CD34+ cell dose predicts relapse and survival after T-cell-depleted HLA-identical haematopoietic stem cell transplantation (HSCT) for haematological malignancies


Professor Barrett BMT Unit, Hematology Branch, Bldg 10, Rm 7C103, National Institutes of Health, 9000 Rockville Pike, Bethesda, MD 20894, USA.


Seventy-eight patients with haematological malignancies, received T-cell-depleted stem cell transplants and cyclosporin followed by delayed add-back of donor lymphocytes to prevent leukaemia relapse. The source of stem cells was bone marrow in 50 patients and granulocyte colony-stimulating factor (G-CSF)-mobilized peripheral blood in 28 patients. In univariate analysis, only the CD34+ cell dose (but not the stem cell source or the T lymphocyte dose) and disease status were predictive for transplant-related mortality, relapse and survival. Patients receiving ≥ 3 × 106 CD34+ cells/kg had an overall actuarial survival of 68% compared with 52%, 35% and 10%, respectively, for cell doses of 2–2.99, 1–1.99 and < 1 × 106/kg. Multivariate analysis of risk factors for relapse identified disease risk and CD34+ cell dose as the only factors. Relapse was 62.5% in 38 patients at high risk of relapse vs. 25% for 40 patients at intermediate or low risk. CD34+ cell doses of ≥ 3 × 106/kg were associated with a 13.5% relapse vs. 48% for recipients of lower doses. This favourable effect of CD34+ cell dose on relapse was apparent in both high- and intermediate- plus low-risk groups. Our results support the potential benefit of a high stem cell dose in lowering transplant-related mortality (TRM) and in reducing relapse after allogeneic marrow or blood stem cell transplants.

With advances in peripheral blood stem cell collection and T-cell depletion techniques, it is now possible to control the stem cell and lymphocyte content of bone marrow and blood allografts. Graft engineering has been further facilitated by the use of CD34+ as a surrogate marker for stem cells ( Roscoe et al., 1994 ). T lymphocytes in the graft are responsible for unfavourable effects from graft vs. host disease (GVHD) as well as a beneficial effect on engraftment and the graft vs. leukaemia (GVL) effect. For unmanipulated transplants, because of the background of these powerful lymphocyte-mediated effects, it has been difficult to detect the effect of CD34+ cells on transplant outcomes. In T-cell-depleted marrow transplants, however, higher CD34+ cell doses have been shown to decrease transplant-related mortality (TRM; Mavroudis et al., 1996 ) and to improve the rate and completeness of engraftment in matched ( Noga et al., 1998 .) and mismatched transplants ( Aversa et al., 1994 ; Reisner & Martelli, 1995). However, the effects of CD34+ cell dose on relapse have not been studied, and the CD34+ and T-cell doses required for optimum transplant outcome are not yet established.

As T-cell depletion decreases the incidence of acute and chronic GVHD but increases the risk of relapse ( Apperley et al., 1986 ; Butturini & Gale, 1988; Marmont et al., 1991 ), we have used a T-cell depletion strategy followed by delayed add-back of T cells to preserve the GVL effect while reducing the risk and severity of GVHD ( Johnson et al., 1993 ; Barrett et al., 1998 ). To investigate the impact of stem cell dose on transplant outcome further, we studied CD34+ cell dose in a series of 78 patients undergoing T-cell-depleted transplants. Here, we describe a powerful association of CD34+ cell dose with TRM and relapse in patients undergoing allogeneic T-cell-depleted bone marrow and peripheral blood stem cell transplantation for haematological malignancies.



