Somatic genetic changes in lung cancer and precancerous lesions

Annals of Oncology 6 (Suppl. J): S27-S32, 1995. © 1995 Kluwer Academic Publishers. Printed in the Netherlands.

Symposium article Somatic genetic changes in lung cancer and precancerous lesions V. Sundaresan,1 A. Heppell-Parton,1 N. Coleman,3 M. Miozzo,2 G. Sozzi,2 R. Ball,4 N. Cary,5 P. Hasleton,6 W. Fowler7 & P. Rabbitts1 'MRC Clinical Oncology and Radiotherapeutics Unit, MRC Centre, Cambridge, U.K.; 2Division of Experimental Oncology, Istituto Nazionale Tumori, Milan, Italy, 3Department of Pathology, University of Cambridge, Cambridge, U.K.; 4Department of Pathology, Norfolk and Norwich Hospital, Norwich, Norfolk, U.K.; 5 Department of Pathology, Papworth Hospital, Papworth Everard, Cambridgeshire, U.K.; 6 Department of Pathology, Regional Cardiothoracic Centre, Wythenshawe Hospital, Wythenshawe, Manchester, U.K.; 1 Bowman Cray School of Medicine, Winston-Salem, North Carolina, U.S.A.

Background: Morphological abnormalities of the bronchial epithelium are associated with lung cancer development and are considered likely to represent the preneoplastic stage of the disease. The association of these lesions with different histological types of lung cancer was reviewed in a series of 97 samples. Lesions associated with squamous cell carcinomas provided the best samples for further study. The objective of this study was to describe the somatic genetic changes which occur in these preinvasive lesions. Among the various candidate somatic genetic changes, loss of heterozygosity on chromosome 3 and changes to the p53 gene were selected as being the most informative. It was demonstrated that these genetic changes, characteristic of fully invasive lung tumours, also occur at the premalignant stage of the disease. In an attempt to take a less

Introduction

Many epithelial tumours develop through a series of morphological changes of increasing disorder. Progressive disarray of the genotype drives the phenotypic abnormalities. Elucidation of these somatic genetic changes is vital for a complete understanding of tumour development and will allow attention to be focused on the preinvasive stage of the disease, enabling both the early detection and identification of new therapeutic targets. Various somatic genetic changes have been identified in lung tumours [1], and this study investigates which of these are detectable at the preinvasive stage of the disease. Preinvasive bronchial lesions as candidate preneoplastic lesions

Bronchial epithelial abnormalities detected in association with fully invasive lung tumours range from squamous metaplasia to severe dysplasia amounting to localized carcinoma. A direct sequential relationship between these various lesions and malignant lung tumours, however, has not been demonstrated in longitudinal studies. The best evidence for an association

directed approach to the comparison of invasive and preinvasive lesions, karyotype analysis was performed on short-term cultures of bronchial cells adjacent to the bronchial margin obtained from patients undergoing lung tumour resection. One such karyotype had a deletion to chromosome 3 (del 3pl3-14) as the single abnormality. Conclusion: It was concluded that genetic damage to p53 and chromosome 3 is involved in the preinvasive stage of lung cancer, and that damage to chromosome 3 is a particularly early event. Key words: dysplasia, lung cancer, morphology, premalignancy, somatic genetic changes

probably comes from cytological studies [2] demonstrating the presence of abnormal epithelial cells in sputum samples from uranium miners, showing increasingly abnormal cells as the patients progressed towards invasive lung tumours. Bronchial abnormalities adjacent to lung tumours The preinvasive pathway from squamous metaplasia through various grades of dysplasia is most commonly associated with lung tumours of the squamous histological subtype. Bronchial epithelium adjacent to invasive lung tumours of all major histologies was assessed for the presence of identifiable morphological lesions. Table 1 summarizes the morphological abnormalities observed in bronchial mucosa in 97 specimens of lung resected for primary lung cancer or reactive benign conditions. Reserve cell hyperplasia and squamous metaplasia were identified in association with both benign and malignant conditions. In contrast, bronchial epithelial 'preneoplastic' abnormalities ranging from mild to severe dysplasia were most commonly seen in association with squamous cell carcinoma (SQC). Similar preneoplastic abnormalities were seen in association with other histological types of lung cancer, though the numbers analysed in this study were

Downloaded from http://annonc.oxfordjournals.org/ at Freie Universitaet Berlin on June 30, 2015

Summary

28 Table 1. Morphological abnormalities of bronchial mucosa observed in surgically resected specimens. Number of cases analysed

Squamous cell carcinoma Adenocarcinoma Large cell, undifferentiated Small cell Carcinoid Reactive/benign disease

a

Reserve cell hyperplasia (%)

