Long-term clinical results of autologous infusion of mobilized adult bone marrow derived CD34+ cells in patients with chronic liver disease.

Evidence is growing in support of the role of stem cells as an attractive alternative in treatment of liver diseases. Recently, we have demonstrated the feasibility and safety of infusing CD34(+) adult stem cells; this was performed on five patients with chronic liver disease. Here, we present the results of long-term follow-up of these patients. Between 1 x 10(6) and 2 x 10(8) CD34(+) cells were isolated and injected into the portal vein or hepatic artery. The patients were monitored for side effects, toxicity and changes in clinical, haematological and biochemical parameters; they were followed up for 12-18 months. All patients tolerated the treatment protocol well without any complications or side effects related to the procedure, also there were no side effects noted on long-term follow-up. Four patients showed an initial improvement in serum bilirubin level, which was maintained for up to 6 months. There was marginal increase in serum bilirubin in three of the patients at 12 months, while the fourth patient's serum bilirubin increased only at 18 months post-infusion. Computed tomography scan and serum alpha-foetoprotein monitoring showed absence of focal lesions. The study indicated that the stem cell product used was safe in the short and over long term, by absence of tumour formation. The investigation also illustrated that the beneficial effect seemed to last for around 12 months. This trial shows that stem cell therapy may have potential as a possible future therapeutic protocol in liver regeneration.

Subcellular distribution of p210(BCR-ABL) in CML cell lines and primary CD34+ CML cells.

We analysed the subcellular distribution of p210(BCR-ABL) protein using a junction-specific anti-BCR-ABL monoclonal antibody and confocal laser scanning microscopy (CLSM). Our studies have shown that p210(BCR-ABL) is arranged in discrete foci in the cytoplasm of cell lines and primary CD34(+) cells but not mononuclear cells suggesting the foci may be a feature of immature chronic myeloid leukaemia cells. We have devised a strategy to score the foci and found the mean number of foci varies between the cell types. The number of foci per cell is directly related to the level of p210(BCR-ABL) expression. CLSM was also used to analyse the distribution and colocalization of CT10 regulator-like (CRKL) p210(BCR-ABL). CRKL-p210(BCR-ABL) foci were completely or partially associated, touching or separate in different regions of the same cell. We also analysed the distribution of phosphorylated CRKL (pCRKL) with p210(BCR-ABL) and unexpectedly found only a small proportion of pCRKL in complex with p210(BCR-ABL). The foci distribution and high levels of uncomplexed p210(BCR-ABL), pCRKL and CRKL protein suggested the possibility of a dynamic equilibrium. Imatinib promoted nuclear transport of p210(BCR-ABL)-positive foci. It also disrupted complex formation between p210(BCR-ABL) and casitas B-cell lymphoma and CRKL but not between p210(BCR-ABL) and GRB2. Our observations of the CRKL and p210(BCR-ABL) complex may be important for understanding the function of CRKL.

Phosphatidylinositol-3 kinase inhibitors reproduce the selective antiproliferative effects of imatinib on chronic myeloid leukaemia progenitor cells.

We investigated the role of the phosphatidylinositol-3 kinase (PI-3K) pathway in regulating the proliferation of primary chronic myeloid leukaemia (CML) progenitor cells by using imatinib to inhibit the activity of p210(Bcr-Abl). The effect of imatinib on the expression of PI-3K pathway proteins was investigated by kinase assays and Western blotting; PI-3K was inhibited by wortmannin or LY294002, Jak2 by AG490 and farnesylation by FTI II; progenitor cell proliferation (self-renewal) was measured by growing myeloid colonies in vitro, then replating them to observe secondary colony formation. Suppression of p210(Bcr-Abl) with imatinib indirectly suppressed the activity of PI-3K and its downstream targets (Erk, Akt and p70S6 kinase), thereby implicating the PI-3K pathway in p210(Bcr-Abl)-mediated signalling in primary CML progenitor cells. The PI-3K inhibitors, wortmannin and LY294002 reproduced the differential effects of imatinib on normal and CML progenitor cell proliferation in vitro by increasing normal cell (P = 0.001) and reducing CML cell proliferation (P = 0.0003). This differential effect was attributable to dysregulated signalling by granulocyte colony-stimulating factor in CML. The responses of individual patient's cells to wortmannin correlated with their responses to imatinib (P = 0.004) but not their responses to AG490 (Jak2 kinase inhibitor) or FTI II (farnesyltransferase inhibitor). Individual responses to wortmannin also correlated with responses to interferon alpha (IFNalpha) (P = 0.016). Imatinib-resistant K562 cells were sensitive to LY294002. Inhibition of the PI-3K pathway may be common to imatinib and IFNalpha and reflect dysregulated cytokine signalling. As imatinib-resistant cells remained sensitive to wortmannin and LY294002, targeting the PI-3K pathway may provide an alternative therapy for imatinib-resistant patients.

