Loss of Heterozygosity in Human Breast Carcinomas in the Ataxia Telangiectasia, Cowden Disease and BRCA1 Gene Regions

Loss of heterozygosity in human breast carcinomas in the ataxia telangiectasia, Cowden disease and BRCA1 gene regions

Fabienne Kerangueven1, FrancËois Eisinger2, Tetsuro Noguchi1, Florence Allione1,3, VeÂ ronique Wargniez1,3, Charis Eng4, Georges Padberg5, Charles Theillet6, Jocelyne Jacquemier7, Michel Longy8, Hagay Sobol1,2 and Daniel Birnbaum1,9

1Laboratoire de Biologie des Tumeurs, Institut Paoli-Calmettes, Marseille, France; 2DeÂpartement d'OncogeÂneÂtique, Institut Paoli- Calmettes, Marseille, France; 3DeÂpartement de MeÂdecine NucleÂaire, Institut Paoli-Calmettes, Marseille, France; 4Dana Farber Cancer Institute, Boston, Massachusetts, USA; 5University Hospital, Nijmegen, The Netherlands; 6IGGM, CNRS URA 9942, Montpellier, France; 7DeÂpartement d'Anatomie-pathologique, Institut Paoli-Calmettes, Marseille, France; 8Fondation BergonieÂ, Bordeaux, France; 9U.119 Inserm, 27 Bd. LeõÈ Roure, 13009 Marseille, France

To appreciate the involvement of known or potential breast tumors and suspected to contain tumor susceptibility genes in sporadic breast tumors, we have suppressor genes are on chromosome arms 1p, 1q, searched for chromosomal deletions by studying loss of 3p, 6q, 7q, 8p, 11p, 11q, 13q, 16q, 17p, 17q, 18q and heterozygosity (LOH) at 43 microsatellite (CA)n 22q (reviewed in Devilee and Cornelisse, 1994 and markers from human chromosomes 10, 11 and 17, in BieÂ che and Lidereau, 1995). Linkage and mutation 115 unselected consecutive samples of breast carcinoma analyses have identi®ed the BRCA1 and BRCA2 genes, with particular emphasis on speci®c regions. No site of located on the long arm of chromosome 17 and 13 consistent LOH was identi®ed on chromosome 10. Five respectively, as major loci for hereditary breast cancer regions of LOH were contained within bands q22-24 of (Stzatton and Wooster, 1996, for review). chromosome 11 for which nearly 50% of the tumors had In addition to BRCA1 and BRCA2, other loci, while LOH at at least one marker. This region is thus a major being involved in breast cancer susceptibility, could site of deletion in breast cancer and several tumor also play a role in the development of sporadic breast suppressor genes seem to be involved. One of them may tumors. Increased susceptibility to breast cancer has be the ataxia telangiectasia (ATM) gene which is located been noted in carriers of constitutional translocations in one of the aected regions. Five regions of LOH, one involving the 11q23 band (Lindblom et al., 1994). The of which is within the BRCA1 gene area, were possible involvement of the recently isolated (Savitsky recognized along chromosome 17. LOH at three of et al., 1995) ataxia-telangiectasia (ATM) gene, located these regions were found in highly proliferative tumors. at 11q22-23, in breast cancer susceptibility has been When combined with a previous study of chromosome 13 suggested (Swift et al., 1987). The importance of with emphasis on BRCA2 and Rb1 genes, this work putative loci from chromosome region 11q22-23 in allowed to distinguish a total of 12 regions of LOH, sporadic breast cancer, suspected after cytogenetic variably aected in breast tumors and correlated with studies (Feri-Passantonopoulou et al., 1991; Hains- prognostic parameters. worth et al.,, 1991), has been recently revealed by LOH analysis (Carter et al., 1994; Hampton et al., 1994a; Keywords: ataxia telangiectasia; BRCA1; breast cancer; Koreth et al., 1995; Windqvist et al., 1995). Breast chromosome 10; chromosome 11; chromosome 17 tumors are also frequently observed in patients with Cowden disease (Starink et al., 1986). The gene for this syndrome has been mapped to chromosome region 10q22-23 (Nelen et al., 1996). Introduction We have looked for LOH at chromosome 10, 11 and 17, with special emphasis on regions hosting known or Mutations in tumor suppressor genes seem important putative breast tumor susceptibility genes, i.e. 10q22- in the development of familial and so-called sporadic 23, 11q22-24 and 17q-12-21. We have cumulated this breast carcinomas. Several chromosome sites, suspected analysis with a previous study of chromosome 13 or known to contain such genes, are involved, often in where the BRCA2 gene is located, in order to evalaute the same tumor. Sites of potential alterations have been the possible role in sporadic tumors of genomic regions identi®ed by studying losses of chromosomal segments which contain loci with either proven or potential or by genetic linkage analysis in cancer prone families. involvement in mammary carcinogenesis and possible Major regions of loss of heterozygosity (LOH) interactions between them. In addition, we have correspond to the location of known cancer genes, attempted a combined quantitative analysis of 12 such as TP53 on the short arm of chromosome 17 and de®ned regions of LOH with respect to histo-clinical E-cadherin on chromosome arm 16q, or of putative parameters. tumor suppressor genes that have not been character- ized yet. Segments of genome frequently deleted in Results Correspondence: D Birnbaum or H Sobol Received 3 June 1996; revised 12 September 1996; accepted 12 For each chromosome, the largest number of September 1996 microsatellites was chosen within or close to the Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 340 regions of interest, i.e. 10q22-23, 11q22-24 and 17q12- The previous results provide evidence against the 21 to estimate the incidence of LOH. The other involvement of Cowden disease region in sporadic markers were used for comparison, to determine