Expression of markers shared between human haematopoietic cells and neuroblastoma cells

Neuroblastoma (NB) is a solid tumor of childhood with a relatively bad prognosis, with the exception of young infants (less than 1 year), in whom spontaneous regression of tumor burden occurs. The reasons for this are still unknown but immune mechanisms may be involved. In this study, we have examined the ability of several monoclonal antibodies (MoAbs), which recognize markers predominantly expressed on human haematopoietic cells, to react with four human neuroblastoma cell lines (UKF-NB 1-4) and SK-N-SH as control cell line. In order to define the phenotype of NB cells, we used a large panel of MoAbs consisting of 2 major groups: a) well characterized MoAbs raised against antigens of neuroectodermal origin from the Kemshead-serie (e.g. UJ 13A, UJ 127.II, UJ 167.11, UJ 181.4, UJ 223.8, A2B5), b) monoclonal antibodies which have been considered to react with haematopoietic cells (HLA-DR and anti-CD-molecules CD1, CD7, CD9, CD10, CD13, CD16, CD19, CD20, CD24, CD57). The phenotypic analyses were performed at various times of culture by an immunoenzymatic procedure (APAAP-technique). Most of the MoAbs used against neuroblastoma cells showed a strong reactivity pattern with the NB cell lines. None of the antibodies against T-lymphocytes bound to any of the NB cells assayed in our study, with the exception of anti-CD 1. On the contrary, B-cell markers BA-2 (CD9) and BA-1 (CD24) cross-reacted with the NB cells just as well as the marker for NK-cells (CD57), but they did not express reactivity with Leu-11b (CD16), anti-CALLA (CD10) and anti-HLA-DR.     

Neuroblastoma during acute lymphoblastic leukemia in remission

Neuroblastoma was diagnosed in a child after a 20-month remission of a pre-B acute lymphoblastic leukemia (ALL). Clumps of atypical cells suggestive of neuroblastoma were seen in the bone marrow. They were positive for monoclonal antibody (MoAb) UJ13A (neuroblastoma cells) and negative for MoAb T29/33 (anti-leucocyte common antigen CD45) with immunocytochemical staining. A right paravertebral mass displacing the kidney was demonstrated by abdominal echotomography, and serum vanilmandelic acid was slightly increased. Despite specific chemotherapy against neuroblastoma and after a transient clinical improvement, the patient died 7 months later of disseminated disease. Immunocytochemical staining on cells frozen at diagnosis of leukemia with MoAb UJ13A and T29/33 was unable to demonstrate neuroblastoma cells and showed the pattern usually observed in leukemia (UJ13A- and T29/33+).     

Monoclonal antibody UJ 127:11 detects a 220,000-240,000 kdal. glycoprotein present on a sub-set of neuroectodermally derived cells

The monoclonal antibody UJ 127-11 was raised following immunization of mice with human foetal brain and subsequent somatic cell hybridization of spleen cells with the mouse myeloma cell line P3-X63-Ag8-653. Studies on normal foetal and adult tissues show that, by indirect immunofluorescence, the antigen recognized by UJ 127:11 is restricted in its expression to cells of neural rather than glial origin. Neural tumours such as neuroblastoma, medulloblastoma and ganglioglioma (neural component) bind the monoclonal antibody whereas malignancies originating from glial cells do not bind UJ 127:11. Biochemically the monoclonal antibody has been shown to bind to a glycoprotein of 220,000-240,000 mol. wt. under reducing and non-reducing conditions. Despite similarities in the molecular weight between human fibronectin and the antigen recognized by UJ 127:11, they have different serological and biochemical characteristics, suggesting that the monoclonal antibody is not binding to either cell or plasma fibronectin.     

Relationship between chemical-induced fragile sites and chromosomal breakpoints in malignant cells in children.

