- Email: [email protected]
RETRACTED: Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
Oral Oncology (2003) 39 497–505
www.elsevier.com/locate/oraloncology
D
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions§
TE
¨kib, Jan Erik Sondresenc, Wanja Kildald, Jon Sudbøa,*, Ari Ristima ¨rn Risbergb, Morten Boysene, Hanna S. Koppangf,g, Albrecht Reithd, Bjo Jahn M. Neslandb, Magne Brynec a
Department of Medical Oncology and Radiotherapy, The Norwegian Radium Hospital, Oslo 0310, Norway Department of Pathology, Helsinki University Central Hospital and Molecular and Cancer Biology Research Program, University of Helsinki, Finland c Department of Oral Biology, Oslo, Norway d Department of Pathology, The Norwegian Radium Hospital, Oslo, Norway e Department of Oto-Rhino-Laryngology, The National Hospital, Oslo, Norway f Department of Clinical Dentistry, University of Oslo, Oslo, Norway g Department of Pathology and Forensic Odontology, University of Oslo, Oslo, Norway
AC
b
KEYWORDS
Summary Emerging data indicate a link between genetic instability and up-regulation of cyclooxygenase-2 (COX-2). To see if individuals at high risk of oral cancer are candidates for treatment with selective COX-2 inhibitors (coxibs), levels of COX-2 expression in healthy, premalignant and cancerous oral mucosa were compared with the occurrence of DNA ploidy status as a genetic risk marker of oral cancer. COX-2 gene product was evaluated immunohistochemically in 30 healthy persons, in 22 patients with dysplastic lesions without previous or concomitant carcinomas, and in 29 patients with oral carcinomas. The immunohistochemical findings were verified by western blotting. COX-2 expression was correlated to DNA content as a genetic risk marker of oral cancer. COX-2 was up-regulated from healthy to premalignant to cancerous oral mucosa. Thus, COX-2 expression was found in 1 case of healthy oral mucosa (3%). All specimens from healthy mucosa had a normal DNA content. In patients with premalignancies. In 29 patients with oral carcinomas, cyclooxygenase-2 expression was observed in 26 (88%), and aneuploidy was observed in 25 cases (94%, P=0.04). Notably, of 22 patients with dysplastic lesions, COX-2 was exclusively expressed in a subgroup of nine patients (41%) identified to be at high risk of cancer by the aberrant DNA content of their lesions. Seven of these patients were followed for 5 years or more. An oral
R
ET
Oral precancer; Premalignancy; Oral carcinoma; Risk markers; Cyclooxygenase-2; Chemoprevention; Coxibs; DNA content; Aneuploidy
R
Received 13 January 2003; accepted 20 January 2003
§
None of the listed authors or acknowledged persons have any conflicts of interest, financially or personally, in relation to the data presented in the article. Presented in part at The Premier International Meeting on Cancer Prevention Research—Frontiers in Cancer Prevention Research, 14—18 October, 2002, Boston, and at the 14th EORTC-NCI-AACR Symposium on ‘‘Molecular Targets and Cancer Therapeutics’’, 19—22 November, 2002, Frankfurt, Germany. * Corresponding author. E-mail address: [email protected] (J. Sudbø). 1368-8375/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S1368-8375(03)00012-5
498
J. Sudbø et al. carcinoma developed in six of them (85%; P=0.02). These findings emphasize the need to determine whether coxibs can reduce the risk of oral cancer in patients with highrisk precancerous lesions. # 2003 Elsevier Science Ltd. All rights reserved.
TE
Patients with oral cancer have a poor prognosis because many have metastatic disease when a diagnosis is reached,13 and they are prone to second primary tumors of the upper aero-digestive tract.48 In developing countries, with limited resources in public health systems, head and neck cancer may account for as much as 50% of all cancers.9 Treating patients having oral precancers with cancer preventive agents is a possible lowcost approach towards an important global health problem that in western countries has proven difficult to control by cost-intensive surgery, radiotherapy, and chemotherapy.1,2,1015 Efficient chemopreventive intervention requires the identification of high-risk individuals and a plausible chemopreventive agent.16 Individuals with aberrant DNA content in precancerous oral white patches represent a subgroup of patients at a considerably increased risk of cancer.17,18 Recently, much attention has been paid to the role of cyxlooxygenase-2 (COX-2) in carcinogenesis.19,20 A large number of published epidemiologic studies provide data on the hypothesis that NSAIDs prevent colorectal cancer.21 COX-2 influences several processes important to cancer development, such as apoptosis, angiogenesis and invasiveness, although the relative importance of each of these effects in tumorigenesis is uncertain. The inducibility of COX-2 is explained, at least in part, by the presence of numerous cis-acting elements in the 50 -flanking region of the COX-gene, such as at the NF-B, NFIL6, and CRE sites.22 Increased stability of COX-2 mRNA also contributes to the induction of COX-2.23 When COX-2 is over-expressed in precancerous lesions, the levels of etheno-adducts are also increased, possibly as a result of increased oxidative stress.2426 Etheno-adduct may bind to DNA and give rise to permanent mutations in tissues if DNA repair mechanisms do not make complete DNA repairs during cell cycles.24,25,27 Permanent mutations by this mechanism is known to occur in headand neck cancers.28 DNA damage restricted to certain human genes may lead to aneuploidy,29 which has been found to be a driving force in several human malignancies.24,30 Previously published data indicate that up-regulation of COX-2 occurs in oral carcinogenesis.31 However, COX-2 expression has
previously not been correlated to other molecular prognostic markers in normal, dysplastic and cancerous oral mucosa. With the main purpose to see if individuals harbouring lesions at high risk of developing into oral cancer are candidates for chemopreventive treatment with coxibs, we investigated COX-2 expression in normal, dysplastic and cancerous oral mucosa, and then correlated these findings to DNA content as a risk molecular marker of oral cancer.
