previously employed by other investigators (6-9). In the present study, the FEMX-I cells were cultured under conditions in which known growth or nutritional factors might have been growth limiting. Most of these factors are of low molecular weight and hence are not retained in dialyzation procedures. The possibility that such factors might have influenced our results seems unlikely because the dialyzation experiments indicated that the rat lung factors mediating the growth stimulation had an M, greater than 50 000. Recently, Nicolson (18) has found that highly metastatic cells respond differentially to organderived paracrine growth factors present in dissimilar amounts in different organs. One such factor, with potent effects on lung-colonizing cells, has an M, of 66 000 and may be related to transferrin. This factor is found in conditioned medium from lung tissues of several species, including rats and humans (19). Factors inhibiting cell growth have also been reported. Thus, Storm-Mathisen et al. (20) described factors of high molecular weight in conditioned medium from mouse lung that inhibited the clonogenic capacity of human hematopoietic cells. It is conceivable that the above-mentioned factors or related factors may mediate the effect on the FEMX-I cells in our systems and that the net response to growthstimulating or growth-inhibiting factors may be different for different tumors. The elucidation of such interactions may be important to help improve the understanding of human tumor metastasis formation.

References

normal and neoplastic cells in vitro. J Nad Cancer Inst 75:303-306, 1985 (7) HORAK E, DARLING DL, TARIN D: Analysis of

organ-specific effects on metastatic tumor formation by studies in vitro. JNCI 76:913-923, 1986 (8) NICOLSON GL, DULSKI KM: Organ specificity

of metastatic tumor colonization is related to organ-selective growth properties of malignant cells. Intj Cancer 38:289-294,1986 (9) NATTO S, GIAVAZZI R, FIDLER U: Correlation

between the in vitro interaction of tumor cells with an organ environment and metastatic behavior in vivo. Invasion Metastasis 7:16-29, 1987 (10)

FODSTAD 0 , KJ0NNIKSEN I, AAMDAL S , ET AL:

Extrapulmonary, tissue-specific metastasis formation in nude mice injected with FEMX-I human melanoma cells. Cancer Res 48:43824388, 1988 (//)

(6) SZANIAWSKA B, MAJEWSKI S, KAMINSKI MJ, ET

AL: Stimulatory and inhibitory activities of lung-conditioned medium on the growth of

1024

Robert J. C. Slebos* Ralph H. Hruban, Otilia Dalesio, Wolter J. Moot, G. Johan A. Offerhaus, Sjoerd Rodenhuis

KJ0NNIKSEN I, STORENG R, PlHL A , ET AL: A

human tumor lung metastasis model in athymic nude rats. Cancer Res 49:5148-5152, 1989 (12) KJ0NNIKSEN I, NESLAND JM, PIHL A, ET AL: A

nude rat model for studying metastasis of human tumor cells to bone and bone marrow. J Natl Cancer Inst 82:408-412, 1990 (13) COURTENAY VD, MILLS J: An in vitro colony

assay for human tumours grown in immunesuppressed mice and treated in vivo with cytotoxic agents. Br J Cancer 37:261-268, 1978 (14) TVEIT KM, FODSTAD 0 , PIHL A: The useful-

ness of human tumor cell lines in the study of chemosensitivity. A study of malignant melanomas. IntJ Cancer 28:403-408, 1981 (15) FOOSTAD 0 , AAMDAL S, PIHL A, ET AL: Ac-

tivity of mitozolomide (NSC 353451), a new imidazoltetrazine against xenografts from human melanomas, sarcomas, and lung and colon carcinomas. Cancer Res 45:1778-1786, 1985 (16) FOOSTAD 0 , AAMDAL S, MCMENAMIN M, ET

AL: A new experimental metastasis model in athymic nude mice, the human malignant melanoma LOX. Int J Cancer 41:442^*49, 1988 (17) PARTRIDGE CS, BODEN J, LEWIS JC, ET AL:

