Tobacco Documents Online

MULTIDISCIPLINARY CARE OF LUNG CANCER PATIENTS ~X529 HE~ATOL ONOOL OLIN N 97 (0]~ D SAUNDERS CO HILL PA MOLECULAR EVENTS IN LUNG CARCINOGENESIS York E. Miller, MD, and Wilbur A. Franklin, MD Lung cancer is the most common cause of cancer death and accounts for the most years of life lost from cancer in the United States. For these parameters, lung cancer is more important than the combination of breast, prostate, colon, and rectal cancer,v° In 1996, an estimated 177,000 new cases of lung cancer will have been diagnosed in the United States, and more than 158,000 individuals will die from lung cancer. Although death rates for men are currently decreasing, those for women are increasing sharply. Current trends in cigarette smoking among teenagers show an increase in smoking rates, emphasizing the continued importance of lung cancer, both now and in the future. Exposure to tobacco smoke through active smoking is the major factor leading to the development of lung cancer, causing 87% and 85% of cases in men and women, respectively. Additional environmental carcinogens that increase the risk for lung cancer include passive smoke exposure, asbestos, ionizing radiation, nitrogen mustard gas, arsenic, cadmium, chromium, nickel, acrylonitrile, chloromethyl ethers and vinyl chloride. The most important intervention in stopping the lung cancer epi- demic is to decrease the use of tobacco products. On the physician- patient level, smoking cessation is an effective intervention; however, now similar numbers of lung cancers are being diagnosed in ex-smokers This work was supported by NCI PS0 CA58187, NCI P30 CA46934, NHLBI RO1 HL45745 and a Merit Review Grant from the Department of Veterans Affairs. From the Divisions of Pulmonary and Critical Care Medicine and Medical Oncology, Veterans Affairs Medical Center (YEM), the University of Colorado Health Sciences Center, and the University of Colorado Cancer Center, Denver, Colorado HEMATOLOGY/ONCOLOGY CLINICS OF NORTH AMERICA VOLUME 11 • NUMBER 2 • APRIL 1997 215 THIS ARTICLE I5 FOR INDIVIDUAL USE ONLY AND MAY NOT BE FURTHER REPRODUCED OR STORED ELECTRONICALLY 14ITHOUT NRITTEN PERMISSION FROM THE COPYRIGHT HOLDER.~ UNAUTHORIZED REPRODUCTION MAY RESULT ~ IN FINANOIAL AND OTHER PENALTIES. ~/

MILLER & FRANKLIN and smokers, underscoring the need to prevent young persons from initially acquiring the addiction to tobacco. The most promising strate- gies to accomplish this goal are at the political and societal levels and include increased taxation, education, and counter-advertising. The tremendous economic power of the tobacco industry makes these changes very difficult. Even if smoking rates among teenagers were to dramatically decrease now, the problem of lung cancer would continue to persist in our society for decades. Because the survival for individuals in whom lung cancer is diag- nosed at an early, asymptomatic stage is better than for those in whom the diagnosis is based on symptoms, screening is a logical approach to decreasing the mortality from lung cancer. In the past, sputum cytology and chest radiographs have been evaluated,z~, ~, ~, 5~ The screening trials conducted in the 1970s and 1980s were not focused on what we now know to be a high-r,.'sk subset of smokers and did not find a benefit in terms of decreasing mortality from lung cancer. Additional screening modalities, such as spiral CT, early identification of malignant cells in sputum by computer-assisted image analysis, or monoclonal antibodies, are now generating interest as potentially useful in screening for lung cancer. A complementary and potentially more attractive strategy is to target premalignant respiratory epithelium, rather than established neoplasms, for early detection and treat~nent. To reliably identify individuals with prema- lignant respiratory epithelium at high risk for progression to lung cancer, an understanding of the preceding molecular events is needed. In addi- tion, new treatment modalities, perhaps based on genetic differences between normal and premalignant respiratory epithelium, are needed. MULTISTEP CARCINOGENESIS Cancer is in large part a mutational disorder and results from the progressive accumulation of mutations causing a loss of normal mechanisms of cellular growth control. This process has been elegantly demonstrated in colon carcinogenesis,s7 A series of genetic changes are found ~ premalignant colonic epithelium, including DNA methylation changes, chromosome 5q deletions, chromosome 18q deletions, and ras