D
1. Introduction
2. Patients and methods
R
ET
R
AC
2.1. Tissue specimens
Tissue specimens from healthy oral mucosa were obtained by surgical resection from 25 healthy patients undergoing dental implant surgery. All patients gave informed consent to use the tissue specimens for research purposes. Archival material was obtained from 27 patients with who originally had received a diagnosis of epithelial dysplasias from white patches of the oral cavity, and from 29 patients with early stage oral squamous cell carcinomas. None of the patients with carcinoma had nodal or distal metastases (clinical staging T12N0M0). All histologic sections were evaluated by four separate pathologists according to guidelines given by WHO.32 Only cases in which complete agreement upon the diagnosis (normal, dysplastic, cancerous) was achieved among all four pathologist were included in the study. Clinical characteristics and follow-up data of the patients were obtained by chart review. All specimens from healthy oral mucosa were obtained from edentulous alveolar ridge, without any sign of mucosal inflammatory conditions, which would have precluded the insertion of a dental implant. This was done to avoid confounding staining from inflammatory conditions such as periodontitis. All specimens of dysplasias were obtained from leukoplakias from the floor of the mouth away from the alveolar ridge. Hence, COX-2 positivity is not likely to stem from inflammatory conditions such as periodontitis. For the same reason, all investigated carcinomas were taken from the lateral border of the tongue, which is the most prominent location of oral carcinomas.
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
D
Western blotting procedures were performed on frozen biopsy tissues from the same patients as the paraffin embedded tissue samples had been obtained from. From each biopsy, a small part was removed and instantly frozen. The presence of the COX-2 gene product was verified by immunoblotting according to a western blot procedure. Briefly, Cells were lysed in ice-cold NP-40 lysis buffer (1% NP40, 10% glycerol, 20 mM Tris—HCl pH 7.5, 137 mM NaCl, 100 mM sodium vanadate, 1 mM phenylmethy sulphonyl fluoride (PMSF) and 0.02 mg/ml each of aprotinin, leupeptin and pepstatin). Lysates were sonicated and clarified by centrifugation. Protein quantitation was done by Bradford analysis and 30 mg protein/lane was resolved by SDS—polyacrylamide gel electrophoresis (SDS— PAGE), and transferred to acetate film according to standard procedures. The presence of COX-2 protein was verified by immunostaining with a human monoclonal antibody towards COX-2 protein (160112, Cayman Chemicals Co., Ann Arbor, MI). Carcinomas with verified massive up-regulation of COX-2 were used as positive controls.
R
ET
R
AC
Paraffin sections (5 mm) were dried for 30 min at 60 C and incubated overnight at 37 C. The sections were dewaxed in xylene and hydrated through graded alcohol to distilled water before quenching in 0.3% H2O2 in methanol for 30 min. For antigen unmasking, sections were immersed in DAKO1 Target Retrieval Solution (DAKO A/S, Glostrup, Denmark) and micro-waved (MW) at approximately 650 W for 45 min with 3 min rest in between each MW-treatment. The specimens were cooled at room temperature for 20 min before equilibrating in phosphate-buffered saline (PBS), followed by application of 5% normal horse serum in PBS for 20 min. Immunostaining was performed with mouse monoclonal anti human COX-2 in a concentration of 1 mg/ml (160112; Cayman Chemical Company, Ann Arbor, MI, USA) in PBS containing 1% bovine serum albumin and 0.1% sodium azide at 4 C overnight. The following day, sections were covered with biotinylated secondary antibody, 1:200 horse anti mouse IgG (Vector Laboratories Inc., Burlingame, CA, USA), and the antibody— antigen reaction was finally developed by avidin— biotin peroxidase complex (ABComplex; Vectastain; Vector Laboratories Inc., Burlingame, CA, USA) and 3,30 diaminobenzidine (Sigma-Aldrich Corp., Stylus, MS, USA). The specificity of the antibody was determined by pre-adsorption of the primary antibody with human COX-2-control-peptide (1—10mg/ml, Cayman Chemical Company, Ann Arbor, MI, USA) for 1 h at room temperature prior to the staining procedure. a-Smooth muscle cell actin peptide (50 mg/ ml; DAKO A/S, Glostrup, Denmark) was used as a non-COX-2-peptide. Negative controls were either PBS or an irrelevant antibody with the same isotype as the primary antibody. For evaluation of the sections we used a Nikon Eclipse 800 microscope (Nikon Corporation, Tokyo, Japan) with a magnification of 200. Sections with staining in more than 1% of the cells were regarded as COX-2 positive, all other sections were regarded as negative. Only intracytoplasmatic staining within epithelial cells was regarded as COX-2 positivity. Care was taken to exclude the possibility of inflammatory cells mistakenly being identified as COX-2 positive epithelial cells. The sections were blindly scored by four independent observers (two of the authors; T.S. and M.Br., in addition to B.D. and B.R., see Acknowledgements), and the final score was reached as a consensus among these four observers.
