Sequential loss of heterozygosity in the progression of squamous cell carcinoma of the lung.

Radiographically occult bronchogenic squamous cell carcinomas are early lung cancers that localize mainly in the bronchial wall, and are thought to be a good model for investigating genetic alterations through lung cancer progression. In order to elucidate sequential genetic changes in lung cancers, we analysed the incidence of allelic losses on chromosome regions 2q33, 3p21, 5q21, 7q31, 9p21 and 17p13 for 40 cases of radiographically occult bronchogenic squamous-cell carcinomas and 40 cases of advanced lung cancers microdissected. In this study we used eight microsatellite dinucleotide polymorphic markers. Frequent loss of heterozygosity (LOH) was observed on 3p21 (53%), 5q21 (44%) and 17p13 (61%) in roentgenographically occult bronchogenic squamous cell carcinomas. 2q, 7q and 9p were lost less frequently in both roentgenographically occult bronchogenic squamous cell carcinomas and advanced lung cancers. These results suggest that several tumour-suppressor genes are associated with lung cancer progression and that genetic changes on 3p21, 5q21 and 17p13 are early events.

study of allelic losses is important to elucidate genetic alterations and in the search for tumour-suppressor genes. Numerous reports have been published concerning allelic losses in advanced lung cancers (Tsuchiya et al. 1992: Field et al. 1996. The first report of allelic losses in preneoplastic lesions of the lung was published by Sundaresan et al (1992) and a few investigators have reported allehic losses in early cancer or precancer of the lung in a few cases (Chung et al. 1995: Hung et al. 1995: Thiberville et al. 1995a). Therefore. for the further elucidation of multistep tumorigenesis of lutn cancer. more cases of early cancer or precancer of the lung must be examined.
Radiographically occult bronchogenic squamous cell carcinomas (ROCs) are early lung cancers that are detected only by sputum cytology, and are located mainly in the bronchial wall (Saito et al. 1992). Non-treated ROCs develop into advanced lung, cancers with radiolooically abnormal shadows (radiographically non-occult squamous cell carcinomas: RNOCs) after several years (Saito et al. 1990).
Accordingly. the ROC is thought to be a good model for the purpose of elucidating the sequential genetic alterations in the progression of lung cancer. In this study. we analyse allelic losses on six chromosomes of 40 cases of ROC and 40 of RNOC.

MATERIALS AND METHODS
Forty cases of resected ROCs and 40 cases of resected RNOCs were examined. All cases were male. All cases of ROCs were classified as stage I. Resected specimens of ROCs were examined pathologically by senial block sectioning (2 mm block thickness) (Nagamoto et al. 1993). Depth and site of maximum invasion were decided by histopathological analysis (Nagamoto et al. 1993).
ROCs were divided into two groups according to depth of invasion: intrabronchial wall invasion (25 cases) and extrabronchial wall invasion (15 cases). RNOCs were also divided into two groups: stage 1 (19 cases) and other stages (stage H-IV. 21 cases).
For RNOCs. tumours and corresponding normal tissues were stored frozen at -80'C until DNA extraction could be performed. DNA was prepared by proteinase K digestion and phenol-chloroform extraction. For ROCs. eight 20-gm-thick sections of tumours and corresponding normal tissues were cut from formalin-fixed. paraffin-embedded blocks. These eight sections were used for microdissection according to the technique described elsewhere (Sundaresan et al. 1992). DNA was obtained by proteinase K digestion and phenol-chloroform extraction.

