Transplant-related mortality and long-term graft function are significantly influenced by cell dose in patients undergoing allogeneic marrow transplantation

Cell dose is recognized as an important positive predictor of outcome in allogeneic bone marrow transplantation (BMT), both in the animal model1,2 and in human beings.3 It is unclear whether the beneficial effect is due to a larger dose of hemopoietic progenitors or other cell subpopulations4: One hypothesis would be that a larger number of grafted hemopoietic progenitors would lead to faster hematologic recovery.5 In this context, mobilized peripheral blood (PB) progenitor cell transplants have called the attention of several investigators to the kinetics of hematologic reconstitution after engraftment. Comparison of PB with bone marrow (BM) transplant recipients has revealed shorter neutropenia and thrombocytopenia in the former group: In a randomized study conducted by the European group for Blood and Marrow Transplantation (EBMT), the median day to reach a granulocyte count of 0.5 × 10⁹/L was 12 versus 15 (P < .0001) for PB versus BM grafts, respectively.6 Similarly, the median day to reach a platelet count of 20 × 10⁹/L was 15 versus 20 (P < .0001).6 Therefore, hemopoietic recovery is significantly faster in PB graft recipients. In the same study, however, transplant-related mortality (TRM) and survival were identical in the 2 groups.6 Most retrospective and prospective studies comparing PB versus BM transplants show that, despite faster hematologic recovery, transplant-related complications and mortality are quite similar.6-16 Therefore, a larger number of progenitor cells in PB grafts leads to faster hematologic recovery, with little influence on transplant mortality.

In the present study we asked whether increasing the cell dose in BM grafts would influence long-term hematologic recovery and possibly TRM. We therefore studied 905 consecutive patients undergoing an allogeneic BMT and analyzed graft function, as indicated by leukocyte and platelet counts and hemoglobin levels, at different time points as well as TRM, survival, and relapse.

Between April 1976 and December 2000, 905 consecutive patients with hematologic disorders underwent an unmanipulated allogeneic BMT at our institution. The median patient age was 30 years (range, 1-66 years); the median follow-up for deceased patients was 124 days (range, 1-6533 days) and 2510 days (range, 88-9094 days) for surviving patients. The donor was an HLA-identical sibling in 735 patients, a one-antigen mismatched related donor in 35, or a matched unrelated donor in 135 patients. The median donor age was 31 years (range, 2-77 years). The median interval between diagnosis and BMT was 326 days (range, 22-6240 days). The diagnoses were marrow failure (n = 94), acute leukemia (n = 397), chronic myeloid leukemia (CML, n = 298), lymphoproliferative disorders (LPDs, n = 55), and myelodysplasia (MDS, n = 61). Of the 811 leukemia patients, 448 patients (55%) received allogeneic BMT in first remission or in first chronic phase, and 363 (45%) were classified as advanced. Patient characteristics are summarized in Table1.

Eighty-five patients (10%) were conditioned with cyclophosphamide (CY) 200 mg/kg alone; 651 patients (72%) received CY (120 mg/kg) and total body irradiation (TBI; 9.9-12 Gy); and 169 patients (18%) were prepared with thiotepa/CY- or busulfan/CY-containing regimens. Acute graft-versus- host disease (GVHD) prophylaxis consisted of cyclosporin A (CyA) with or without methotrexate (MTX). Intravenous CyA (1-5 mg/kg daily) was administered until day +20 and then switched to the oral route (12.5 mg/kg daily). Methotrexate was administered at doses of 10 to 15 mg/m² on day +1 after BMT and 8 to 10 mg/m² on days +3, +6, and +11 after BMT. Acute and chronic GVHD were graded according to standard criteria.17 

Bone marrow (BM) was collected using small-volume (2 mL) aspirations, 20 to 25 mL/kg of recipient body weight,18under general anesthesia; it was processed through small-gauge needles (20 G) to avoid losing cells in the infusion filters. In cases of major ABO incompatibility, the BM was not depleted of red cells, but patients' isohemagglutinins were reduced to a titer of 1:16 or less in saline, with 1 or 2 plasma exchanges. BM was administered intravenously after completion of the conditioning regimen.

Platelet and white blood cell (WBC) counts were used to evaluate graft function, as were hemoglobin levels.

Transplant-related mortality (TRM) is defined as death due to causes unrelated to the underlying disease: Patients relapsing are censored as surviving at the time of relapse. Disease-free survival is defined as the probability of being alive free of disease; events are death in remission, relapse, and death due to the underlying disease, whichever occurs first. The Student t and Mann-Whitney tests were used for continuous variables and the χ² test for 2 × 2 tables. Actuarial probabilities of transplant-related mortality and overall survival were calculated with the Kaplan-Meier method,19 and the log-rank test was used to evaluate the differences between curves. The following factors were studied in multivariate Cox analysis for potential effect on transplant-related mortality rates: donor type, donor/recipient age, type and disease phase, cell dose, and conditioning regimen. The number-cruncher software (NCSS, version 5.0; JL Hintze, Kaysville, UT) was used to perform the analysis.