if breast cancer. To further investigate the possible this incidence was prominent for the given chromo- participation of chromosome region 10q22-23 in breast some or if other regions exhibited LOH at a carcinogenesis, this time in genetic predisposition to comparable or dierent frequency. The number of breast cancer, and in other conditions than the Cowden tumor samples tested varied from 67 to 115 depending disease itself, we set up a linkage analysis of breast on the marker. cancer families. Seven families with breast cancer only syndrome, previously selected on the basis of absence of linkage to BRCA1 and BRCA2 linkage (Kerangueven et LOH at markers from chromosome 10 al., 1995b) were tested with markers D10S201, D10S532 Fourteen microsatellite markers were used to search for and D10S541 as described (Kerangueven et al., 1995b). LOH at chromosome arms 10p and 10q. Six of them, Total cumulative lod scores were negative at several from D10S201 to D10S185, are closely linked in a 15 cM values of y (not shown). Lod scores were weakly interval (Gyapay et al., 1994; Dib et al., 1996; Moshonas positive (0.20) for D10S201 at y40.2 but this was et al., 1996) in the recently de®ned Cowden disease dependent on one family only. Even if a lod score of 72 region, at 10q22-23. In linkage analyses (Nelen et al., (hypothesis of independence) was not reached, the 1996), the maximum lod score was obtained with marker results are not in favor of the hypothesis of the D10S573 and the most probable location of the diesease Cowden disease gene being a BRCA gene. gene was de®ned between D10S215 and D10S564. Markers D10S532 and D10S541, which in our hands, gave better informativity, were used instead of D10S573 and D10S215, respectively (Gyapay et al., 1994). Figure 1 presents a summary of the LOH results on this chromosome. Taking into account only informative cases, the percentage of LOH at each marker ranged from 13 ± 27% (Figure 1). Overall, this incidence was close to the general background we observed in previous studies (Kerangueven et al., 1995a,b,c). LOH at markers from the Cowden region varied from 16 ± 23%, not above the average incidence noted all along the chromosome. Half of this was due to monosomies but some interstitial deletions were observed. However, they were distributed over the two arms and no precise region of consistent loss could be identi®ed.

Figure 2 Illustration of LOH at chromosome 11 markers. Computer representation of the polyacrylamide gel analysis by Figure 1 Schematic representation of the distribution of LOH at automatic sequencing apparatus of polymerase chain reaction chromosome 10 in representative breast carcinomas. At left is a products from selected tumors (lower rows) and corresponding diagram of chromosome 10, with the location of the microsatellite blood samples (upper rows). The microsatellite alleles are (CA)n markers and genes potentially involved in tumorigenesis, as represented by two (heterozygote) peaks except in the case of known to date. Numbers at far right represent the ratio of tumors homozygosity (non informative cases). LOH at one allele (arrows) with LOH vs the informative cases (the total number of analysed corresponds to a signi®cant decrease in one of the two peaks. The tumors varied from 48 to 115, depending on the marker) higher molecular weight allele is on the left in each tracing Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 341 Five regions of consistent loss, designated a to e, all LOH at markers from chromosome 11 located in bands 11q22-24 (Van Heyningen and Little, A total of 19 markers from chromosome 11 were studied. 1995), were identi®ed. They were centered on markers Figure 2 shows representative LOH and Figure 3 presents D11S1818 in the ATM gene area, D11S940 just a summary of the results obtained in the 11q22-24 region. centromeric, and D11S1356, D11S925 and D11S934, Taking into account only informative cases, the telomeric to ATM. Regions were de®ned on the basis of percentages of LOH at each marker ranged from 24 ± involvement of several tumors showing LOH and 26% at D11S901 and D11S917, to 41 ± 51% at D11S1356, retention of heterozygosity at adjacent markers. For D11S528 and D11S910 (Figure 3). In contrast, LOH at example, in Figure 3, tumors 1288, 5809 and 5411 are markers from chromosome arm 11p did not exceed the used to de®ne region a, tumors 5411, 5510 and 5362 to 20% range (not shown) which, as previously mentioned de®ne region e. The distance between these regions is above for chromosome 10 markers, correspond in our still important, and this certainly calls for a re®ned hands to a background of sporadic losses which does not study. Importantly, region b contains the ATM locus, allow the identi®cation of a particular region (Keran- de®ned by markers D11S1343 to D11S2106 (Savitsky et gueven et al., 1995a,b,c). The incidence of 50% is in close al., 1995). agreement with ®ndings by other groups (Carter et al., 1994; Hampton et al., 1994b; Koreth et al., 1995). LOH at markers from chromosome 17 The tumor samples could be separated into two distinct groups, according to their pattern of LOH. In Ten markers from chromosome 17 were studied to search a ®rst group of tumors, LOH was observed at all for LOH at both arms, with emphasis on the BRCA1 gene chromosome arm 11q markers, suggesting that region, at 17q21. Representative results are shown in inactivation of one allele of a putative tumor Figure 4 and summarized in Figure 5. Five regions of suppressor gene could have occurred in these tumors LOH, two on 17p, three on 17q, were individualized. One by large deletion or, in a few cases distinguished by region of 17q, deleted in almost half of the tumor samples, LOH at chromosome arm 11p markers as well, by was located within or near the BRCA1 gene area. processes resulting in monosomy. A second group (as Incidence of LOH in region e was low and identi®cation detailed in Figure 3) comprised tumors with LOH of this region could have been neglected, since as it was detected with only a restricted number of 11q markers, mentioned for chromosome 10 markers, it was close to the i.e. with interstitial deletions. background value. It was only legitimized by the existence