2.3. Western blotting
TE
2.2. Quantitation of cyclooxygenase-2 expression
499
2.4. Measurement of DNA content as a risk marker of oral cancer
The methods for measuring DNA content have been presented elsewhere.17 Briefly, paraffinembedded tissue samples fixed in 4% buffered formaldehyde were sectioned and dysplastic areas were identified and retrieved with a microdissective technique. Two 50-mm sections were cut and enzymatically digested (type XXIV protease, Sigma Chemical, St. Louis) to yield isolated nuclei and subsequently a monolayer.33 Adjacent sections stained with hematoxylin and eosin were analyzed in order to verify the dysplastic content of each tissue sample. The DNA content of nuclei stained with Feulgen’s stain and periodic acid—Schiff stain was measured and analyzed with use of a specifically designed system for image analysis (Fairfield Imaging, Kent, United Kingdom), coupled to a Zeiss Axioplan II microscope (Zeiss, Oberkochen, Germany) with 40 lens, and 0.65 objective, and a final magnification of 1600. The number of homologous chromosomes was determined as a function of light absorption in the stained nuclear DNA, according to an established protocol.34
2.5. Statistical analysis Comparison of cyclooxygenase-2 expression in healthy, dysplastic and cancerous tissues and its
500
J. Sudbø et al.
3. Results 3.1. Patient characteristics
of 29 carcinomas (88%; P=0.04) (Table 2). There was no correlation between cyclooxygenase-2 expression levels and the histological grading of epithelial dysplasia (P=0.34) (Table 2). There was consensus between observers in all nine aneuploid cases with respect to quantification of cyclooxygenase-2 staining. Thus, the difference in negative versus moderate to strong cyclooxygenase-2 staining was unequivocal.
D
3.3. Cyclooxygenase-2 expression and its relation to DNA-content as a risk marker of cancer There was a signficant difference in the distribution DNA content between the the cyclooxygenase2 expressing and non-expressing lesions (Table 3). Of 30 cases with biopsies from healthy mucosa, all had a normal (diploid) DNA content, and in one specimen cyclooxygenase-2 expression was found. This case was, however, diagnosed as hyperplastic (Table 4). In 22 cases with dysplasia, nine patients had an aberrant (aneuploid), 12 had a normal (diploid) and one case was classified with an intermediate (tetraploid) DNA content (Table 4). Cyclooxygenase-2 expression was observed in lesions from 9 of the 22 patients with premalignant lesions, and exclusively in a subgroup of lesions with an aberrant (aneuploid) DNA content (Table 4). By contrast, none the 13 cases with diploid or tetraploid dysplasias showed cyclooxygenase-2 expression (P=0.02; Table 4). Of the 26 cyclooxygenase-2 positive carcinomas, 24 (92%) were classified as aneuploid and 2 (8%) were DNA tetraploid (Table 4).
TE
association with the DNA content was done by Fisher’s exact test, Student’s t-test, or the Wilcoxon—Mann—Whitney test, as appropriate. Disease-free and overall survival rates were computed by the Kaplan—Meier method, and the results compared with the use of the log-rank test. The logistic-regression analyses reflected the variables age, sex, DNA content, cyclooxygenase-2 expression levels, use of tobacco and alcohol, and use of non-steroidal anti-inflammatory drugs (NSAIDs) and coxibs. In a subanalysis, this model was applied to the patients with premalignant lesions. All P values were two-sided, and the threshold for significance was set at 0.05. All statistical analyses were performed with SPSS statistical software package (SPSS, Chicago).
R
ET
R
AC
Table 1 summarizes the main characteristics of the study subjects. None of the persons included in the study were under current treatment with a selective cyclooxygenase-2 inhibitor at the time of biopsy, or had been so within the previous 2 months (Table 1). None of the subjects had used a non-selective inhibitor of cyclooxygenase continuously for more than 3 weeks during the 3 years previous to the time when biopsies were obtained. There were no significant differences between the five groups of patients (with or without cyclooxygenase-2 expressing lesions, and with aneuploid, tetraploid or diploid lesions) with respect to clinical characteristics (Table 1). The age distribution among the persons with healthy oral mucosa (64.9 7.8 years) was not significantly different from the age distribution among patients with dysplasias or carcinomas (66.8 6.9 and 67.0 6.3 years, respectively; P=0.47). The median observation time of the patients with premalignant and malignant lesions was 43 months. After the histological re-evaluation, 5 of the 11 cases initially classified as mild dysplasia were typed as healthy or at least non-dysplastic. Hence, 30 cases were grouped as having healthy oral mucosa.