Squamous metaplasia

Mild-tosevere dysplasia

53 22

19 11

29 4

27 7

6 3 4

4 2 1

2 1 1

2 2 -

9

2

3

2a

97

39

40

Loss of heterozygosity on chromosome 3

40

Focal evidence of mild bronchial epithelial abnormalities.

small. Gazdar has reported an association between epithelial dysplasia and adenocarcinomas [3]. In the authors' experience, however, the dysplasia associated with adenocarcinomas generally tends to be of low grade, so lesions associated with SQC have been the focus of attention. Morphological abnormalities, amounting at most to mild dysplasia, were present in the bronchial epithelium from 2 of 9 patients without lung tumours (see Table 1). No bronchial epithelial abnormalities were identified in patients with carcinoid tumours.

Candidate genetic lesions Cytogenetic analysis of lung tumour cell lines and molecular genetic analysis of DNA isolated from biopsies and cell lines has revealed a number of consistent somatic genetic changes in lung cancer [4, 5]. Most studies have investigated deletions and loss of heterozygosity of chromosome 3 (three distinct regions are involved) [6], loss of heterozygosity of 17pl3 associated with mutation in the p53 gene, deletions of 13ql4, and point mutations and larger rearrangements of the Rb gene. Recently 9p has been intensively studied as homozygous deletions have been identified in small cell lung cancer cell lines in the region of the interferon gene cluster. A candidate tumour suppressor gene, multiple tumour suppressor gene 1 (MTS1), has been isolated from within this region and is currently being evaluated for its contribution to lung cancer. The region of 5q encompassing the familial adenomatous polyposis/MCC gene cluster is often lost in lung cancer samples, and two regions on chromosome l i p are involved, particularly in non-small cell lung cancer, though the Wilms' tumour gene itself does not appear to be implicated. Apart from regions of chromosomal loss, point mu-

In the early 1980s, deletions of the short arm of chromosome 3 were detected in small cell lung cancer, and the use of restriction fragment length polymorphism (RFLP) analysis later confirmed this observation [4,5]. In fact, almost every small cell lung cancer appears to have genetic damage to 3p. The situation for non-small cell lung cancer is slightly more equivocal. In the authors' study, about 70% of non-small cell lung cancers showed 3p loss [7]. Non-small cell lung cancer is more difficult to study than small cell lung cancer, as the tumours carry a much higher proportion of normal cells which obscure the tumour genotype. As the tumour samples were only examined macroscopically, the possibility existed that the proportion of non-small cell lung cancer tumours with loss of alleles on 3p had been underestimated. Since performing the original study, the authors have had the opportunity of examining 25 SQC samples obtained after microdissection of tumour specimens. This treatment ensures a tumour sample containing at least 50% of tumour cells. All of these samples showed allele loss on 3p (Table 2). Some of the analyses have been previously published [8, 9]. Two other studies identified 3p loss in all the SQC samples investigated [10, 11]. It can, therefore, be concluded that 3p loss occurs in almost all SQC and is thus a good candidate for involvement in preinvasive bronchial lesions. K-ras gene There is general agreement that K-ras mutations are an indicator of poor prognosis in lung cancer, but opinion has diverged concerning the correlation between mutation and histological subtype. Early studies indicated that K-ras mutation was confined to adenocarcinomas [12], but a recent study from Rosell et al. [13] found Kras mutations occurring with a higher frequency in SQC (eight patients out of 38) than in adenocarcinoma (three patients out of 22). If K-ras mutations occur in SQC they might also be detected in preinvasive lesions associated with SQC. Because of the restricted range of mutations in the K-ras gene (93% occur in codon 12), this would be a particularly useful marker for early disease detection. This prompted an examination of the frequency of K-ras mutations in a cohort of resected squamous cell carcinoma tumours. A total of 25 squamous cell carcinoma samples was analysed by nucleotide sequence analysis following polymerase chain reaction to amplify

Downloaded from http://annonc.oxfordjournals.org/ at Freie Universitaet Berlin on June 30, 2015

Total

Morphological abnormalities observed in bronchial epithelium

tation or gene amplification in oncogenes is also important in lung cancer development. Members of the myc family and Her-2/neu show deregulated expression, and point mutations have been detected in the K-ras gene [1]. From within this wide range of somatic genetic changes, three were chosen as possible candidates for involvement at the preinvasive stage of the disease.