Figure 3 Schematic representation of the distribution of LOH at chromosome arm 11q in human breast tumors. At far left is a diagram of a portion of chromosome arm 11q, with the location of the microsatellite (CA)n markers as it is known to date (Gyapay et al., 1994; Van Heyningen and Little, 1995). The middle panel is a schematization of allelic losses in representative tumors of the panel. Black lanes indicate LOH; absence of black lane means that the region is unaected. Numbers at far right represent the ratio of tumors with LOH vs the informative cases (the total number of analysed tumors was 115) for the selected markers (in the ATM region, numbers correspond to D11S1818 and D11S2106). Vertical bars on the right indicate common regions of minimal loss. Tumors with an ampli®cation of genes from the 11q13 band are designated with a black square on top Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 342 of several tumors (5411, 1062, 5743 and 5510 in Figures 4 Cumulative regional allelotyping and 5) with a pattern of LOH in the distal portion of chromosome arm 17q. Even if some tumors presented interstitial deletions with markers of chromosome 10, the Cowden disease area, or any other portion of chromosome 10, was not considered as a major region of consistent loss. Our LOH analysis of three chromosomes with 43 markers thus allowed to de®ne 10 regions aected in breast tumors on chromosomes 11 and 17. A previous study of chromosome 13 with eight markers, allowed us to identify two additional regions of LOH, possibly associated with BRCA2 and Rb1 genes, respectively (Kerangueven et al., 1995a). We looked for correlations between clinico-pathological parameters and LOH at a given region, and performed a combined analysis on all 12 regions from chromosomes 10, 11, 13 and 17. This analysis, designated `cumulative regional allelotyping' (CRA), puts emphasis on quantitative data (Kerangueven et al., 1996). The scoring of regions of LOH was done on 31 tumors tested with all markers (including chromosome 13 markers) as follows: scoring was done on two contiguous markers per de®ned region; a region was positive if LOH occurred at at least one marker, negative if the two alleles were present for each marker, or if one of the marker was non informative and the other has retained heterozygosity; the region was non informative in a given tumor if the two markers were non informative (accordingly, the proportion of tumors aected for each region varies slightly from the proportion aected for each marker). CRA on a total of 12 regions of LOH distributed over the four chromosomes provided information (shown in Figures 6 and 7) on both the distribution of aected regions per tumor, and the proportion of aected tumors per region of LOH. The proportion of aected tumors (31=100% of LOH) varied according to the region considered (Figures 6 and 7). No region presented less than 30% or more than 60% of LOH (Figure 6a). Two regions were deleted in the 30 ± 40% range, as shown in Figures 6a and 7. The other 10 regions were aected Figure 4 Illustration of LOH at chromosome 17 markers. See in more than 40% of the cases. LOH at a given region also legend of Figure 2 of a chromosome arm was independent of LOH at

Figure 5 Schematic representation of the distribution of LOH at chromosome 17 in representative breast carcinomas. At far left is a diagram of chromosome 17, with the location of the microsatellite (CA)n markers and of tumor suppressor genes, as known to date. The middle panel is a schematization of allelic losses in tumors of the panel. Results are represented as follows: open squares, retention of heterozygosity; black squares, LOH; gray squares, non informative; nd, no data. Numbers at far right represent the ratio of tumors with LOH vs the informative cases. Vertical bars on the right indicate ®ve regions of minimal loss Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 343 A B

Figure 6 Cumulative analysis of LOH in a panel of 31 breast carcinomas with LOH at 51 markers from chromosomes 10, 11, 13 and 17: respective distributions of aected regions and tumors. A. number of aected regions of consistent loss as a function of tumors (31=100%). B. number of aected tumors as a function of regions (15=100%)