Choriocarcinoma xenografts in the nude rat. LabAnim 18:261-264, 1984 (IS) NICOLSON GL: Organ specificity of cancer metastasis is determined, in part, by tumor cell properties and cytokines expressed at particular organ sites. Proc Am Assoc Cancer Res 31:506-507,1990 (19) CAVANAUGH PG, NICOLSON GL: Purification

(/) LIOTTA LA: Mechanisms of cancer invasion and metastasis. In Important Advances in Oncology (DeVita VT Jr, Hellman S, Rosenberg S, eds). Philadelphia: Lippincott, 1985, pp 2 5 41 (2) HART IR: 'Seed and soil' revisited: Mechanisms of site-specific metastasis. Cancer Metastasis Rev 1:5-16, 1982 (3) NICOLSON GL: Cancer metastasis: Tumor cell and host organ properties important in metastasis to specific secondary sites. Biochim Biophys Acta 948:175-224, 1988 (4) ZETTER BR: The cellular basis of site-specific rumor metastasis. N Engl J Med 332:605-612, 1990 (5) PAGET S: The distribution of secondary growths in cancer of the breast. Cancer Metastasis Rev 8:98-101, 1989

Relationship Between K-ras Oncogene Activation and Smoking in Adenocarcinoma of the Human Lung

and some properties of a lung-derived growth factor that differentially stimulates the growth of tumor cells metastatic to the lung. Cancer Res 49:3928-3933,1989 (20) STORM-MATHISEN I, HELGESTAD J, LIE SO: In-

hibitory activity in mouse lung-conditioned medium studied in the agar assay for bone marrow colony-forming cell: Removal by vitamin C. Scand J Clin Lab Invest 45:67-76, 1985

To investigate a possible relationship between the exposure to tobacco smoke and the presence of ras point mutations, we examined lung adenocarcinoma samples from 27 smokers and from 27 nonsmokers. Activating point mutations in K-ras (also known as KRAS2) and N-ras (also known as NRAS) were determined by using the polymerase chain reaction and oligonucleotide hybridization to detect the mutated sequences. Mutations were more often found in adenocarcinomas obtained from smokers (eight of 27) than in adenocarcinomas obtained from nonsmokers (two of 27) (P = .044, Fisher's exact test). All mutations were present in K-ras codon 12. None of the other parameters examined differed significantly between the ras-positive and ras-negative groups. We conclude that exposure to carcinogenic agents in

Received February 22, 1991; revised April 15, 1991; accepted April 25, 1991. Supported by grants NKI 87-15 and NKI 91-08 from the Netherlands Cancer Society. Dr. Offerhaus is supported by the Netherlands Organization of Scientific Research and in part by the Netherlands Digestive Disease Foundation. R. J. C. Slebos, O. Dalesio, W. J. Mooi, S. Rodenhuis, Division of Experimental Therapy and Departments of Pathology and Medical Oncology, The Netherlands Cancer Institute, Amsterdam, The Netherlands. R. H. Hruban, G. J. A. Offerhaus, Department of Pathology, The Johns Hopkins University Hospital, Baltimore, Md. We thank Professor Dr. L. den Engelse for critically reading the manuscript. This study was based upon information from The Johns Hopkins Tumor Registry. Correspondence to: Robert J. C. Slebos, Ph.D., Division of Experimental Therapy H6, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands.

Journal of the National Cancer Institute

tobacco smoke is an important factor in the induction of point mutations in K-ras in human lung adenocarcinomas, but that K-ras mutations may also infrequently occur in tumors of nonsmokers. [J Natl Cancer Inst 83:1024-1027,1991]

