|Home | About | Journals | Submit | Contact Us | Français|
Double cord blood transplantation (DCBT) may overcome the slow hematopoietic recovery and engraftment failure associated with infusion of a single cord blood unit. In DCBT, only one unit typically contributes to long-term hematopoiesis, but little is known about factors affecting cord predominance. As results from a phase I trial suggested that order of infusion may affect cord predominance, we analyzed the effect of preinfusion variables on chimerism patterns of 38 patients enrolled in the initial study and a subsequent phase II trial. All patients were treated with a reduced-intensity conditioning (RIC) regimen of fludarabine, melphalan and thymoglobulin followed by DCBT. Byday100, 66% of patients had hematopoiesis derived from a single cord blood unit. Higher post-thaw total nucleated cell and CD34+ cell dose were associated with cord predominance and in 68% of patients (P = 0.03); the predominant cord blood unit was infused first. Only the post-thaw CD34+ cell dose of the predominant unit predicted time to both neutrophil and platelet engraftment. Although based on a small number of patients, our results identify parameters that may affect cord predominance and engraftment in the setting of DCBT following RIC and suggest possible strategies for selecting infusion order for cord blood units.
Umbilical cord blood (UCB) transplantation is a viable option for both adults and children with hematologic malignancies.1–4 Advantages of UCB include lower rates of GVHD and greater flexibility in HLA matching.5 The major disadvantage of UCB transplantation in adults is that the low cell dose (10-fold lower than a bone marrow harvest) results in delays and failures of hematopoietic recovery.1,3 To overcome this limitation, we and others have attempted infusion of multiple UCB products. In general, these studies showed promise in terms of disease-free survival and demonstrated the feasibility of double cord blood transplants.6–10
Barker et al.7 reported in a double cord blood transplantation (DCBT) study using myeloablative conditioning that 76% of patients at day 30 and 100% of patients at day 100 had only a single cord contributing to hematopoiesis. In a recent study of 93 patients receiving a reduced-intensity conditioning (RIC) regimen followed by DCBT, only one unit was detected in 91% of the patients at day 100 and all the patients beyond 1 year.10 No studies have reliably identified preinfusion variables that significantly influence cord predominance.10–12
In an initial phase I study of DCBT with an RIC regimen, we unintentionally found the unit infused first was more likely to predominate (P = 0.05).6 To better assess preinfusion variables that predict cord predominance and engraftment in the setting of DCBT, we analyzed all patients treated with this regimen between 2003 and 2006. While the number of patients is small, our analysis helps to identify into factors leading to cord predominance and engraftment in the setting of reduced-intensity DCBT.
The data presented include patients enrolled in two RIC DCBT protocols approved by the Dana-Farber/Harvard Cancer Center Office for the Protection of Research Subjects. The first trial was phase I and the second was phase II.6 All consecutive subjects enrolled between 2003 and 2006 were included except those with early engraftment failure (that is, never achieved neutrophil engraftment by day 35) or death prior to obtaining a day 30 chimerism result. Written informed consent was obtained from all patients prior to participation. Eligible patients were hematopoietic stem cell transplant candidates between the ages of 18 and 65 with no available 6/6 or 5/6 HLA-A, B and DRβ1 high-resolution matched related or 6/6 unrelated adult donor. An ECOG score of 0, 1 or 2 was required. Patients with chemotherapy-sensitive relapsed non-Hodgkin’s lymphoma (NHL), Hodgkin’s lymphoma (HL), AML or ALL in CR 2 or greater, or CR 1 with high-risk cytogenetics, CLL having failed two prior chemotherapy regimens, CML in accelerated or second stable phase and myelodysplasia (MDS) were eligible.