Figure 7 Cumulative regional allelotyping of chromosomes 10, 11, 13 and 17. Relative frequencies (in %) of LOH at 15 regions (this presentation has correspondence in 6a). As indicated below, regions can be classi®ed in three classes according to frequency of involvement. No region is involved in less than 30 and more than 60% of tumors other regions of the same arm and independent of The proportion of aected regions (12=100% of LOH on a dierent arm or chromosome except in a LOH) varied from tumor to tumor. Every tumor but few cases. LOH at regions 17b and 17c (P=0.013), at three presented LOH at at least one region. Twelve 13a and 13b (P=0.0001), and two regions from tumors presented LOH at one, two or three regions dierent chromosomes, such as 11a and 17d only. Fourteen tumors showed LOH at more than 60% (P=0.005) were signi®cantly associated. of the regions (Figure 6b). This distribution showed When speci®c regions were considered, the propor- clearly the existence of two groups of tumors, one with tions of LOH were as follows: the ATM gene region a low level of LOH (0 to 30%, in Figure 6b) and one was involved in 48% of tumors, the BRCA1 gene with a high level of LOH (from 40 to more than 90%). region in 48% and the BRCA2 gene region in 52%. The tumors with a low level of LOH were character- These ®gures for the three regions are strikingly close ized by a lower mitotic index (P=0.017). These two and may represent a maximum of incidence for a given groups of tumors may be caused by dierent molecular gene. alterations and may have distinct clinical evolutions. If Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 344 this turns out to be the case, this type of cumulative al., 1993; Davis et al., 1996; Gabra et al., 1996) and analysis would be particularly helpful in assessing the stomach (Baa et al., 1996) carcinomas, and in initial prognosis. melanomas (Tomlinson et al., 1993; Herbst et al., 1995), and thus may host at least one so-called multiple tumor suppressor gene, the activity of which Correlation with clinico-histological parameters is not restricted to breast epithelial cells. Further A statistical analysis of the correlations between LOH studies will aim at re®ning the localization of this and clinico-pathological parameters was performed. important locus. Unfortunately, because of the Among the 31 tumors, the ones with LOH at 13a variability in the choice of the markers studied, it is (BRCA2 region) were associated with the presence of dicult to determine if the 11q regions of most metastatic lymph nodes (P=0.007) and with an age frequent LOH are the same in the dierent types of below 50 (P=0.015). Tumors with LOH at regions tumors. A survey of the various reports (also 17a, 17b (P53 region) and 17c were highly prolif- summarized in Davis et al., 1996 and in Hui et al., erative. LOH at regions 17a ± c was indeed correlated 1996) indicates that regions d and e reported here with a high grade, due to a high mitotic index could be identical to two regions of LOH observed in (P=0.02 to 0.005). However, LOH at region 17d ovarian cancers (Davis et al., 1996; Gabra et al., 1996) (BRCA1) was not signi®cantly correlated with high while regions b and d, which have already been grade, although we have observed a similar correlation evidenced in breast cancer by other authors (Negrini between high mitotic index and BRCA1 gene germ et al., 1995) are apparently also aected in bladder line mutations (Eisinger et al., 1996). It is possible (Shaw and Knowles, 1995), cervical (Hampton et al., that dierent mechanisms of BRCA1 inactivation lead 1994a), lung (Rasio et al., 1995), nasopharyngeal (Hui to tumors with distinct clinico-pathological para- et al., 1996) and ovarian (Gabra et al., 1996) cancers. meters. Alternatively, the 11q22-23 region may contain several With respect to 11q markers, no strong correlation tumor suppressor genes variably involved in dierent with any of the parameters was evidenced. Winqvist et types of cancer. There are several potential candidates. al. (1995) found that LOH at 11q23 is predictive of Since the ATM gene has been suspected to play a role aggressive postmetastatic course. Clinical follow-up of in cancer susceptibility (Swift et al., 1987), its tumors from our panels will be done to assess this participation may be suspected. The involvement of point. Ampli®cation of the 11q13-14 region has been the MLL/ALL1 gene (Rowley, 1993) located close to detected in 15 ± 25% of breast carcinomas (Gaudray et D11S1356 (Van Heyningen and Little, 1995) and al., 1992). Such a process may alter the structure of which has been found to be deleted in a gastric chromosome arm 11q and lead to the loss of the carcinoma cell line (Baa et al., 1995), or of remaining portion of the chromosome, creating LOH unidenti®ed tumor suppressor genes, potentially also which could be falsely interpreted as the inactivation of involved in familial cases (Lindblom et al., 1994; a tumor suppressor gene. Upon examination of cases Sobol et al., 1994), should also be investigated. One of with both types of alteration (Figure 3), LOH at 11q23 the putative tumor suppressor genes may be a did not seem to be associated with ampli®cation (as metastasis suppressor gene (Phillips et al., 1996). The tested by Southern blot hybridization with several zebra pattern observed in this area is intriguing and probes from 11q13 band -not shown) since loci needs further investigation with additional markers to telomeric of 11q13 (D11S919, D11S917) are retained obtain a better de®nition of the aected regions. In in most cases. Furthermore, the possibility that the addition to being due to several potential tumor 11q13 band contains a tumor suppressor gene has been suppressor genes, special instability in the region may evoked (Zhuang et al., 1995) but was not apparent in be responsible for the complexity of the LOH pattern. our study. The same was true for chromosome arm 17q Five regions of loss, two of them on 17p12-13, have and ERBB2 ampli®cation. been identi®ed by several authors in breast carcinomas and other cancers (Cropp et al., 1993; Kirchweger et al., 1994; Nagai et al., 1995; Schultz et al., 1996) and Discussion coarsely correspond to the ones de®ned here. Our second goal was to determine whether the Our ®rst goal was to determine regions of LOH on region containing known or putative breast tumor speci®c chromosomes, i.e. chromosomes 10, 11 and 17, susceptibility