Purging and Haemopoietic Progenitor Cell Selection by CD34 Cell Separation

Purging and haemopoietic progenitor cell selection by CD34؉ cell separation

¨ ¨ W Kruger1, M Gruber1, S Hennings1, N Fehse1, B Fehse1, K Gutensohn2, N Kroger1 and AR Zander1

1Bone Marrow Transplantation Unit, 2Blood Transfusion Service, University Hospital Eppendorf, Hamburg, Germany

Summary: High-dose chemotherapy supported by autologous stem cell reinfusion for treatment of metastatic and high-risk female breast cancer is currently under investigation in several Tumour cell contamination of autologous peripheral American and European studies.1,2,3 Tumour cell contami- blood stem cell samples (PBSC) and bone marrow (BM) nation of autologous stem cell products has been described is frequent. Enrichment of CD34+ stem cells is a promis- using immunocytochemistry, reverse transcriptase PCR and ing approach to purging tumour cells from autografts cell culture techniques in approximately one third of har- without damaging progenitor cells. Breast cancer cells vests with a range from 0% to 100%.4–7 The clonogenic were seeded (10؊3؊10؊7) into mononuclear cells from and metastatic potential of tumour cells isolated from blood G-CSF-mobilised PBSC and BM harvests from patients and autografts was demonstrated in cell culture assays and without breast cancer. CD34+ cells were enriched from in nude mice.8–10 The metastatic potential of accidentally mixtures either by immunomagnetic separation (Isolex- retransplanted tumour cells could be investigated by gene 50, and MiniMACS) or by biotin-streptavidin immu- marking of autografts prior to infusion followed by the noaffinity columns (Ceprate-LC). CD34؉ cell fractions detection of marked cells after relapse.11 However, due to were determined by FACS, cancer cells were detected the poor transduction efficacy of epithelial cancer cells by immunocytochemically with an anti-pancytokeratin gene transfer, a negative result would not exclude the possi- antibody. The CD34؉ cells were enriched with a median bility of relapse induction by accidentally reinfused tumour Isolex-50), 96.5% cells. Different approaches have been described to purge) (17 ؍ purity of 92.2% (43.5–96.1) (n MiniMACS) and 77.9% (31.4– autografts of tumour cells.12 A major problem of purging) (17 ؍ n) (99.2–66.6) -Ceprate-LC) from PBSC and BM har- procedures using cytotoxic agents is the damage to haemo) (15 ؍ n) (93.6 vests. The percentages of median recovery of CD34+ poietic progenitor cells with subsequent delay of cells were 30.8% (18.6–71.8) (Isolex-50), 69.9% (39.1– engraftment, an increased incidence of severe infections 100) (MiniMACS) and 42.9% (23.7–100) (Ceprate-LC). due to prolonged neutropenia and bleeding complications The median tumour cell reductions in log steps were due to thrombocytopenia.13–15 -Haemopoietic cells necessary for recovery after high (13 ؍ Isolex-50), 3.5 (2.6–4.3) (n) (13 ؍ n) (4.3–2.9) 3.7 -Ceprate-LC). dose therapy and progenitor support express the CD34 anti) (17 ؍ MiniMACS) and 1.5 (0.9–2.9) (n) Results were compared statistically by univariate analy- gen on their surface.16 These cells can easily be enriched sis. Purity was significantly (P Ͻ 0.05) better after Mini- by immunological methods using anti-CD34 antibodies and MACS selection. Recovery rates were significantly dif- separation by biotin-streptavidin immunoaffinity or mag- ferent between all devices tested. Tumour cell purging netic separation with paramagnetic microbeads. Rapid and was superior after immunomagnetic separation (P Ͻ safe engraftment