3.2. Cyclooxygenase-2 expression in healthy, premalignant and cancerous oral mucosa Weak staining was observed in one case of healthy oral mucosa (3%), as compared with strong staining in 9 of 22 dysplastic lesions (41%) and in 26
3.4. Cyclooxygenase-2 expression in oral dysplasias and its relation to grading of the lesions We found no significant correlation between the histological grading of the dysplasias and COX-2 expression, when the dysplasias were graded as mild, moderate and severe, according to guidelines given by WHO. The interobserver variation as calculated by kappa statistics was 0.53 (range 0.39— 0.63)
3.5. Disease-free survival in patients with dysplasia Of nine patients with cyclooxygenase-2 expressing precancers, seven patients had a follow-up time for more than 60 months. Six of these seven patients (85%) developed a carcinoma after a median observation time of 67 months (range 61—
Median age—yearc Male sex—no. (%) Tobacco use—no. (%) No history of use Former use Current use Cigarettesd Smokeless tobaccod No information available
Subjects with COX-2 negative biopsies (N=45)
Subjects with aneuploid biopsies (N=34)
65.9 7.8 28 (67)
66.6 6.8 27 (60)
66.8 7.0 22 (65)
23 (51) 8 (18) 8 (18) 7 (16) 3 (7) 6 (13)
16 (47) 6 (18) 8 (24) 6 (18) 4 (12) 4 (12)
1 1 1 1
(25) (25) (25) (25) 0 1 (25)
23 (53) 8 (19) 8 (19) 6 (14) 5 (12) 4 (9)
3 (7) 13 (29) 21 (47) 8 (18)
2 (6) 8 (24) 20 (58) 4 (12)
0 0 2 (50) 2 (50)
2 (5) 16 (37) 17 (40) 8 (19)
18 (40) 16 (36) 11 (24)
4 (12) 17 (50) 13 (38)
3 (75) 1 (25) 0
15 (35) 15 (35) 13 (30)
26 (76) 8 (24) 0
4 (100) 0 0
37 (86) 6 (14) 0
17 (47) 7 (19) 9 (25) 6 (17) 6 (17) 3 (8)
1 (3) 11 (30) 18 (50) 6 (17)
Use of NSAID-no. (%)f No history of use Previous useg Current use
4 (11) 17 (47) 15 (42)
b c d e f
26 (72) 10 (28) 0
Subjects with tetraploid biopsies (N=4)
Subjects with diploid biopsies (N=43)
65.8 6.3 4 (100)
65.8 6.0 29 (68)
41 (91) 4 (9) 0
Because of rounding, percentages may not total 100. COX-2 indicates cyclooxygenase-2. Plus—minus values are S.D. No differences between groups were statistically significant. Because of overlap in tobacco-habits, percentages total more than 100. Not quantified. NSAID denotes non steroidal anti-inflammatory drugs.
501
ET
a
R
Alcohol use—no. (%) No history of use Former use Current usee No information available
Use of coxibs—no. (%) No history of use Previous useg Current use
D
Subjects with COX-2 positive biopsiesb (N=36)
AC TE
Characteristics
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
Table 1 Characteristics of the 81 study subjectsa
(26) (6) (12) (9) (74)
R
9 2 4 3 25
(64) (29) (9) (9) (11) (7)
b
a
29 13 4 4 5 3
ET (3)d (25) (6) (11) (8) (72) 1 9 2 4 3 26 Healthy oral mucosac Dysplasia—no. (%) Mild Moderate Severe Carcinoma—no. (%)
Patients with COX-2 positive biopsiesb (N=36) Characteristics
R
D
70 months) (P=0.02). The remaining two patients have been observed for 17 and 23 months without developing a carcinoma. The 13 patients with diploid or tetraploid lesion that did not express cyclooxygenase-2 have not developed a carcinoma during a median follow-up time of 63 months (range 49—72 months). The specificity and sensitivity of DNA content as a prognostic marker in oral premalignancies was 85 and 97%, respectively. None of the patients diagnosed with oral premalignancies have developed a carcinoma in other sites of the upper aero-digestive tract during follow-up.
TE
3.6. Overall survival in patients with carcinoma A minimum of 5-years of follow-up was available for all the 29 patients with stage I—II carcinomas (median observation time 88 months, range 108—217 months). Most of the patients were treated with radical surgery and postoperative radiotherapy up to 50 Gy, with fractions of 2 Gy. Of the 26 patients that had cyclooxygenase-2 positive aneuploid carcinomas, 13 had a relapse during the observation time. The patient with the cyclooxygenase-2 negative and aneuploid carcinoma died from progression of disease after an observation time of 96 months. The other two cases have not had a relapse after initial therapy, and the patients with diploid and tetraploid lesions have been observed for 128 and 203 months, respectively.