genes were aected by LOH in sporadic in breast tumors. In prostate cancer, LOH at 10q22-24, tumors. This question has already been evoked and in a region slightly telomeric of the Cowden locus, has partly answered with respect to BRCA1 and BRCA2 been evidenced (Gray et al., 1995; Ittman, 1996) and gene regions (Cleton-Jansen et al., 1995; Cropp et al., the role of PAC1, on 10pter-q11, and Mix1 at 10q25, 1993; Kerangueven et al., 1995a; Kirchweger et al., two potential tumor suppressor genes, has been 1994; Nagai et al., 1995). For the Cowden disease investigated (Gray et al., 1995; Sanchez et al., 1996). region, the low frequency of LOH and the absence of However, in breast cancer, no evidence for the linkage to familial breast tumors both suggest that it is participation of genes from chromosome 10 could be not prominently involved in cancer. However, it obtained from the LOH data. remains possible that the gene is involved in sporadic LOH at markers of chromosome region 11q22-23 breast cancer by other ways than deletion. Our study has been observed in several types of cancer, including of Cowden disease region in seven hereditary breast bladder (Shaw and Knowles, 1995), cervical (Hampton cancer families did not provide evidence for linkage, et al., 1994b; Bethwaite et al., 1995) colorectal but it remains very preliminary and needs to be (Keldysh et al., 1993), lung (Iizuka et al., 1995), strengthened by an extensive study with more families nasopharyngeal (Hui et al., 1996), ovarian (Foulkes et and more markers. Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 345 We also provided evidence for the possible D10S1644, D10S541, D10S564, D10S185, D10S534, involvement of the ataxia telangiectasia region in D10S186 and D10S212; chromosome 11: D11S922, breast tumorigenesis. However, other regions with D11S904, D11S903, D11S901, D11S919, D11S917, similar or greater frequency could be identi®ed D11S940, D11S1343, D11S2179, D11S1818, D11S2106, nearby, and it is by no means certain that the gene D11S1347, D11S1356, D11S528, D11S925, D11S1328, D11S934, D11S910 and D11S969; chromosome 17: involved in ataxia telangiectasia is the same as the D17S1174, D17S786, IG P53, D17S933, D17S855, one(s) involved in breast carcinogenesis. Finding so D17S1323, D17S1327, D17S806, D17S785 and D17S928. many sites of deletion in this region remains intriguing. Markers D10S532, D10S1644, D10S541 and D10S564 are It is possible that not all sites correspond to a bona ®de located within or close to the Cowden locus region (Nelen tumor suppressor gene and that some deletions are due et al., 1996). Markers D11S1343, D11S2179, D11S1818, to an overall greater instability of the region. D11S2106 and D11S1347 are located within or close to the The third goal was to stress on the fact that a large ATM gene region (Savitsky et al., 1995). Primers were number of genomic regions may host tumor suppressor designed according to Gyapay et al. (1994) and Dib et al. genes involved in breast cancer and to suggest a way to (1996), except for intragenic P53 IGP53 5'-AGGATAC- analyse cumulative quantitative data. In contrast, our TATTCAGCCCGAGGTG-3',5'-ACTGCCACTCCTTG- CCCCATTC-3'). Microsatellite analysis was performed goal was not directed towards the re®nement of using ampli®cation by polymerase chain reaction using a identi®ed regions, something that may be better Perkin-Elmer Cetus thermal cycler, model 9600, and the performed on concentrating eorts on selected sites. resulting products were visualized and analysed with automated ¯uorescent ABI 373 or 377 sequencing apparatus, as described (Kerangueven et al., 1995b,c). Materials and methods Statistical analysis Tumor samples Signi®cances of dierences for contingency tables (2*2) One hundred and ®fteen unselected primary breast were assessed by Chi-square (with Yate's correction) or carcinomas, and blood samples from the same patients, Fisher's exact test when an expected cell value was less were collected at Institut Paoli-Calmettes in Marseille (48 than n=5. Statistical analyses were performed using the tumors) and Institut Paul Lamarque in Montpellier (67 EPI-INFO package, version 5.01. For contingency tables tumors). The following clinical and biological parameters (R*C) with sparse data, the exact p value for Kruskal- were studied in this panel: tumor size, histoprognostic Wallis test (*,*) (Hollander and Wolfe, 1973; Agresti et al., grade, lymph node (N) status (47% and 41% of N positive 1990) was computed using the StatXact package (Cytel tumors, for the two panels, respectively), hormonal Software corp.). receptors (63% and 68% of estrogen receptor positive, 53% and 68% progesterone receptor positive), DNA index (70% with aneuploidy), S-phase, ERBB2, and p53 status. Genomic DNAs were extracted as previously described (Theillet et al., 1989). Acknowledgements We are grateful to Drs D Maraninchi and C Mawas, for discussions and their encouragement and to Drs C BonaõÈ ti- Genetic markers analysis PellieÂ and L Essioux for help and advice. This work was DNAs were examined for LOH with the following supported by Institut Paoli-Calmettes, Inserm and grants microsatellites (CA)n markers oriented according to recent from: Association pour la Recherche sur le Cancer, consensus maps (Gyapay et al., 1994; Matise et al., 1994; ComiteÂ s des Bouches-du-RhoÃ ne, des Alpes de Haute- Van Heyningen and Little, 1995; Dib et al., 1996; Provence et du Var de la Ligue Nationale Contre le Moshonas et al., 1996): Chromsome 10: D10S558, Cancer, FEGEFLUC, and FeÂ deÂ ration Nationale des D10S191, D10S199, D10S561, D10S201, D10S532, centres de Lutte Contre le Cancer.