after reinfusion of enriched CD34+ cell 0.001). Tumour cell purging is a main objective of fractions has been described by several investigators after CD34+ selection in the autologous setting. Our in vitro allogeneic and autologous transplantation.17,18 The data clearly indicate that immunomagnetic separation immunologic selection of CD34-positive cells has no toxic is more efficient in the prevention of accidental rein- side-effects on haemopoietic stem cells and progenitors fusion of contaminating tumour cells compared to without prolongation of cytopenia after reinfusion. immunoaffinity. However, it is not yet known if the Furthermore, CD34 positive-cell selection is a promising same results can be obtained with fresh contaminating approach to reducing the incidence or mitigating the sever- tumour cells. ity of graft-versus-host-disease in the allogeneic setting by Keywords: CD34+ cell selection; micrometastases; purg- adjusted or nearly complete T cell depletion. However, the ing; breast cancer major rationale for CD34-positive cell selection in autologous marrow and stem cell transplantation is a reduction or removal of contaminating tumour cells accidentally cohar- vested during leukapheresis to avoid their reinfusion after high-dose therapy.19,20 + + ¨ Correspondence: W Kruger, Bone Marrow Transplantation Unit, Univer- In this report we compare CD34 cell enrichment, CD34 sity Hospital Eppendorf, Martinistraße 52, 20246 Hamburg, Germany cell recovery, and the efficacy of purging of experimentally Received 22 September 1997; accepted 11 November 1997 seeded breast cancer cells from bone marrow and G-CSF- Purging by CD34+ cell selection ¨ W Kruger et al 666 mobilised peripheral blood stem cell collections between Only gravity was used for cell flow through selection three devices using immunoaffinity and magnetic columns. microbeads for separation of anti-CD34 antibody-labelled + + CD34 cells. We show that CD34 enrichment leads to a FACS analysis decrease in the total number of contaminating tumour cells up to 4.3 log steps with reasonably high recovery of CD34- The flow cytometric analyses were performed before and positive progenitor cells. after CD34+ cell selection on a FACScan flow cytometer (Becton Dickinson, Heidelberg, Germany). The instrument settings were established for linear amplification of light Material and methods scatter and logarithmic amplification of fluorescence chan- nels. The cells were stained following standard protocols with a phycoerythrin (PE)-coupled antibody (HPCA-2; Bone marrow and leukapheresis samples Becton Dickinson) not interfering with the antibodies used + Aliquots of 21 bone marrow samples and 37 leukaphereses for CD34 enrichment and recommended by the suppliers were obtained after informed consent. Nineteen volunteer of selection devices. Controls were performed with an anti- donors and two patients with acute myeloid leukaemia isotype IgG1-PE antibody (Becton Dickinson). Signals were underwent bone marrow harvests from the iliac crest under presented graphically as a dot plot and data were analysed general or spinal anaesthesia without prior growth factor with Lysis-II software (Becton Dickinson).21 mobilisation. Thirty-seven leukapheresis aliquots were obtained from 18 patients after preceding 5-day G-CSF Immunocytochemistry mobilisation with a daily dose of 10–24 ␮g/kg body weight injected subcutanously. The underlying diagnoses of the Cytospin slides were prepared with a Shandon cytospin stem cell donors were non-Hodgkin’s lymphoma (n = 8), centrifuge (Shandon, Runcorn, UK). Two hundred thou- multiple myeloma (n = 3), sarcoma (n = 2) and Hodgkin’s sand cells were spun onto each slide and one to five slides were prepared per sample. The number of slides available disease. Four patients were volunteer donors for allogeneic + transplantation. Mononuclear cells were isolated from mar- depended on the amount of recovered cells after CD34 cell + row and stem cell samples by Ficoll centrifugation follow- separation. Due to the limited CD34 cell count available for immunocytochemistry, we could test for samples spiked ing standard protocols. Most marrow and stem cell aliquots Ϫ Ϫ were too small to split them into three aliquots to run the with 10 6 or 10 7 cells only for relative tumour cell three devices. enrichment. Slides were air-dried, fixed, and stored at Ϫ20°C. Tumour cells on cytospin slides were detected by the anti- Cell lines and tumour cell spiking cytokeratin antibody KL1 (Boehringer Mannheim, Mannheim, Germany) and subsequent APAAP (alkaline The breast cancer cell lines MCF-7 and MDA-MB453 were phosphatase anti-alkaline phosphatase) stain. Stained cyto- used for tumour cell spiking experiments. The cell lines spin slides were evaluated by light microscopy, and the were purchased from the American Type Culture Collec- KL1-positive cells were counted.22 tion (ATCC), Rockville, MD, USA, and cultured and main- tained under the conditions recommended by ATCC. For spiking experiments, the cells were trypsinised to obtain Data analysis single cell suspensions. Suspensions were checked by light Data were analysed using the computer software Excel microscopy. Density of tumour cells was calculated using (Microsoft, Munich, Germany) and WinSTAT (Kalmia Co, a Neubauer chamber and mononuclear cell fractions of Cambridge, MA, USA). For the comparison of the two marrow, and leukapheresis samples were counted after Fic- groups, the independent t-test or the Mann–Whitney U test oll separation using a Coulter counter (Coulter, Krefeld, were used, and processing data from all devices were com- Germany). Leukapheresis and bone marrow samples were pared by univariate analysis. spiked with tumour cells MCF-7 and MDA-MB453 in logarithmic dilutions from 10Ϫ3 (0.1% tumour cells) to 10Ϫ7 (0.00001% tumour cells). For purging experiments tumour Results cells from 10Ϫ3 to 10Ϫ7 were performed for each device at least twice – once with each cell line – with higher concen- Twenty-one bone marrow and 37 leukapheresis samples trations being tested up to six times. Breast cancer cell lines were available to be spiked with tumour cells and purged were not mixed for experiments. by CD34+ selection devices. A statistical overview of MNC counts and percentage of CD34+ cells is shown in Table CD34+ enrichment 1. The absolute MNC count of leukapheresis samples was significantly higher with a median value of 600 (range 98– The CD34+ cell enrichment systems Ceprate-LC, Isolex- 1200) × 106 than that of bone marrow samples with 110 50 and MiniMACS were purchased from CellPro (Munich, (42–390) × 106 mononucleated cells (P Ͻ 0.001). One mil- Germany), Baxter (Munich, Germany) and Miltenyi Biotec lilitre of bone marrow harvest contained at median (Bergisch Gladbach, Germany), respectively. The systems 2.67 × 106 cells vs 104.67 × 106 MNC per ml leukapheresis were used according to the manufacturers’ instructions. sample (P Ͻ 0.001). The percentage of CD34+ cells per ml Purging by CD34+ cell selection ¨ W Kruger et al 667 Table 1 Cell count (1 × 106) and CD34+ content (%) of bone marrow and leukaphereses samples