Many fragile sites in the human genome occur at or near chromosomal breakpoints reportedly involved in translocations of DNA material in neoplastic cells. This fact has led some investigators to postulate that fragile sites have a pathogenic role in human neoplasia. To learn whether caffeine-induced fragile sites relate to breakpoints found in the neoplastic cells of an individual patient, we studied lymphocytes from the peripheral blood of 32 patients in remission from malignant disease. Lymphocytes were cultured in medium containing either 5-Fluoro-2'-deoxyuridine (FdU) or FdU plus caffeine, and G-banded metaphases were examined for nonrandom breaks. Analyses of completely G-banded malignant cell chromosomes from 31 of the 32 patients were available for comparison. In only once case, a 5-year-old child with acute lymphoblastic leukemia, did a caffeine-induced fragile site (1q44) coincide with a breakpoint in the neoplastic cells [dup(1)(q21-->q44)]. Our findings suggest that chromosomal abnormalities in childhood malignancies cannot generally be explained by the presence of FdU- or FdU plus caffeine-induced fragile sites.     

Relationship Between Chemical-Induced Fragile Sites and Chromosomal Breakpoints in Malignant Cells in Children

Many fragile sites in the human genome occur at or near chromosomal breakpoints reportedly involved in translocations of DNA material in neoplastic cells. This fact has led some investigators to postulate that fragile sites have a pathogenic role in human neoplasia. To learn whether caffeine-induced fragile sites relate to breakpoints found in the neoplastic cells of an individual patient, we studied lymphocytes from the peripheral blood of 32 patients in remission from malignant disease. Lymphocytes were cultured in medium containing either 5-Fluoro-2'-deoxyuridine (FdU) or FdU plus caffeine, and G-banded mataphases were examined for nonrandom breaks. Analyses of completely G-banded malignant cell chromosomes from 31 of the 32 patients were available for comparison. In only once case, a 5-year-old child with acute lymphoblastic leukemia, did a caffeine-induced fragile site (1q44) coincide with a breakpoint in the neoplastic cells [dup(1)(q21 $ q44)]. Our findings suggest that chromosomal abnormalities in childhood malignancies cannot generally be explained by the presence of FdU- or FdU plus caffeine-induced fragile sites.     

Retinoic acid downregulates Rae1 leading to APC(Cdh1) activation and neuroblastoma SH-SY5Y differentiation

In neuroblastoma cells, retinoic acid induces cell cycle arrest and differentiation through degradation of the F-box protein, Skp2, and stabilization of cyclin-dependent kinase inhibitor, p27. However, the mechanism responsible for retinoic acid-mediated Skp2 destabilization is unknown. Since Skp2 is degraded by anaphase-promoting complex (APC)(Cdh1), here we studied whether retinoic acid promotes differentiation of human SH-SY5Y neuroblastoma cells by modulating Cdh1. We found that retinoic acid induced the nuclear accumulation of Cdh1 that paralleled Skp2 destabilization and p27 accumulation. The mRNA and protein abundance of Rae1-a nuclear export factor that limits APC(Cdh1) activity in mitosis-decreased upon retinoic acid-induced inhibition of neuroblastoma cell proliferation. Furthermore, either Rae1 overexpression or Cdh1 inhibition promoted Skp2 accumulation, p27 destabilization and prevented retinoic acid-induced cell cycle arrest and differentiation. Conversely, inhibition of Rae1 accelerated retinoic acid-induced differentiation. Thus, retinoic acid downregulates Rae1, hence facilitating APC(Cdh1)-mediated Skp2 degradation leading to the arrest of cell cycle progression and neuroblastoma differentiation.     