oncogene mutations. In the instances of chromosome 5q and 18q dele- tions, the tumor suppressor genes that are inactivated have been defined: the APC and DCC genes, respectively. Germline mutation of the APC gene results in familial adenomatous polyposis. Thus, a syndrome with a genetic predisposition to colon cancer is due to a germline mutation of a gene that is often somatically mutated in nonhereditary colonic neoplasia. Colon cancer has been an ideal and highly productive model for elucidating multistep carcinogenesis, because precursor lesions are endoscopically visible for biopsy and molecular analysis, and clearly defined hereditary cancer syndromes are described. The sequential genetic changes in lung cancer have only recently begun to be defined. Several factors have contributed to this delay: the o absence of a macros, the lack of widely ~ small cell lung canc~ and distal lung par~ familial lung cance~ chain reaction have sputum, demonstrat netic research to cli~ GENETIC SUSCEP" LUNG DISEASE An understand; to the development nonfamiiial cases. T complicated by. the disease expression ~ making colleq~ o Familial "$~r e$ recently, a segregat gene determining s if an individual sr predicted to be es disease (69% for pz for patients diagno useful for estimati specific (autosomal determine whether hereditary cases or each accounting for sis seems more lik~ including proto-on. for enzymes involv gens, any of which ity-depending on tl Smokers unqu~ lung disease, inclu~ nary disease) and/~ varies among smo apparent in a sub implicated genetic milial component i suspected.'~ Smoke from threefold to with smoker~a~,~'itl exposure.-~. ~ M known to contrib~

MOLECULAR EVENTS IN LUNG CARCINOGENESIS 217 )ung persons from ~t promising strate- and societal levels :er-advertising. The 'astry makes these ; teenagers were to :er would continue ~ng cancer is diag- for those in whom ogical approach to t, sputum cytology Pne screening trials on what we now ot find a benefit in tditional screening malignant ceils in oclonal antibodies, screening for lung strategy is to target ~h~izeoplasms, for dW with prema- ;ion to lung cancer, ~s needed. In addi- ~enetic differen, ces ium, are needed. and ~esults from a loss of normal ~as been elegantly ~netic changes are DNA methylation deletions, and ras 5q and 18q dele- ~ave been defined: Cation of the APC a syndrome with ;ermline mutation ~ereditary colonic productive model cursor lesions are lysis, and clearly ave only recently to this delay: the i absence of a macroscopically identifiable precursor lesion in the airway, the lack of widely accepted precursor lesions for adenocarcinoma and small cell lung cancer (SCLC), the relative inaccessibility of the bronchi and distal lung parenchyma, and the difficulties inherent in identifying familial lung cancer. Molecular diagnostic tests based on polymerase chain reaction have allowed the detection of mutated ras oncogenes in sputum, demonstrating one potential translation of basic molecular ge- netic research to clinical practice.~° GENETIC SUSCEPTIBILITY TO TOBACCO-INDUCED LUNG DISEASE An understanding of the mechanism of a rare genetic susceptibility to the development of malignancy often applies to the more common nonfamilial cases. The analysis of genetic susceptibility to lung cancer is complicated by the near-absolute requirement of cigarette smoking for disease expression and by the poor survival of the affected individuals, making collection of extended families for genetic analysis difficult. Familial aggregation of lung cancer has been des.cribed.~* More recently, a segregation analysis has supported the existence of a major gene determining susceptibility to lung cancer, which is penetrant only if an individual smokes7~ A genetic basis for lung cancer has been predicted to be especially frequent among patients with early-onset disease (69% for patients diagnosed under age 50 compared with 22% for patients diagnosed after age 70). Although segregation analysis is useful for estimating the fraction of cases caused by genes with a • specific (autosomal dominant) mode of inheritance, it is not possible to determine whether mutations at one locus account for the majority of hereditary cases or if many different lung cancer susceptibility loci exist, each accounting for a fraction of all hereditary cases. The latter hypothe- sis seems more likely, given the number and types of candidate genes, including proto-oncogenes, tumor suppressor genes, and genes coding for enzymes involved in the metabolism of procarcinogens and carcino- gens, any of which could have a major effect on lung cancer susceptibil- ity depending on the genotype of the carrier. Smokers unquestionably have an increased risk for tobacco-induced lung disease, including airways obstruction (chronic obstructive pulmo- nary disease) and/or lung cancer; however, susceptibility to lung disease varies among smokers, with a more rapid loss of pulmonary function apparent in a subset of 15% to 25%.7 Many studies have proven or implicated genetic factors influencing susceptibility, and a common fa- milial component in lung cancer and airways obstruction has long been suspected.