Between September 1993 and September 1998, 78 patients with haematological malignancies received a T-lymphocyte-depleted bone marrow or peripheral blood stem cell transplant from an HLA-identical sibling. Patients were treated under National Institutes of Health protocols 93-H-0212 (chronic myelogeneous leukaemia), 94-H-0092 (acute myelogeneous leukaemia and myelodysplastic syndrome), 94-H-0182 (multiple myeloma) and 95-H-0099 (peripheral blood stem cell transplant for haematological malignancies) approved by the local Institutional Review Board. Recipient age ranged between 10 and 58 years with a median of 38 years. Thirty-six patients had chronic myeloid leukaemia (CML), of whom 24 were in chronic phase (CP), seven in accelerated phase (AP) and five in blastic crisis (BC). Twenty-eight patients had other myeloid malignancies: 14 myelodysplastic syndrome in transformation (tMDS), 13 acute myeloblastic leukaemia (AML) (seven in remission and six in relapse) and one patient with eosinophilic leukaemia. Ten patients had multiple myeloma, two had acute lymphoblastic leukaemia (ALL) in second remission, one patient had chronic lymphocytic leukaemia (CLL), and another had non-Hodgkin's lymphoma (NHL) in relapse. Twenty-eight patients (CML-CP and AML in first remission) were designated as standard risk for transplant complications and relapse. Twelve with CML-AP and acute leukaemias in second or subsequent remission were designated as intermediate risk, and the remaining 38 patients with more advanced leukaemia (refractory AML, CML-BC), MDS, CLL, relapsed NHL or multiple myeloma were designated as high risk.

Preparative regimen

Seventy-two patients received a total-body irradiation (TBI)-based regimen of 120 mg/kg cyclophosphamide and fractionated TBI (CyTBI). Sixty-nine patients received 1360 cGy in eight fractions over 4 days, and four (UPN 1, 2, 3 and 5) received 1200 rads in six fractions over 3 days, together with intravenous antithymocyte globulin (ATG) (Upjohn, Kalamazoo, MI, USA) at 40 mg/kg × 4 days between days −5 and −2 before BMT. In six patients who had received previous radiation, BuCy (busulphan at 16 mg/kg and cyclophosphamide at 200 mg/kg) were given over 4 days.

Collection and processing of bone marrow and peripheral blood progenitor cells

For the first 38 patients, who received T-cell-depleted bone marrow, donor bone marrow was concentrated on the CS3000 cell separator (Baxter, Roundlake, IL, USA) and elutriated as described previously ( Mavroudis et al., 1996 ). A T-cell dose of 2 × 105 CD3+ cells/kg recipient weight was achieved by using the rotor-off elutriation fraction, with supplementation as needed, from the 140 ml/min and 110 ml/min elutriation fractions. For the next 12 patients, who received granulocyte colony-stimulating factor (G-CSF)-stimulated donor bone marrow, donors received filgrastim 10 μg/kg/day subcutaneously for 5 days before bone marrow harvest. The harvested marrow was then processed by CD34-positive selection on the Ceprate SC system (CellPro, Bothell, WA, USA), which resulted in a T-cell-depleted bone marrow product ( Cottler-Fox et al., 1995 ). The last 28 patients received T-cell-depleted peripheral blood progenitor cells (PBPCs). Donors for these patients were mobilized with filgrastim 10 μg/kg/day subcutaneously for 6 days, with leukapheresis (15-l volume processed) on the Baxter CS3000Plus on days 5 and 6. PBPC collections were processed either fresh or as a pool of two collections using the CeprateTCD T Cell Depletion System (CellPro), which included a CD34-positive selection on the CeprateSC column, followed by a CD2-negative selection using a similar but smaller column. The target T-cell dose for this patient group was 1 × 105/kg CD3+ cells.

GVHD prophylaxis

To prevent GVHD, patients received a BMT depleted of T cells. As further GVHD prophylaxis, all patients received cyclosporin 3 mg/kg intravenously, starting on day −4 until an oral dose was tolerated. Doses were adjusted to maintain a cyclosporin plasma level between 200 and 400 μg/ml. Cyclosporine was continued at least to day 180 after BMT and longer if chronic GVHD developed.