Chromosome Fragility of Lymphocytes from Breast Cancer Patients in Relation to Epidemiologic Data

The chromosomal fragility of peripheral blood lymphocytes from 50 women, undergoing operations for breast tumors (47 carcinomas, 2 intraductal papillomatoses and 1 malignant lymphoma) was studied to ascertain the association between chromosome fragility and epidemiologic data, such as a family or personal history of cancer, hormonal status, etc. Under conditions of folic acid and thymidine depletion, the average number of gaps and breaks on the patients' lymphocyte chromosome was 6.02±5.28 and that in the control medium was 2.0±2.0 while those of healthy controls were 5.8±5.5 and 1.36±1.22. These gaps and breaks were mostly seen in group A chromosomes (4.1±2.6) in 24 patients, including the 2 with benign tumors and the 1 with the lymphoma as well as 11 healthy controls. They were frequent in group B (3.0±0) in 3 patients, in group C (4.3±2.9) in 11 patients, and in groups D (2.0±1.0) and E (3.0±1.0) in 3 patients from each. This different distribution of gaps and breaks correlated neither with the patients' age nor with their tumor's histology, but patients having a late menarche were distributed in non-A groups. There was low inducibility of breaks in patients with a family history of breast cancer and/or relatively rare cancers. The availability of common fragile sites for studying an individual's susceptibility to cancer is discussed. One patient showed a bromodeoxyuridine-requiring heritable 10q25 fragile site. Another, with triple primary cancers, showed a constitutional translocation of t(5;19)(q15;q13).     

Chromosome fragility of lymphocytes from breast cancer patients in relation to epidemiologic data

The chromosomal fragility of peripheral blood lymphocytes from 50 women, undergoing operations for breast tumors (47 carcinomas, 2 intraductal papillomatoses and 1 malignant lymphoma) was studied to ascertain the association between chromosome fragility and epidemiologic data, such as a family or personal history of cancer, hormonal status, etc. Under conditions of folic acid and thymidine depletion, the average number of gaps and breaks on the patients' lymphocyte chromosome was 6.02 +/- 5.28 and that in the control medium was 2.0 +/- 2.0 while those of healthy controls were 5.8 +/- 5.5 and 1.36 +/- 1.22. These gaps and breaks were mostly seen in group A chromosomes (4.1 +/- 2.6) in 24 patients, including the 2 with benign tumors and the 1 with the lymphoma as well as 11 healthy controls. They were frequent in group B (3.0 +/- 0) in 3 patients, in group C (4.3 +/- 2.9) in 11 patients, and in groups D (2.0 +/- 1.0) and E (3.0 +/- 1.0) in 3 patients from each. This different distribution of gaps and breaks correlated neither with the patients' age nor with their tumor's histology, but patients having a late menarche were distributed in non-A groups. There was low inducibility of breaks in patients with a family history of breast cancer and/or relatively rare cancers. The availability of common fragile sites for studying an individual's susceptibility to cancer is discussed. One patient showed a bromodeoxyuridine-requiring heritable 10q25 fragile site. Another, with triple primary cancers, showed a constitutional translocation of t(5;19)(q15;q13).     

Construction of mouse A9 clones containing a single human chromosome (X/autosome translocation) via micro-cell fusion

Cell hybrids between hypoxanthine guanine phosphoribosyl transferase (HGPRT)-deficient mouse cell lines (A9 or RAG) and each of 12 different human fibroblasts (GM cells) containing various X/autosome translocations were formed, selected and isolated. Several human chromosomes including an X/autosome translocation carrying HGPRT locus were found in these hybrid cells. To construct A9 cell clones that contain a single X/autosome translocation, micro-cell fusion was undertaken to transfer these chromosomes from the hybrids to A9 cells. Karyotype analysis revealed that most of the resulting micro-cell hybrids contain, in a background of mouse chromosomes, only the human X/autosome translocations which were present in the GM cells used for cell hybridization. Sublines of A9 cells were established containing the following autosomal segments: 1q23----1qter; 1q12----1pter; 3p12----3pter; 3q21----3qter; 11q13----11qter; 11q13----11pter; 11p11----11qter; 11q23----11pter; 12q24----12pter; 16q24----16pter; 17q11----17pter.     