MNC BM MNC BM/ml MNC LP MNC LP/ml CD34+ % BM CD34+ %LP n 21 21 37 37 20 36 mean 134.8 3.6 684.3 116.3 2.9 1.3 standard deviation 83.4 3.6 287.1 49.6 1.4 1.4 minimum 42.0 1.0 98.0 37.2 0.6 9.0e–02 maximum 390.0 17.7 1200.0 216.7 5.8 5.2 median 110.0 2.7 600.0 104.7 2.9 0.8

MNC BM = absolute mononucleated cell count in bone marrow samples; MNC BM/ml = mononucleated cell count per ml of bone marrow samples; MNC LP = absolute mononucleated cell count in leukaphereses samples; MNC LP/ml = mononucleated cell count per ml of leukaphereses samples; CD34+ %BM= percentage of CD34+ cells of mononucleated bone marrow cell fraction; CD34+ %LP= percentage of CD34+ cells of mononuclear leukapheresis cell fraction. was significantly higher in marrow samples than in stem Table 2 Purity and recovery (in %) of CD34+ cell fraction separated cells with a median value of 2.94 (0.58–5.83)% vs 0.80 from bone marrow and leukapheresis samples, factor of CD34+ cell enrich- (0.09–5.17)% (P Ͻ 10Ϫ4). Overall 20, 21 and 17 separation ment and purging of breast cancer cells in logs (absolute tumour cell reduction) experiments were done with Ceprate-LC, MiniMACS and Isolex-50. The numbers may differ in subsequent data Ceprate-LC MiniMACS Isolex-50 because not all experiments were evaluable. Purity Purity of enriched CD34+ cells n 15 17 17 mean 72.9 94.5 82.6 The median purity of enriched CD34+ cells was best after standard deviation 18.0 7.5 15.5 minimum 31.4 66.6 43.5 separation by MiniMACS (96.5%, 66.6–99.2%), followed maximum 93.6 99.2 96.1 by Isolex-50 (92.2%, 43.5–96.1%) and Ceprate-LC (77.9%, median 77.9 96.5 92.2 31.4–93.6%). The differences were significant between Ͻ Recovery MiniMACS and the other two devices (P 0.05), but not n 15 17 17 between Isolex-50 and Ceprate-LC. The data are shown in mean 51.6 72.3 36.5 Table 2. Figure 1a shows the corresponding box and whis- standard deviation 25.0 25.0 13.9 ker diagram. minimum 23.7 39.1 18.6 maximum 100.0 100.0 71.8 median 42.9 69.9 30.8 + Recovery of enriched CD34 cells Enrichment factor + n 15 17 17 Recovery of immunologically enriched CD34 cells was mean 96.4 129.7 74.1 best using the MinMACS device (69.9%, 39.1–100%), fol- standard deviation 98.2 199.8 55.7 lowed by Ceprate-LC (42.9%, 23.7–100%) and Isolex-50 minimum 17.9 17.0 18.6 (30.8%, 18.6–71.8%). The differences were significant maximum 353.5 854.6 191.2 between MiniMACS and the other two systems (P Ͻ 0.05) median 49.1 58.1 57.8 (Table 2 and Figure 1b). Purging (log) n 17 13 13 mean 1.6 3.4 3.6 Enrichment factor standard deviation 0.5 0.5 0.4 minimum 0.9 2.6 2.9 The CD34+ cell count per volume unit after selection div- maximum 2.9 4.3 4.3 ided by the CD34+ cell count before selection was defined median 1.5 3.5 3.7 as the enrichment factor. A large enrichment factor is a parameter for a small graft volume after selection. Con- sidering all samples, the median enrichment was significantly (P Ͻ 0.001) better from leukaphereses (108-fold, log, 2.6–4.3); the differences were not significant. Purging 19–855) than from marrow (30-fold, 17–69); the differ- by Ceprate-LC was significantly less effective with a ences between the devices were small and not significant. median breast cancer cell depletion of 1.5 log (0.9–2.9). The median enrichment factor was 49.1 (17.9–353.5) for The differences between immunomagnetic and immuno- Ceprate-LC, 57.8 (18.6–191.2) for Isolex-50 and 58.1 affinity separation were highly significant (P Ͻ 0.001). (17.0–854.6) for MiniMACS (Table 2 and Figure 1c). Data are shown in Table 2 and Figure 1d. For statistical reasons, experiments without recovered tumour cells after CD34+ selection were excluded from further analysis. This Purging was observed, as expected, at dilutions of 10Ϫ6 and 10Ϫ7. Tumour cell purging from marrow and leukapheresis For ethical reasons it was impossible to increase the sample samples was highest after separation by Isolex-50 with a volume as would be necessary for a statistical evaluation median of 3.7 log (2.9–4.3) followed by MiniMACS (3.5 of tumour cell dilutions of 10Ϫ6 and 10Ϫ7 and higher. Thus, Purging by CD34+ cell selection ¨ W Kruger et al 668 a b Purity(%) Recovery(%) 100 100

80 80

60 60

40 40

20 Ceprate-LC MiniMACS Isolex-50 Ceprate-LC MiniMACS Isolex-50 Separation Separation c d Enrichment factor TC depletion (log) 4 600

3 400

2 200

1 0

Ceprate-LC MiniMACS Isolex-50 Ceprate-LC MiniMACS Isolex-50 Separation Separation Figure 1 (a) Purity of enriched CD34+ cells, (b) recovery (%) of CD34+ cells after separation, (c) CD34+ cell enrichment, and (d) tumour cell depletion (log steps) by CD34+ selection. The boxes indicate the standard deviation around the mean value, the line in the boxes the median and the whiskers the range.

