to what extent other factors, including other measures of disease burden (eg, soluble CD52), influence antibody elimination. As an alternative, use of “in the bag” administration of alemtuzumab¹¹  also merits further investigation because concentrations of antibody may be less prone to variation compared with an in vivo strategy.

Independent of tumor burden, it is clear that there remain significant variations in alemtuzumab concentrations that are unexplained. The mechanisms underlying clearance of alemtuzumab from the circulation and extracellular space are not known, although presumably they may be influenced by hepatic function and the monocyte-macrophage system. In this study, we did not test whether patients with prior exposure to alemtuzumab had preexisting anti–rat globulin responses that may have enhanced antibody clearance. However, the reported rates are very low and occur primarily after repeated subcutaneous administration, making this unlikely in our cohort of patients.²³  In any case, despite substantial variation in levels, we were unable to demonstrate a relationship between peak concentration attained and any clinical outcome. This suggests that other pharmacokinetic parameters may be of greater importance in predicting responses. It will be important also to consider how host factors might affect drug pharmacodynamics when designing future studies. For example, cellular susceptibility to antibody-dependent cellular cytotoxicity might be relevant to response. Although variations in antibody-dependent cellular cytotoxicity–mediated killing have been linked to Fc receptor polymorphisms in patients treated with rituximab,²⁴  a recent study showed no relationship between responsiveness of CLL to alemtuzumab and FCGR3A and FCGR2A polymorphisms,²⁵  but there may be other polymorphisms that affect on outcome.

Although lymphocyte recovery at 12 months was more rapid in the 30-mg group than the higher dose cohorts, early lymphocyte recovery, the rate of infections, and the incidence of full donor chimerism were similar in all groups. Thus, we identified no optimal dose that enabled dramatic improvements in immune reconstitution without inducing GVHD. This implies that efforts to bolster early immune reconstitution with the strategy of in vivo T-cell depletion at the time of transplantation will continue to require add-back of therapeutic T cells in the posttransplantation period. For example, we and others, have demonstrated the potential of transferring antigen-specific T cells as a preemptive or prophylactic strategy to accelerate anti-CMV immune reconstitution.^26-28  This strategy is being extended to other pathogens, including Epstein-Barr virus, adenovirus, and Aspergillus.^29,30  Transfer of unselected DLI with the use of an escalated dose regimen can correct mixed chimerism or eradicate persistent disease, with only a low risk of GVHD in the HLA-identical setting.⁹  In this study, nearly one-half of the patients received DLI, and significant GVHD was only observed in 16% of patients overall, emphasizing the general applicability of this approach despite substantial reductions in alemtuzumab dose. Other approaches such as CD8 depletion of DLI may also be of value in reducing the risk still further, especially in patients in whom the intention is to correct mixed chimerism.^31,32 

In conclusion, this prospective study has shown that a substantial dose reduction of alemtuzumab to 30 mg (given on day −1 as a single infusion) is possible in the context of reduced intensity conditioned HLA-identical sibling transplantations without increasing NRM or the incidence of GVHD while improving immune reconstitution in terms of lymphocyte numbers. In future United Kingdom collaborative studies that involve HLA-identical sibling transplantations, we therefore intend to adopt the dose of 30 mg as standard. Although this trial was not primarily designed to evaluate differences in PFS, an important question for future studies that involve a less heterogenous group of patients will be whether deescalation of the alemtuzumab dose will improve initial disease control and reduce the risk of relapse.

We thank Chris Cox and other staff members at BioAnaLab (Millipore) who carried out the enzyme-linked immunoabsorbent assay studies.

Contribution: G.O. collated and verified data from individual trial centers; M.R. performed the statistical analyses; J.S. performed the pharmacokinetic analyses; A.F., P.K., D.M., M.C., C.C., P.J., A.C., A. Parker, A. Pettitt, A.B., R.P., J.S., R.C., K.P., K.T., and E.M. were trial investigators at each of the clinical centers participating in the study and contributed to the writing of the paper; G.H. oversaw the enzyme-linked immunoabsorbent assay studies; R.C. oversaw the collation of the data, performed the analysis, and wrote the paper; and S.M. designed the trial, oversaw the analysis, and wrote the paper.

Conflict-of-interest disclosure: S.M. has received an unrestricted educational grant from Bayer-Schering. The remaining authors declare no competing financial interests.

Correspondence: Ronjon Chakraverty, Transplantation Immunology Group, Department of Hematology, Royal Free and University College Medical School, Rowland Hill St, London, NW3 2PF; e-mail: r.chakraverty@medsch.ucl.ac.uk.