T-cell add-back

To reconstitute immune function, patients received fresh or cryopreserved, thawed mononuclear cells from the donor, obtained by leukapheresis. Three add-back schedules were evaluated: (1) 2 × 106 CD3+ cells/kg on day 30 followed by 5 × 107 CD3+ cells/kg on day 45 (total 5.2 × 107/kg); (2) 1 × 107 CD3+ cells/kg on day 30 after BMT; (3) 1 × 107 CD3+ cells/kg on day 45 followed by 5 × 107 CD3+ cells/kg on day 100 (total 6 × 107/kg). An exception was made to withhold T-cell add-back to standard-risk patients (at low risk of relapse) developing grade II or more a-GVHD after transplant and any patient being treated for grade ≥ II a-GVHD.

Post-transplant management

All patients received fluconazole and norfloxacin until day 30 for antifungal and antimicrobial prophylaxis, followed by Bactrim weekly from day 30 to day 180 as prophylaxis against pneumocystis infection. CMV antigen was monitored weekly until day 100 after BMT, and gancyclovir with high-dose immunoglobulin was initiated when there was evidence of CMV reactivation ( Couriel et al., 1996 ). From May 1995, patients also received high-dose acyclovir as prophylaxis against CMV and herpes virus reactivation ( Prentice et al., 1994 ). Acute GVHD ≥ grade II was treated with high-dose methylprednisolone.

Haematological recovery and disease status

This was monitored by bone marrow examination on days 21 and 100, at 6 months, 1 year and then annually.

CD34+ and T-cell doses

These were measured by automated cell counting and flow cytometric quantification of the CD34+ and CD3+ fractions using a Becton-Dickinson FACScan instrument as described previously ( Mavroudis et al., 1996 ).

Statistical analysis

Actuarial probabilities of TRM, relapse and survival were calculated according to the method of Kaplan & Meier (1958). Differences between outcomes were compared using Peto and Peto's modification of Wilcoxon's rank-sum analysis ( Peto & Peto, 1972). Fisher's exact test was used to compare differences in the grade of GVHD. Cox multivariate analysis was used to determine relative risk for the independent risk factors. The following factors were entered into the model: CD34+ cell dose, CD3+ cell dose, risk status (high vs. intermediate + low), age, donor sex match (female into male recipient vs. other combinations) and acute GVHD.



Thirty-six patients have survived at between days 56 and 1836 with an actuarial survival of 34.3 ± 6.8% at a median follow-up of 18.8 months. Forty-two patients died: 17 from leukaemic relapse or progressive myeloma and 25 from transplant-related causes. Death from relapse occurred between 45 and 842 (median 159) days after transplantation. Transplant-related deaths occurred 11–771 (median 78) days after transplantation; deaths from infection occurred in 14 patients (five from CMV interstitial pneumonia, four from aspergillosis, two from bacterial infections and two from RSV pneumonia), acute GVHD in three patients, veno-occlusive disease in two patients and pulmonary haemorrhage in two patients. Two deaths were associated with cytopenias (intracranial haemorrhage and allergic blood transfusion reaction): one died from ARDS and one from a transplant-associated Guillain–Barré syndrome. Disease-free survival (DFS) was significantly higher in standard-risk and intermediate-risk patients compared with high-risk patients (58.2 ± 10.3% and 48.6 ± 17.6% vs. 8.5 ± 6.9%, P < 0.0001; Fig 1).

Figure 1.


Factors affecting TRM, survival and disease-free survival

Table I shows the results of univariate analysis of factors associated with transplant outcome. Disease status (early + intermediate vs. advanced disease) and CD34+ cell dose were the only factors with a significant effect on TRM. The effect of CD34+ cell dose on survival and DFS was incremental, with the highest survival and DFS being associated with CD34+ cell doses over 3 × 106/kg.

Table 1. Table I. Risk factors for survival, DFS and TRM in univariate analysis. Thumbnail image of

To investigate the effect of CD34+ cell dose further, patient characteristics for CD34+ cell doses above and below the median of 3.0 × 106CD34+/kg were compared ( Table II). Stem cell source was closely linked to CD34+ cell dose with peripheral blood transplants accounting for the majority of recipients of ≥ 3 × 106 CD34+ cells/kg. The dose of CD3+ cells at transplant correlated inversely with CD34+ cell dose, such that significantly more recipients of cell doses of < 3 × 106 CD34+ cells/kg received T-cell doses higher than 1 × 105 CD3+ cells/kg. However, this did not result in differences in the incidence of acute GVHD in the high and low CD34+ cell dose populations. More recipients of low CD34+ cell doses were female donor into male recipient transplants (P = 0.047).

