¹Laboratory of Neuroblastoma Research, Advanced Biotechnology Center, 16132 Genoa, Italy

²Laboratory of Population Genetics, National Institute for Cancer Research (IST), 16132 Genoa, Italy

³Laboratory of Molecular Genetics, Gaslini Children's Hospital, 16148 Genoa, Italy

⁴Department of Paediatrics and Child Health, University of Birmingham, Birmingham B15 2TT, UK

⁵Department of Oncology, Biology and Genetics, University of Genoa, 16132 Genoa, Italy

⁶Department of Research, AI duPont Hospital for Children, Wilmington, Delaware, DE 19899, USA

⁷Department of Hematology and Oncology, Gaslini Children's Hospital, 16148 Genoa, Italy

⁸Laboratory of Immunogenetics, National Institute for Cancer Research (IST), 16132 Genoa, Italy

⁹Department of Internal Medicine, Cardioangiology and Hepatology, University of Bologna, 40138 Bologna, Italy

Correspondence to: G P Tonini, Laboratory of Population Genetics, National Institute for Cancer Research (IST), L.go R. Benzi, 10, 16132 Genoa - Italy; E-mail: tonini@cba.unige.it

^aAuthors have equally contributed to the present work

The most frequent genetic alterations described in neuroblastoma (NB) are amplification of MYCN oncogene and deletion of chromosome 1p, although somatic deletions have been demonstrated at other chromosomal intervals. Since loss of heterozygosity (LOH) at distal 4p has been observed in about 20-29% of neuroblastomas, we have evaluated deletions in 41 Italian NB samples by LOH analysis at loci mapping to 4p as follows: pter-D4S2936-D4S412-D4S2957-D4S432-D4S3023-D4S431-cen. Our analysis showed allele losses in eight out of 41 samples (19.5%) and allowed the identification of a smallest region of overlapping deletion (SRO) of 3.0 cM, delimited by D4S412 and D4S3023. Two of these tumors with 4p LOH are from patients belonging to a family with recurrent NB. Interestingly the genotyping of this family revealed an identical haplotype that includes the nonrecombinant loci D4S412, D4S2957 and D4S432 shared by all affected children and demonstrated that this haplotype is retained in the two tumors carrying somatic deletions from patients of this family. Furthermore linkage analysis was performed in two NB families and yielded an overall lod-score of 3.0 in the interval including the haplotype. This provides a confirmatory indication that the region delimited by D4S2936 and D4S3023, which also includes the new defined SRO, may harbor NB predisposing gene/s.

Oncogene (2002) 21, 8356-8360. doi:10.1038/sj.onc.1206009

neuroblastoma; chromosome 4p; LOH; linkage analysis; tumor suppressor genes

Neuroblastoma (NB) is a neural crest derived tumor that substantially contributes to childhood cancer mortality. Although extensive molecular studies have demonstrated that tumor aggressiveness greatly depends on the genetic abnormalities found in NB cells, no NB-related gene has been discovered until now. The pattern of genetic aberrations in NB is complex and heterogeneous and the molecular events leading to the development of the tumor are largely unclear. The most frequent genetic alterations described in NB tumors and derived cell lines are amplification of MYCN oncogene (Schwab et al., 1984; Brodeur et al., 1984), deletion of the short arm of chromosome 1 (1p) (White et al., 1995; Schwab et al., 1996; Martinsson et al., 1997) and amplification of genetic material at 17q (Bown et al., 2001), all of which are associated with a severe clinical course. Moreover somatic deletions at other chromosomal intervals (Srivatsan et al., 1993; Takita et al., 1995) suggest the involvement of several genes in the pathogenesis of NB.

Loss of heterozygosity (LOH) has been observed at distal 4p in sporadic neuroblastomas with an incidence ranging from 20 to 29% (Caron et al., 1996; Oude Luttikhuis et al., 2001). Deletions at 4p have also been demonstrated by comparative genomic hybridization (CGH) analysis in sporadic (Vandesompele et al., 1998, 2001) and familial NB cases (Altura et al., 1997; McConville, personal communication). The role of chromosome 4 in the development of NB is also suggested by experiments in transgenic mice, which overexpress MYCN in neuroectodermal cells and develop neuroblastomas in 28% of cases. Remarkably 25% of these tumors carry a deletion of chromosome 5, orthologous with human chromosome 4 (Weiss et al., 1997). All these data are consistent with the hypothesis that a NB-related gene maps to distal 4p.