~4 Smokers with airflow obstruction or chronic bronchitis have from threefold to fivefold increased rates of lung cancer as compared with smokers with normal pulmonary function and the same tobacco exposure.3~. 7~. ~4 Mutations causing alpha-1 antiprotease deficiency are kno~vn to contribute to emphysema; however, these mutations are re-

218 MILLER & FRANKLIN" sponsible for a small percentage of cases, and there is no evidence for an increased risk for lung cancer in these individuals. In addition, morphometric analysis of inflammatory cell infiltration in smokers with and without airway obstruction did not show differences, suggesting that genetic factors other than those influencing airway inflammation are responsible for the variation in susceptibility to tobacco-induced disease among smokers.4 Specific germline mutations have been impli- cated in lung cancer; but the mutations so far identified, notably in the p53 and Rb tumor suppressor genes, account for only a small fraction of lung cancer, and there is no evidence for increased risk of airways obstruction in gene carriers.37,55 Genes encoding enzymes that metabolize procarcinogens and carcinogens, including two p450 isozym.es, CYP2D6 and CYP1A1, the glutathione S-transferase locus, and DT-diaphorase, have been implicated in lung cancer susceptibility, but they are unlikely to account for the correlation between lung cancer and airways obstruc- tion and may be mbre specific to lung carcinogenesis.2, 8, 9, 18, 37, ~, 86 Recently, a preliminary report has suggested that individuals with high levels of bombesin-like peptides may be more susceptible, to tobacco- induced lung disease, should they smoke.53 Studies are underway to determine whether this trait is inherited. Clinical features (airflow obstruction, positive family history for lung cancer, particularly with young age of onset) that suggest a genetic susceptibility to lung cancer are easily determined and can be used to define high-risk groups for study of early diagnosis and intervention. An example of the power of using clinical features to select a high-risk group is a report that current or ex-smokers with airflow obstruction have a greater than 25% incidence of moderate or greater atypia on sputum cytology, compared with a less than 2.5% incidence of such atypia in the less-high-risk group studied in a screening study performed in the 1970s.38 This report underscores the need to target truly high-risk groups for the study of premalignant dysplasias, rather than middle- aged smokers, a lower-risk group. MOLECULAR GENETIC ANALYSIS OF LUNG CANCER There is considerable and growing evidence that molecular path- ways eventuating in SCLC and non-SCLC (NSCLC) are similar. SCLC and NSCLC share a number of features, although there are consistent molecular genetic differences between lung cancer cell types. In addition, a high degree of plasticity of cell type is observed among lung tumors. This includes the frequent occurrence of mixtures of squamous cell, small cell, and adenocarcinomas, the ability to alter lung cancer cell line phenotype by transfecting specific genes, and the ability of lung cancer cell lines to spontaneously change phenotype.16. ~ 49 Therefore, it is useful to consider genetic analysis of all major cell types of lung cancer to- gether, keeping in mind that characteristic differences between cell types occur. J I ] .! !- Autocrine Growth Much interest growth factors by factors that act thr~ receptors appear to stimulating tyrosin~ receptor family are bombesin-like pepti the most studied in mal growth factor, tion in NSCLC. Son and receptors for tht have demonstrated adapted to grow in additives, and that bombesin-like pepti conditions. Cell surf/~pe regulate the effects surface peptidase th. received particular by normal pulmon; undetectable levels Moreover, neutral e~. induced signal tran~ combinant neutral xenografts in vivo h~ is not likely a viable the principle that m the growth of prem~ Additional peptidas~ low or undetectable normal respiratory e Defined Genetic Ait A number of ch tumors and cell lin~ summarized in Tabk One emerging p alterations of protein sot gene is central i~ cycle. Rb under~oes inactive in ar~ag Most SCLC tu'~Srs mechanism for this i:

ce is no evidence for :iduals. In addition, tion in smokers .with fferences, suggesting ~irway inflammation to tobacco-induced ~ns have been impli- tiffed, notably in the ~nly a small fraction ~sed risk of airways ,'mes that metabolize ) isozymes, CYP2D6 and DT-diaphorase, ~ut they are unlikely ~nd airways obstruc- enesis.2, s, 9, 18, 37, ~, s~ dividuals with high ceptible to tobacco- ,~s are underway to ly history for tat'~ggest a genetic and can be used to is and intervention. to select a high-risk airflow obstruction c greater atypia on , incidence of such ng study performed rget truly high-risk ather than middle- ICER at molecular path- are similar. SCLC here are consistent t types. In addition, nong lung tumors. of squamous cell, ~ng cancer cell line lity of lung cancer ~erefore, it is useful of lung cancer to- nces between cell I ! i i l t MOLECULAR EVENTS IN LUNG CARCINOGENESIS 219 Autocrine Growth Factors Much interest has been focused on the production of autocrine growth factors by lung tumors and cell lines. Neuropeptide growth factors that act through G protein-coupled serpentine transmembrane receptors appear to be predominant in SCLC, whereas growth factors stimulating tyrosine kinase receptors of the epidermal growth factor receptor family are particularly expressed and active in NSCLC. The bombesin-like peptide family of neuropeptide growth factors have been the most studied in SCLC, and transforming growth factor alpha, epider- mal growth factor, and the heregulins have received the greatest atten- tion in NSCLC. Some NSCLC cell lines, however, express both ligands and receptors for the bombesin-like peptides.2. Siegfried and co~vorkersr4 have demonstrated that the A549 adenocarcinoma cell line' can be adapted to grow in RPMI 1640 medium alone without serum or other additives, and that the cell line upregulates gastrin-releasing peptide, a bombesin-like peptide most often associated with SCLC, under these conditions. Cell surface peptidases have the capability to degrade and thus regulate the effects of growth factors. Neutral endopeptidase, a cell surface peptidase that degrades a number of peptide growth factors, has received particular attention2z ~. 7~ Neutral endopeptidase is expressed by normal pulmonary epithelial cells, but it is expressed at low or undetectable levels by most SCLC and NSCLC cell lines and tumors. Moreover, neutral endopeptidase inhibition can increase neuropeptide- induced signal transduction and cell growth.~2. '~- Administration of re- combinant neutral endopeptidase to cell lines in vitro and to tumor xenografts in vivo has been demonstrated to slow growth; although this is not likely a viable therapeutic approach, it does demonstrate proof of the principle that manipulation of peptidase expression mav modulate the growth of premalignant and malignant respiratory epit[~elial cells2 Additional peptidases, including carboxypeptidase M, are expressed at low or undetectable levels in lung cancer, but they are expressed by normal respiratory epithelial cells and may be growth suppressors.'3 Defined Genetic Alterations in Lung Cancer A number 0f characteristic genetic changes can be detected in lung tumors and cell lines, with some specificity for cell type. These are summarized in Tables 1 and 2. One emerging pattern is a difference bet~veen SCLC and NSCLC in alterations of proteins that control the cell cycle. The Rb tumor suppres- sor gene is central in the control of a cell's progression through the cell cycle. Rb undergoes reversible phosphorylation; phosphorylated Rb is inactive in arresting the cell cycle and dephosphorylated Rb is active. Most SCLC tumors and cell lines do not express the Rb protein; the mechanism for this inactivation is most commonlv deletion or mutation