Table 2. Table II. Risk factors of patients grouped according to CD34+ cell dose given.Thumbnail image of

Variables affecting relapse

The relapse rate was significantly higher with high-risk disease, but there was no significant difference between low- and intermediate-risk disease (Fig 2) and, in subsequent analyses, low- and intermediate-risk disease were therefore grouped together. In Cox regression analysis, only disease risk and the CD34+ cell dose were significantly associated with relapse ( Table III). Relapse probability for the 57 recipients of < 3 × 106 CD34+ cells/kg was 48% vs. 13.5% for the 21 recipients of ≥ 3 × 106 CD34+ cells/kg (P = 0.05). The relative relapse risk for low vs. high CD34+ cell doses was 0.24. The effect of CD34+ cell dose on relapse was then examined for low- and intermediate-risk disease and for high-risk disease (Fig 3A and B). In high-risk disease, the actuarial probability of relapse was 62.6%. There was a significantly higher relapse rate for patients receiving < 3 × 106 CD34+ cells/kg compared with those receiving higher doses (75.8 vs. 20%, P = 0.028). For low- and intermediate-risk disease, the overall relapse rate was 25.1%. A significant difference in relapse rate was found in patients receiving < 1 or ≥ 1 × 106 CD34+ cells/kg (80 vs. 10.1%, P = 0.0005).

Figure 2.

Fig 2. Kaplan–Meier estimates of cumulative incidence of relapse in patients in low-, intermediate- and high-risk groups (P = 0.0001).

Table 3. Table III. Risk factors for relapse in multivariate analysis. Thumbnail image of
Figure 3.

 × 106/kg (P = 0.028). (B) Low-risk patients, ≥ 1 × 106/kg vs. < 1 × 106/kg (P = 0.0005).

Outcome for patients with CML in chronic or accelerated phase

To study CD34+ cell does in a single homogeneous group, we analysed the subset of 31 patients with CML who were transplanted in first chronic phase or accelerated phase (low- and intermediate-risk disease). Again, there was a significant correlation between CD34+ cell dose, relapse and DFS, such that patients receiving higher doses of CD34+ cells had a more favourable relapse rate and higher leukaemia-free survival (P = 0.00047 and P = 0.0066 respectively). The beneficial effect of CD34+ cell dose was most apparent for doses of ≥ 1 × 106 CD34+ cells/kg (Fig 4A and B).

Figure 4.


Effect of T-cell add-back

Fifty-four patients surviving until day 100 after transplant were studied to detect the effect of total T-cell add-back on relapse. Three patients did not receive T-cell add-back because of prior GVHD and, in the remaining 51 patients, T-cell add-back dose ranged between 0.2 and 22.8 × 107/kg. The relapse rates of 26 patients who received T-cell doses of between 0 and 2 × 107/kg (median 1 × 107/kg) and 28 patients who received T-cell doses of between 5 and 22.8 × 107/kg (median 5.1 × 107/kg) were 33.9% and 47%, respectively, a difference that was not statistically significant (P = 0.75).


In this study, two factors, CD34+ cell dose and disease risk category, were significantly associated with TRM and disease-free survival. TRM was especially high in patients receiving less than 1 × 106 CD34+ cells/kg, extending our previous observation in a smaller series of marrow transplant recipients ( Mavroudis et al., 1996 ). However, the favourable effect of higher CD34+ cell doses on survival and DFS was not explained only by an improvement in TRM, as the relapse rate was also much lower in patients receiving > 3 × 106 CD34+ cells/kg. Multivariate analysis confirmed the CD34+ cell dose and disease risk group to be the only factors examined affecting relapse. Furthermore, a favourable effect of high CD34+ cell dose on relapse was seen in both low + intermediate risk and high-risk patients, although at different dose levels, as well as in the subgroup of patients with chronic or accelerated phase CML.