RESULTS
The allelotyping of all 80 cases was shown in Table 1. The average frequency of LOH was 40%. The frequency of LOH of ROCs and RNOCs is shown in Figure 2. In all groups, 3p, 5q and 17p showed frequent LOH. Moreover, allelic loss on 17p was more frequent in RNOCs (70%) than in ROCs (49%). On the other hand, 2q, 7q and 9p showed loss less frequently in both ROCs and RNOCs.
The average fractional allelic loss (FAL) (Vogelstein et al, 1989) of all cases, ROCs and RNOCs was 0.4, 0.39 and 0.42 respectively. Ratio of cases with FAL >0.5 increased gradually according to cancer progression (Table 2). In ROCs with intrabronchial wall invasion, six cases had LOH on only one locus, and six cases had LOH on two loci. Of these six cases, four cases had loss on 3p2l and any other locus, two cases had loss on 17pl3 and any other locus, one case had loss on 3p21 and 17pl3.    Howx exer. the number of cases -studied x% as not enough to a.scertain the statistical sicnificance of differences in incidence. Our present studv shox-ed a constant. hich incidence of LOH on 5q2I in all four groups. These results suggLeSt that LOH on 5q2 I is related to an early step in SQLC progression The TP53 polymorphic marker used in this present studv exists on p53 tumour-suppressor gene locus. A frequent p53 aberration xw as observed in many cancers including lungcancers (Monica et al. 1991 . Recentlv. some groups reported that LOH on 17pl3 occurred in dx splasia and CIS of the luno in a fexx cases Sozzi et al. 1992: Sundaresan et al. 1992 Our results showxed a high frequencv of LOH on the p53 locus even in intrabronchial xwall invxasion of ROC. and the frequenc\ of LOH increased gradually according to the degree of cancer progression. These data suggest that the p53 gene is related to an early step of SQLC progression and also correlated with the depth of invasion. Concerning this suggestion, positive immnunostaining of p53 was observed to be significantly correlated with the depth of invasion in colorectal cancer (leda et al, 1996). Kohno et al (1994) reported that a homozygous deletion was detected on chromosome 2q33 in a human small-cell lung carcinoma cell line, and suggested the presence of a novel tumour-suppressor gene there. The frequency of LOH on 2q in several reports (Tsuchiya et al. 1992;Shiseki et al, 1994;Kohno et al, 1994) ranged from zero to 63% in advanced lung cancers. Our examination showed a constant, low frequency of LOH on 2q33 in tumour progression. Based on our results, we conclude that LOH involving 2q33 is less important for SQLC progression.
LOH on 7q31 was seen frequently in head and neck squamous cell carcinomas (Zenklusen et al, 1995). Some investigators reported that 9p21 frequently showed LOH even in dysplasia and CIS of the lung (Thiberville et al, 1995a). Others reported mutation of p16 to be more frequent in metastatic lesions than in primary lung cancers (Okamoto et al, 1995). However, our results showed no relationship between LOH on 7q31 or 9p21 and SQLC progression.
The average FAL of ROCs was lower than that of RNOCs, and the ratio of cases with FAL >0.5 increased gradually according to the degree of cancer progression. These results suggest an accumulation of genetic alterations linked to SQLC progression. Among six cases of ROC with intrabronchial wall invasion having LOH on two loci, five cases had LOH on 3p2l or 17pl3 and only one case had LOH on 3p2l and 17pl3, which suggests that loss on 3p2l or 17pI3 plays an important role in lung cancer progression and occurs at the early stage of tumorigenesis. It also suggests that loci other than 3p2l or l7pl3 may also play an important role.
Systems other than LOH [methylation error (Merlo et al, 1995) loss of imprinting (Kondo et al, 1995), for example]) may play an important role in SQLC progression.
In summary, we analysed many cases of early and advanced SQLCs, and presented evidence that genetic alterations on 3p2l. 5q21 and l7pl3 are related to the progression of SQLCs.
The alterations on 3p2l, 5q21 and l7pl3 occurred frequently even in the stage of intrabrnchial wall invasion of ROCs, which seems to be an early step of tumour progression. Moreover, an allelic loss on 17pl3 is also related to the late stage of tumorigenesis. On the other hand, genetic changes on 2q33, 7q31 and 9p21 were few and not related to the progression of SQLCs. In this study, materials were limited to SQLCs. The number of cases studied was not enough for a real statistical analysis. Further studies on many cases of premalignant lesions are needed to determine more precisely the sequential genetic changes in lung cancer progression.