220 MILLER & FRANKLIN Table 1, CHROMOSOME DELETIONS AND TUMOR SUPPRESSOR GENE INACTIVATION IN LUNG CANCER Chromosomal Corresponding Tumor Deletion Suppressor Gene 3p Multiple candidates: SemalV and V, GNAI2, APH, ACY1, UBE1L, FHIT. VHL infrequently mutated 5q ? APC, MCC 9p INK p15, p16. Likely other genes also 11q Unknown 13q Rb (-100% SCLC, -20% NSCLC) 17p p53 (-90% SCLC, ~60% NSCLC) of the Rb gene.31, 32 On the other hand, Rb is expressed by most NSCLC; however, Rb is usually highly phosphorylated and inactive in NSCLC. Two synergistic mechanisms for this have been demonstrated. Cyclin D, which activates cyclin dependent kinases (cdk) 2, 4, and 6, is expressed at high levels, leading to Rb phosphorylation.68 The cdk inhibitors, p15 and p16, are often not expressed in NSCLC owing to deletion or methylation, further promoting Rb phosphorylation91 (summarized in Table 3). These divergent strategies for evading cell cycle control may have important ramifications for the design of new pharmaceutical ap- proaches to therapy; synthetic cdk inhibitors may be useful in NSCLC, but they would not be expected to be efficacious in SCLC. Another copsistent difference between SCLC and NSCLC is the presence of activating k-ras mutations. These are not reported in SCLC but occur in 25% to 50% of NSCLC. Of interest, transfection of activated ras into SCLC cell lines has been reported to alter differentiation, with the emergence of NSCLC features.= Chromosomal Alterations in Lung Cancer The karyotype of lung cancer cell lines and tumors is highly com- plex. In spite of much study, no consistent translocations have been described that might pinpoint the location of genes potentially important in lung cancer biology. Table 2. ONCOGENE MUTATIONS OR OVEREXPRESSION IN LUNG CANCER Oncogene SCLC NSCLC k-ras 0 30% to 50% activating mutations myc family Frequent overexpression Rare prad (cyclin D) 0 Frequent overexpression; inactivates Rb her2/neu 0 30% overexpression, gene amplification rare as mechanism kit Frequent overexpression Rare | 1 I .] ill Table 3. ALTERATIONS IN C LUNG CANCER Control Mechanism Rb tumor suppressor Cyclin D (prad oncogene) Kinase inhibitors (p15, p16) The most consisten of genetic material on this finding initially en, pressor genes in this rc the large region delet~ cases). Microcell-mediat some 3p into mouse A9 nicity; these results sup this region.2° In additiol deletions involvi1~ch~ regions, suggesti~tha on chromosome 3p.l~-21 deleted are much more erozygous deletion, mu being evaluated. The only complett suppressor gene on ch gene. Mutations affecti ~esting that it is not suppressor genes inclu kinase, and GNAI2 (a tially involved in the d and UBEtL) are encod low or undetectable le inappropriate prolongr has been demonstrated ,,.. ,.7 Missense point m~ been described in a S( codes a ubiquitin-degr 3p21.3, but it is over~ might act to retard intr~ ubiquination. The FHI |:t lit encodes a protei apparent mRNA isofo., merase chain reaction mutations is currently Frequent loss o~ been described, [ 1

V, GNAI2, APH, ~luently mutated | | | MOLECULAR EVENTS IN LUNG CARCINOGENESIS 221 Table 3. ALTERATIONS IN CELL CYCLE CONTROL MECHANISMS IN LUNG CANCER Control Mechanism ' SCLC NSCLC Rb tumor suppressor Absent Expressed, but inactivated by phosphorylation Cyclin D (prad oncogene) Not overexpressed Increased expression Kinase inhibitors (p15, p16) Detectable Absent or decreased ;ed by most NSCLC; inactive in NSCLC. ~onstrated. Cyclin D, . and 6, is expressed The cdk inhibitors, [~n~utO deletion or mmarized in e control may pharmaceutical ap- .e useful in NSCLC, SCLC. and NSCLC is .the ~t reported in SCLC sfection of activated :tifferentiation, ~vith nors is highly com- 9cations have been 9tentially important _UNG CANCER NSCLC activating mutations }verexpression; es Rb • xpression, gene ~tion rare as mechanism I The most consistent cytogenetic abnormality in lung cancer is a loss of genetic material on the short arm of chromosome 3?. ~6. ~4. ~8, ~8 Although this finding initially encouraged investigators to search for tumor sup- pressor genes in this region, progress has been slow, mainly owing to the large region deleted (approximately 1% of the genome in some cases). Microcell-mediated transfection of genetic material from chromo- some 3p into mouse A9 cells has resulted in cells with reduced tumorige- nicity; these results support the presence of a tumor suppressor gene in this region.~° In addition, several groups have documented homozygous deletions involving chromosome 3p12-3, 3p14.2, 3p21.31, and 3p21.33 regions, suggesting that more than one tumor suppressor gene resides on chromosome 3p.19-=~. aa. 4s, s6. 