The ras oncogenes have provided the first link between chemical carcinogens and oncogene activation (7). Three different ras oncogenes have been identified: Harvey-, Kirsten-, and N-ras. All code for three highly homologous 21-kd proteins (p21-ras) that can bind guanosine triphosphate and guanosine diphosphate and that possess guanosine triphosphatase activity with a putative role in cellular signal transduction (2). The oncogenic potential of the ras genes is unleashed by point mutations, which usually result in substitution of one of the amino acids at position 12, 13, or 61 of p21-ras (2). Activated ras oncogenes are frequently found in animal tumors induced by chemical carcinogens (J). In humans, mutational activation of Kras (also known as KRAS2) is commonly found in carcinomas of the pancreas and colon and in adenocarcinomas of the lung, while N-ras (also known as NRAS) oncogene activation is more often encountered in hematopoietic malignancies (4). We have previously shown that point mutations in K-ras codon 12 occur in about one third of lung adenocarcinomas, and we have speculated that these point mutations might result from direct mutation by carcinogenic ingredients of tobacco smoke (5,6). When the results from all our previous studies were combined, no point mutations in codon 12 of K-ras were found in 12 adenocarcinomas from nonsmokers, but 27 samples from 90 smoking patients were positive for this mutation, which is a statistically significant difference (P = .023, Fisher's exact test) (7). In these retrospective studies, however, the numbers of patients remained rather small because lung cancer is rare in nonsmokers. To confirm or dismiss the presumed strong association between smoking and ras mutations, we studied archival tumor samples from 54 patients, whose detailed smoking history was available, for point mutations in the K-ras and N-ras onVol. 83, No. 14, July 17, 1991

cogenes. The H-ras oncogene was not studied because point mutations in this gene are infrequent in human lung adenocarcinomas (6).

Patients and Methods Patient selection and clinical parameters. Thirty-five nonsmokers, who had primary adenocarcinoma of the lung diagnosed between 1980 and 1989, were selected from the files of The Johns Hopkins Tumor Registry. The archival material from six patients contained less than 10% tumor cells. Information available from the medical records and from contacting the patients or their relatives (possible in 22 out of 35 cases) revealed that two patients were ex-smokers. Each of the 27 remaining nonsmokers was matched by age, race, and sex with a smoker who had adenocarcinoma of the lung diagnosed during this same 19801989 period. In all cases, the patients were also matched for tumor type (bronchioloalveolar or not). The total group of 54 patients consisted of 44 women and 10 men, ranging in age from 40 to 87 (median 66) years at the time of diagnosis; 44 were Caucasian, and 10 were not (nine were black; one was Oriental). The average age difference between the matched sets of nonsmokers and smokers was 3 years. Matching for race was possible in all except two cases. The 27 patients with a history of smoking had smoked between 15 and 120 (median 48) pack-years (packs per day times the number of years the patient smoked). When available, the following parameters were collected from the medical records: family history, pack-years of smoking, years of smoking, cigarettes per day, occurrence of previous tumors of other origin, TNM class, tumor stage, histological classification, tumor differentiation, and age at first diagnosis of the disease. All pathology slides were reviewed by two diagnostic pathologists (R. H. Hruban and G. J. A. Offerhaus). Statistical analysis was performed using these parameters separately and in a multivariate analysis. Two-sided P values of less than .05 were considered statistically significant. Detection of point mutations in Kand N-ras. DNA was isolated from a 25\im section using a rapid lysis procedure

(5). The paraffin-embedded tumor sections were incubated in 250 \xL buffer, containing 10 nW Tris-HCl (pH 7.5), 50 mW KC1, 2.5 mW MgCl2, 0.45% Tween 80, 0.1 mg/mL gelatin, and 100 |ig/mL proteinase K, at 37 °C for 20-24 hours. The remaining debris was removed by centrifugation at 12 000 rpm for 3 minutes, and the supernatant was used directly for amplification in vitro (9). Conditions for the polymerase chain reaction and the detection of point mutations with mutation-specific oligonucleotides have been described previously (10). All samples were screened for point mutations in codons 12, 13, and 61 of the Kand N-ras oncogenes.