Expression of folate sensitive and aphidicolin induced chromosomal fragile sites in familial neuroblastoma

The expression of folate sensitive and aphidicolin induced fragile sites in the blood lymphocyte chromosomes of affected and unaffected members from 2 neuroblastoma families were studied. The subjects included 4 neuroblastoma patients, and 9 of their clinically healthy first degree relatives and corresponding number of age and sex matched controls. Lymphocytes cultured in folate deprived culture medium showed rare fragile sites at band p13.1 of chromosome 1, in a frequency of 3%-5% in all the 4 neuroblastoma patients. In aphidicolin treated cultures, the patients and unaffected members in neuroblastoma families, showed hypersensitivity to aphidicolin, as evidenced by the significant increase in percentage of aberration/cell (ab/c) and damaged cells (dc), over that of controls (P < 0.01). Aphidicolin induced fragile sites were more pronounced in chromosomes 1 and 2. A larger number of subjects have to be studied to prove whether altered fragile site expression may be a cytogenetic evidence for an individual or familial cancer predisposing genetic constitution.     

Chromosome Imbalances and Alterations of AURKA and MYCN Genes in Children with Neuroblastoma

Background: Neuroblastoma (NB), like most human cancers, is characterized by genomic instability,manifested at the chromosomal level as allelic gain, loss or rearrangement. Genetics methods, as well asconventional and molecular cytogenetics may provide valuable clues for the identification of target loci andsuccessful search for major genes in neuroblastoma. We aimed to investigate AURKA and MYCN generearrangements and the chromosomal aberrations (CAs) to determine the prognosis of neuroblastoma.Methods: We performed cytogenetic analysis by G-banding in 25 cases [11 girls (44%) and 14 boys (66%)]and in 25 controls. Fluorescence in situ hybridization (FISH) with AURKA and MYCN gene probes was alsoused on interphase nuclei to screen for alterations. Results: Some 18.4% of patient cells exhibited CAs., with asignificant difference between patient and control groups in the frequencies (P<0.0001). Some 72% of the cellshad structural aberrations, and only 28% had numerical chnages in patients. Structural aberrations consistedof deletions, translocations, breaks and fragility in various chromosomes, 84% and 52% of the patients havingdeletions and translocations, respectively. Among these expressed CAs, there was a higher frequency at 1q21,1q32, 2q21, 2q31, 2p24, 4q31, 9q11, 9q22, 13q14, 14q11.2, 14q24, and 15q22 in patients. 32% of the patients hadchromosome breaks, most frequently in chromosomes 1, 2, 3, 4, 5, 8, 9, 11, 12, 19 and X. The number of cells withbreaks and the genomic damage frequencies were higher in patients (p<0.001). Aneuploidies in chromosomes X,22, 3, 17 and 18 were most frequently observed. Numerical chromosome abnormalities were distinctive in 10.7%of sex chromosomes. Fragile sites were observed in 16% of our patients. Conclusion: Our data confirmed thatthere is a close correlation between amplification of the two genes, amplification of MYCN possibly contributingsignificantly to the oncogenic properties of AURKA. The high frequencies of chromosomal aberrations andamplifications of AURKA and MYCN genes indicate prognostic value in children with neuroblastomas and maypoint to contributing factors in their development.     

Chromosome rearrangements in fumigant appliers: possible relationship to non-Hodgkin's lymphoma risk.