6~. ~s. 92 Because the regions homozygously deleted are much more informative than the previously described het- erozygous deletion, multiple candidate tumor suppressor genes are now being evaluated. The only completely characterized and generally accepted tumor suppressor gene on chromosome 3p is the von Hippel-Lindau (VHL) gene. Mutations affecting the VHL gene are rare in lung cancer, sug- gesting that it is not critical in pathogenesis,z° Candidate 3p tumor suppressor genes include Sema IV and V (semaphorins IV and V), 3p kinase, and GNAI2 (a G protein subunit).~s, ~9. ~s Three enzymes poten- tially involved in the degradation of intracellular proteins (APH, ACY1 and UBEIL) are encoded by chromosome 3p21-3 and are expressed at low or undetectable levels in SCLC; a tumor suppressor effect of the inappropriate prolongation of half-life of specific intracellular proteins has been demonstrated in other systems and is possible in SCLC.1~, 43. ~, 60, 6~ Missense point mutations involving the ACY1 coding region have been described in a SCLC cell line2~ The UNPH oncogene, which en- codes a ubiquitin-degrading enzyme, is also localized to chromosome 3p21.3, but it is overexpressed in SCLC; UNPH overexpression also might act to retard intracellular protein degradation by impeding protein ubiquination. The FHIT gene (3p14.2) has recently received attention,s~ FHIT encodes a protein with hydrolase activity and exhibits multiple apparent mRNA isoforms on RT-PCR. Whether or not these are poly- merase chain reaction artifacts or alternative splicing or inactivating mutations is currently under investigation. Frequent loss of heterozygosity involving chromosome 9p21 has been described f. -~ The p16 cdk inhibitor gene has been mapped to a

222 MILLER & FRANKLIN deleted region and is frequently inactivated in NSCLC, as is the p15 cdk inhibitor gene.+t Additional candidate regions have been shown to be homozygously deleted, suggesting that multiple tumor suppressor genes are encoded by chromosome 9p2+, ~9 Three regions on chromosome 11q frequently exhibit loss of hetero- zygosity in lung cancer.+4 Presumably, several tumor suppressor genes reside in these regions. IDENTIFICATION AND BIOLOGY OF PREMALIGNANT RESPIRATORY EPITHELIUM Classic studies by Saccomanno6+ in smoking uranium miners have demonstrated a progression of alterations in sputum cytology prior to the development of invasive carcinoma. For squamous cell carcinoma, a well-accepted progression of premalignant epithelial changes has been defined (Fig. 1). A histologic premalignant precursor lesion for adenocar- cinoma, atypical alveolar hyperplasia, has been described2~, 42 The corres- ponding sputum cytologic abnormalities for the latter have not been described. It is unlikely that cells from these small peripheral lesions would be shed into the sputum in detectable numbers. Atypical alveolar hyperplasia lesions are not routinely searched for in resection specimens, nor is the entity universally accepted as a precursor to adenocarcinoma. In patients with adenocarcinoma of the lung, coexistent squamous dys- plasias frequently occur, suggesting that a "field effect" exists. No pre- cursor lesion for SCLC has been accepted. The rare disorder, idiopathic diffuse hyperplasia of pulmonary neuroendocrine cells, has been de- scribed as including carcinoid tumors in its spectrum, but patients have been followed for many years without the development of SCLC or NSCLCY A variety of immunohistochemical techniques may be applied to the study of premalignant respiratory epithelium.2. Growth fraction, as measured by Ki-67 immunoreactivity, is often increased, even in histologically unremarkable tissue (Fig. 2). Transferrin receptor, a prolif- eration marker that is absent from normal epithelium, is often expressed by premalignant epithelial cells. Epidermal Growth Factor (EGF) recep- tor, which is found only on basilar cells in normal mucosa, is expressed throughout the full thickness of the bronchial mucosa in premalignant epithelium. Finally, neuroendocrine differentiation can be detected by immunostaining for markers such as chromogranin A, calcitonin gene- related peptide, bombesin, or neural cell adhesion molecule.+t Overex- pression of neuroendocrine markers is not well documented in premalig- nant epithelium, but preliminary results suggest that premalignant bronchial epithelial cells may overexpress one or more of these neuroen- docrine markers, p53 immunoreactivity is commonly used to °detect cells in which the p53 tumor suppressor gene is mutated. Because this technique is neither completely sensitive nor specific for p53 mutation,. confirmatory molecular genetic studies are needed. Figure 1. Morphological squ~ous carcinoma. On t~ bilayer of cells, the ill-define~ cytoplasm which underlie ar have basally oriented nuclei. with many different types of of an increase in the depth Squ~ous metaplasia occur flattened superficial epitheli~ degrees of atypia (moderat~ and reduction in nuclear c~ nuclear cytoplasmic ratio an, situ (lower strip of bronchial A major obstacle nant airway epitheliur biopsy. We have perfc high-risk individuals carcinoma), with the ~,~roups have found a chial biopsy, however been devel~ Controlled clir