The two UCB products selected were compatible, based on allele-level high-resolution typing, with the recipient and each other at a minimum of 4/6 HLA-A, B and DRβ1 loci. The products had to have a minimum combined pre-cryopreservation total nucleated cell (TNC) dose of 3.7 × 107 per kg. Conditioning consisted of fludarabine (30 mg/m2 per day on days −8, −7, −6, −5, −4 and −3; total 180 mg/m2), melphalan (100 mg/m2 per day on day −2) and rabbit antithymocyte globulin (thymoglobulin) (1.5 mg/kg per day on days −7, −5, −3, −1; total dose 6.0 mg/kg). UCB products were obtained from national and international cord blood banks, shipped frozen and stored locally until completion of conditioning. CBUs were thawed using the technique of Rubinstein et al.13 The second CBU was thawed only after confirmation that the first unit had been infused successfully leading to the two units being administered between 1 and 6 h apart (94% >4 h apart). In the first cohort of 17 patients, as there was no available data on cord unit selection for infusion order, the units were chosen randomly for infusion. On the basis of our results suggesting order of infusion may impact CBU predominance, for the second cohort of 21 patients, the unit with the higher TNC dose (based on collection center TNC) was infused first.6
Measurement of post-thaw unit CD34+ cell content was performed using a dual platform method (flow cytometry and TNC measurement). Measurement of post-thaw unit TNC content was performed using the Coulter ActDiff (Beckman Coulter, Miami, FL, USA).
The two protocols differed in the GVHD prophylaxis strategies. In the first cohort (17 patients), patients received continuous intravenous infusion cyclosporine starting on day −1 and mycophenolate mofetil (MMF) 15 mg/kg intravenously twice daily beginning on day 0. On day +4, the cyclosporine was infused at 3 mg/kg until the patient could tolerate oral medication. After day +5, the cyclosporine was dosed by level (target 200–300 ng/ml). In the absence of GVHD, the MMF was tapered between days +30 and +60, and the cyclosporine was tapered between days +100 and +180 post transplant. In the second cohort (21 patients), GVHD prophylaxis consisted of sirolimus (12mg loading dose on day −3 followed by a 4mg daily maintenance dose) and tacrolimus (0.05 mg/kg twice a day starting on day −3).14 Target sirolimus trough levels were 3–12 ng/dl and target tacrolimus levels were 5–10 ng/dl. In the absence of GVHD, the tacrolimus and sirolimus were tapered on approximately day 100 (±2 weeks) post transplant. The tapering schedule was left to the treating physician’s discretion.
Peripheral blood samples after transplantation were collected for donor chimerism studies. Total leukocyte chimerism was determined by analysis of short tandem repeat loci (Profiler Plus, Applied Biosystems, Foster City, CA, USA). On the basis of the sensitivity levels for the assay, full donor chimerism was defined as ≥95%. Chimerism was analyzed at three time points: day 30 (±10 days), day 60 (±10 days) and day 100 (±10 days). We defined the predominant unit as either the only CBU engrafting or, in the case of two units engrafting, the unit contributing to >60% hematopoiesis at day 100. In patients who died before day 100, the unit contributing to >60% hematopoiesis as of the last chimerism result was considered predominant. There was not sufficient data on the majority of patients for analysis of T-cell and/or myeloid chimerism.
Myeloid engraftment was defined as an ANC of >500 cells per μl for 2 consecutive days. Short-term platelet engraftment was defined as two consecutive platelet counts on different days >20 000 cells per μl, without transfusion in the preceding 7 days. Long-term platelet engraftment was defined as two consecutive counts above 100 000 cells per μl. White blood cell and platelet counts were monitored daily during hospitalization, weekly until day 100 and then at least monthly.