Appliers of pesticides (n = 18) who are exposed to the fumigant phosphine or who have a mixed exposure to other pesticides and phosphine demonstrate a significant increase in chromosome rearrangements in G-banded chromosomes from peripheral blood compared to control subjects (n = 26). Appliers who had discontinued using phosphine for at least 8 months prior to specimen collection (n = 5) do not demonstrate significant increases in chromosome rearrangements compared to controls. Breakpoint analysis of 6,138 metaphases from all subjects demonstrates 196 breaks per 3605 metaphases in exposed subjects and 102 breaks per 2,533 metaphases in control subjects. Bands with significantly more breaks than expected based on band length in all study subjects were 1q32, 3p14, 7p15, and 14q11. Three of these four bands had significantly more breaks than expected in the exposed group, and all four bands had a significant excess of breaks in the control group. There are four bands with a significant excess of breaks in the exposed group and no breaks in the control group; each of these occurs in a known protooncogene region. These are 1p13 (NRAS), 2p23 (NMYC), 14q32 (ELK2), and 21q12 (ETS-2). Most breaks at bands 1p13, 14q32, and 21q22 are associated with chromosome rearrangements and occurred in appliers who have a mixed exposure to phosphine and other pesticides. Cytogenetic abnormalities, i.e., rearrangements and/or deletions involving bands 1p13, 2p23, and 14q32, are associated with non-Hodgkin's lymphoma. We speculate that these findings could relate to the risk of evolution of a neoplastic clone in these workers. Epidemiological studies of similarly exposed workers indicate an excess of non-Hodgkin's lymphoma.     

Therapeutic application of a radiolabelled monoclonal antibody in nude mice xenografted with human neuroblastoma: tumoricidal effects and distribution studies

Monoclonal antibody UJ13A radiolabelled with isotopes of iodine has been shown to selectively localize to human neuroblastoma xenografts. When 131I-UJ13A conjugates were given to nude mice at high doses (100-150 microCi), tumours temporarily disappeared, only to regrow. No selection for neuroblastoma cells that were UJ13A - negative was observed. Distribution studies on mice receiving radiolabelled UJ13A demonstrated the antibody is rapidly lost from the blood of animals. This cannot be accounted for by selective uptake into xenografts or any other mouse organ examined. We concluded there is a rapid equilibration of isotope between intra- and extravascular spaces in the animal. The rapid, biphasic loss of UJ13A from the blood of mice may explain why so little injected antibody can target to the human tumour xenografts.     

食管癌遗传病因的研究：Ⅴ.林县食管癌高癌家族成员淋巴...

The chromosome fragile sites of cultured peripheral lymphocytes from 40 members of 4 high risk cancer families and 10 members of 4 low risk cancer families in Linxian County were analysed. The results showed that 46 fragile sites in 7045 lymphocytes expression at 502 times (7.13%) were found in high risk cancer families and 8 fragile sites in 1053 lymphocytes expression at 26 times (2.47%) were found in low risk cancer families. There was a significant difference between the two groups (P less than 0.01). In 46 fragile sites carried by 40 members of high risk cancer families, 27 were common, 5 rare, 12 provisional and 2 new fragile sites. Among them, the fragile sites at 1p22-p36 and 4q21-q31 were detected in members of high risk cancer families and in patients with esophageal cancer, meanwhile, uniform breakpoint in chromosome deletion and rearrangement was also found in 4 esophageal cancer cell lines. Therefore, the author conjectures that these fragile sites at 1p13-p36 and 4q21-q31 may be fragile site-specific for high risk cancer families and patients with esophageal cancer, and they may be breakpoint-specific for esophageal cancer cells. These fragile sites may play an important role in esophageal carcinogenesis in high risk cancer families.     

Immunophenotype profile of childhood medulloblastomas and supratentorial primitive neuroectodermal tumors using 16 monoclonal antibodies

