Allelic Losses at Chromosome 8p21–23 Are Early and Frequent Events in the Pathogenesis of Lung Cancer¹ Free

Lung cancer is the most frequent cause of cancer deaths in both men and women in the United States (1), and tobacco smoking is accepted as the major cause (2). Lung cancer is classified into two major groups, SCLC³ and NSCLC (3). Squamous cell carcinoma, adenocarcinoma, and large cell carcinoma are the major histological types of NSCLC (4). As with other epithelial malignancies, lung cancers are believed to arise after a series of progressive histopathological changes (preneoplastic lesions) in the bronchial epithelium (5). However, the sequence of histological changes has been well established only for squamous cell carcinoma (5). These morphological preneoplastic steps include hyperplasia, squamous metaplasia, dysplasia, and CIS.

Many mutations, especially those involving recessive oncogenes, have been described in invasive lung cancers (3). However, relatively little is known about the molecular events preceding the development of such tumors. Recently, we and others (6, 7) have reported a very high incidence of molecular abnormalities (LOH and microsatellite alterations) in the normal and abnormal bronchial mucosa of cancer patients and of former and current smokers. Similarly, studies of molecular abnormalities in preneoplastic lesions associated with lung cancer have shown that allelic losses (LOH) at chromosomal regions 3p and 9p occur early during the multistage development of invasive lung cancer, followed by losses at the TP53 gene (17p; Refs. 8, 9, 10, 11, 12, 13).

Allelic losses on the short arm of the chromosome 8 (8p) have been reported as a frequent event in several cancers, including lung, breast, colon, prostate, hepatocellular carcinoma, and head and neck (14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25). The cumulative data strongly indicate that chromosome 8p may harbor one or more TSGs, and at least three 8p regions (8p12–21, 8p21, and 8p22) potentially harboring such genes have been identified in several neoplasms (14, 15, 16, 17, 18, 19, 20, 22, 23, 24). Using a limited number of polymorphic markers, frequent allelic losses (∼50%) involving the 8p21–22 region have been detected previously in NSCLC cell lines (21) and primary tumors (22, 24, 26). However, no data about the stage at which such abnormalities occur during the multistage development of lung cancer are available.

In the present study, we use allelotyping with multiple markers in tumors and preneoplastic lesions to further define the critical chromosome 8 region that may harbor one or more TSGs and determine the stage the deletion first appears.

Thirty paired lung cancer cell lines, of which 13 were SCLCs and 17 were NSCLCs, and their corresponding B-lymphoblastoid (BL) cell lines were used in this study, as described previously (21). All of the lung cancer cell lines as well as most of the BL lines were initiated by the authors or coworkers at the National Cancer Institute and Hamon Cancer Center (Table 1). The NSCLC lines consisted of 12 adenocarcinomas, 2 large cell carcinomas, 2 squamous carcinomas, and 1 adenosquamous carcinoma. The cell lines tested are listed in Fig. 1. They are deposited at the American Type Culture Collection.

Paraffin-embedded materials from 68 surgically resected primary lung carcinomas representing the three major histological types of lung cancer were analyzed. They consisted of 22 small cell, 25 squamous cell, and 21 adenocarcinomas. They were obtained from cases resected between 1980 and 1997 at Parkland Hospital (Dallas, TX) and M. D. Anderson Cancer Center (Houston, TX). The clinicopathological stage was determined after surgery using standard criteria. The patients consisted of 30 women and 38 men ranging in age from 30 to 84 years (mean, 62 years). Most of the tumors were stage I (49 of 68 cases, 72%), although they varied from stage I to IV. All of the patients were smokers, and most of them were heavy smokers (mean, 39 pack-years; range, 15–120 pack-years). Other relevant clinical and pathological information is presented in Table 1.

Serial 5-μm sections were cut from archival, formalin-fixed, paraffin-embedded tissue. All slides were stained with H&E, and one of the slides was coverslipped. The coverslipped slide was used as a guide to localize regions of interest for microdissection of the other slides.