AC
Because of rounding, percentages may not total 100. COX-2 indicates cyclooxygenase-2. c In addition to the 25 specimens from patients undergoing dental implant surgery, five cases were added after the histological classification was revised in 5 of the 11 cases with mild dysplasia. In these five cases, the typing of dysplasia was altered to a non-dysplastic, non-premalignant diagnosis. d This case was classified as hyperplasia and belonged to the group of five reclassified specimens, initially classified and graded as mild dysplasia.
0 1 (25) 0 1 (25) 0 3 (75) 0
30 12 4 3 5 1
Patients with tetraploid biopsies (N=4) Patients with aneuploid biopsies (N=34) Patients with COX-2 negative biopsies (N=45)
Table 2 Cyclooxygenase-2 expression and DNA content as related to histological typing and grading of oral mucosaa
(70) (28) (9) (7) (12) (2)
J. Sudbø et al.
Patients with diploid biopsies (N=43)
502
3.7. Multivariate analysis In univariate analysis, current use of tobacco, DNA content and cyclooxygenase-2 expression levels were significant prognostic factors (P=0.05, P=0.01, and P=0.03, respectively). However, when these three factors were fitted in a multiple-regression model applied to the complete set of data, only DNA content remained a significant prognostic factor (P=0.01), while cyclooxygenase-2 expression and the use of tobacco and were not (P=0.10 and P=0.07, respectively). In a corresponding univariate analysis, which only included the patients with premalignant lesions, current use of tobacco, DNA content an cyclooxygenase-2 expression levels were found to be significant prognostic factors (P=0.05, P=0.005, and P=0.01, respectively). In the following multivariate analysis, DNA content and cyclooxygenase-2 expression levels remained significant prognostic factors (P=0.01 and P=0.03, respectively).
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
503
Table 3 The DNA content and status of cyclooxygenase-2 expression in obtained tissue specimensa Characteristics
Patients with COX-2 positive biopsiesb (N=36)
Patients with COX-2 negative biopsies (N=45)
Diploid biopsies—no. (%) Tetraploid biopsies—no. (%) Aneuploid biopsies—no. (%)
1 (3) 2 (6) 33 (92)
42 (93) 2 (4) 1 (2)
a
Because of rounding, percentages may not total 100. In addition to the 25 specimens from patients undergoing dental implant surgery, five cases were added after the histological classification was revised in 5 of the 11 cases with mild dysplasia. In these five cases, the typing of dysplasia was altered to a non-dysplastic, non-premalignant diagnosis.
D
b
COX-2-positive biopsiesa
a
COX-2-negative bopsies
Aneuploid
Tetraploid
Diploid
Aneuploid
Tetraploid
Diploid
0 9 24
0 0 2
1c 0 0
0 0 1
0 1 1
29 12 1
AC
Healthy oral mucosab Dysplasia Carcinoma
TE
Table 4 Distribution of cyclooxygenase-2-expression and DNA content among individuals with healthy, premalignant and cancerous oral mucosa
4. Discussion
R
COX-2 indicates cyclooxygenase-2. In addition to the 25 specimens from patients undergoing dental implant surgery, five cases were added after the histological classification was revised in 5 of the 11 cases with mild dysplasia. In these five cases, the typing of dysplasia was altered to a non-dysplastic, non-premalignant diagnosis. c This case was classified as hyperplasia and belonged to the group of five reclassified specimens, initially classified and graded as mild dysplasia. b
R
ET
The novel finding of the present work is the link between overexpression of COX-2 and DNA aneuploidy in dysplastic oral lesion. We found that cyclooxygenase-2 was frequently expressed in oral premalignant and malignant lesions, but not in healthy mucosa. In premalignant lesions, cyclooxygenase-2 was exclusively expressed in lesions identified to be at considerably increased risk of carcinoma by their aberrant DNA content. It is known that COX-2 is up-regulated in oral cancers, but there is a lack of publications including data on the upregulation in oral precancers that have been prospectively followed over a longer period of time. Cyclooxygenase-2 is generally not found in healthy epithelial surfaces, but increases in response to mitogens, proinflammatory cytokines, and growth factors and has been linked to carcinogenesis.3537 The cancer-promoting effect of cyclooxygenase-2 may involve the formation of endogenous mutagens, immortalization of cells, promotion of local invasion and metastasis, or enhanced microcirculation.38 If the cancer preventive effect of coxibs were independent of epithelial cyclooxygenase-2
activity, elevated levels of cyclooxygenase-2 would not call for the use of coxibs. However, we found that cyclooxygenase-2 was only up-regulated in premalignant lesions with a particular high risk of malignant transformation, as given by their aberrant DNA content. This indicates that up-regulation of cyclooxygenase-2 is related to the early development of oral carcinomas. Accordingly, inhibition of cyclooxygenase-2 activity may revert the process leading to these cancers. Even if the primary up-regulation of cyclooxygenase-2 occurs in stromal cells, systemic treatment with a coxib would target these cell clones. The value of DNA content as a prognostic marker may lie in the fact that it is a harbinger of a multitude of early and significant events in cancer development,39 and its specificity and sensitivity as a prognostic marker of oral cancer found in this study (85 and 97%, respectively) confirms previous findings by us in a different group of patients.17 The fact that cyclooxygenase-2 expression was a significant prognostic marker only in patients with premalignant lesions, could reflect a cancer promoting effect of cyclooxygenase-2 in early stage of malignant transformation. Subsequent clonal
504
J. Sudbø et al. Acknowledgements
References
D
We are indebted to Inger Liv Nordli for preparation of tissue slides; Trond Storesund, Sadia Kahn, Saleh Ibrahimi, and Olav Schreurs for performing cyclooxygenase-2 staining of the tissue specimens; Ruth Punthervold and Signe Eastgate for preparation of monolayers for measuring DNA content; Professor Eva Skovlund for statistical support; and to Drs. Gary Kelloff and Steinar Aamdahl for helpful comments on the manuscript.