Results To investigate a possible relationship between smoking and the activation of ras oncogenes in human lung adenocarcinomas, we determined point mutations in the K- and N-ras oncogenes in tumors from smoking and nonsmoking patients. From the total group of 54 adenocarcinomas, 10 tested positive for a mutation in codon 12 of K-ras, while no other Kras or N-ras point mutations were detected. The results of an analysis for Kras codon 12 point mutations in 16 tumor specimens are shown in Fig. 1. Adenocarcinomas from two nonsmokers were positive for a point mutation in K-ras codon 12: the normal GGT sequence was mutated to CGT in one case and to GTT in the other, leading to a glycine to arginine and a glycine to valine amino acid substitution in the proteins, respectively. The glycine to arginine mutation in codon 12 of K-ras, while relatively frequent in pancreatic adenocarcinomas (77,72), has not been found previously in a primary neoplasm of the lung in humans. Eight of the 27 adenocarcinomas from smokers harbored a point mutation in codon 12 of K-ras: six had the sequence TGT, one had GTT, and one had GAT instead of the normal codon GGT, replacing the normal glycine with cysteine, valine, and aspartic acid, respectively. In the group of smokers, the frequency of the point mutation in K-ras codon 12 was 30% (eight of 27), while the incidence of the K-ras point mutation was 7% (two of 27) in the group of nonsmokers, a frequency that is significantly less than that found in REPORTS 1025

Table 1. Distribution of point mutations in codon 12 of K-ras between nonsmokers and smokers*

K12-gly Nonsmokers Smokers Total

K12-cvs

K-ras negative

K-ras positive

Total

25 19 44

2 8 10

27 27 54

•Fisher's exact test: P = .044.

Table 2. Distribution of point mutations in codon 12 of K-ras among nonsmokers, smokers who had smoked up to 50 pack-years, and smokers who had smoked over 50 pack-years* K-ras negative

K-ras positive

Total

25

2

27

11 8

4 4

15 12

44

10

54

Nonsmokers Smokers 15-50 pack-years 51-120 pack-years Total

K12-arg

K12-val

K12-asp

Fig. 1. Determination of point mutations in codon 12 of the K-ras oncogene. Eighteen amplified DNA fragments were dotted onto seven different membranes, and each membrane was hybridized to oligonucleotide probes with each detecting one of the possible K-ras codon 12 sequences. Shown are hybridizations with K12-gly, the wild-type glycine encoding (GGT) sequence; K12-cys, encoding cysteine (TGT); K12-arg, encoding arginine (CGT); and K12-val, encoding valine ( G i l ) at codon 12 of K-ras. The K-ras point mutation-positive samples from nonsmokers are in row 1, column 1 and in row 1, column 6. One of the K12-asp-positive samples came from a smoker who was matched with one of the ex-smokers who was excluded from the analysis. Control samples are in row 3, column 5 (lung adenocarcinoma cell line NCI-H23 with a glycine to cysteine mutation) and in row 3, column 6 (no DNA added).

smokers (P - .044, Fisher's exact test) (Table 1). One sample that was estimated to have at least 80% tumor cells gave a very weak signal with the wild-type K12gly probe, but did show hybridization with the mutation-detecting K12-cys probe (Fig. 1, row 2, column 6). This result strongly suggests that, in this sample, the tumor cells had lost their normal K-ras allele and that the detected wild-type signal could be attributed com102KS

pletely to the normal cells infiltrating into the tumor. If one considers the point mutation in K-ras codon 12 to be a direct result of the exposure to carcinogens in tobacco smoke, an increase in the frequency of K-ras point mutations in the group of heavy smokers might be expected. When all patients in our rather small group of 27 smokers were classified as either nonsmokers, as smokers having only a 15-50 pack-year history, or as smokers having a 51-120 pack-year history, a trend toward an increased incidence of ras activation emerged (P = .037, chi-square test for trend) (Table 2). This effect could, however, be attributed completely to the differences in mutation frequency between nonsmokers and smokers. Analysis of the other parameters revealed no significant differences between the patients with K-ras-positive and K-ras-negative adenocarcinomas with respect to age at diagnosis, sex, race, previous tumors of other origin, or tumor stage and differentiation (Table 3). In a multivariate analysis, all other parameters lost their predictive value after smoking was included in the model.