Progenitor cells from patients with advanced phase chronic myeloid leukaemia respond to STI571 in vitro and in vivo.

STI571 targets p210(BCR-ABL) in chronic myeloid leukaemia (CML). In vitro, STI571 reduces self-replication (replating ability) by chronic-phase CML CFU-GM. Here, we studied CFU-GM in advanced-phase (accelerated and blast crisis) CML. The numbers and self-replication of CFU-GM in advanced phase were greater than in the chronic phase. Self-replication by CFU-GM from advanced phase patients was reduced by STI571 or IFN alfa to the same extent as in the chronic phase. The reduced replating ability induced by STI571 correlated with that induced by IFN alpha (r=0.73). STI571 treatment in vivo also reduced replating ability and the numbers of CFU-GM/ml of blood.

The influence of INK4 proteins on growth and self-renewal kinetics of hematopoietic progenitor cells.

This study investigated the influence of expression of proteins of the INK4 family, particularly p16, on the growth and self-renewal kinetics of hematopoietic cells. First, retrovirus-mediated gene transfer (RMGT) was used to restore p16(INK4a) expression in the p16(INK4a)-deficient lymphoid and myeloid cell lines BV173 and K562, and it was confirmed that this inhibited their growth. Second, to sequester p16(INK4a) and related INK4 proteins, cyclin-dependent kinase 4 (CDK4) was retrovirally transduced into normal human CD34(+) bone marrow cells and then cultured in myeloid colony-forming cell (CFC) assays. The growth of CDK4-transduced colonies was more rapid; the cell-doubling time was reduced; and, upon replating, the colonies produced greater yields of secondary colonies than mock-untransduced controls. Third, colony formation was compared by marrow cells from p16(INK4a-/-) mice and wild-type mice. The results from p16(INK4a-/-) marrow were similar to those from CDK4-transduced human CFCs, in terms of growth rate and replating ability, and were partially reversed by RMGT of p16(INK4a). Lines of immature granulocytic cells were raised from 15 individual colonies grown from the marrow of p16(INK4a-/-) mice. These had a high colony-forming ability (15%) and replating efficiency (96.7%). The p16(INK4a-/-) cell lines readily became growth factor-independent upon cytokine deprivation. Taken together, these results demonstrate that loss of INK4 proteins, in particular p16(INK4a), increases the growth rate of myeloid colonies in vitro and, more importantly, confers an increased ability for clonal expansion on hematopoietic progenitor cells.

Peripheral blood progenitor cell mobilisation alters myeloid, but not erythroid, progenitor cell self-renewal kinetics.

Transplantation of progenitor cells which have been mobilised into the bloodstream (PBPC) following the administration of G-CSF results in more rapid neutrophil recovery than transplantation of bone marrow (BM). The reasons for the accelerated neutrophil engraftment are not clear, but would be explained by increased self-replication of myeloid progenitor cells (CFU-GM). We have used a CFU-GM replating assay to investigate myeloid progenitor self-replication, and quantification of subcolony formation during erythroid burst formation to quantify erythroid progenitor self-renewal. Secondary colony formation by CFU-GM, grown from PBPC and then replated was increased compared with secondary colony formation by BM CFU-GM (P = 0.0001); erythroid subcolony formation was not altered. There was no difference between the replating abilities of PBPC CFU-GM derived from allogeneic donors (normal individuals) and autologous donors (patients with malignant disease) although differences were found between subgroups of autologous donors. The increased replication of PBPC could not be accounted for by a reduction in progenitor cell apoptosis; PBPC CFU-GM contained slightly fewer apoptotic CD34+ cells than BM CFU-GM. The increased replication by PBPC CFU-GM was reversible because it declined when CFU-GM colonies were passaged through three sequential CFU-GM replating cycles. This decline in self-replication was more rapid than the decline seen in replated BM CFU-GM. The self-replication of PBPC CFU-GM, and subcolony formation by BFU-E could be further enhanced by exposure to cytokines in vitro. We conclude that mobilisation alters the replication kinetics of myeloid, but not of erythroid, progenitor cells, that mobilisation-induced events are of limited duration and that in vitro exposure to cytokines may modify PBPC progenitor cell kinetics.