The Wilcoxon rank-sum test and Fisher’s exact test were used to examine differences in baseline characteristics between the two protocols. McNemar’s test, the Wilcoxon signed-rank test and a one-sample test of binomial proportion were used to identify cord blood factors associated with the predominant unit. All analyses involving TNC and CD34+ cell dose were based on post-thaw results. A log-linear model was used to determine the independent effects related to individual unit contributions of TNC and CD34+ cell dose and order of infusion on cord predominance. The times to ANC and platelet engraftment were based on the cumulative incidence estimated in the presence of death, relapse or late graft failure as competing risks. The relationship between engraftment time and continuous cord blood parameters was analyzed using the Gray test to compare the cumulative incidence curves of two groups defined by the median value as a threshold.15 The Gray test was stratified by protocol to control for cord blood effects on engraftment due to GVHD prophylaxis. A time-varying covariate was used to model the effect of chimerism group in a proportional hazards model of engraftment time. Patients who had not engrafted or failed due to a competing risk were censored at the last follow-up date their blood counts were checked. All P-values for these effects are based on two-sided hypothesis tests. The statistical analysis was computed using SAS 9.1, except R version 2.5.1 was used to perform the competing risks analysis.
A total of 43 patients were enrolled between 2003 and 2006. Of those enrolled, three patients failed to engraft and two patients died prior to obtaining day 30 chimerism results. These five patients were not included in the analysis as the major end points required sufficient chimerism data.
The characteristics of the 38 patients and the cord blood units (CBUs) used for their transplants are shown in Tables 1 and and2.2. There were 17 patients from the first protocol (GVHD prophylaxis of MMF and cyclosporine) and 21 from the second (GVHD prophylaxis of sirolimus and tacrolimus). The only difference in baseline characteristics of patients was a trend for disproportionately fewer female patients in the cyclosporine/MMF compared to the sirolimus/tacrolimus protocol. In regard to CBUs infused, there was a trend for units from the cyclosporine/MMF group to contain a lower post-thaw TNC dose on average compared to the sirolimus/tacrolimus protocol. Otherwise, there were no significant differences in CD34+ cell dose or age of the CBUs as well as in HLA, ABO or sex matching between each CBU and the recipient. Given the identical conditioning regimen as well as eligibility and cord blood selection criteria, the two protocols were combined for subsequent analyses to obtain greater statistical power.
We assigned patients to one of three groups based on the pattern of chimerism observed on days 30, 60 and 100 following stem cell infusion. Group 1 consisted of patients in whom a single CBU contributed to ≥95% of hematopoiesis. In group 2, both CBUs contributed to hematopoiesis. The third group consisted of patients with contributions of one CBU and the recipient. The different chimerism patterns at day 30 (n = 35 patients: three patients did not have chimerism tested), day 60 (n = 34 patients: two patients died and two patients did not have chimerism tested) and day 100 (n = 29 patients: three additional patients died, one lost the cord blood graft and three did not have chimerism tested) are depicted in Figure 1. At day 30, 29% of patients had single CBU hematopoiesis (group 1). As time progressed, patients transitioned to complete single unit chimerism with 50% in this group by day 60 and 66% in this group by day 100.
Sixteen patients have follow-up for over 1 year. Fourteen had chimerism from a single CBU by day 100 and all have retained complete single donor chimerism (five of these over 2 years and two over 3 years). Only one patient with complete single donor chimerism at any time point, in the absence of graft failure or autologous reconstitution, lost complete single cord unit chimerism over time. He reached 95% single donor chimerism on day 65. On day 177, the last chimerism result before the patient died a month later from post transplant lymphoproliferative disorder, the predominant cord’s contribution to hematopoiesis dropped to 86% with the remaining 14% of hematopoiesis coming from the second CBU. The two remaining patients followed at least 1 year (one for 487 days and one for 713 days) continued to have contributions of both CBUs to hematopoiesis with 83 and 66% chimerism of the predominant CBU.
We defined the predominant unit as either the only CBU engrafting or, in the case of two units engrafting, the unit contributing to >60% hematopoiesis at day 100. In patients who died before day 100, the unit contributing to >60% hematopoiesis as of the last chimerism result was considered predominant. As there is no accepted standard for defining unit predominance, we based our choice on the observation that once a CBU contributed >60% of hematopoiesis at day 60 this fraction was maintained or increased, in the absence of graft failure or autologous reconstitution, over long-term follow-up.
