Rearrangement of chromosome 1p in breast cancer correlates with poor prognostic features.

In a cytogenetic study of breast cancer biopsies, clonal abnormalities of chromosome 1p were identified in 56% (14) of 25 informative patients. Translocations predominated, involving 1p22 (n = 1), 1p35 (n = 1) or 1p36 (n = 10) breakpoints. Chromosome 1p abnormalities were associated with estrogen receptor (ER) negativity (P = 0.03, 2-tailed Fisher Exact Probability test), high histological grade (P = 0.02, 2-tailed Mann-Whitney U-test) and an unfavourable Melbourne Prognostic Score (NEPA P = 0.02, SEPA P = 0.04, 2-tailed Mann-Whitney U-tests). These findings are consistent with the possibility that a gene located on chromosome 1p is implicated in tumour progression.

Whilst cytogenetic studies in the haematological malignancies have proved to be invaluable in both research and patient management, the same cannot be said for the common solid tumours. However, the finding that a locus on chromosome 5q appears to be involved in colon cancer, at least in patients with familial polyposis coli (Bodmer et al., 1987;Solomon et al., 1987) is important, since the initial lead for this investigation was the cytogenetic observation of a chromosome 5q deletion in a single patient with Gardner's syndrome (Herrera, 1986).
One of the problems with the cytogenetic study of breast cancer is the morass of complex chromosomal changes which have been repeatedly described (for review see Hainsworth & Garson, 1990) which is in sharp contrast to the single chromosome events often seen in the leukaemias. A possible approach towards defining those events which are important in tumour progression is to look for chromosomal changes which correlate with a poor prognosis.
In the course of studying breast cancer karyotypes (Hainsworth et al., 1991) two chromosomes appeared to be of importance. The 'earliest' change observed, based on its occurrence in 'operable' tumours with diploid-range karyotypes, was translocation or deletion of the long arm of chromosome #16 involving a 16q22 breakpoint. However, the most frequently observed rearrangements involved the short arm of chromosome #1, which form the basis of this report.

Materials and methods
Surgical biopsy specimens (n = 144) were received from 143 patients with primary breast cancer, one of whom had bilateral tumours, treated between April 1987 and March 1989. Of the 144 specimens, banded analyses were possible in 31 (22%). In five cases, both normal and abnormal metaphases were observed but only the normal metaphases could be karyotyped. Thus, meaningful karyotypes were obtained in 26 patients. In the remaining 113 cases, insufficient metaphases were obtained to enable analysis.
Cytogenetic analysis Cytogenetic data were obtained using a direct technique (n = 24), synchronised short-term culture (n = 1) or both techniques (n = 1). Full details of the methodology have been published elsewhere (Hainsworth et al., 1991). Briefly, fresh macroscopic tumour was transported to the laboratory in RPMI 1640 medium (Commonwealth Serum Laboratories, Melbourne) containing penicillin and streptomycin and mechanically disaggregated using scalpels.
In the direct technique (Mark, 1975) 1 ml of single cell suspension was incubated with 5 ml 0.075 M potassium chloride and colcemid (final concentration 1.6 to 4.0 pg ml-') at 37°C for 30 min. The cells were fixed in methanol/acetic acid (3:1) and conventional air-dried slides prepared. If Giemsa stained slides demonstrated the presence of metaphases further slides aged at 60°C were G-banded (Seabright, 1971). Metaphases were photographed under oil-immersion using 50 ASA monochrome film.
In three cases a modified synchronised culture technique was used (Webber & Garson, 1983).
Interpretation and analysis The International System for Human Cytogenetic Nomenclature was used throughout (ISCN, 1985). Because of the complex chromosomal changes seen, it was unusual for more than one cell to have exactly the same karyotype. Nevertheless, particular chromosomal abnormalities were frequently present in the majority of cells analysed. Structural changes affecting two or more cells were considered clonal, whereas losses were considered clonal only if a chromosome was missing from at least three cells in which all remaining chromosome were identifiable. No attempt was made to characterise chromosomal gains.
Associations between chromosome lp abnormalities and several staging and prognostic factors were sought. The parameters investigated were age, tumour size, nodal status, joint UICC/AJCC tumour staging (Hutter, 1987), histological grade (Bloom & Richardson, 1957), oestrogen and progesterone receptor (ER and PR) levels and the previously described (Bryan et al., 1986) and validated (Alexander et al., 1987) Melbourne Prognostic Index. The presence or absence of lp abnormalities was compared with non-normally distribution continuous data (e.g. tumour size) and ordered categorical data (e.g. UICC stage) using the Mann-Whitney U-Test, and with binary variables (e.g. node positivity) using the Chi squared or Fisher Exact Probability Test as appropriate.
Since patients possessing cytogenetic data constituted a small subgroup, they were compared with those lacking cytogenetic data for the above prognostic factors using the same 132 P.J. HAINSWORTH et al.

tests.
The level of significance was set at P = 0.05 throughout.