We selected 11 cases of squamous cell carcinoma (Table 1) that contained multiple foci of various preneoplastic changes. All compartments of the respiratory tree were examined. The microslides were examined by two pathologists (A. F. G. and I. I. W.) and scored using published criteria for the histological identification of epithelial preneoplastic lesions of lung (5). Histopathological diagnoses were categorized as: (a) normal respiratory epithelium; (b) “mildly abnormal epithelium”: hyperplasia (goblet cell or basal cell type) or simple squamous metaplasia without dysplasia; (c) dysplasia: because of limited numbers, we did not divide dysplasias into mild, moderate, or severe categories; and (d) CIS: although minor atypical changes arising in hyperplastic respiratory epithelium were identified, dysplastic changes were only diagnosed in abnormal metaplastic epithelium. All noninvasive lesions, including hyperplasia, squamous metaplasia, and dysplasia were referred to as “preneoplasias.” CIS lesions were referred to as “noninvasive” neoplasia.

We identified a total of 95 histologically discrete foci each consisting of at least 800 cells. They included samples from the 11 invasive carcinomas, 52 preneoplasias (24 hyperplasias, 10 metaplasias, and 18 dysplasias), 12 CIS, and 20 samples of histologically normal epithelium. One or more foci of histologically normal or mildly abnormal epithelium were present in all 11 cases, and one or more discrete CIS lesions were identified in 10 cases. All nontumoral lesions were located in centrally located large bronchi (lobar, segmental, and subsegmental).

We studied 58 biopsy specimens obtained by fluorescence bronchoscopy as described (27) from 31 subjects, 13 current smokers (mean and median, 2 samples) and 18 former smokers (mean, 2 samples; median, 1.7 samples; Table 1). All subjects were recruited by S. L. at the British Columbia Cancer Agency (Vancouver, British Columbia, Canada) as part of an Institutional Review Board-approved clinical trial to study the effect of smoking on the respiratory epithelium. All participants gave written informed consent. Subjects were categorized as to smoking status as published previously (6). All smokers had smoked more than 20 pack-years, except for one subject (10 pack-years). Most former smokers (15 of 18, 83%) had ceased smoking for 5 years or longer (mean, 22 years). Other relevant subject information is presented in Table 1. Pathological diagnoses were categorized as stated previously. The samples included 6 histologically normal epithelium, 13 mildly abnormal epithelia, 34 dysplasias, and 5 CIS.

Microdissection from archival paraffin-embedded tissues was performed either by laser capture microdissection (28) or manually using a micromanipulator (9) from multiple microslides of each sample. DNA extraction was performed as described previously (9). Dissected lymphocytes or stromal cells from the same slides were used as a source of constitutional DNA from each case. After DNA extraction, 5 μl of the proteinase K-digested samples, containing DNA from at least 100 cells, were used for each multiplex PCR reaction.

To evaluate LOH, we used primers flanking 26 dinucleotide and multinucleotide microsatellite repeat polymorphisms spanning the entire length of chromosome 8. Nineteen markers spanned the short arm (8p), and 7 markers were used to examine the long arm (8q). The microsatellite markers used are listed in Table 2. For analysis of lung cancer cell lines, all 26 polymorphic markers spanning both chromosome 8 arms were used. Subsets of markers spanning 8p21–23 regions were used for analysis of primary tumors (15 markers) and normal and abnormal epithelium (8 markers; Table 2; Fig. 1). Primer sequences were obtained from the Genome Database.

For cell lines, PCR-LOH analysis was performed directly on genomic DNA (21), whereas in microdissected samples, a two-round PCR strategy was used as described previously (29). LOH was scored by visual detection of complete absence of one allele (Figs. 1 and 3).