Post-thaw TNC dose, post-thaw CD34+ cell dose, HLA match, sex match, ABO match, order of infusion and age differences between the CBUs were considered potential predictors of cord predominance (Table 3; Figure 2). Univariate analysis revealed that a higher CD34+ cell dose (P < 0.01) or TNC dose (P = 0.01) was a significant predictor of predominance. In addition, the predominant unit was the first one infused in 26 out of 38 of patients (68%, P = 0.03).
In the first cohort (cyclosporine/MMF), the first CBU infused was randomly selected. As a result, the unit with the higher collection center TNC dose was infused first in 11 out of 17 patients and the unit with the higher post-thaw TNC and CD34+ cell dose was infused first in 10 out of 17 patients. In the second cohort (sirolimus/tacrolimus), on the basis of our preliminary results, the unit with the higher collection center TNC dose was infused first (20 out of 21 patients).6 This practice led to the unit with the higher post-thaw TNC and CD34+ cell dose infused first in 17 and 15 out of 21 patients, respectively. As such, we controlled for the potential confounding of cell dose with order of infusion using multivariate analysis. In the multivariate model, order of infusion (P = 0.03), post-thaw CD34+ cell dose (P = 0.03) and post-thaw TNC dose (P = 0.06) were all independent predictors of cord predominance. Age differences between the CBUs and HLA, ABO and sex match between units and the patient did not impact predominance.
Median engraftment times were 20 days for neutrophils, 43 days for platelets > 20 000 and 125 days for platelets > 100 000. The effects of post-thaw-combined CD34+ cell dose, post-thaw-combined TNC dose and age of the CBUs were analyzed as potential contributors to faster engraftment. Table 4 shows the trend for a higher post-thaw CD34+ cell dose leading to shorter time to ANC > 500 (P = 0.13) as well as faster long-term engraftment of platelets > 100 000 (P = 0.12). Further analysis demonstrated that it was only a higher post-thaw CD34+ cell dose of the predominant unit that correlated with faster ANC (P = 0.10) and long-term platelet (P = 0.05) recovery. Chimerism patterns over time (group 1 versus group 2 and 3) did not affect engraftment.
DCBT offers a unique opportunity to investigate competitive transplants in humans and learn more about CBU parameters that could lead to improved outcomes. This study represents one of the largest detailed analyses of cord blood chimerism in adult patients receiving RIC followed by DCBT.
Our study has several limitations. First, we studied two protocols with differences in GVHD prophylaxis regimens and in the order cord units were infused. However, patients received an identical conditioning regimen and there were identical patient eligibility and CBU selection criteria. Given the similarities, we chose to combine patient chimerism data to investigate the effects of preinfusion variables. Larger studies will help delineate how differences in GVHD prophylaxis might have affected chimerism patterns and outcomes. Although in the second protocol we attempted to infuse the unit with the higher collection center TNC dose first, multivariate analysis confirmed that order of infusion and cell dose are independent predictors of CBU predominance.
Second, the data represent a small cohort, follow-up is limited and lineage-specific chimerism results were not available for most patients. The results, however, are consistent. Only the post-thaw CD34+ cell dose of the predominant unit predicted time to engraftment of both platelets and neutrophils, and both post-thaw CD34+ cell and TNC doses predicted cord predominance. Taken together, the data are intriguing but still need to be confirmed in larger trials.
Our results support previous analyses indicating that a single CBU becomes predominant in the setting of DCBT.7,10 Compared with these prior studies, it took longer for full single donor chimerism to be achieved in our trials as 34% of patients by day 100 and 12% beyond 1 year still had contributions of both CBUs to hematopoiesis. These findings may be due to differences in conditioning regimens.