The median nucleated cells infused was 3.4 × 10⁸/kg (25th percentile 2.4 × 10⁸/kg, 75th percentile 5 × 10⁸/kg). Patients were stratified into 3 groups according to low, intermediate, or high cell dose: ≤ 2.4 × 10⁸/kg (n = 247); > 2.4 × 10⁸/kg and ≤ 5 × 10⁸/kg (n = 452), and > 5 × 10⁸/kg nucleated cells (n = 206).

Table 2 outlines the clinical characteristics of patients in the 3 groups. Patients in the high cell dose group were significantly older (P = .0005), had significantly older donors (P = .001), and comprised significantly more alternative donor transplants (P = .05) and fewer patients receiving TBI (P = .000 01). Notably, all of these factors have a negative influence on TRM in univariate χ² analysis: recipient age more than 30 years (TRM 38% vs 29%, P = .004), donor age 30 or more years (TRM 38% vs 28%, P = .003), alternative donor transplants (TRM 44% vs 30%, P = .0004), and use of non-TBI regimens (TRM 36% vs 32%, P = .2).

Platelet recovery is outlined in Figure1 in patients stratified according to the cell dose received (low, intermediate, or high): Patients in the latter group had higher platelet counts at all time points up to 1 year after transplantation, and the difference was significant at theP < .01 level. Similar results were obtained when looking at WBC counts (Figure 2) on days +20, +50, and +180 after BMT (P < .01). Hemoglobin levels were comparable at different time points after transplantation.

The actuarial 5-year transplant-related mortality (TRM) was 41% versus 36% versus 28%, respectively, (P = .01) in patients receiving low, intermediate, or high cell dose (Figure3). Crude TRM within day +100 was 24% for patients receiving a low cell dose, 19% for intermediate, and 16% for a high cell dose (P = .008). Crude TRM for patients surviving 100 days was 23%, 17%, and 11%, respectively, in the 3 groups (P = .01). The cell dose effect on TRM was more pronounced in patients with advanced-phase disease (47% vs 33% vs 26%, P < .008) as compared with patients with early-phase disease (32% vs 31% vs 21%, P = .1). The same was true for older patients (age > 30 years; 56% vs 35% vs 27%, P < .0001) as compared with younger patients (age ≤ 30 years; 31% vs 32% vs 20%, P = .1). The cell dose effect on TRM was seen in patients grafted from an alternative donor (62% vs 44% vs 33%, P = .004) as well as in patients grafted from an HLA-identical sibling (37% vs 31% vs 22%,P = .006). When patients were stratified according to diagnosis, the cell dose effect on TRM was most significant in patients with CML (52%, 39%, 18%, respectively, for low, intermediate, high cell dose; P = .001), MDS (39%, 35%, 25%;P = .006), and chronic lymphoproliferative disorders (70%, 22%, 20%; P = .01) as compared with patients with acute leukemia (29%, 26%, 21%; P = .3) and marrow failure (52%, 42%, 32%; P = .2). Acute GVHD was scored as grade 0, I, II, III, and IV, respectively, in 92 (10%), 332 (37%), 342 (38%), 104 (11%), and 35 (4%) patients. Acute GVHD grade III to IV was 17% and 16% in patients receiving low or intermediate cell dose, and this is higher when compared with patients grafted with the high cell dose (11%; P < .05). Chronic GVHD was scored as absent, limited, or extensive, respectively, in 672 (74%), 186 (21%), and 47 (5%) patients and was not different in the 3 groups (P = .2).

The actuarial 5-year overall survival was significantly different in patients receiving low, intermediate, or a high cell dose: 45% versus 51% versus 56% (P = .0008; Figure4). Deaths due to GVHD and infections were 28% versus 23% versus 16%, respectively, in the 3 groups,P = .009. There was also a difference in the risk of being infected on day +30 after transplantation: Bacterial infections were seen in 16%, 14%, and 10% in the 3 groups (P = .1); fungal infections in 32%, 11%, and 9% (P = .001); and viral infections in 13%, 5%, and 4% (P = .01), respectively, in the low, intermediate, and high group. Causes of TRM other than GVHD and infections were comparable (P = .5). The number of deaths due to leukemia was similar in the 3 groups (17%, 14%, 14%; P = .8), and this was confirmed in the Kaplan-Meier analysis.

The actuarial 5-year disease-free survival was significantly different in patients receiving low, intermediate, or high cell dose: 41% versus 42% versus 48% (P = .02).

The actuarial risk of relapse at 5 years in patients who received low, intermediate, and high cell dose was, respectively, 30% versus 33% versus 35% (P = .3); for patients in first remission the actuarial risk is 19%, 23%, and 26% (P = .4).