References

AgrestiA,MehtaCandPatelN.(1990).Exactinferencefor Cleton-Jansen AM, Collins N, Lakhani SR, Weissenbach J, contingency tables with ordered categories. J. Am. Statist. Devilee P, Cornelisse C and Stratton M. (1995). Loss of Assoc., 85, 543 ± 558. heterozygosity in sporadic breast tumours at the BRCA2 Baa R, Negrini M, Mandes B, Rugge M, Ranzani G, locus on chromosome 13q12-q13. Br. J. Cancer, 72, 1241 ± Hirohashi S and Croce C. (1996). Loss of heterzygosity for 1244. chromosome 11 in adenocarcinoma of the stomach. Cropp C, Champeme MH, Lidereau R and Callahan R. Cancer Res., 56, 268 ± 272. (1993). Identi®cation of three regions on chromosome 17q Baa R, Negrini M, Schichman S, Huebner K and Croce C. in primary human breast carcinomas which are frequently (1995). Involvement of the ALL-1 gene in a solid tumor. deleted. Cancer Res., 53, 5617 ± 5619. Proc. Natl. Acad. Sci. USA, 92, 4922 ± 4926. Davis M, Hitchcock A, Foulkes W and Campbell I. (1996). Bethwaite P, Koreth J, Herrington C and McGee JO'D. Re®nement of two chromosome 11q regions of loss of (1995). Loss of heterozygosity occurs at the D11S29 locus heterozygosityinovariancancer. CancerRes.,56,741 ± 744. on chromosome 11q23 in invasive cervical carcinoma. Br. Devilee P and Cornelisse C. (1994). Somatic genetic changes J. Cancer, 71, 814 ± 818. in human breast cancer. Biochem. Biophys. Acta, 1198, BieÁ che I and Lidereau R. (1995). Genetic alterations in breast 113 ± 130. cancer. Genes Chromosom. Cancer, 14, 227 ± 251. Dib C, FaureÂ S, Fizames C, Samson D, Drouot N, Vignal A, Carter S, Negrini M, Baa R, Gillum D, Rosenberg A, Millasseau P, Marc S, Hazan J, Seboun E, Lathrop M, Schwartz G and Croce C. (1994). Loss of heterozygosity Gyapay G, Morissette J, Weissenbach J. (1996). A at 11q22-q23 in breast cancer. Cancer Res., 54, 6270 ± comprehensive genetic map of the human genome based 6274. on 5,264 microsatellites. Nature, 380, 152 ± 154. Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 346 Eisinger F, Stoppa-Lyonnet D, Longy M, Kerangueven F, Kerangueven F, Eisinger F, Allione F, Wargniez V, Parc P, Noguchi T, Bailly C, Vincent-Salomon A, Jacqueimer J, Jacquemier J, Noguchi T, Sobol H and Birnbaum D. Birnbaum D and Sobol H. (1996). Germ line mutation at (1996). Cumulative regional allelotyping of human breast BRCA1 aects the histoprognostic grade in hereditary carcinomas: application to chromosomes 1 to 5. Int. J. breast cancer. Cancer Res., 56, 471 ± 474. Oncol., 8, 1155 ± 1163. Ferti-Passantonopoulou A, Panani A and Raptis S. (1991). Kerangueven F, Essioux L, Dib A, Noguchi T, Allione F, Preferential involvement of 11q23-24 and 11p15 in breast Geneix J, Longy M, Lidereau R, Eisinger F, Pebusque cancer. Cancer Genet. Cytogenet., 51, 183 ± 188. M-J, Jacquemeir J, Bonaiti-Pellie C, Sobol H and Foulkes W, Campbell I, Stamp G and Trowsdale J. (1993). Birnbaum D. (1995b). Loss of heterozygosity and linkage Loss of heterozygosity and ampli®cation on chromo- analysis in breast carcinoma: indication for a putative some 11q in human ovarian cancer. Br. J. Cancer, 67, third susceptibility gene on the short arm of chromosome 268 ± 273. 8. Oncogene, 10, 1023 ± 1026. Gabra H, Watson V, Taylor K, Mackay J, Leonard R, Steel Kerangueven F, Noguchi T, Adelaide J, Jacquemier J, Sobol M, Porteous D and Smyth J. (1996). De®nition and H and Birnbaum D. (1995c). Allelic loss at markers of re®nement of a region of loss of heterozygosity at 11q23.3- chromosomal band 7q31 is not a frequent event in human q24.3 in epithelial ovarian cancer associated with poor breast cancer. Oncology Rep., 2, 89 ± 90. prognosis. Cancer Res., 56, 950 ± 954. Kirchweger R, Zeillinger R, Schneeberger C, Speiser P, Gaudray P, Szepetowski P, Escot C, Birnbaum D and Louason G and Theillet C. (1994). Paterns of allele losses Theillet C. (1992). DNA ampli®cation at 11q13 in human suggest the existence of ®ve distinct regions of LOH on cancer: from complexity to perplexity. Mutat. Res., 276, chromosome 17 in breast cancer. Int. J. Cancer, 56, 193 ± 317 ± 328. 