Discussion Epidemiological studies leave no doubt that most human lung cancers are caused by exposure to tobacco smoke (75). Tobacco smoke contains a large number of carcinogens, cocarcinogens, and tumor promoters, any of which could (in principle) be involved in the multistep process toward the development of a bronchial neoplasm. The oncogenes of the ras family have provided the first link between the action of carcinogens and the activation of oncogenes, which in the case of the ras oncogenes takes place by a point mutation in the protein-coding domain (7,5). Prompted by our previous observations, we initiated this study to

*Chi-square test for trend: P = .037. Table 3. Characteristics of 54 adenocarcinoma patients with or without a point mutation in K-ras codon 12 of their tumors* K-ras negative (N = 44)

K-ras positive (N = 10)

67(40-87)

62(44-73)

Sex Male Female

9 35

1 9

10 44

Race Caucasian Black Oriental

35 8 1

9 1 0

44 9 1

15

3 1 2 3 1

18 3 18 10 5

3 5 2 0

13 22 17 "%

8 0 0 0

40 4 1 3

2

6

Age, y. median (range)

Stage I II III IV Unstaged

t 16 7 4

Degree differentiated Poorly 10 Moderately 17 Well 15 2 Not determined Previous tumor of other origin None 32 4 Benign Hematopoietic 1 3 Adenocarcinoma 4 Epidermoid carcinoma

Total (N = 54)

•Statistically, none of these characteristics differed significantly between the two groups.

seek confirmation of the suspected linkage of K-ras mutation with exposure to tobacco smoke. We found a significantly higher incidence of K-ras point mutations in adenocarcinomas of the lung obtained from smokers (30%) than in adenocarcinomas of the lung obtained from nonsmokers (7%) (P - .044, Journal of the National Cancer Institute

Fisher's exact test). Thus, this independent series confirms the linkage. We have, however, now encountered the first two patients (in our experience) who did not smoke, but who did have a point mutation in their lung tumors. Although the role of tobacco smoke exposure in the induction of lung neoplasms in humans has been convincingly demonstrated, it has proven difficult to identify the specific components responsible (13,14). It is striking that exposure to such a cocktail of potentially mutagenic chemicals induces only one of a series of possible ras mutations: a point mutation in codon 12 of K-ras, mostly a guanine to thymine transversion (6,15). This fact could indicate that a single specific component of tobacco smoke is responsible for the mutations. An alternative explanation could be that only the K-ras codon 12 point mutation, and not the other types of ras activation, can contribute to carcinogenesis or tumor progression in bronchial epithelial cells. In human lung cancer, the K-ras codon 12 oncogene activation is almost exclusively found in the adenocarcinoma subtype, but virtually never in small-cell or epidermoid lung carcinomas (75). Even within codon 12 of K-ras, the guanine to thymine transversion in position 1 is by far the most frequent mutation found in lung adenocarcinomas (75) but not in colonic (16) or pancreatic adenocarcinomas (11,17). Also, in the p53 gene, a gene that is frequently altered in human lung tumors, guanine to thymine transversions are common point mutations (18). It is of interest that the two K-ras-positive adenocarcinomas from nonsmokers did not contain this common mutation. One of these adenocarcinomas contained a guanine to cytosine transver-

Vol. 83, No. 14, July 17, 1991

sion in position 1 of codon 12 of K-ras, a type of mutation that has thus far not been found in a primary human lung carcinoma. The other had a guanine to thymine transition in position 2 of K-ras codon 12. One possible explanation for this new type of mutation is that the guanine to thymine transversion is the result of one of the direct carcinogens in tobacco smoke and occurs early during tumorigenesis, while the other types of point mutations might arise as a result of exposure to other components. In the group of smokers, a doseresponse relationship between the level of tobacco smoke exposure and the frequency of the point mutation in codon 12 of K-ras was absent. This lack of a relationship suggests that the induction of point mutations in K-ras is only one step in the complex process of carcinogenesis. It is possible, and even likely, that other genetic alterations can be caused by tobacco carcinogens and that mutations in other genes may substitute for K-ras oncogene activation in tumorigenesis. Mutations in the putative effector molecule for the ras proteins or in genes that encode other guanosine triphosphate-binding proteins may be prime candidates for such a role. Further study is required for the identification of these substrates and of the carcinogens involved in their activation.