To determine whether the deletions in 8p21–23 regions were progressive in individual foci, we determined frequencies of loss of individual foci using an 8p fractional allelic loss (8p FAL) index as follows:

Because heterozygosity at the different loci varied between subjects, the number of chromosomal regions tested in subjects varied. Thus, an index was used to compare chromosome 8p allelic losses between smoking subjects (current and former smokers). The FAL index for all biopsy specimens from an individual subject (chromosome 8p FAL-subject) was calculated as follows:

Statistical analyses was performed using the nonparametric Wilcoxon and Fisher Exact tests. The cumulative binomial test (30) was used to examine the likelihood that the occurrence of a particular event (loss of the same allele in the invasive carcinoma and an associated epithelial sample) occurs at a particular probability when observed in repeated trials. When the results are compared with a chance occurrence or nonoccurrence, the particular probability of comparison is 0.5. Probability values of P < 0.05 were regarded as statistically significant.

Thirty lung cancer cell lines (13 SCLCs and 17 NSCLCs) were analyzed for LOH in both chromosome 8 arms, using 26 microsatellite markers (Fig. 1,a and Table 2). Allelic losses involving sites at 8p21–23 were detected frequently in both SCLC (8 of 13, 62%) and NSCLC (13 of 17, 77%) cell lines (Fig. 1,a and Table 2). In some cell lines, especially NSCLC, all or most of the short arm was deleted. In other cell lines, the 8p deletions were small and discontinuous (Fig. 1,a). Deletions involving 8q and the proximal part of 8p (8p11–12) were relatively infrequent (0–36%; Table 2) and focal.

Sixty-eight microdissected primary lung tumors were analyzed for LOH at chromosome 8p21–23 using 15 microsatellite markers (Fig. 1,b and Table 2). Allelic losses at this region were frequent in the three major types of lung cancer (SCLC: 19 of 22, 86%; squamous cell: 25 of 25, 100%; and adenocarcinoma: 17of 21, 81%). As with lung cancer cell lines, all or almost all of this extensive region was deleted in some tumors, especially squamous cell carcinomas (Fig. 1 b). In other cases, the losses were small and discontinuous. There was no correlation between chromosome 8p allelic loss and extent of smoking exposure, sex, or age (data not shown).

We microdissected a total of 84 histologically discrete foci of normal-appearing epithelia and precursor lesions from 11 surgically resected specimens. We limited this analysis to foci accompanying squamous cell carcinoma because the sequence of histological changes in this cancer has been well established, and because 8p deletions were most frequent in this tumor type. On the basis of the experience with lung cancer cell lines and microdissected tumors, a panel of eight highly informative microsatellite markers spanning 8p21 to 8p23 regions was used (listed in Table 2).

Although no 8p allelic losses were detected in histologically normal epithelium, one or more regions having allelic loss were detected in five (15%) of 34 foci of mildly abnormal epithelium (Fig. 2). Increasing severity of histological change was characterized by increasing frequencies of deletions at chromosome 8p regions. Thus, 50% of dysplastic lesions, 92% of CIS, and 91% of invasive tumors demonstrated allelic loss at one or more chromosome 8p regions. No defined pattern or sequence of allelic losses at these regions were detected during the multistage development of squamous cell carcinomas. However, the extent of the 8p21–23 deletions in histologically normal and mildly abnormal foci (mean 8p FAL index, 0.003) and dysplastic lesions (mean 8p FAL index, 0.36) was significantly smaller (P < 0.002) and more discrete than in the more advanced foci (CIS and invasive carcinoma: mean 8p FAL index, 0.79). Examples of the progression in the size of the 8p21–23 deletions during the multistage of seven resected squamous cell carcinomas and their accompanying preneoplastic and noninvasive lesions are shown in Fig. 3.