in order to achieve an approximation for these conditions, were correlated using the Spearman rank test. Positive cor- we calculated regression curves from the available data relations were found between the CD34+ cell percentage of shown in Figure 2a–c. This analysis suggests that the kin- unselected samples and the purity after separation etics of depletion are more independent of tumour cell con- (correlation coefficient (coc) = 0.40, P = 0.002), and centration for the Ceprate-LC than for the MiniMACS or between the purity of selected CD34+ cells and the percent- Isolex-50 devices. The purging efficacy of all devices age of recovery (coc = 0.32, P = 0.014). A negative corre- examined is better when samples with a higher tumour cell lation was found between the CD34+ cell content of load were processed. unselected samples and the CD34+ cell enrichment after selection (coc = Ϫ0.97, P Ͻ 0.001). Comparison of marrow and blood stem cell processing Comparing the processing data of all marrow and leukaph- Discussion eresis samples, only the CD34+ cell enrichment was significantly better from leukapheresis samples than from bone Currently, at least three devices are available for the iso- marrow (P Ͻ 0.05). Data are shown in Table 3. Subanalysis lation and enrichment of CD34-positive cell populations for each device confirmed this significance for Ceprate-LC from bone marrow and peripheral blood stem cell collec- and MiniMACS subgroups. Because of the small number tions. The devices work either on an immunomagnetic basis of experiments, results for blood and bone marrow have or by immunoaffinity. The device separating CD34-positive not been compared further. cells by biotin-streptavidin immunoaffinity has been used widely for clinical graft processing. Magnetic bead-based + systems are awaiting permission for clinical use. The ques- Correlation of depletion, purity, recovery and CD34 cell tion of which system is best for which purpose has not yet enrichment been examined. In this paper, we compare the capacities of Tumour cell depletion, purity of selected CD34+ cell frac- three systems for CD34-positive cell selection and for tion, recovery of CD34+ cells and CD34+ cell enrichment tumour cell purging. Purging by CD34+ cell selection ¨ W Kruger et al 669 a b 4.4 3.0 4.2

2.5 4.0

2.0 3.8

3.6 1.5 3.4 Isolex-50 Ceprate-LC 1.0 3.2 TC depletion (log) TC depletion (log)

0.5 3.0

2.8 0.0 34567 34567 TC dilution (10–x) TC dilution (10–x)

c 4.5

4.0

3.5

3.0 MiniMACS

TC depletion (log) 2.5

2.0 34567 TC dilution (10–x)

Figure 2 Regression curves of tumour cell depletion by (a) Ceprate-LC, (b) Isolex-50, and (c) MiniMACS. x = dilution factor (log).

Table 3 Bone marrow and leukapheresis processing data

Purity (%) Recovery (%) Enrichment TC depletion (log)

Bone marrow n 20 20 20 16 mean 86.5 52.6 37.3 2.7 standard deviation 16.1 23.9 17.1 1.0 minimum 31.4 24.0 17.0 1.0 maximum 99.1 100.0 69.3 4.3 median 93.1 44.6 29.9 2.8 Leukaphereses n 29 29 29 27 mean 81.8 54.3 143.6 2.8 standard deviation 16.8 28.0 159.3 1.2 minimum 40.2 18.6 18.6 0.9 maximum 99.2 100.0 854.6 4.3 median 87.0 43.2 108.0 3.3