References

(5) RODENHU1S S , VAN DE WETER1NG M L , MOOI

WJ, ET AL: Mutational activation of the K-ras oncogene. A possible pathogenetic factor in adenocarcinoma of the lung. N Engl J Med 317:929-935,1987 (6) ROOENHUIS S, SLEBOS RJC, BOOT AJM, ET AU

Incidence and possible clinical significance of K-ras oncogene activation in adenocarcinoma of the human lung. Cancer Res 48:5738-5741, 1988 (7) RODENHUTS S, SLEBOS RJC, KIBBELAAR RE, ET

AL: Mutational activation of the Kirsten-ras oncogene is associated with early relapse and poor survival in adenocarcinoma of the lung. ProcASCO 9:881, 1990 (8) HIGUCHI R: Rapid, efficient DNA extraction for PCR from cells or blood. Perkin Elmer Cetus Amplifications 2:1-3, 1990 (9) MULLJS KB, FALOONA FA: Specific synthesis

of DNA in vitro via a polymerase-catalyzed chain reaction. Methods Enzymol 155:335350, 1987 (10) VERLAAN-DE VRIES M, BOGAARD ME, VAN

DEN ELST H, ET AL: A dot-blot screening procedure for mutated ras oncogenes using synthetic oligodeoxynucleoudes. Gene 50:313-320, 1986 ( / / ) SMIT VT, BOOT AJ, SMITS AM, ET AL: K-ras

codon 12 mutations occur very frequently in pancreatic adenocarcinomas. Nucleic Acids Res 16:7773-7782, 1988 (12) GRUNEWALD K, LYONS J, FROLJCH A, ET AL:

High frequency of Ki-ras codon 12 mutations in pancreatic adenocarcinomas. Int J Cancer 43:1037-1041,1989 (13) LOEB LA, ERNSTER VL, WARNER KE, ET AL:

Smoking and lung cancer An overview. Cancer Res 44:5940-5958, 1984 (14) HECHT SS, HOFFMANN D: Tobacco-specific

nitrosamines, an important group of carcinogens in tobacco and tobacco smoke. Carcinogenesis 9:875-884, 1988 (15) RODENHUIS S, SLEBOS RJC: The ras oncogenes

in human lung cancer. Am Rev Respir Dis 142:S27-S3O,1990 (16) VOGELSTEIN B, FEARON ER, HAMILTON SR, ET AL: Genetic alterations during colorectal-tumor development. N Engl J Med 319:525-532, 1988

(/) BARBACID M: Oncogenes and human cancer: Cause or consequence? Carcinogenesis 7:10371042, 1986 (2) BARBACID M: ras Genes. Annu Rev Biochem 56:779-827,1987

(17) ALMOGUERA C, SHIBATA D, FORRESTER K, ET

(3) GUERRERO I, PELLICER A: Mutational activa-

Mutations in the p53 gene are frequent in primary, resected non-small cell lung cancer. Lung Cancer Study Group. Oncogene 5:16031610,1990

tion of oncogenes in animal model systems of carcinogenesis. Mutat Res 185:293-308, 1987 (4) Bos JL: ras oncogenes in human cancer A review. Cancer Res 49:4682^689, 1989

AL: Most human carcinomas of the exocrine pancreas contain mutant c-K-ras genes. Cell 53:549-554, 1988 (18) CHIBA I, TAKAHASHI T, NAU MM, ET AL:

REPORTS 1027