Cytogenetic abnormalities in tumours
The cytogenetic features of the 26 primary breast cancers are summarised in Figures 1 and 2. Apparent discrepancies in numbers between these figures result from the fact that individual tumours may display multiple clonal abnormalities affecting the same chromosome. Full karyotypic details are to be found in Hainsworth et al. (1991). Figure 1 shows that 25 tumours were informative for chromosome #1 and these cases form the subject of this paper. The clinico-pathological features of those with and without cytogenetic data for chromosome #1 are shown in Table I.
Abnormalities of the short arm of chromosome #1 were found in 14 (56%) of the 25 primary breast tumours (Table  II). In four cases more than one abnormality of chromosome lp was present in a single tumour. Translocations predominated, involving lp22 (n = 1), lp35 (n = 1) or lp36 (n = 10) breakpoints (Figures 3 and 4). Because of limitations in the chromosomal quality, only one of the translocation partners was defined (case 104). Deletions were observed with breakpoints at lpl2 (n = 1), lp22 (n = 2) and lp33 (n = 1), and one inversion was identified with lp22-lp36 breakpoints.   Amongst patients with tumour karyotypes, the presence of chromosome lp rearrangements was significantly associated with ER negativity, high histological grade and high Melbourne Prognostic NEPA and SEPA Scores, all signifying an unfavourable prognosis (Table III).
With the exception of one lobular tumour (case 104), tumours with Ip abnormalities were all invasive ductal carcinomas.
In the comparison of those with and without chromosome #1 data, those informative for chromosome #1 were more likely to be PR negative (Table I). For all other clinicopathological factors assessed, those with chromosome #1 data exhibited no significant differences when contrasted with the rest of the study group.

Discussion
At a cytogenetic level, little attempt has previously been made to correlate chromosomal abnormalities in breast cancer with clinical behaviour, no doubt because of the enormous technical difficulties experienced in producing analysable metaphases from breast tissue (Pathak, 1979;Limon et al., 1986;Sandberg et al., 1988) and the marked complexity and heterogeneity of karyotypic data obtained (Rodgers et al., 1984;Hill et al., 1987;Gebhart et al., 1986;Hainsworth et al., 1991).
At a molecular level, the prognostic associations for loss of heterozygosity at some loci have been sought. Deletion affecting the Harvey-ras locus (lipl5) has been linked with poor prognosis (Theillet et al., 1986;Mackay et al., 1988). Genuardi et al. (1989) reported that distal deletion of a chromosome lp36 locus was more common in those with early age of diagnosis, strong family history and multifocal disease than in patients with none of the characteristics of hereditary tumours (Genuardi et al., 1989). However, no associations with standard staging and prognostic factors were observed.
The data presented here show that chromosome lp rearrangements, predominantly distal translocations, were cytogenetically recognised in 14 (56%) of 25 primary breast cancers. A preponderance of distal lp changes has not been noticed by other authors. Mitchel and Santibanez-Koref (1990) report involvement of chromosome lpl3 breakpoints in 6/14 of their own breast cancers and in 17/99 specimens (56 tumour biopsies and 43 pleural effusions) from the University of Lund computerised Cancer Chromosome Registry.
The assocation of chromosome lp abnormalities with four of the prognostic factors studied suggests that rearrangement at this site may correlate with tumour progression. In this context it should be noted that chromosome #1 alterations are frequently observed in both solid and haematological malignancies (Heim & Mitelman, 1987). Teleologically, this suggests a broad role for chromosome #1 abnormalities in carcinogenesis, not confined to breast cancer.
In this study there were proportionately far more translocations than deletions of chromosome lp. Based on these results, it would be highly speculative to propose a specific genetic mechanism operating at chromosome #1 which could be implicated in tumour progression. These findings are however in keeping with the occurrence of allelic deletion at the D1Z2 locus (mapping to chromosome lp36) in 41% of 37 informative tumours (Genuardi et al., 1989). The latter is consistent with the notion that a suppressor gene near the D1Z2 locus may be implicated in the pathogenesis of ductal breast cancer.
The limitations of this analysis are recognised. Chromosome #1 data was only available for 25 tumours. These obviously represent a highly selected subgroups of the patients treated during this period although comparison with those lacking karyotypes suggested little bias. It is also conceivable that the occurrence of lp abnormalities merely represents an increase in genetic instability which happens to be associated with features of poor prognosis. However the frequency with which the distal portion of the p arm is singled out indicates that some sort of selective process is at work conveying an advantage to clones possessing distal lp rearrangements.