The cell dose was analyzed in multivariate Cox analysis for potential effect on TRM together with 6 other clinical factors: donor type, donor age, recipient age, type and disease phase, conditioning regimen, and year of transplantation. In multivariate Cox analysis on TRM, cell dose was a significant predictor (P = .002; relative risk 0.6) together with donor type (P = .0001), year of transplantation (P = .0001), conditioning regimen (P = .02), and recipient age (P = .02; Table3).

We have shown in the present study that graft function is improved both in the short and long term after transplantation of a high dose of allogeneic bone marrow cells: This results in significantly higher white blood cell counts up to 6 months after engraftment and higher platelet counts at all time intervals up to 1 year after transplantation. Although this would sound expected, we could not find this information in studies comparing different cell doses or different sources of stem cells.6-16 Many reports concentrate on duration of neutropenia or thrombocytopenia and disregard peripheral blood counts beyond day +20. In the present study, instead, we could show that graft function was particularly improved beyond day +20 and even beyond day +100, suggesting a long-lasting effect of a high marrow cell dose: One-year postengraftment median platelet counts for patients receiving low, intermediate, or high marrow cell dose were 130 × 10⁹/L, 167 × 10⁹/L, and 191 × 10⁹/L, respectively.

The second finding is the strong correlation between cell dose and transplant mortality, in keeping with other studies20-25: In multivariate analysis patients receiving a high cell dose (> 5 × 10⁸/kg) had a relative risk of dying of transplant-related complications of 0.6 when compared with patients receiving a low cell dose (< = 2.4 × 10⁸/kg). This was almost entirely due to a reduction of lethal infections and GVHD: The risk of death caused by infection/GVHD (often related) was 28%, 23%, and 16%, respectively, in patients receiving a low, intermediate, or high cell dose (P = .009). Other studies have shown a positive effect of cell dose on graft-versus-host disease,20,21,26 and the fact that we find reduced GVHD with improved graft function in patients receiving a high cell dose is in keeping with the recent demonstration of a strong correlation between severity of acute GVHD and platelet counts.27 In the present study the actuarial 5-year TRM in patients receiving low, intermediate, or high cell dose was, respectively, 41%, 36%, and 28% (P = .01). This could not be explained by a selection of patients with good prognosis, because the high cell dose group contained more patients with risk factors such as older age and alternative donor transplants, and this is confirmed in the multivariate Cox analysis.

The reduction of transplant-related mortality was so strong that it produced improved overall survival. We did not find a significant difference in the leukemia relapse between the 3 groups of patients, notwithstanding a recent report on a possible favorable effect of a high cell dose.25 However, if GVHD and TRM are reduced, it is unlikely that leukemia relapse would also be reduced. In keeping with these observations, we find that the disease-free survival was significantly reduced in patients grafted with low cell dose in comparison with intermediate and high dose (41% vs 42% vs 48%;P = .02). The cell dose effect on TRM was more evident in patients at high risk for dying of transplant complications, such as patients older than 30 years of age, with advanced disease, or with CML/MDS. In patients with CML, the transplant mortality was 52%, 39%, 18%, respectively, for low, intermediate, or high cell dose, in keeping with a recent report.28 

One question that comes up concerns the cell component important in reducing transplant-related mortality. Is it stem cells, hemopoietic progenitors, lymphocytes, or other cells such as stromal or mesenchymal cells?29 Stem cell numbers would seem important, as shown by experimental data on radioprotection30 and clinical data in human beings on progenitor cell content and quality of hematologic reconstitution.5,31 The question is how other cells influence stem cell function. The high lymphocyte content of peripheral blood grafts, producing significant acute and chronic GVHD, may counterbalance the positive effect of cell dose and produce an overall negative effect, mostly when the total cell count exceeds 9 × 10⁸/kg.32 This was also seen in a recent study on CD34⁺ selected allografts, with a cutoff for increased TRM of 3 × 10⁶/kg CD34⁺peripheral blood cells33: Thus, CD34⁺ cells per se or their progeny, including antigen-presenting cells,34could have a promoting role on graft-versus-host disease and thus impair graft function. Other accessory cells, such as mesenchymal stem cells (MSCs), instead, could play a positive role on graft function, by virtue either of a suppressive effect on GVHD35 or of a direct promoting effect on stem cells36: These MSCs are found in bone marrow but not in peripheral blood harvest.4 

In conclusion, this study indicates a very strong cell dose effect on graft function when using bone marrow as a stem cell source, resulting in lower transplant mortality. Data from the literature and our own institution (A.B., unpublished data, 2000) would suggest that this is not the case for peripheral blood grafts. Therefore, graft function and low transplant toxicity is not solely the result of a large CD34⁺ cell dose and is probably influenced by other cell subpopulations. Some of these, such as mesenchymal stem cells, are being studied as candidates for the cell dose effect. At present we would recommend using a high marrow cell dose for best graft function and low transplant mortality: How to manipulate peripheral blood grafts to mimic this effect remains to be determined.

The great work of our nursing staff is gratefully acknowledged.