We used LOH analysis to search for 4p deletions in Italian NB cases. Forty-one samples of primary tumors at onset from patients in different stages of disease (15 patients in stage 1, five in stage 2, six in stage 3, 11 in stage 4 and four in stage 4s) were chosen on the basis of at least 80% tumor cell content to reduce nonneoplastic tissue contamination. Thirty-nine were sporadic cases and two were familial cases. As previously described (Perri et al., 1997), in each patient DNA was isolated from both tumor and blood samples. We evaluated LOH at distal 4p through the analysis of six microsatellite markers (Figure 1) belonging to the Genethòn human linkage map (http://www.genethon.fr). Microsatellite analysis has been performed using two methods. DNA samples were amplified using primers labeled with FAM and HEX fluorescent amidites, PCR products loaded in a ABI 377 (Applied Biosystems, Foster City, CA, USA) automated sequencer and allele size defined by GeneScan^Ô and Genotyper^Ô softwares. A silver staining detection was adopted to reveal polymorphisms typed by amplification with unlabeled primers. LOH assessment threshold was set at 60% decrease in allele peak or signal intensity of the tumor DNA sample with respect to the constitutional one.

LOH at distal 4p occurred in eight out of 41 (19.5%) informative samples, in agreement with the LOH incidence found by Caron et al. (1996). Two out of eight tumors had terminal deletions (1149 and 1425), four had a large deletion (1068, 1083, 565 and 736) and two (820 and 1045) showed interstitial deletions (Figure 1). It is interesting to note that the deletion from case 1045 encompassing loci D4S2957 and D4S432 identifies a smallest region of overlapping deletion (SRO) of 3.0 cM, delimited by D4S412 and D4S3023 (Figure 1). These findings allow to redefine and narrow down the minimal deleted region of 34 cM previously described (Caron et al., 1996).

Patients whose tumor sample showed LOH (Figure 1) were one stage 1 patient (case no. 1068), two stage 2B (no. 565 and 820), two stage 3 (no. 736 and 1083) and three stage 4 (no. 1045, 1149 and 1425). MYCN amplification was found in the tumor samples of a stage 3 (no. 1083) and a stage 4 (no. 1149) patients and 1p36 deletion in the tumor of a stage 2B patient (no. 820). These data suggest that 4p deletion may occur independently of MYCN amplification, 1p36 deletion and stage of disease. Five out of eight cases carrying the 4p deletion were stages 3 and 4, which suggests that 4p deletion may be more frequent in advanced stages of disease. Furthermore, given that the analysed NB cases were chosen on the basis of tumor cell content, they cannot be considered representative of the entire Italian NB population in terms of stage distribution. Therefore, we believe that the 19.5% 4p deletion found in the present study might underestimate the true proportion of LOH. As a matter of fact in a previous report the CGH analysis carried out on patients with advanced stage of disease showed 4p losses in seven out of 22 cases (31.8%) (Lo Cunsolo et al., 2000).

Two tumor samples showing allelic losses (Figure 1, cases 736 and 1425) are from patients belonging to a family with recurrent NB, where four affected children (first and second cousins) were identified in the last generation (Figure 2; Family IGG-E). This confirms that 4p deletion can be found in both sporadic and familial NB cases. A chromosome 4p deletion was also found in a patient from the English family UB-NB1 (Figure 3).

The LOH analysis carried out on the two available tumor samples from family IGG-E revealed a terminal deletion on the paternally derived chromosome of patient IV-23 (Figure 2; see also Figure 1, case 1425) and a large deletion including all analysed markers in patient IV-25, that occurred on the chromosome inherited from his mother (Figure 2; see also Figure 1, case 736). The latter result is in keeping with the monosomy of chromosome 4 previously revealed by CGH on tumor sample from patient IV-25 (data not shown). In family UB-NB1 a terminal deletion was found on the tumor from patient III-1 (Figure 3). The deletion occurs on the maternally derived chromosome and includes the genomic interval between D4S2936 and D4S394. Although NB usually occurs as a sporadic tumor, about 1-1.5% of all diagnosed cases show familial recurrence with autosomal dominant inheritance and incomplete penetrance. Knudson and Strong (1972) suggested that the 'two-hit hypothesis' proposed for retinoblastoma could be applied to NB and they estimated that 22% of sporadic neuroblastomas might result from a germinal mutation involving different genes.