TE
expansions, which may be reflected in gross genomic aberrations, could render pathways other than cyclooxygenase-2 activity equally important for the propagation of disease.40 The finding that histological grading did not correlate to DNA ploidy status or COX-2 expression confirms previous studies by our group and others, where it is demonstrated that histological grading of oral dysplasias may have limited prognostic value.17,4143 Nevertheless, quantitative methods for histological evaluation of tissue architecture might reveal some correlation between the degree of dysplasia, COX-2 expression and DNA ploidy status.41,44 It is not known whether coxibs are most effective as cancer preventive drugs when used alone or in combination with other agents. Phase 3 trials have documented that retinoids can reverse malignant development in the oral mucosa,45 but these agents have adverse effects that limit their longterm use for cancer prevention in asymptomatic individuals.4547 One effect of retinoids is to inhibit cyclooxygenase-2. Thus, a recent study48 demonstrated epidermal growth factor (EGF)-mediated induction of cyclooxygenase-2 in oral squamous cell carcinoma cells, and the suppression of this upregulation by retinoids. A synergistic effect of retinoids and cyclooxygenase-2 inhibitors therefore seems reasonable.49,50 Whether a combination of coxibs, retinoids or other inhibitors of the EGF pathway could be used as cancer preventive drugs in doses that would not give significant side effects must be settled from a dose-ranging study. The follow-up time for the patients with precancerous lesions was limited to a median of less than 4 years, and only for 15 out of 22 cases did we have a follow-up time of more than 5 years. However, available data indicate that a majority of carcinomas arise between 2 and 3 years after initial diagnosis of premalignancy.17,40,51 All patients with premalignant lesions included in this study were observed for more than 2 years. Therefore, the data presented may be an accurate estimate of the rate of malignant transformation in patients with such lesions. Since cyclooxygenase-2 is not a specific marker of malignancy, treating all patients having cyclooxygenase-2-containing lesions with a coxib could lead to overtreatment. By contrast, our data indicate that including only patients with concomitant aneuploid and cyclooxygenase-2 expressing lesions would not lead to undertreatment. It may be that patients with aneuploid lesions without cyclooxygenase-2 activity are not candidates for treatment with a coxib. Nevertheless, our findings emphasize the need to determine whether treating patients at high risk of oral carcinomas with coxibs can reduce the risk of cancer.
R
ET
R
AC
1. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics 1999. CA Cancer J Clin 1999;49(1):8—31. 2. Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer statistics 2000. CA Cancer J Clin 2000;50:7—33. 3. Silverman SJ, Gorsky M. Epidemiologic and demographic update in oral cancer: California and national data—1973 to 1985. J Am Dent Assoc 1990;120:495—499. 4. Slaughter DP, Southwick HW, Smejkal W. Field cancerization in oral stratified squamous epithelium: clinical implications of multicentric origin. Cancer 1953;6:963—968. 5. Lippman SM, Hong WK. Second malignant tumors in head and neck squamous cell carcinoma: the overshadowing threat for patients with early-stage disease. Int J Radiat Oncol Biol Phys 1989;17:691—694. 6. Scully C, Porter S. ABC of oral health. Swellings and red, white, and pigmented lesions. BMJ 2000;321:225—228. 7. Scully C, Porter S. ABC of oral health. Oral cancer. BMJ 2000;321:97—100. 8. Boysen M, Loven JO. Second malignant neoplasms in patients with head and neck squamous cell carcinomas. Acta Oncol 1993;32:283—288. 9. Macfarlane GJ, Boyle P, Evstifeeva TV, Robertson C, Scully C. Rising trends of oral cancer mortality among males worldwide: the return of an old public health problem. Cancer Causes Control 1994;5:259—265. 10. Callery CD, Spiro RH, Strong EW. Changing trends in the management of squamous carcinoma of the tongue. Am J Surg 1984;148:449—454. 11. de Vries N, van Zandwijk N, Pastorino U. Chemoprevention in the management of oral cancer: EUROSCAN and other studies. Eur J Cancer B Oral Oncol 1992;28B:153—157. 12. Kelloff GJ, Boone CW, Steele VK, Perloff M, Crowell J, Doody LA. Development of chemopreventive agents for lung and upper aerodigestive tract cancers. J Cell Biochem Suppl 1993;17F:2—17. 13. Schantz SP, Ostroff JS. Novel approaches to the prevention of head and neck cancer. Proc Soc Exp Biol Med 1997;216: 275—282. 14. Kelloff GJ, Sigman CC, Greenwald P. Cancer chemoprevention: progress and promise. Eur J Cancer 1999;35:1755— 1762. 15. Kelloff GJ, Crowell JA, Steele VE, Lubet RA, Boone CW, Malone WA, et al. Progress in cancer chemoprevention. Ann N Y Acad Sci 1999;889:1—13. 16. Lee JJ, Hong WK, Hittelman WN, Mao L, Lotan R, Shin DM, et al. Predicting cancer development in oral leukoplakia: ten years of translational research. Clin Cancer Res 2000;6: 1702—1710. 17. Sudbø J, Kildal W, Risberg B, Koppang HS, Danielsen HE,
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
23.