We sought confounding factors that might explain the differences in relapse rates in these groups. In multivariate analysis, we excluded a competing effect of T cells given either at the time of transplant or at add-back. Furthermore, acute GVHD (which should be associated with a lower relapse rate from an accompanying GVL mechanism) was similar in recipients of higher and lower CD34+ cell doses. We also considered the possibility that the primary factor determining relapse was the marrow cell source (marrow vs. peripheral blood), as the favourable CD34+ cell doses were highly correlated with peripheral blood as the stem cell source. Against this is the absence of any significant reduction in relapse rate in two large comparative studies of bone marrow vs. peripheral blood stem cell transplants ( Bensinger et al., 1996 ; Bacigalupo et al., 1998 ). Finally, it remains possible that factors other than those studied here may have influenced relapse. Further confirmation in other transplant series will be required in order to substantiate further the role of CD34+ cells on transplant outcome.

A relationship between CD34+ cell dose and transplant outcome was first shown in bone marrow autografts, in which a 106 CD34+ cells/kg dose was identified as a safe minimum dose for engraftment ( Bensinger et al., 1994), whereas > 2.5 –5 × 106/kg is necessary to achieve the most rapid recovery of neutrophil and platelet counts ( Bensinger et al., 1995 ; Weaver et al., 1995 ). The CD34+ cell dose also has an impact on the outcome of marrow allografts. In our earlier study of T-cell-depleted marrow transplants, TRM was higher in patients receiving > 106 CD34+ cells/kg ( Mavroudis et al., 1996 ). It has also been shown that increasing the CD34+ cell content of the graft overcame delayed engraftment and graft failure ( Noga et al., 1998 ).

A possible antileukaemic effect of CD34+ cells has been described recently in a murine transplant model for leukaemia ( Hartung et al., 1998 ). Mice receiving G-CSF plus stem cell factor (SCF)-mobilized peripheral blood stem cell allografts had a higher DFS than those receiving G-CSF-stimulated transplants. The only difference in the composition of the transplants was a higher haematopoietic precursor cell content of the G-CSF + SCF allografts. An effect of CD34+ cell dose on relapse has not been described previously in man. A favourable effect of nucleated cell dose on relapse was described for CML patients undergoing identical sibling transplants ( Ringden & Ringden, 1985). Additionally, higher marrow cell dose was associated with better leukaemia-free survival in patients undergoing matched unrelated donor transplants ( Sierra et al., 1997 ). However, the composition of these grafts was unknown. Therefore, it was not possible to differentiate between higher lymphocyte dose resulting in higher GVHD-related GVL effect vs. a GVL effect from a higher stem cell dose. In our study, the lack of correlation between relapse and initial T-cell dose and the uniformity of T-cell add-back dose between high and low CD34+ cell groups eliminated the possibility that the results observed resulted from a T-cell effect.

It is not known how CD34+ cell dose may affect relapse. One possible mechanism is that higher CD34+ cell doses favour more rapid establishment of donor-derived antigen-presenting cells presenting leukaemia-associated antigens to the donor immune system. Alternatively, higher CD34 cell doses might favour more rapid immune recovery, enhancing a GVL effect. We are currently studying these possibilities.

Our findings suggest that transplanting over 3 × 106 CD34+ cells/kg is an important treatment objective in T-cell-depleted allogeneic transplants of bone marrow and peripheral blood, especially for high-risk patients. High CD34+ cell doses can easily be achieved using G-CSF-mobilized peripheral blood progenitor transplants. Further studies are needed to determine the optimum CD34+ cell dose above which there is no additional beneficial effect on relapse.


We thank all BMT Unit coordinators and nurses for their dedicated care of our patients, and all the members of the Bone Marrow Transplant Team for their constant support of the programme. We also thank the Intensive Care Unit for their exemplary care of these patients.