Immunophenotype analysis of 17 childhood medulloblastoma (MED) and supratentorial primitive neuroectodermal tumors (SPNET) was performed on frozen sections using 16 monoclonal antibodies (MoAb) with the biotin-streptavidin alkaline phosphatase immunohistochemical technique. Neuroectodermal associated antigens, reacting with MoAb UJ13/A, UJ127.11, UJ167.11, and UJ223.8 were detected on greater than 10% of the cells in 15 of 17 MED/SPNET. Thy-1 was present on 14 of 17 tumors and absent on two of three SPNET. Neuronal (NF) and glial (GFAP) differentiation markers were evaluated. NF-H was demonstrated in 15 of 17, NF-M in six of 17 and NF-L in one of 17 tumors; GFAP was positive in nine of 17 patients. In nine of 17 MED/SPNET both proteins were present within the same tumor. Common leukocyte antigen was demonstrated on greater than 50% of the cells in four of 14 tumors as were shared tumor/leukocyte markers using monoclonal antibodies Thy-1, PI153/3, UJ308. The most frequent MED immunophenotype analysis was UJ 13/A+, UJ 127.11+, UJ 167.11+, UJ223.8+, PI 153/3+, A2B5+, GFAP+, NF-H+, and CLA-, NF-M-, NF-L-, 215-, 275-, 282.1-. The authors conclude that MED and SPNET are heterogeneous for expression of 16 markers and have similar immunophenotype analysis profiles, supporting the concept of their common, neuroectodermal origin. Common leukocyte antigen on both tumor cells and leukocytes precludes identification of tumor infiltrating leukocytes using monostaining techniques.     

Increased fragile sites and sister chromatid exchanges in bone marrow and peripheral blood of young cigarette smokers

Cigarette smoking is considered to be the single most important acquired cause of cancer mortality. Studies of chromosome aberrations, sister chromatid exchanges, and fragile sites in peripheral blood or bone marrow are useful methods to detect the effects of the environmental mutagens or carcinogens found in cigarette smoke. The effects of smoking on the immature cells in the bone marrow have not been studied. Here, we examine the peripheral blood and bone marrow in 18 smokers (15 females and 3 males) with a median age of 25 years (range, 21-40) and an average cigarette use corresponding to 6 pack years. In both bone marrow cells and peripheral blood lymphocytes, we were able to show a significantly increased frequency of sister chromatid exchanges in smokers with a 5 or more cigarette pack year history, but not in those who smoked less than 5 pack years. We also found a higher frequency of sister chromatid exchanges in peripheral blood lymphocytes than in bone marrow cells. In addition, the peripheral lymphocytes of smokers demonstrated (a) a significantly higher frequency of fragile sites, (b) an increased number of metaphases with extensive breakage; and (c) elevated expression of fragile sites at the cancer breakpoints 3p14.2, 11q13.3, 22q12.2, and 11p13-p14.2 and at the oncogene sites bcl 1, erb B, erb A, and sis. Our results suggest that chromosomal DNA of peripheral blood lymphocytes is sensitive to cigarette smoking. Studies of the chromosomal changes in these cells provide an index of the mutagenic damage caused by these exogenous agents in individual patients and the ability of individuals to repair that damage, and might predict susceptibility to malignant events.     

The distribution of alternative agents for targeted radiotherapy within human neuroblastoma spheroids

This study aims to select the radiopharmaceutical vehicle for targeted radiotherapy of neuroblastoma which is most likely to penetrate readily the centre of micrometastases in vivo. The human neuroblastoma cell line NB1-G, grown as multicellular spheroids, provided an in vitro model for micrometastases. The radiopharmaceuticals studied were the catecholamine analogue metaiodobenzyl guanidine (mIBG), a specific neuroectodermal monoclonal antibody (UJ13A) and beta nerve growth factor (beta NGF). Following incubation of each drug with neuroblastoma spheroids, autoradiographs of frozen sections were prepared to demonstrate their relative distributions. mIBG and beta NGF were found to penetrate the centre of spheroids readily although the concentration of mIBG greatly exceeded that of beta NGF. In contrast, UJ13A was only bound peripherally. We conclude that mIBG is the best available vehicle for targeted radiotherapy of neuroblastoma cells with active uptake mechanisms for catecholamines. It is suggested that radionuclides with a shorter range of emissions than 131I may be conjugated to benzyl guanidine to constitute more effective targeting agents with potentially less toxicity to adjacent normal tissues.     