29 Table 2. Genotype of squamous carcinomas at five loci on chromosome 3. Patient

Sample

D3S30

D3S2

D3F1S52

THRB (Dra 1)

THRB (Msp 1)

VS 15

N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N T

1,2 2 1,2 2 2

1

1,2 1 1,2 2 1,2 2 1,2 2 1,2

1

2

1

2

1

2

1,2 1 F 1 1,2 2 2

1,2 2 F

1,2 2 1,2 1,2 F

2

2

1

2

1

1,2 2 F

1,1 F

2

1

ND

ND

1

2

F

F 2

1,2 1 1,2 2 1,1

VS41 VS42 VS43 VS44 VS46 VS47 VS48

VS51 VS 55 VS 57 VS58 VS59 VS60 VS64 VS67 VS69 VS 71 VS 72 VS 110 VS 111 VS 112 VS 113 VS 115

1,2 2 1,2 1 1,2 1 1 1,2 1 1,2 2 1,2 1 1,1

1 1,2 1 F 1,2 1 F

1,2 2 1,2 2 1,2 1 1 1

1,1

1,2 1 1,2 1 2,2

1,2 1 1

1,2 2 2

ND 1,2 1 1,2 1 F

1,2 F 1,2 2 1,2 1 2

ND

2

2,2

ND ND

ND

1,2 2 2,2

1,2 2 1,2 F 1,2 1 ND

ND

2,2

2,2

1,2

2,2

2,2

1,2 1 1,2 1

1,2 2 2,2

2,2 1,2 1 1,2 2 1,2 2 F 1.2 F 1,2 1 ND 1,1

1

Somatic genetic changes in preinvasive bronchial lesions

1

1,2 1,2 F

1,2 F 1,2 2 ND

N = normal DNA; T - tumour DNA; ND - not done; F - failed.

relevant exons, but a mutation in codon 12 was detected in only one of these samples. Although more samples are required to exclude totally the possibility of SQC carrying K-ras mutations at the frequency reported by Rosell et al., these results, together with those from other laboratories, led to the conclusion that Kras mutations are uncommon in SQC and are, therefore, unlikely to be useful markers for preinvasive bronchial lesions. p53gene Unlike 3p loss, genetic damage to the p53 gene is not associated with every SQC of the lung. The degree of

In preliminary studies, severe dysplasia was used as a candidate preneoplastic lesion, and a number of samples were assessed for loss of heterozygosity on chromosome 3 and involvement of the p53 gene using immunohistochemistry. These analyses convincingly demonstrated that the same abnormalities as occurred in fully malignant tumours could also be detected at the preinvasive stage of the disease [8]. Three samples with severe dysplasia were obtained from patients without detectable invasive disease, and all three showed 3p allele loss and two of the three also had involvement of p53. More recently, bronchial lesions believed to represent earlier stages in the disease process have been examined and have also proved to carry somatic genetic changes; changes to chromosome 3 appear to precede p53 mutations (Chung and Rabbitts, unpublished results). Relationship between dysplasia and neoplasia in the lung

The evidence that bronchial dysplasia precedes neoplasia is strong but circumstantial and is most convincing for SQC, as mentioned above. The 'earlier' lesions might be expected to carry fewer genetic abnormalities than malignant tumours. In the authors' studies, only two candidate lesions were used, and for these the invasive tumours and preinvasive lesions from the same patient carried an identical genetic change. If a greater variety of markers had been used, it might have been possible to identify abnormalities that occur in tumours but not preinvasive lesions. An alternative to the 'candidate' approach is to examine the whole genome and compare the karyotype of malignant tumours with that of preinvasive lesions. This was achieved by obtaining lung tumour specimens at resection together with adjacent macroscopically normal bronchus. The bronchus was sectioned and each section divided to obtain a sample for histological diagnosis and a sample for short-term culture. It was anticipated that some of these sections would contain areas of dysplasia, and in this way a short-term culture of dysplastic epithelium would be obtained. The karyo-

Downloaded from http://annonc.oxfordjournals.org/ at Freie Universitaet Berlin on June 30, 2015

VS49

1,2 1 2

1,2 1 1,2 1 1,2 1 1,2 1 1,2 1 1

apparent involvement varies between studies, but at least 50% of SQCs have p53 mutations [1]. The advantages of p53 as a marker of premalignancy are that mutations can be detected directly by sequence analysis of DNA from a candidate morphological lesion, and that antibodies against the p53 protein can be used to deduce the presence of mutant p53. The latter approach can be used on fixed sections, allowing determination of somatic genetic changes without loss of tissue detail.