Previous studies of different bronchial lesions from the same individual patient demonstrated that loss of parental alleles at any one locus was not random, and that there was a strong tendency for the identical allele to be lost in all nonneoplastic and neoplastic foci examined (8, 9, 10). We refer to this phenomenon as ASMs (8, 9, 10). We determined the frequency of ASMs in the 35 foci harvested from 11 patients with lung cancer and demonstrated one or more sites of 8p allelic loss. For all 89 comparisons involving eight chromosome 8p microsatellite markers, the same parental allele was lost in all 89 (100%; Fig. 3). The possibility of this happening by chance are extremely remote, as tested by the cumulative binomial test (P = 1.6 × 10⁻²⁷).

Our previous studies demonstrated that allelic losses occurring during the multistage development of squamous cell carcinoma are not random (8). The earliest and most frequent regions of allelic loss occurred at several 3p regions and 9p21, followed by deletions at the TP53 gene locus (8). To determine the role of 8p21–23 deletions in the sequential molecular changes involved in the development of squamous cell lung carcinoma, we analyzed the pattern of deletions at regions 3p, 9p21, and 8p21–23 in the invasive tumors and their accompanying normal and abnormal epithelia (Table 3). Because 54 epithelial specimens from the 11 lung cancers had been analyzed previously for 3p and 9p21 regions (8), only those specimens were considered. Three patterns of allelic loss were discerned in the preneoplastic (n = 42), noninvasive (n = 12), and invasive foci (n = 11; Table 3): (a) “negative” (wild-type) pattern, in which no allelic loss at any region was noted; (b) “LOH without 8p loss” pattern, with losses at other regions but no deletion at 8p; and (c) “LOH with 8p loss” pattern, with deletions at 8p and/or deletions at other regions. The relationships between histological diagnoses and patterns of allelic loss are shown in Table 3. The “negative” pattern was detected in nearly half of normal/mildly abnormal epithelia and dysplastic lesions. Although the “LOH without 8p loss” pattern was seen only in normal and mildly abnormal epithelia, the “LOH with 8p loss” pattern was detected in nearly half of dysplastic lesions and was the only pattern detected in advanced lesions (CIS and invasive carcinoma). Of great interest, 8p21–23 deletions were never detected alone and always accompanied 3p or 3p and 9p21 allelic losses.

We studied 58 biopsy specimens obtained by fluorescence bronchoscopy from 31 subjects, 13 current smokers, and 18 former smokers (Table 1). The same panel of microsatellite markers spanning 8p21 to 8p23 regions that was used to examine epithelial samples accompanying lung carcinomas was used (Table 2). Similar to samples obtained from cancer patients, there was a progressive increase in 8p LOH frequency with increasing severity of histopathological changes (Fig. 2). Surprisingly, the frequencies of chromosome 8p LOH detected in bronchial samples from smokers without lung cancer were slightly higher in all histological categories than those found in cancer patients. However, no differences in the pattern of LOH at 8p21–23 regions were detected between epithelial samples examined from resected lung cancer and smokers’ biopsy specimens. Of interest, 20 of the 31 (65%) smokers without cancer (10 current and 10 former smokers) demonstrated LOH at one or more 8p chromosomal regions examined. Although the 8p FAL index was higher in current than former smokers (Fig. 4), we found no statistically significant differences in the frequencies and patterns of LOH between the two groups, and multiple 8p deletions were found in biopsy specimens from former smokers who had quit 10–48 years previously.

We studied the role of allelic losses at chromosome 8 in the multistep pathogenesis of lung cancer using a four-step analysis that included a study of lung cancer cell lines, microdissected primary lung tumors of the three major histological types, and normal and abnormal respiratory epithelium from lung cancer patients and smokers without lung cancer.