199. Gray I, Phillips S, Lee S, Neoptolemos J, Weissenbach J Koreth J, Bethwaite P and McGee JO'D. (1995). Mutation at and Spurr NK. (1995). Loss of the chromosomal region chromosome 11q23 in human non-familial breast cancer: a 10q23-25 in prostate cancer. Cancer Res., 55, 4800 ± microdissection microsatellite analysis. J. Pathol., 176, 4803. 11 ± 18. Gyapay G, Morissette J, Vignal A, Dib C, Fizames C, Lindblom A, Sandelin K, Iselius L, Dumanski J, White I, Millasseau P, Marc S, Bernadi G, Lathrop M and Nordenskjold M and Larsson C. (1994). Predisposition Weissenbach J. (1994). The 1992 ± 94 GeÂ neÂ thon human for breast cancer in carriers of constitutional translocation genetic linkage map. Nature Genet., 7, 246 ± 339. 11q;22q. Am. J. Human Genet., 54, 871 ± 876. Hainsworth P, Raphael K, Stillwell R, Bennett R and Moschonas N, Spurr N and Mao J. (1996). Report of Garson M. (1991). Cytogenetic features of twenty-six the ®rst international workshop on human chromo- primary breast cancers. Cancer Genet. Cytogenet., 52, some 10 mapping 1995. Cytogenet. Cell Genet., 72, 205 ± 218. 99 ± 112. Hampton G, Mannermaa A, Winquist R, Alavaikko M, NagaiM,MedeirosA,BrentaniM,BrentaniR,Marques Blanco G, Taskinen P, Kiviniemi H, Newsham I, L, Mazoyer S and Mulligan L. (1995). Five distinct Cavanee W and Evans G. (1994a). Loss of heterozygos- deleted regions on chromosome 17 de®ning dierent ity in sporadic human breast carcinoma: a common subsets of human primary breast tumors. Oncology, 52, region between 11q22 and 11q23.3. Cancer Res., 54, 448 ± 453. 4586 ± 4589. NegriniM,RasioD,HamptonG,SabbioniS,RattanS, Hampton G, Penny L, Baergen R, Larson A, Brewer C, Liao Carter S, Rosenberg A, Schwartz G, Shiloh Y Cavenee W S, Busby-Earle R, Williams A, Steel C, Bird C, Stanbridge and Croce C. (1995). De®nition and re®nement of E and Evans G. (1994b). Loss of heterozygosity in cervical chromosome 11 regions of loss of heterozygosity in carcinoma: subchromosomal localization of a putative breast cancer: identi®cation of a new region at 11q23.3. tumor-suppressor gene to chromosome 11q22-q24. Proc. Cancer Res., 55, 3003 ± 3007. Natl. Acad. Sci. USA, 91, 6953 ± 6957. Nelen M, Padberg G, Peeters E, Lin A, Van der Helm B, Herbst R, Larson A, Weiss J, Cavenee W, Hampton G and Frants R, Coulon V, Goldstein A, Van Reen M, Easton Arden K. (1995). A de®ned region of loss of hetero- D, Eeles R, Hodgson S, Mulvihill J, Murday V, Tucker zygosity at 11q23 in cutaneous malignant melanoma. M, Mariman E, Starink T, Ponder B, Ropers H, Kremer Cancer Res., 55, 2494 ± 2496. H, Longy M and Eng C. (1996). Localisation of the gene Hollander M and Wolfe D. (1973). The one way out. In: Non for Cowden disease to chromosome log 22 ± 23. Nature parametric statistical methods. M Hollander and D Wolfe Genet., 13, 114 ± 116. (eds). John Wiley and sons, NY pp 115 ± 132. Phillips K, Welch D, Miele M, Lee JH, Wei L and Weissman Hui A, Lo KW, Leung SF, Choi PHK, Fong Y, Lee JCK and B. (1996). Suppression of MDA-MB-435 breast carcinoma Huang DP. (1996). Loss of heterozygosity on the long arm cell metastasis following the introduction of human of chromosome 11 in nasopharyngeal carcinoma. Onco- chromosome 11. Cancer Res., 56, 1222 ± 1227. gene, 56, 3225 ± 3229. Rasio D, Negrini M, Manenti G, Dragani T and Croce C. Iizuka M, Sugiyama Y, Shiraishi M, Jones C and Sekiya T. (1995). Loss of heterozygosity at chromosome 11q in lung (1995). Allelic losses