In unprocessed samples, the median percentage of dilutions from 10Ϫ3 to 10Ϫ7. All dilution steps were exam- CD34+ cells was more than three times higher in bone mar- ined up to six times with each device. Tumour cells were row than in leukapheresis samples. The median mono- searched on APAAP-stained cytospin slides after CD34- nuclear cell count per volume unit was nearly six-fold positive selection. Slides without visible tumour cells were higher in leukapheresis samples than in bone marrow. excluded from statistical analysis. Purging by immunomag- Fifty-eight samples were used for preparing tumour cell netic devices was significantly better with a median of 3.6 Purging by CD34+ cell selection ¨ W Kruger et al 670 and 3.5 log steps than by immunoaffinity with 1.5 log steps. grafts is possible by CD34-positive cell selection needs Differences between both immunomagnetic devices were clarification. small and non-significant. We did not determine the capacity of the devices to Anderson et al23 described a 2.5 log reduction of breast deplete T lymphocytes for allogeneic graft processing. The cancer cells from bone marrow grafts by mafosfamide minor differences in CD34+ cell enrichment and the differ- purging with a concentration of 80 ␮g/ml. However, this ences in purity of selected cells between the three devices mafosfamide concentration also killed 99% of haemopo- indicate only slight differences in T cell depletion due to ietic progenitor cells in the in vitro study. A delay of variations of graft volume after processing. Beyond that, engraftment is common after reinfusion of mafosfamide- the concept of a complete T cell depletion to preserve a purged autografts after high-dose therapy for acute myeloid graft-versus-leukaemia effect and to generate a sufficient leukaemia.15 AML patients are commonly more intensively host-immune system after allogeneic transplantation has pre-treated than breast cancer patients. However, Shpall et been abandoned.26,27 However, to investigate purging effi- al described the same phenomenon after mafosfamide purg- cacy for processing aphereses from patients with non- ing of breast cancer autografts. In a subsequent clinical Hodgkin’s lymphoma, these experiments should be phase 1 trial, a significant delay in leukocyte recovery could repeated determining the percentage of B and T lympho- be correlated with the mafosfamide concentration used for cytes before and after selection. The necessity of investigat- tumour cell purging.24 The combination of mafosfamide ing these questions was recently supported by a paper from Hawkins et al,28 who described – in contrast to our results – incubation with immunotoxin treatment was examined by + O’Briant et al25 in an in vitro study. They described 2.2– the highest purity of CD34 cells selected from materials 5.4 log tumour cell reduction from a 1:10 breast cancer obtained from CML patients after immunoaffinity selection. cell dilution in irradiated marrow cells. The mafosfamide Our results demonstrate that effective breast cancer cell concentration of 40 ␮g/ml in combination with immunotox- purging from autografts is possible by immunomagnetic ins reduced CFU-GM growth to 37% after purging. Similar CD34-positive selection. The median tumour cell depletion results obtained by combined purging were published by by immunobead selection (3.5 log and 3.7 log) is inferior Anderson et al.23 A one log step tumour cell reduction after to the best results obtained by combined mafosfamide etherlipid incubation in an experimental model was pub- immunotoxin treatment (5.4 log), but since the clinical sig- lished by Dietzfelbinger et al.13 nificance of accidental tumour cell reinfusion has not been A problem of all in vitro purging studies is the choice clarified, patients should not be endangered by mafosfamide-induced prolongation of engraftment. Hopefully, the of the ratio of tumour:progenitor cells. Most investigators + have worked with 10Ϫ1–10Ϫ2 dilutions. These ratios allow CD34 selection can in future be combined with other purging techniques, eg antibody-based negative selection, to excellent