Combination of interferon alpha with either Ara-C or ATRA in vitro reduces the selective action of interferon against CML CFU-GM.

Although interferon (IFN)-alpha has no specific inhibitory effect on the plating efficiency of granulocyte-macrophage colony-forming cells (CFU-GM) from patients with chronic myeloid leukaemia (CML), it does selectively inhibit the replating ability (secondary colony formation) of CML CFU-GM. Thus, amplification of CFU-GM may be a target for IFN-alpha and other agents used in the treatment of CML. Here we examined whether cytarabine (Ara-C) or all-trans retinoic acid (ATRA) exert similar effects and whether they might in combination with IFN-alpha enhance its efficacy. We found that Ara-C preferentially inhibits the formation of CML CFU-GM compared to normal CFU-GM, but this inhibition was not increased by addition of IFN-alpha. When Ara-C was added to cultures containing IFN-alpha, the inhibition of replating by CML progenitors was abrogated. ATRA increased significantly the plating efficiency of normal CFU-GM. The addition of IFN-alpha to ATRA had no effect on CML or normal colony numbers. However, addition of ATRA to cultures containing IFN-alpha reversed the selective inhibition of CML CFU-GM replating seen in cultures containing IFN-alpha alone. In four IFN-alpha/Ara-C experiments, secondary CML patient-derived colonies were examined by fluorescence in situ hybridisation (FISH). All of them were Ph chromosome positive. No significant effects on CFU-GM production were observed when CML primitive haemopoietic progenitor cells were investigated in a delta (delta) assay. Thus we conclude that combining IFN-alpha with Ara-C or ATRA neutralises the effect of IFN-alpha on CML CFU-GM. This observation provides a rationale for treating patients with alternating courses of IFN-alpha and Ara-C or ATRA, rather than giving either of these two agents in combination with IFN-alpha.

The tyrosine kinase inhibitor STI571, like interferon-alpha, preferentially reduces the capacity for amplification of granulocyte-macrophage progenitors from patients with chronic myeloid leukemia.

OBJECTIVE: To determine whether the compound STI571 (formerly known as CGP571418B), a selective inhibitor of the protein tyrosine kinase (PTK) activity of ABL and BCR-ABL proteins, preferentially reduces the capacity for amplification of granulocyte-macrophage progenitors (CFU-GM) from patients with chronic myeloid leukemia while sparing normal CFU-GM and to compare responses of CML and normal cells with STI571 and IFN-alpha. MATERIALS AND METHODS: Chronic phase CML and normal CFU-GM were grown with and without STI571, IFN-alpha, or the two agents in combination. Colonies were plucked and replated in 96-well microtiter plates. Secondary colonies were scored, and the results were expressed as the area-under-the-curve (AUC) of the distribution of secondary colony numbers per primary CFU-GM. This value gives an overall measure of the replating ability or amplification of the original CFU-GM population. RESULTS: STI571 selectively inhibits the formation of granulocyte-macrophage colony-forming cells (CFU-GM) from CML patients. It also significantly inhibits the amplification of CML CFU-GM (p = 0.002) as measured by secondary colony formation after replating primary CFU-GM colonies. In contrast, amplification of normal CFU-GM was enhanced (p = 0.001) at low concentrations (0.1 microM) of STI571 with a return to baseline at 10 microM STI571. Addition of interferon (IFN)-alpha to STI571 abolished the increase in normal CFU-GM amplification seen with either agent alone. There was a highly significant correlation between the in vitro response to STI571 and the in vitro response to IFN-alpha (r = 0.74 for CML cells, and 0.77 for normal cells). CONCLUSION: We conclude that STI571, like IFN-alpha, preferentially suppresses amplification of CML CFU-GM while sparing normal CFU-GM.

Selective elimination of leukemic CD34(+) progenitor cells by cytotoxic T lymphocytes specific for WT1.

Hematologic malignancies such as acute and chronic myeloid leukemia are characterized by the malignant transformation of immature CD34(+) progenitor cells. Transformation is associated with elevated expression of the Wilm's tumor gene encoded transcription factor (WT1). Here we demonstrate that WT1 can serve as a target for cytotoxic T lymphocytes (CTL) with exquisite specificity for leukemic progenitor cells. HLA-A0201- restricted CTL specific for WT1 kill leukemia cell lines and inhibit colony formation by transformed CD34(+) progenitor cells isolated from patients with chronic myeloid leukemia (CML), whereas colony formation by normal CD34(+) progenitor cells is unaffected. Thus, the tissue-specific transcription factor WT1 is an ideal target for CTL-mediated purging of leukemic progenitor cells in vitro and for antigen-specific therapy of leukemia and other WT1-expressing malignancies in vivo.