Both arms of chromosome 8 were examined for allelic loss in lung cancer cell lines, and deletions involving the distal part of the short arm (8p21–8p23) were frequently detected in both SCLC and NSCLC lines. Deletions involving the proximal part of 8p and all of 8q were relatively rare. High frequencies were also noted in DNA from microdissected tumors representing the three major types of lung cancers. There was no correlation between chromosome 8p allelic losses and extent of smoking exposure, sex, or age. In some instances, the deletions involved all or almost all of the arm (especially in squamous cell carcinomas), whereas in other cases, they were relatively small and discontinuous. Of interest, high incidences of LOH at 8p21 or 8p22 regions have been documented in several cancer types, including colon, breast, prostate, head and neck, urinary bladder, and hepatocellular carcinomas (15, 16, 18, 19, 20, 22, 24, 31, 32, 33, 34), and two genes in the 8p22 region, platelet-derived growth factor receptor β-like gene and N33, have been proposed as candidate TSGs (35, 36). However, we were unable to identify the smallest region(s) of overlapping deletions in lung cancers. These findings suggest that either one or more tumor suppressor genes involved in the pathogenesis of lung cancer reside in this chromosomal region, or that the deletions represent evidence of genomic instability affecting the distal part of 8p without specifically targeting one or more genes.

Our present findings confirm and extend previous observations that mutations accumulate early during the multistage pathogenesis of lung cancer (6, 7, 8, 9, 10, 11). We limited the present study to the pathogenesis of squamous carcinoma arising in the setting of smoking-damaged respiratory epithelium because the sequence of histological changes in this cancer is well established. Our results indicate that allelic losses at chromosome 8p21–23 regions occur as an early event in the multistage development of this neoplasm, commencing in mildly abnormal epithelium (hyperplasia or squamous metaplasia). There was a progressive increase of allelic loss frequency and of the extent of the deletions with increasing severity of histopathological changes. Of interest, 8p allelic losses have been detected at a relatively early stage during the pathogenesis of head and neck carcinomas (25), another smoking-related tumor.

Our previous studies demonstrated that allelic losses occurring during the multistage development of squamous cell carcinoma are not random (8). The earliest and most frequent regions of loss occurred at several 3p regions and 9p21, followed by deletions at the TP53 gene locus (8). Our findings suggest that although 8p21–23 deletions are a relatively early event (commencing at normal/mildly abnormal epithelia stage) in the pathogenesis of lung cancer, they usually follow 9p deletions and always follow 3p losses. By examining all our material for 8p, 3p, and 9p allele loss, we suggest that the order of events is normally either 3p→9p→8p or 3p→8p→9p (Table 3).

Of interest, the same parental allele was always lost in precursor lesions, as in the corresponding invasive carcinomas. We and others have documented and discussed this phenomenon, known as ASMs, both in the respiratory epithelium of smokers and in patients with lung cancer (8, 9, 10, 11).

We and others (6, 7) have reported a very high incidence of allelic losses at several chromosomal regions frequently deleted in lung cancer (3p, 9p, RB, and TP53 loci) in the normal and abnormal bronchial epithelium of former and current smokers. Our findings of frequent chromosome 8p21–23 deletions in bronchial epithelium of former and current smokers confirm the findings that 8p deletions commence early during the multistage pathogenesis of lung cancer. The chromosome 8p21–23 allelic loss frequencies detected in bronchial samples from smokers were slightly higher in all histological categories than the figures from cancer patients. In the smoking population, fluorescence bronchoscopy was used to select biopsy sites (based on areas of abnormal fluorescence), possibly accounting for the higher mutational frequency. In addition, the biopsies obtained by fiberoptic bronchoscopy are relatively small and have limited numbers of epithelial cells. In surgically resected samples, extensive areas of normal epithelium were frequently present, and the relatively large microdissected foci may have contained more than one clonal or subclonal population.

As with the other allelic deletions discussed previously, we found no statistically significant differences between current and former smokers in the frequencies of 8p21–23 allelic losses, and deletions were found in subjects who had quit smoking 10–48 years previously. These findings are consistent with the sustained increased risk of lung cancer in former smokers (37).

In conclusion, our findings indicate that deletions at chromosome 8p21–23 regions are an important and early event in the pathogenesis of lung cancers. Thus, chromosome 8p21–23 allelic losses may be useful markers in smoking-damaged epithelium for risk assessment and for monitoring the efficacy of chemopreventive regimens.