30

X

10

7

15

14

21

16

17

22

type of one such sample, which was successfully established in short-term culture, is shown in Fig. 1. This illustrates an almost normal human karyotype, but with a deletion in the short arm of chromosome 3 (del 3 p l 3 14). The parallel histology of this sample unexpectedly showed normal rather than dysplastic bronchial epithelium. Taken at face value, this suggests that normal bronchial epithelium can harbour cells that have already undergone somatic genetic changes. However, we cannot exclude the possibility that the sample used for short-term culture contained dysplastic epithelium which did not extend into the tissue retained for histological assessment. Without molecular analysis of a large proportion of the cells used to establish the shortterm culture, it is not possible to determine whether the loss of genetic material from 3p was a feature of most of the cells in the sample or only of those few that underwent division in short-term culture. The large cell carcinoma of this patient did not establish in shortterm culture, but in RFLP analysis it showed allele loss from chromosome 3. More recently, Sozzi et al. [14] have successfully established samples of severe bronchial dysplasia in short-term culture; again, pseudodiploid karyotypes were observed. Two cases carried 17pl3 deletions indicative of involvement of the p53 gene. This was confirmed for one of the samples by detection of a mutation by partial sequence analysis of the p53 gene [14]. Comparison of the pattern of 3p allele loss or p53 mutation in both dysplasia and tumour from the same

18

11

19

12

20

Fig. 1. Karyotype analysis of cells grown in short-term culture from a bronchial epithelium sample with normal histology. The only clonal chromosomal abnormality is deletion 3pl3-14.

patient almost always detected the same genotype. This may be because the samples of dysplasia used were sufficiently close to the tumour sample that they were likely to share a common clonal origin. Recently, we have studied the pattern of 3p loss in tumours and associated dysplasias of low grade. We have observed several examples in which the tumours show allele loss at all informative loci, whereas the early dysplasias only show partial loss (manuscript in preparation). If the tumours share the same clonal origin as the dysplasia, this suggests that damage to chromosome 3 may be sequential. It is a particularly striking feature of lung tumours that the genetic damage to chromosome 3 often takes the form of large deletions or even whole homologue loss. It has recently become clear that at least three separate regions of chromosome 3 are involved in malignant lung tumours, and it may be that more than one gene on chromosome 3 needs to be inactivated for full expression of malignancy, explaining why large deletions are so prevalent. The authors' observations, both molecular and cytogenetic, would be consistent with proximal loss as an early event, with loss of a more distal gene being required for development of the fully malignant phenotype. It is equally likely, however, that the dysplasia and tumour in the same patient were of independent origin, thus explaining their genotypic differences. Isolation of the genes on 3p involved in the development of lung cancer will be important tools in the examination of the relationship between dysplasia and tumour.

Downloaded from http://annonc.oxfordjournals.org/ at Freie Universitaet Berlin on June 30, 2015

13

6

31 Future prospects

References 1. Minna JD, Bader S, Bansel A et al. Molecular genetics of lung cancer: Multiple genetic lesions are involved in the pathogenesis of lung cancer. In Motla G (ed): Lung Cancer, Frontiers in Science and Treatment. Genoa: Grafica 1993; 25-47. 2. Saccomano G, Archer VE, Auerbach O, Saunder RP, Brennan LM. Development of carcinoma of the lung as reflected in exfoliated cells. Cancer 1974; 33: 256-70. 3. Gazdar AF. Molecular changes preceding the onset of invasive lung cancers. Lung Cancer 1994; 11 (Suppl 2): 16-17. 4. Whang-Peng J, Bunn PA, Kao-Shan CS et al. A non-random chromosomal abnormality, del 3p(14-23), in human small cell lung cancer (SCLC). Cancer Genet Cytogenet 1982; 6: 11934. 5. Brauch H, Johnson B, Hovis J et al. Molecular analysis of the short arm of chromosome 3 in small cell and non-small lung carcinoma of the lung. N Engl J Med 1987; 317:1109-13. 6. Hibi K, Takahashi K, Yamakawa R et al. Three distinct regions involved in 3p deletions in human lung cancer. Oncogene 1992; 7:445-9. 7. Rabbitts P, Douglas J, Daly M et al. Frequency and extent of allelic loss in the short arm of chromosome 3 in non-small cell lung cancer. Genes Chromosomes Cancer 1898; 1: 95-105. 8. Sundaresan V, Ganly P, Hasleton P et al. Paraffin wax-embedded material as a source of DNA for the detection of somatic genetic changes. J Pathol 1993; 169: 43-52. 9. Sundaresan V, Ganly P, Hasleton P et al. p53 and chromosome 3 abnormalities, characteristic of malignant lung tumours, are detectable in preinvasive lesions of the bronchus. Oncogene 1992; 7:1989-97. 10. Tsuchiya E, Nakamura Y, Weng SY et al. Allelotype of nonsmall cell lung carcinoma - comparison between loss of heterozygosity in squamous cell carcinoma and adenocarcinoma. Cancer Res 1992; 52: 2478-81. 11. Yokoyama S, Yamakawa K, Tsuchiya E et al. Deletion mapping on the short arm of chromosome 3 in squamous cell carcinoma and adenocarcinoma of the lung. Cancer Res 1992; 52: 873-7. 12. Slebos RJC, Kibbelaar EE, Dalesio O et al. K-ras oncogene activation as a prognostic marker in adenocarcinoma of the lung. N Engl J Med 1990; 323: 561-5. 13. Rosell R, Li S, Skacel Z et al. Prognostic impact of mutated Kras gene in surgically resected non-small cell lung cancer patients. Oncogene 1993; 8: 2407-12.