Contact-mediated inhibition of human haematopoietic progenitor cell proliferation may be conferred by stem cell antigen, CD34.

INTRODUCTION: The function of CD34, a transmembrane sialomucin expressed by human haematopoietic progenitor cells, is poorly understood. Its structure suggests it may act as a cell adhesion and signalling molecule. MATERIALS AND METHODS: KGIa cells and primary CD34-positive marrow cells were tested for their ability to aggregate in the presence of the anti-CD34 antibody QBEND10; CFU-GM colonies were grown using standard methods and tested for their content of colony-forming cells by replating; 'haematons' were isolated from marrow by filtration; the phosphorylation of CD34 was investigated by immunoprecipitation and Western blotting DISCUSSION: CD34-positive cells in human bone marrow, like KG1a cells, aggregate when incubated with QBEND10. Staining aggregates with anti-CD34-FITC revealed that aggregation involved co-localisation of CD34 at intercellular binding sites. We examined myeloid colonies (CFU-GM) grown from normal human bone marrow cells, and multicellular aggregates ('haematons') separated from freshly aspirated marrow by filtration, and found CD34-positive cells bound together with co-localisation of the CD34 at the binding sites. This finding shows that CD34-positive cell-cell adhesion occurs physiologically in vitro and in vivo. QBEND10-induced aggregation of KG1a and CD34-positive cells was enhanced by staurosporine (a protein kinase C inhibitor) and inhibited by genistein (a protein tyrosine kinase inhibitor). Moreover, aggregated cells had increased phosphorylation of tyrosine on CD34 and translocation of protein kinase C (PKC) to the cytoplasm, compared with non-aggregated cells. We used the ability of primary colonies to produce secondary colonies on replating as a functional parameter and found that the replating ability of the colonies was increased by treatment with genistein (P=0.003). In addition, the ability of individual samples of primary CD34-positive cells to undergo QBEND10-induced aggregation and the ability of CD34-positive cell-derived colonies to produce secondary clones on replating were inversely related (r=0.86). CONCLUSION: Our results suggest that homotypic aggregation of haematopoietic progenitor cells may be an important mechanism for preventing inappropriate proliferation in vivo. Thus, regulation of expression of the CD34 molecule may play an important role in maintaining the normal level of haematopoietic activity by contact-mediated inhibition of progenitor cell proliferation.

Evidence for a continuous decline in haemopoietic cell function from birth: application to evaluating bone marrow failure in children.

There are considerable differences in haemopoietic activity between young children and adults on the one hand, and between adults and the elderly on the other. A fundamental unanswered question is whether these differences relate to discrete stages or are part of a continuous process. We have sought to define aspects of the haematological ageing process, and have found that results from children with bone marrow failure syndromes differ from age-matched reference values. Haemopoietic cells were obtained from umbilical cord blood, from blood and bone marrow of healthy individuals and from the blood of young patients with bone marrow failure syndromes. Clonogenic myeloid progenitors (CFU-GM) were grown in semi-solid medium to measure their frequency; the proliferative capacity of myeloid progenitors was measured by replating colonies and observing secondary colony formation. We found that the frequency of CFU-GM in normal marrow increased and their proliferative capacity decreased exponentially with age. The proliferative capacity of CFU-GM in normal blood also decreased exponentially with age. This relationship extrapolated back to the levels of proliferation measured for cord blood CFU-GM (age = 0). The proliferative capacities of CFU-GM from children with bone marrow failure syndromes were severely reduced compared with age-matched reference values. These results indicate that a decline in haemopoietic progenitor cell function begins at birth and continues throughout life. This decline may occur prematurely in childhood marrow failure syndromes with a predisposition to leukaemia.

Cell biology of CML cells.

At the cellular level, expansion of haemopoiesis in chronic myeloid leukaemia (CML) must involve some imbalance in cell production along the myeloid maturation pathway. The relevant kinetic parameters are cell loss by apoptosis and differentiation and cell gain by proliferation (self-renewal). In spite of the predominance of the BCR-ABL-positive leukaemic cells, some BCR-ABL-negative, presumably normal, progenitor cells remain for long periods in chronic phase CML. Thus, understanding the kinetics of CML and normal progenitor cells may lead to therapeutic strategies capable of reducing malignant cell growth and reactivating normal haemopoiesis.