MOLECULAR EVENTS IN LUNG CARCINOGENESIS 223 CLC, as is the p15 cdk ve been shown to be ~.mor suppressor genes exhibit loss of hetero- nor suppressor genes NANT Jranium miners have um cytology prior to tous cell carcinoma, a ial changes has been r lesion for adenocar- :ribed.39, 42 The corres- latter have not been d/ipheral lesions ers'~/~typical alveolar resection specimens, c to adenocarcinoma. stent squamous dys- ;fect" exists. No pre- ' disbrder, idiopathic cells, has been de- m, but patients have ~pment of SCLC or • may be applied to .24 Growth fraction, 'increased, even in • in receptor, a prolif- n, is often expressed Factor (EGF) recep- nucosa, is expressed 9sa in premaligna~mt can be detected by A, calcitonin gene- molecule.~ Overe\- ~ented in premalig- that premalignant re of these neuroen- nly used to detect Jtated. Because ttm:- c ~53 mutation, 1 I I Figure 1. Morphological changes in the bronchial mucosa thought to precede invasive squamous carcinoma. On the left (A) is histologically normal epithelium consisting of a bilayer of cells, the ill-defined lower layer of which is composed of small ceils with scanty cytoplasm which underlie and interdigitate with larger columnar mucociliary cells which have basally oriented nuclei. Reserve cell hyperplasia (B) is a frequent finding in patients with many different types of bronchial pathology, both benign and malignant, and consists of an increase in the depth and cellularity of the basal zone of the bronchial mucosa. Squamous metaplasia occurs when the mucociliary cell layer is completely replaced by flattened superficial epithelium. Squamous metaplasia may be accompanied by variable degrees of atypia (moderate atypia shown in C) characterized by nuclear pleomorphism and reduction in nuclear cytoplasmic ratio. Finally, severe nuclear abnormalities, high nuclear cytoplasmic ratio and poor maturation of epithelial cells characterize carcinoma in situ (lower strip of bronchial mucosa in D). A major obstacle to the understanding of the biology of premalig- nant airway epithelium has been the difficulty of identifying lesions for biopsy. We have performed blind bronchoscopic biopsies on a series of high-risk individuals (mostly current or ex-smokers with suspected lung carcinoma), with the identification of few histologic dysplasias. Other groups have found a significant yield of dysplasias on blind endobron- chial biopsy, however. Recently, an ultraviolet fluorescence device has been developed that may improve the detection of dysplastic lesions.~7 Controlled clinical trials are ongoing.

224 MILLER & FRANKLIN Figure 3. Hyperplastic bronchial epithelium (A) before and (B) after microdissection to remove epithelial cells from basement membrane (BM). Microdissected cells purified by microdissection are then used for genetic analysis. Figure 2. Se~ion of bronchial mucosa immunostained for Ki-67. Darkly staining nuclei indicate cycling cells and large numbers of such nuclei indicate a proliferation focus in the bronchial mucosa. These foci are not distinguishable without the aid of immunohistochemis- t~. MOLECULAR ANALY LESIONS Dysplastic squam epithelial cells. Biopsk epithelium but also connective tissue. The not either physically s, of the specimen or de~, of mutant and wild ty~ epithelium has been 3). The resulting DN,~ (PCR)-based methods. Several classes of Point tnutations affectir documented by a vari( sis of PCR products ( phism analysis of PCI in which mutat/ oc ing ras mutatio]"~C--al possible mutation site: at a molecular level b, in dysplastic tissue (I~ analyzing several alle! some of interest. To c( initially analyzed on informative. Fluoresce far information on in~ repeats may undergo the presence of multil- dysplastic DNA is ar repeats, a phenomen~ methylation, one mech, technique involving ct Rabbitts and colk methods to the analw tected loss of hetero~\ tissue found in proxi recently, Hung and co tion specimens. Dyspi chromosome 3p mar adjacent tumor. These that lung cancers aris one or more mutation heterozygosity of chr( tumor suppress/~l~en. have been infe.r~b, v lar means in dyspla~i.