Correspondence to: Dr. P. Rabbitts MRC Clinical Oncology and Radiotherapeutics Unit MRC Centre Hills Road Cambridge CB2 2QH, U.K.

Discussion Professor Stan Kaye (University of Glasgow): I assume you were talking about lung cancer associated with smoking. Do you have data from non-smokers that allow you to say whether the chromosome 3p changes are also involved in non-smokers? Dr. Rabbitts: No. We have enough difficulty obtaining our samples without asking difficult questions like that. Professor Sidney Lowry (Belfast City Hospital): What you are describing appears to be a very early change on 3p and it seems to be more extensive the more advanced the malignancy. This is similar to what we have seen in ovarian cancer. Could we make the generalization that the greater the genetic damage, the greater the degree of malignancy? Dr. Rabbitts: I do not know if we can say that yet. Professor Sidney Lowry (Belfast City Hospital): We see loss of an entire chromosome; is that true of 3p? Dr. Rabbitts: It can be. We believe that the extent of damage to chromosome 3 in lung tumours indicates that several genes on chromosome 3 have to be knocked out to produce the full malignant potential; the most economical way to knock out several genes is to remove the whole chromosome. Dr. Nicholas Reed (Western Infirmary, Glasgow): Are these changes permanent? Would you expect to see devolution if a smoker stopped smoking? Dr. Rabbitts: The changes are permanent in these cells. When squamous metaplasias resolve the damaged cells disappear rather than reverting to normal. Dr. Nicholas Reed (Western Infirmary, Glasgow): In pre-malignant cervix you see a devolution back to normal cells, but that presumably is with dysplastic rather than pre-malignant changes. Dr. Rabbitts: Can you be sure that is what is happening? There could be re-population by normal cells after death of the dysplastic cells. In my opinion, a cell that has genetic loss cannot become a normal cell. Professor Jim Carmichael (City Hospital, Nottingham): You have detected these small deletions in the early, pre-malignant situations. Have you looked for them in smokers with apparently normal lungs?

Downloaded from http://annonc.oxfordjournals.org/ at Freie Universitaet Berlin on June 30, 2015

The failure of current treatments for lung cancer, which often presents with metastatic disease, emphasizes the need to detect the disease at an earlier stage, preferably before it becomes invasive. The observation of genetic changes in preinvasive lesions of the bronchus indicates a possible new approach to the detection of early disease. In the past, attempts have been made to identify malignant change in exfoliated material from patients at high risk of developing lung cancer. The attraction of using somatic genetic changes as targets for early disease screening is that they may be detected using molecular biological techniques that display high selectivity and sensitivity. A much more complete understanding of the natural history of lung cancer development and the relationship between dysplasia and invasive tumour, however, is required before this possibility could become reality.

14. Sozzi G, Miozzo M, Donghi R et al. Deletions of 17p and p53 mutations in preneoplastic lesions of the lung. Cancer Res 1992; 52: 6079-82.

32

tic damage to other chromosomes. As we chose this candidate approach we also tried to take a non-candidate approach, a more global approach, growing out Professor Hilary Calvert (Newcastle General Hospital): bronchial epithelium and looking at the cytogenetics; in One would expect the carcinogens from cigarette those circumstances we saw very little damage other smoke to be non-specific chemicals that would react than that to chromosome 3. Obviously there are probwith all chromosomes and all sections of DNA. How is lems in growing out epithelium, so we now plan to it that the chromosomal damage is consistently local- move away from a microdissection approach, to use ized to a few sites? interface cytogenetics, with regional chromosome Dr. Rabbitts:We cannot be sure that it is. We have inves- paints, to ask questions about single cells lining the tigated chromosome 3 and p53, but there may be gene- bronchus. Dr. Rabbitts: Again, we do not have the samples to do that.

Downloaded from http://annonc.oxfordjournals.org/ at Freie Universitaet Berlin on June 30, 2015