Optimal timing for processing and cryopreservation of umbilical cord haematopoietic stem cells for clinical transplantation.

Some of the factors that may influence the number and quality of cord blood haematopoietic progenitor cells available for transplantation have been investigated including site of collection, delayed processing after collection and cryopreservation protocol. We used the granulocyte-macrophage progenitor (CFU-GM) and erythroid burst-forming unit (BFU-E) assays to quantify progenitors. The capacity of CFU-GM to produce secondary colonies was used as a measure of progenitor cell quality. We found that: (1) there were no significant differences in total nucleated cells (TNC), mononuclear cells (MNC), CFU-GM or BFU-E numbers in paired specimens from the umbilical vein or veins at the base of the placenta. The potential of the CFU-GM to produce secondary colonies from the two sites was similar; (2) storing cord blood at room temperature or at 4 degrees C resulted in a significant reduction in progenitor cell numbers beyond 9 h; and (3) cryopreservation following either controlled rate freezing or passive cooling reduced MNC numbers, viability and CFU-GM survival insignificantly but the potential of CFU-GM to produce secondary colonies was significantly reduced post cryopreservation (P = 0.04). We conclude that the yield of CB progenitor cells is not affected by the site of collection, but is adversely affected by delays between collection and cryopreservation. Furthermore, cryopreservation reduced the CFU-GM potential to produce secondary colonies. Measures of progenitor cell quality as well as quantity may be relevant to assessing CB blood collections.

BCR-ABL-positive progenitors in chronic myeloid leukaemia patients in complete cytogenetic remission after treatment with interferon-alpha.

To determine the source of residual disease detected in patients with chronic myeloid leukaemia (CML) in complete cytogenetic remission (n=8) after treatment with interferon-alpha (IFN-alpha), we have tested CFU-GM colonies grown from bone marrow mononuclear cells or from plastic-adherent (Pdelta) cells for BCR-ABL mRNA using a nested multiplex RT-PCR. We compared our results with those obtained by analysis of colonies from newly diagnosed patients (n=4) and patients achieving no cytogenetic response (n=1) or incomplete cytogenetic response to treatment with IFN-alpha (n=5). A total of 1239 informative colonies were analysed. A small proportion of BCR-ABL-positive colonies was detected in all eight patients in complete cytogenetic remission, suggesting the persistence of leukaemia that could potentially lead to relapse. The overall proportion of BCR-ABL-positive colonies in patients achieving a cytogenetic response to IFN-alpha correlated with the levels of BCR-ABL transcripts detected in the peripheral blood by competitive RT-PCR (P=0.004). We conclude that residual disease detected in the peripheral blood of complete cytogenetic responders to IFN-alpha is at least partly derived from clonogenic myeloid cells. It is probable that the leukaemia clone in CML is only very rarely or never entirely eradicated by treatment with IFN-alpha.

A two-color BCR-ABL probe that greatly reduces the false positive and false negative rates for fluorescence in situ hybridization in chronic myeloid leukemia.

The t(9;22) translocation resulting in the fusion of BCR and ABL genes is pathognomonic in chronic myeloid leukemia (CML) and may be investigated at the molecular level using fluorescence in situ hybridization (FISH). Two-color BCR-ABL probes visualizing one fusion signal (1F FISH) have high false positive rates (FPR) and false negative rates (FNR). The FPR is a result of the random spatial association of probe signals within normal interphase cells so that some cells appear to contain the BCR-ABL fusion gene. The FNR of 1F FISH probes depends on the distance between the BCR and ABL probes hybridized to the BCR-ABL fusion gene (< or =368 kb); the "gap" between the signals causing the cell to be interpreted as normal. To overcome these difficulties, a two-color probe was used, employing four yeast artificial chromosome (YAC) sequences that span the breakpoint regions of the BCR and ABL genes and that visualize the two fusion signals BCR-ABL and ABL-BCR in CML cells (2F FISH). The FNR for the 2F FISH probes was assessed on clonal Ph+ granulocyte-macrophage-colony-forming cell (CFU-GM) derived colonies and was reduced to 0.4% (2/450), compared with an FNR of 13.5% (111/823) with 1F FISH. The FPR in normal mononuclear cells for the 2F FISH was 0. 19 +/- 0.12% (3/1,700), whereas the FPR using 1F FISH was 4.5 +/- 2.3% (63/1,294). The 2F FISH can thus be used to evaluate very small frequencies of BCR-ABL-positive and -negative interphase cells and may be of use in the clinical monitoring of CML.