in human chromosome 11 in lung adenocarcinoma: identi®cation of three independent cancers. Genes Chrom. Cancer, 13, 40 ± 46. regions. Cancer Res., 55, 3988 ± 3991. Ittman M. (1996). Allelic loss on chromosome 10 in prostate Rowley J. (1993). Rearrangements involving chromosome adenocarcinoma. Cancer Res., 56, 2143 ± 2147. band 11q23 in acute leukaemia. Semin. Cancer Biol., 4, Keldysh P, Dragani T, Fleischman E, Konstantinova L, 377 ± 385. Perevoschikov A, Pierotti M, Della Porta G and Kopnin Sanchez Y, Lovell M, Marin M, Wong P, Wolf-Ledbetter E, B. (1993). 11q deletions in human colorectal carcinomas: McDonnell T and McNeill Killary A. (1996). Tumor cytogenetics and restriction fragment length polymorph- suppression and apoptosis of human prostate carcinomas ism analysis. Genes Chrom. Cancer, 6, 45 ± 50. mediated by a genetic locus within human chromosome Kerangueven F, Allione F, Noguchi T, Adelaide J, Sobol H, 10pter-q11. Cancer Res., 93, 2551 ± 2556. Jacquemier J and Birnbaum D. (1995a). Patterns of loss of heterozygosity at loci from chromosome arm 13q suggest a possible involvement of BRCA2 in sporadic breast tumors. Genes Chrom. Cancer, 13, 271 ± 274. Sporadic breast cancer and susceptibility gene regions F Kerangueven et al 347 Savitsky K, Bar-Shira A, Gilad S, Rotman G, Ziv Y, Theillet C, Le Roy X, deLapeyrieÁ re O, Grosgeorges J, Vanagaite L, Tagle D, Smith S, Uziel T, Sfez S, Ashkenazi Adnane J, Raynaud S, Simony-Lafontaine J, Goldfarb M, M, Pecker I, Frydman M, Harnik R, Patanjali S, Simmons Escot C, Birnbaum D and Gaudray P. (1989). Amplifica- A, Clines G, Sartiel A, Gatti R, Chessa L, Sanal O, Lavin tion of FGF-related genes in human tumors: possible M, Jaspers N, Taylor M, Arlett C, Miki T, Weissman S, involvement of HST in breast carcinomas. Oncogene, 4, Lovett M, Collins F and Shiloh Y. (1995). A single ataxia 915 ± 922. telangectasia gene with a product similar to PI-3 kinase. Tomlinson I, Gammack A, Stickland J, Mann G, Mackie R, Science, 268, 1749 ± 1753. Keord R and McGee JO'D. (1993). Loss of hetero- Schultz D, Vanderveer L, Berman D, Hamilton T, Wong A zygosity in malignant melanoma at loci on chromosomes and Godwin A. (1996). Identi®cation of two candidate 11 and 17 implicated in the pathogenesis of other cancers. tumor suppressor genes on chromsome 17q13.3. Cancer Genes Chrom. Cancer, 7, 169 ± 172. Res., 56, 1997 ± 2002. Van Heyningen V and Little PFR. (1995). Report of the Shaw M and Knowles M. (1995). Deletion mapping of fourth international workshop on human chromsome 11 chromosome 11 in carcinoma of the bladder. Genes mapping 1994. Cytogenet. Cell Genet., 69, 128 ± 158. Chrom. Cancer, 13, 1±8. Winqvist R, Hampton G, Mannermaa A, Blanco G, Sobol H, Birnbaum D and Eisinger F. (1994). Evidence for a Alavaikko M, Kiviniemi H, Taskinen P, Evans G, Wright third breast-cancer susceptibility gene. Lancet, 344, 1151 ± F, Newsham I and Cavenee W. (1995). Loss of 1152. heterozygosity for chromosome 11 in primary human Starink TM, Van Der Veen JP, Arwert F, De Waal LP, De breast tumors is associated with poor survival after Lange GG, Gille JJ and Eriksson AW. (1986). The metastasis. Cancer Res., 55, 2660 ± 2664. Cowden syndrome: a clinical and genetic study in 21 Zhuang Z, Merino M, Chuaqui R, Liotta L and Emmert- patients. Clin. Genet., 29, 222 ± 233. Buck M. (1995). Identical allelic loss on chromosome Stratton M and Wooster R. (1996). Hereditary predisposi- 11q13 in microdissected in situ and invasive human breast tion to breast cancer. Current Op. Genet. Develop., 6, 93 ± cancer. Cancer Res., 55, 467 ± 471. 97. Swift M, Reitnauer P, Morrell D and Chase C. (1987). Breast and other cancers in families with ataxia-telangiectasia. New Engl. J. Med., 316, 1289 ± 1294.