A number of other suppressor genes have been implicated in breast cancer.

Nm23, originally characterized as a metastasis suppressor gene, has been reported to be uniformly expressed in all epithehal cells of benign breast lesions (35 cases that included fibrocystic disease and fibroadenomas but were not further described) and noncomedo DCIS (18 cases including solid, cribriform, and micropapillary types), but some cells were negative for Nm23 in all 7 cases of comedo DCIS (Royds et al., 1993). For invasive cancers, the percent of negative cases increased with grade (Royds et al., 1993), supporting observations that comedo DCIS is associated with more aggressive cancers.

An immunohistochemical study of retinoblastoma susceptibility gene (RBI) expression reported that 16 of 56 invasive cancers contained some negative cells, but all epithelial cells in benign breast lesions were uniformly positive (Varley et al., 1989). The benign lesions were not described but fibrocystic disease and fibroadenomas were included and no DCIS cases were examined.

If LOH occurs randomly, most events that result in LOH probably are irrelevant to the development of breast disease and cancer.

The baseline incidence for LOH appears to be less than 5% based on a study using randomly selected probes, that is, probes for chromosomal regions not known to be associated with cancer (Chen, L.C. et al., 1992). Significantly increased incidences of LOH have been described for more than half of the chromosome arms (Table 1), and more than one locus has been implicated on several arms in invasive breast cancer (Callahan et al., 1993; Cleton-Jansen et al., 1994; Kirchweger et al., 1994; Merlo et al., 1994; Dorion-Bonnet et al., 1995; Gudmundsson et al., 1995; Nagai et al., 1995; Radford et al., 1995; Yaremko et al., 1995; Munn et al., 1996; Sheng et al., 1996; Spirin et al., 1996). Thus, a great number of suppressor genes may be involved in breast cancer.

To determine which are important in early stages of breast cancer development, studies have compared LOH in low grade invasive carcinomas to LOH in high grade cancers or examined DCIS, sometimes comparing comedo to noncomedo histologic types. Only a few studies have utihzed benign breast lesions.

LOH is frequently observed in DCIS. Low nuclear grade DCIS averaged 1.2 chromosomal arms lost per case whereas intermediate and high grade DCIS averaged a loss of 5.6 chromosomal arms (Fujii et al., 1996b). Losses of 16q and 17p were common in all grades of DCIS suggesting a role in initiation of the DCIS stage (Fujii et al., 1996b). Microdissection and microsateUite length polymorphism analysis revealed loss of 16q in 89% of DCIS cases with the most frequent loss in the region q23.3-24.1 (20 of 26 informative cases; Chen, T. et al., 1996).

In a number of studies of cases of mixed DCIS and invasive cancer, LOH was concordant in the two lesions as one would expect if DCIS is a precursor for the invasive component (O’Connell et al., 1994; Koreth et al., 1995; Zhuang et al., 1995). Additional chomosomal arms are lost and heterogeneity develops with progression (Fujii et al., 1996a, b). Other analyses of DCIS indicate frequent LOH at 8p, 13q, and 17p as well as 16q (Radford et al., 1995; Stratton et al, 1995). The latter two studies consisted primarily of comedo DCIS. A recent study restricted to noncomedo DCIS found LOH in 40% of cases at 3p, 1 Iq, 13q, and 16q (Man et al., 1996).

There were two areas involved on 1 Iq and the 3p loss appeared to be loss of the FHIT gene (fragile histidine triad), which is of interest because FHIT as well as another gene located in the chromosmal band 3pl4.2, PTPRG (human receptor tyrosine phosphatase Ϫ gene), were deleted in three samples of atypical hyperplasia but not in fibroadenoma as assessed by RT-PCR (Panagopoulos et al., 1996).