24.
25.
26.
27.
28.
29. 30. 31.
R
32.
33.
34.
D
22.
TE
21.
35. DuBois RN, Abramson SB, Crofford L, Gupta RA, Simon LS, Van De Putte LB, et al. Cyclooxygenase in biology and disease. FASEB J 1998;12:1063—1073. 36. Herschman HR. Prostaglandin synthase 2. Biochim Biophys Acta 1996;1299:125—140. 37. Taketo MM. Cyclooxygenase-2 inhibitors in tumorigenesis (Part II). J Natl Cancer Inst 1998;90:1609—1620. 38. Li M, Wu X, Xu XC. Induction of apoptosis by cyclo-oxygenase-2 inhibitor NS398 through a cytochrome C-dependent pathway in esophageal cancer cells. Int J Cancer 2001; 93:218—223. 39. Ghadimi BM, Sackett DL, Difilippantonio MJ, Schrock E, Neumann T, Jauho A, et al. Centrosome amplification and instability occurs exclusively in aneuploid, but not in diploid colorectal cancer cell lines, and correlates with numerical chromosomal aberrations. Genes Chromosomes Cancer 2000;27:183—190. 40. Sudbø J, Kildal W, Johannessen AC, Koppang HS, Sudbø A, Danielsen HE, et al. Gross genomic aberrations in precancers: clinical implications of a long-term follow-up study in oral erythroplakias. J Clin Oncol 2002;20:456—462. 41. Karabulut A, Reibel J, Therkildsen MH, Praetorius F, Nielsen HW, Dabelsteen E. Observer variability in the histologic assessment of oral premalignant lesions. J Oral Pathol Med 1995;24:198—200. 42. Sudbø J, Bryne M, Johannessen AC, Kildal W, Danielsen HE, Reith A. Comparison of histological grading and large scale genomic status (DNA ploidy) as prognostic tools in oral dysplasia. Journal of Pathology 2001;194:303—310. 43. Sudbø J. Pathology in Disgrace? Journal of Pathology 2002; 196:244—245. 44. Sudbø J, Bankfalvi A, Bryne M, Marcelpoil R, Boysen M, Piffko J, et al. Prognostic value of graph theory-based tissue architecture analysis in carcinomas of the tongue. Lab Invest 2000;80:1881—1889. 45. Lotan R, Xu XC, Lippman SM, Ro JY, Lee JS, Lee JJ, et al. Suppression of retinoic acid receptor-beta in premalignant oral lesions and its up-regulation by isotretinoin. N Engl J Med 1995;332:1405—1410. 46. Lippman SM, Heyman RA, Kurie JM, Benner SE, Hong WK. Retinoids and chemoprevention: clinical and basic studies. J Cell Biochem Suppl 1995;22:1—10. 47. Lippman SM, Spitz MR, Huber MH, Hong WK. Strategies for chemoprevention study of premalignancy and second primary tumors in the head and neck. Curr Opin Oncol 1995;7: 234—241. 48. Mestre JR, Subbaramaiah K, Sacks PG, Schantz SP, Tanabe T, Inoue H, et al. Retinoids suppress epidermal growth factorinduced transcription of cyclooxygenase-2 in human oral squamous carcinoma cells. Cancer Res 1997;57:2890—2895. 49. Torrance CJ, Jackson PE, Montgomery E, Kinzler KW, Vogelstein B, Wissner A, et al. Combinatorial chemoprevention of intestinal neoplasia. Nat Med 2000;6:1024— 1028. 50. Gupta RA, DuBois RN. Combinations for cancer prevention. Nat Med 2000;6:974—975. 51. Sudbø J, Ried T, Bryne M, Kildal W, Danielsen HE, Reith A. Abnormal DNA content predicts the occurence of carcinomas in non-dysplastic oral white patches. Oral Oncol 2001; 37:558—565.
AC
20.
R
19.
ET
18.