We found the unit infused first or with the higher post-thaw cell dose had an advantage in engraftment. One possible hypothesis is that the first unit infused was able to fill the hematopoietic stem cell niche reducing the micro-environmental space for the second unit.16–19 The unit with the higher dose of stem cells might also have an advantage. Recent data suggest that the hematopoietic stem cell niche may be less supportive of cord blood than adult hematopoietic stem cells.20 The more stem cells infused, the greater the probability that an amount sufficient to promote engraftment would arrive at and survive occupation of the niche space.
Other investigators have not found that higher TNC/CD34+ cell dose or order of infusion affected predominance. 7,10 Apart from differences in conditioning regimens, in these studies both CBUs were infused in rapid succession while in our study 94% of units were infused at least 4 h apart. Nilsson et al.17 have demonstrated that stem cells can home to the endosteal region in under 5 h giving ample time for the stem cells of the first unit to occupy the limited niche space. Whether the timing of unit infusion affects other aspects of DCBT warrants further research.
The role of histocompatibility in establishing dominance of a given donor is unclear. We found no relationship in this study, although the number of patients examined may have been too small to ascertain an effect of HLA. In our study 82% of CBUs infused were 4/6 matches with the recipient by high-resolution typing for both class I and class II. Most prior studies only utilized low-resolution typing for class I.7,10 It is possible that the HLA similarity of the CBUs may have muted any potential immunologic factors affecting cord predominance allowing the effect of cell dose and order of infusion to become apparent.
Another important finding is that shorter time to ANC and long-term platelet engraftment correlated with only the post-thaw CD34+ cell dose of the predominant unit. A consistent trend was also present for platelets > 20 000. Neither characteristics of the non-predominant unit nor chimerism pattern affected engraftment. Data to support DCBT are inferred from a small number of observational studies.8,10 One might reasonably expect that higher CD34+ cell doses of both the predominant and non-predominant CBUs or an early contribution of both cords or one cord and the recipient to hematopoiesis might lead to faster engraftment. Our observation that only the predominant unit appears to affect engraftment time raises questions about the functional effects of the second unit. Our study, however, did not investigate all potential characteristics of the non-predominant unit that may lead to improved engraftment (for example, T-cell subsets). A randomized trial is required to truly answer whether two cords are superior to one and such a study is underway in the pediatric population.
If only the CD34+ cell dose of the winning CBU affects engraftment, a potential clinical strategy is to try to insure the predominance of the unit with the higher CD34+ cell dose by infusing it first. Based on results from the first protocol, we attempted such a strategy for patients in the second protocol. As we did not know the post-thaw CD34+ cell count for both units until after they were infused, we used the collection center TNC count as a surrogate.
In the second protocol, in 20 of the 21 transplants (95%) the unit with the higher collection center TNC dose was infused first. In 14 of these 20 transplants (70%), this strategy led to the unit with higher CD34+ cell dose, on the basis of our post-thaw determination, also being infused first. In 10 of these 14 transplants (71%), the unit with the higher post-thaw CD34+ cell dose was predominant. If we had used collection center CD34+ cell content to determine order of infusion, the unit with the higher post-thaw CD34+ cell dose would have been infused first only 65% of the time. These results reflect the imperfect correlation between TNC and CD34+ cell counts and the variability in methods for determining cellular content of products.21 A method to accurately determine CD34+ cell dose prior to infusion might be beneficial to insure that the unit with the higher dose is infused first. In this way, that unit would be more likely to predominate with subsequent faster engraftment.
Our results identify preinfusion parameters that may affect engraftment. Future studies should focus on other preinfusion factors that may affect DCBT outcomes. For example, preliminary results indicate that the number of NK cells in the UCB product may affect predominance.22 In-depth analysis of graft makeup in DCBT should provide a unique opportunity to study the basic biology of hematopoietic stem cell engraftment.
This study was supported in part by the National Heart, Lung and Blood Institute (P01 HL070149).