If supported by linkage information, the molecular analyses of tumor samples carrying somatic mutations such as deletions may provide evidence of the second hit necessary for the inactivation of a tumor suppressor gene. On the basis of LOH results we carried out linkage analysis in the two NB families to evaluate whether the chromosomal region affected by somatic deletions is in linkage with the disease.

Linkage analysis is a powerful tool that has allowed the identification of susceptibility genes in specific clusters of familial cancers. Since families with recurrent NB are rarely large enough to have a sufficient number of meiotic recombinant events to reach statistical significance, additional linkage information needs to be obtained from a collection of several families by adding up lod-scores across them. In the first instance we carried out the genotyping of the two NB families at distal 4p. Haplotype reconstruction interestingly revealed that all four patients from family IGG-E shared an identical haplotype including the three nonrecombinant loci D4S412, D4S2957 and D4S432, delimited by D4S2936 and D4S3023 (Figure 2; black box). This genomic interval covers the SRO of 3.0 cM identified by our LOH analysis (Figure 1). Moreover, in tumors from patients IV-23 and IV-25 of family IGG-E allele losses occur on the chromosome homolog to that which carries the co-segregating haplotype (Figure 2; grey shadow). These observations are consistent with the assumption that a disease-gene carrying a germline mutation might be located at the genomic interval encompassing the three nonrecombinant markers and the somatic deletion at the target interval may represent the second hit necessary for the inactivation of a recessive gene according to Knudson's hypothesis (Knudson, 1971).

Then, linkage analysis was performed by means of the lod-score method and the 'affected-only' approach, assuming dominant transmission of the putative susceptibility gene, a disease allele frequency of 0.0001 and equal marker allele frequencies. The 'affected-only' approach is independent of assumptions about the unknown penetrance of the disease phenotype. In this analysis all unaffected individuals likely to carry the disease gene are coded as unknown at the disease phenotype. Therefore, only affected individuals provide information about co-segregation of the disease and marker alleles, whereas unaffected individuals only contribute information on marker phase. Multipoint analysis using markers D4S2936, D4S412, D4S432 and D4S3023 and the programs VITESSE (O'Connell and Weeks, 1995) for family IGG-E and GENEHUNTER (Kruglyak et al., 1996) for family UB-NB1 yielded a maximum of 3.0 overall lod-score over the whole interval including the two nonrecombinant markers D4S412 and D4S432. This further reinforces the hypothesis of a NB predisposing region delimited by loci D4S2936 and D4S3023.

In the case of rare disorders such as NB, confirmatory evidence for linkage at a given genomic interval can be found in patients with a chromosomal abnormality at/or encompassing that location. In the present study somatic deletions have been demonstrated in tumor samples from children affected by sporadic and familial NB. In one family it has been observed that in the two tumors carrying somatic deletions the haplotype shared by all affected children is retained. This suggests that genetic information from distal 4p may play a role in a group of neuroblastomas where a germline mutation segregates in family members. The identification of a SRO of 3.0 cM mapping within the predisposing region promotes the latter as a good candidate likely to harbor tumor suppressor gene/s.

Acknowledgements

This work was supported by Fondazione Italiana per la Lotta al Neuroblastoma. We are grateful to surgeons, clinicians and pathologists of the Italian Cooperative Group for Neuroblastoma and AIEOP (Associazione Italiana di Ematologia e Oncologia Pediatrica) supplying tumor samples and clinical data, to Dr Katia Mazzocco for MYCN amplification and 1p deletion data and to Dr Roberta Bricchetto for language revision.

Altura RA, Maris JM, Li H, Boyett JM, Brodeur GM, Look AT. (1997). Genes Chromosomes Cancer, 19: 176-184.

Bown N, Lastowska M, Cotterill S, O'Neill S, Ellershaw C, Roberts P, Lewis I, Pearson AD. (2001). Med. Pediatr. Oncol., 36: 14-19.

Brodeur GM, Seeger RC, Schwab M, Varmus HE, Bishop JM. (1984). Science, 224: 1121-1124. MEDLINE

Caron H, van Sluis P, Buschman R, Pereira do Tanque R, Maes P, Beks L, de Kraker J, Voute PA, Vergnaud G, Westerveld A, Slater R, Versteeg R. (1996). Hum. Genet., 97: 834-837. Article MEDLINE

Knudson JrAG. (1971). Proc. Natl. Acad. Sci. USA, 68: 820-823. MEDLINE

Knudson AG, Strong LC. (1972). Am. J. Hum. Genet., 24: 514-532.