statistical analyses after purging experiments, but optimise depletion. A final comparison of different breast they do not reflect the clinical situation. Tumour cell con- Ϫ Ϫ cancer cell purging procedures requires identical experi- tamination in vivo seems to occur in ratios of 10 5 to 10 6 mental settings and processing of realistic cancer cell con- or less. The purging kinetics – especially of immunological centrations in progenitor cells. approaches – could be dependent to a greater or lesser Our data clearly indicate the superiority of magnetic bead degree on tumour cell concentration. The difficulty of selection as compared to immunoaffinity selection for suf- examining purging approaches in realistic dilutions in vitro ficient tumour cell purging in this model system. However, is the number of cells necessary for a reliable statistical these data require proof with fresh leukaphereses samples analysis. For ethical reasons, it is impossible to use marrow contaminated with tumour cells and eventually testing in or leukapheresis cells for large scale spiking experiments. clinical studies. Buffy coat cells of red cell donors could be used as an alternative, but probably not for CD34+ selection experiments to be compared with analyses of leukaphereses. Thus, we decided to examine dilutions of cancer cells in References marrow or leukapheresis samples in log steps from 10Ϫ3 up to 10Ϫ7. Due to the relatively small number of cells 1 Hurd DD, Peters WP. Randomized, comparative study of available for experiments, we excluded cytospin analyses high-dose (with autologous bone marrow support) versus low- dose cyclophosphamide, cisplatin, and carmustine as consoli- from higher dilutions without visible tumour cells on cytos- dation to adjuvant cyclophosphamide, doxorubicin, and fluo- pin slides from statistical analysis, but we could exclude a rouracil for patients with operable stage II or III breast cancer relative tumour cell enrichment per slide by this practice. involving 10 or more axillary lymph nodes (CALGB Protocol Regression curves of tumour cell depletion against the 9082). Cancer and Leukemia Group B. J Natl Cancer Inst tumour cell concentration for each device were calculated. Monogr 1995; 41–44. ¨ ¨ The purging kinetics of CD34+ cell selection seem to be 2 Zander AR, Kruger W, Kroger N et al. High dose mitoxan- more independent of tumour cell concentration for cell trone with thiotepa, cyclophosphamide and autologous stem selection by immunoaffinity than for immunomagnetic sep- cell rescue for high risk stage II and stage III breast cancer. aration. However, regression graphs show, for all devices German GABG-4/EH-93-Study. Bone Marrow Transplant 1996; 18 (Suppl. 1): S24–S25. examined, that purging by CD34-positive cell selection is 3 Bezwoda WR, Seymour L, Dansey RD. High-dose chemo- most effective for high tumour cell concentrations and less therapy with hematopoietic rescue as primary treatment for effective with declining tumour cell concentration. The metastatic breast cancer: a randomized trial. J Clin Oncol question of whether a real negative purging of clinical 1995; 13: 2483–2489. Purging by CD34+ cell selection ¨ W Kruger et al 671 4 Datta YH, Adams PT, Drobyski WR et al. Sensitive detection (+) marrow progenitor cells provide effective reconstitution of occult breast cancer by the reverse-transcriptase polymerase for breast cancer (CA) and non-Hodgkin’s lymphoma (NHL) chain reaction. J Clin Oncol 1994; 12: 475–482. patients receiving high-dose chemotherapy with autologous ¨ 5 Kruger W, Krzizanowski C, Holweg M et al. Reverse bone marrow support (HDC/ABMS): recombinant granulocyte transcriptase/polymerase chain reaction detection of cytokera- colony-stimulating factor (G-CSF) accelerates hematopoietic tin-19 mrna in bone marrow and blood of breast cancer recovery (meeting abstract). Proc Am Soc Clin Oncol 1995; patients. J Cancer Res Clin Oncol 1996; 122: 679–686. 