Double minute chromosomes. A bone marrow indicator of neuroblastoma metastasis and relapse: two case reports

Two cases of childhood neuroblastoma are presented. Case 1 was diagnosed as Stage IV with metastasis to the bone marrow. During remission, histologic studies of bone marrow aspirate and biopsy showed a normocellular marrow with no evidence of malignant cells. Concurrent cytogenetic studies of the bone marrow showed the majority of the cells to contain double minute chromosomes (DM). The chromosome findings indicated the presence of neuroblastoma cells in the marrow prior to histologic evidence of relapse. Case 2 was diagnosed as Stage I neuroblastoma with no metastasis to the bone marrow. Subsequent cytogenetic studies showed DM present in a small number of cells and a deletion of chromosome 1 (1p-) in a single cell. The chromosome findings indicated an advanced stage of malignancy which was not evident with histologic techniques. These findings suggest that cytogenetic analysis of bone marrow can be a valuable aid to the early diagnosis, prognosis, and treatment of neuroblastoma.     

Similarities in glycosylation of human neuroblastoma tumors and cell lines

The presence of fucosyl residues linked alpha 1----3(4) to N-acetylglucosamine was demonstrated on the oligosaccharides from glycoproteins of 11 human neuroblastoma tumors from ten different patients. This finding is in complete agreement with the previous report that human neuroblastoma cell lines contained an unusually large proportion of metabolically incorporated L-[3H]fucose in this specific linkage (U. V. Santer and M. C. Glick, Cancer Res., 43:4159-4166, 1983). Furthermore, the glycopeptides derived from the neuroblastoma tumors had a low percentage of fucose-containing biantennary oligosaccharides as determined by affinity to concanavalin A-Sepharose and in this characteristic were similar to glycopeptides from virus transformed and other tumor cells. To obtain these results, the tumor cells were labeled metabolically for 48 h with L-[3H]fucose. The cells were harvested and digested with Pronase, and the glycopeptides were isolated and treated with alpha-L-fucosidase from almonds, specific for the release of fucose linked alpha 1----3(4) to N-acetylglucosamine. A portion of the glycopeptides was characterized by serial affinity chromatography on immobilized concanavalin A and lentil lectin. The phenotypic similarity of the tumor cells to the cell lines, particularly CHP-134, included the paucity of biantennary oligosaccharides and the presence of fucosyl residues on the multiantennae of the glycopeptides.     

Immunological detection of neuroblastoma cells in bone marrow harvested for autologous transplantation

In about 50% of patients with stage IV neuroblastoma, micrometastases are present in the bone marrow when it is harvested for an autograft to follow induction therapy, and the risk of graft contamination by neuroblastoma cells has been the rationale for the use of a purging procedure. However, bone marrow metastases are detected with trephine biopsies which only explore the sites biopsied and do not reflect potential contamination of the pooled marrow harvested for autograft. A two-colour fluorochrome labelling method is described which permits as few as 1 neuroblastoma cell in 100,000 normal bone marrow cells from the autograft to be detected. Three monoclonal antibodies (UJ13A, H11 and 11.14) which react with neuroblastoma cells are used as single reagent in combination with a fourth anti-panleucocyte antibody. This method requires only 2 h for the analysis of three million marrow cells from the autograft, and is more effective than alkaline phosphatase staining with the same monoclonal antibodies. Results were compared with conventional techniques (four biopsies and four aspirates) carried out at the same time in 34 consecutive patients. Of 18 cases with negative aspirates and biopsies, neuroblastoma cells were detected in two autografts by the immunological method. Of 16 cases with positive aspirates and/or biopsies, 10 autografts were positive by the immunological method and six were negative. Thus, marrow micrometastases were detected in 16 of the 34 patients, but the autograft contained malignant cells in only 12 of these patients and the immunological analysis demonstrated that the use of a purging procedure allowed the elimination of neuroblastoma cells from the autograft before its reinjection to the patients.