Kruglyak L, Daly MJ, Reeve-Daly MP, Lander ES. (1996). Am. J. Hum. Genet., 58: 1347-1363.

Lo Cunsolo C, Bicocchi MP, Petti AR, Tonini GP. (2000). Br. J. Cancer, 83: 1295-1300.

Martinsson T, Sjöberg RM, Hedborg F, Hallstenson K, Nordling M, Kogner P. (1997). Eur. J. Cancer, 33: 1997-2001. MEDLINE

O'Connell JR, Weeks DE. (1995). Nat. Genet., 11: 402-408. MEDLINE

Oude Luttikhuis ME, Powell JE, Rees SA, Genus T, Chughtai S, Ramani P, Mann JR, McConville CM. (2001). Br. J. Cancer, 85: 531-537.

Perri P, Pession A, Mazzocco M, Scaruffi P, Strigini P, Iolascon A, Albergoni MP, Basso G, Tonini GP. (1997). Eur. J. Cancer, 33: 1949-1952.

Schwab M, Varmus HE, Bishop JM, Grzeschik KH, Naylor SL, Sakaguchi AY, Brodeur G, Trent J. (1984). Nature, 308: 288-291.

Schwab M, Praml C, Amler LC. (1996). Genes Chromosomes Cancer, 16: 211-229. MEDLINE

Srivatsan ES, Ying KL, Seeger RC. (1993). Genes Chromosomes Cancer, 7: 32-37.

Takita J, Hayashi Y, Kohno T, Shiseki M, Yamaguchi N, Hanada R, Yamamoto K, Yokota J. (1995). Oncogene, 11: 1829-1834. MEDLINE

Vandesompele J, Van Roy N, Van Gele M, Laureys G, Ambros P, Heimann P, Devalck C, Schuuring E, Brock P, Otten J, Gyselinck J, De Paepe A, Speleman F. (1998). Genes Chromosomes Cancer, 23: 141-152.

Vandesompele J, Speleman F, Van Roy N, Laureys G, Brinkschmidt C, Christiansen H, Lampert F, Lastowska M, Bown N, Pearson A, Nicholson JC, Ross F, Combaret V, Delattre O, Feuerstein BG, Plantaz D. (2001). Med. Pediatr. Oncol., 36: 5-10.

Weiss WA, Aldape K, Mohapatra G, Feuerstein BG, Bishop JM. (1997). EMBO J., 16: 2985-2995. MEDLINE

White PS, Maris JM, Beltinger C, Sulman E, Marshall HN, Fujimori M, Kaufman BA, Biegel JA, Allen C, Hillard C, Valentine MB, Look AT, Enomoto H, Sakiyama S, Brodeur GM. (1995). Proc. Natl. Acad. Sci. USA, 92: 5520-5524. MEDLINE

Figure 1 (A) Delimitation of an SRO at 4p16. LOH analysis in 41 NB samples shows eight deleted cases. Sample 1045 carries an interstitial deletion encompassing markers D4S2957 and D4S432, which redefines a SRO of 3.0 cM. Tumor samples 736 and 1425 belong to patients IV-25 and IV-23, respectively, affected by familial NB (Figure 2). Black circles indicate LOH, white circles retention of heterozygosity and hatched circles not informative samples. (B) Electropherogram results at loci D4S412, D4S2957, D4S432 and D4S3023 from NB sample 1045 reveal an interstitial deletion, which defines the SRO. Alleles detected on constitutional (upper panel) vs tumor (lower panel) DNA were sized in base pairs and peak area calculated by GeneScan^Ô software. LOH is exhibited by 61 and 69% decrease in one allele peak of the tumor DNA at D4S2957 and D4S432, respectively. Residual allele peaks are due to the presence of 10% stroma (data supplied by pathologists) in the tumor sample

Figure 2 Pedigree of family IGG-E showing genotyping results and haplotype reconstruction for markers at 4p16. The haplotype for the three nonrecombinant loci D4S412, D4S2957 and D4S432, shared by the four affected children, is marked in a black box. Alleles found to be deleted in tumor samples from patients IV-23 and IV-25 are marked with a grey shadow

Figure 3 Pedigree of family UB-NB1, genotyping results and haplotype reconstruction for markers at 4p16. A somatic deletion involving all analysed markers occurs in the tumor sample from patient III-1 (alleles marked with a grey shadow)