11: A61. ¨ 6 Passos-Coelho JL, Ross AA, Moss TJ et al. Absence of breast 18 Stockschlader M, Hassan HT, Zeller W et al. Allogeneic trans- cancer cells in a single-day peripheral blood progenitor cell plantation with CD34+-selected cells. Leuk Lymphoma 1996; collection after priming with cyclophosphamide and granulo- 25: 145–151. cyte–macrophage colony-stimulating factor. Blood 1995; 85: 19 Kvalheim G, Wang MY, Pharo A et al. Purging of tumor cells 1138–1143. from leukapheresis products: experimental and clinical 7 Hempel D, Oruzio D, Brockmeyer C et al. Immunomagnetic aspects. J Hematother 1996; 5: 427–436. cell selection for removal and enhanced detection of MRD 20 Shpall EJ, Jones RB, Bearman SI et al. Transplantation of from PBSC harvest in breast cancer. Blood 1996; 88 (Suppl. enriched CD34-positive autologous marrow into breast cancer 1): 251A. patients following high-dose chemotherapy: influence of 8 Joshi SS, Kessinger A, Mann SL et al. Detection of malignant CD34-positive peripheral blood progenitors and growth fac- cells in histologically normal bone marrow using culture tech- tors on engraftment. J Clin Oncol 1994; 12: 28–36. ¨ niques. Bone Marrow Transplant 1987: 1: 303–310. 21 Hassan HT, Zeller W, Stockschlader M et al. Comparison 9 Ross AA, Cooper BW, Lazarus HM et al. Detection and between bone marrow and G-CSF-mobilized peripheral blood viability of tumor cells in peripheral blood stem cell collec- allografts undergoing clinical scale CD34+ cell selection. Stem tions from breast cancer patients using immunocytochemical Cells 1996; 14: 419–429. and clonogenic assay techniques. Blood 1993; 82: 2605–2610. 22 Thorban S, Roder JD, Pantel K, Siewert JR. Immunocytoch- 10 Joshi SS, DeBoer JM, Strandjord SJ et al. Characterization of emical detection and prognostic significance of epithelial a newly established human Burkitt’s lymphoma cell line, tumor cells in bone marrow of patients with pancreatic carci- OMA-BL-1. Int J Cancer 1991; 47: 643–648. noma. Zentralbl Chir 1996; 121: 487–489; discussion 490– 11 Brenner MK, Rill DR. Gene marking to improve the outcome 492. of autologous bone marrow transplantation. J Hematother 23 Anderson IC, Shpall EJ, Leslie DS et al. Elimination of malig- 1994; 3: 33–36. nant clonogenic breast cancer cells from human bone marrow. ¨ ¨ 12 Kruger W, Stockschlader M, Hassan HT, Zander AR. Purging Cancer Res 1989; 49: 4659–4664. in der Knochenmark- und Stammzelltransplantation. J Lab 24 Shpall EJ, Jones RB, Bast RC Jr et al. 4-Hydroperoxycyclo- Med 1996; 4: 210–220. phosphamide purging of breast cancer from the mononuclear 13 Dietzfelbinger HF, Kuhn D, Zafferani M et al. Removal of cell fraction of bone marrow in patients receiving high-dose breast cancer cells from bone marrow by in vitro purging with chemotherapy and autologous marrow support: a phase I trial. ether lipids and cryopreservation. Cancer Res 1993; 53: J Clin Oncol 1991; 9: 85–93. 3747–3751. 25 O’Briant KCO, Shpall EJ, Houston LL et al. Elimination of 14 Beran M, Zander AR, Andersson BS, McCredie KB. clonogenic breast cancer cells from human bone marrow. Can- Regrowth of granulocyte-macrophage progenitor cells (GM- cer 1991; 68: 1272–1278. CFC) in suspension cultures of bone marrow depleted of GM- 26 Champlin R. T-cell depletion to prevent graft-versus-host dis- CFC with 4-hydroperoxycyclophosphamide (4-HC). Euro J ease after bone marrow transplantation. Hematol Oncol Clin Haematol 1987; 39: 118–124. North Am 1990; 3: 687–698. 15 Damon LE, Rugo HS, Ries CA, Linker CA. Delayed 27 De Gast GC, Gratama JW, Verdonck LF et al. The influence engraftment of 4-hydroperoxycyclophosphamide-purged auto- of T-cell depletion on recovery of T-cell proliferation to logous bone marrow after induction treatment containing herpesviruses and Candida after allogeneic bone marrow mitoxantrone for acute myelogenous leukemia. Bone Marrow transplantation. Transplantation 1989; 48: 111–115. Transplant 1996; 17: 93–99. 28 Hawkins TE, Marley SB, O’Brien SG et al. CD34+ cell selec- 16 Benedetti F. CD34+ cells: biological aspects. Tumori 1996; 82 tion in chronic phase chronic myeloid leukaemia: a compari- (Suppl. 2): S3–S13. son of laboratory grade columns. Bone Marrow Transplant 17 Shpall EJ, Jones RB, Johnston C et al. Purified CD34-positive 1997; 20: 409–413.