Treatment with interferon-alpha preferentially reduces the capacity for amplification of granulocyte-macrophage progenitors (CFU-GM) from patients with chronic myeloid leukemia but spares normal CFU-GM.

The biological target for interferon (IFN)-alpha in chronic myeloid leukemia (CML) is unknown, but one possibility is that amplification of granulocyte-macrophage colony-forming cells (CFU-GM) is reduced. Replating CFU-GM colonies and observing secondary colony formation provides a measure of CFU-GM amplification. Amplification of CML, but not normal, CFU-GM in vitro was significantly inhibited by IFN-alpha (P = 0.02). In 5 out of 15 CML cases studied by fluorescence in situ hybridization, in vitro treatment with IFN-alpha increased the proportion of CFU-GM, which lacked BCR-ABL. The ability of patients' CFU-GM to amplify, and suppression of this ability by IFN-alpha, predicted responsiveness to IFN-alpha therapy in 86% of cases. Investigation of patients on treatment with IFN-alpha showed a threefold reduction in CFU-GM amplification in responders (P = 0.03) but no significant change in nonresponders (P = 0.8). We conclude that IFN-alpha preferentially suppresses amplification of CML CFU-GM to varying degrees. The differing in vitro sensitivities to IFN-alpha and growth kinetics of individual patients' cells could help differentiate those who will or will not benefit from treatment with IFN-alpha.

Interleukin 3 (IL-3), but not stem cell factor (SCF) increases self-renewal by human erythroid burst-forming units (BFU-E) in vitro.

Interleukin 3 (IL-3) and stem cell factor (SCF) are both important regulators of early haemopoietic cell development. Here, we have compared their effects or the kinetics of erythroid burst formation by BFU-E in normal adult bone marrow. We grew the BFU-E in the presence of erythropoietin (Epo) alone, Epo + IL-3 or Epo + SCF and scored the numbers of subcolonies in individual bursts after 14 days. The data were plotted as the cumulative distribution of the numbers of subcolonies per erythroid burst then linearised by logarithmic transformation. Analysis of the data revealed that IL-3 increases the numbers of subcolonies in BFU-E whilst SCF increases the size of the subcolonies themselves. Experiments involving combinations of Epo + IL-3 + SCF and the delayed addition of IL-3 or SCF indicated that the actions of IL-3 and SCF are largely independent of one another. We conclude that: (1) IL-3 acts at an earlier stage of erythroid differentiation than SCF, and (2) it may be possible to classify haemopoietic growth factors according to their effects on cell kinetics in vitro.

CD34+ cell selection in chronic phase chronic myeloid leukaemia: a comparison of laboratory grade columns.

CD34 positive (CD34+) cell selection is increasingly used for a number of important applications including gene therapy studies, ex vivo expansion and purging. However there are no data regarding the use of different technologies for CD34+ cell selection in chronic myeloid leukaemia (CML). We therefore compared the performance of three laboratory grade CD34+ selection columns (MiniMACS, Cellpro Ceprate LC and Baxter Isolex 50), using CML chronic phase peripheral blood (PB) and bone marrow (BM). With different CML samples the CD34+ purity from the three columns was equivalent, but comparing five paired samples the Ceprate purity was greater than MiniMACS, at 92.5 and 80.9%, respectively, P = 0.04. Combining results from paired and unpaired CML samples, MiniMACS (n = 7) gave a higher CD34+ yield than Ceprate LC (n = 8) or Isolex 50 (n = 4) with a mean of 51.1%, 24.3% and 13.2% respectively, (P = 0.04 and 0.01). Cell losses with all columns were similar. Attempts to improve the yield from the Ceprate LC columns by modifying the method were unsuccessful. Following MiniMACS and Ceprate LC separation the clonogenic potentials of CD34+ cells in the pre- and positive cell fractions were the same. The proportion of CD34+ 38- or CD34+ DR- cells was unchanged following column separation. These data suggest that the MiniMACS column may be the best column for CD34+ cell selection in CML but these results must be confirmed using large scale clinical columns once the MiniMACS column is licensed. It is possible that variations in CD34+ cell yields between the different columns reflect differences in antibody binding affinity to CML cells, or differences in column technologies.