Reith A. DNA content as a prognostic marker in patients with oral leukoplakias. N Engl J Med 2001;344: 1270—1278. Lippman SM, Hong WK. Molecular markers of the risk of oral cancer editorial. N Engl J Med 2001;344:1323—1326. Wong NA, Pignatelli M. Beta-catenin—a linchpin in colorectal carcinogenesis? Am J Pathol 2002;160:389—401. Dempke W, Rie C, Grothey A, Schmoll HJ. Cyclooxygenase2: a novel target for cancer chemotherapy? J Cancer Res Clin Oncol 2001;127:411—417. Vainio H. Is COX-2 inhibition a panacea for cancer prevention? Int J Cancer 2001;94:613—614. Inoue H, Yokoyama C, Hara S, Tone Y, Tanabe T. Transcriptional regulation of human prostaglandin-endoperoxide synthase-2 gene by lipopolysaccharide and phorbol ester in vascular endothelial cells. Involvement of both nuclear factor for interleukin-6 expression site and cAMP response element. J Biol Chem 1995;270:24965—24971. Sheng H, Shao J, Dixon DA, Williams CS, Prescott SM, DuBois RN, et al. Transforming growth factor-beta1 enhances Haras-induced expression of cyclooxygenase-2 in intestinal epithelial cells via stabilization of mRNA. J Biol Chem 2000; 275:6628—6635. Bartsch H. Studies on biomarkers in cancer etiology and prevention: a summary and challenge of 20 years of interdisciplinary research. Mutat Res 2000;462:25579. Nair J, Barbin A, Velic I, Bartsch H. Etheno DNA-base adducts from endogenous reactive species. Mutat Res 1999; 424:59—69. Bartsch H, Nair J. Ultrasensitive and specific detection methods for exocylic DNA adducts: markers for lipid peroxidation and oxidative stress. Toxicology 2000;153:105— 114. Bartsch H, Nair J. Exocyclic DNA adducts as secondary markers for oxidative stress: applications in human cancer etiology and risk assessment. Adv Exp Med Biol 2001;500: 675—686. Schmezer P, Rupprecht T, Tisch M, Maier H, Bartsch H. Laryngeal mucosa of head and neck cancer patients shows increased DNA damage as detected by single cell microgel electrophoresis. Toxicology 2000;144:149—154. Oberley TD. Oxidative damage and cancer. Am J Pathol 2002;160:403—408. Sen S. Aneuploidy and cancer. Curr Opin Oncol 2000;12: 82—88. Chan G, Boyle JO, Yang EK, Zhang F, Sacks PG, Shah JP, et al. Cyclooxygenase-2 expression is up-regulated in squamous cell carcinoma of the head and neck. Cancer Res 1999;59:991—994. Pindborg JJ, Reichart PA, Smith CJ, Waal Jvan der. Histological typing of cancer and precancer of the oral mucosa. 2nd ed. London: Springer; 1997. Hedley DW. DNA analysis from paraffin-embedded blocks. Methods Cell Biol 1994;41:231—240. Haroske G Giroud F, Reith A Bo ¨cking A. 1997 ESACP consensus report on diagnostic DNA image cytometry. Part I: basic considerations and recommendations for preparation, measurement and interpretation. European Society for Analytical Cellular Pathology Anal Cell Pathol 1998;17: 189—200.
505
www.elsevier.com/locate/oraloncology
D
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions§
TE
¨kib, Jan Erik Sondresenc, Wanja Kildald, Jon Sudbøa,*, Ari Ristima ¨rn Risbergb, Morten Boysene, Hanna S. Koppangf,g, Albrecht Reithd, Bjo Jahn M. Neslandb, Magne Brynec a
Department of Medical Oncology and Radiotherapy, The Norwegian Radium Hospital, Oslo 0310, Norway Department of Pathology, Helsinki University Central Hospital and Molecular and Cancer Biology Research Program, University of Helsinki, Finland c Department of Oral Biology, Oslo, Norway d Department of Pathology, The Norwegian Radium Hospital, Oslo, Norway e Department of Oto-Rhino-Laryngology, The National Hospital, Oslo, Norway f Department of Clinical Dentistry, University of Oslo, Oslo, Norway g Department of Pathology and Forensic Odontology, University of Oslo, Oslo, Norway
AC
b
KEYWORDS
Summary Emerging data indicate a link between genetic instability and up-regulation of cyclooxygenase-2 (COX-2). To see if individuals at high risk of oral cancer are candidates for treatment with selective COX-2 inhibitors (coxibs), levels of COX-2 expression in healthy, premalignant and cancerous oral mucosa were compared with the occurrence of DNA ploidy status as a genetic risk marker of oral cancer. COX-2 gene product was evaluated immunohistochemically in 30 healthy persons, in 22 patients with dysplastic lesions without previous or concomitant carcinomas, and in 29 patients with oral carcinomas. The immunohistochemical findings were verified by western blotting. COX-2 expression was correlated to DNA content as a genetic risk marker of oral cancer. COX-2 was up-regulated from healthy to premalignant to cancerous oral mucosa. Thus, COX-2 expression was found in 1 case of healthy oral mucosa (3%). All specimens from healthy mucosa had a normal DNA content. In patients with premalignancies. In 29 patients with oral carcinomas, cyclooxygenase-2 expression was observed in 26 (88%), and aneuploidy was observed in 25 cases (94%, P=0.04). Notably, of 22 patients with dysplastic lesions, COX-2 was exclusively expressed in a subgroup of nine patients (41%) identified to be at high risk of cancer by the aberrant DNA content of their lesions. Seven of these patients were followed for 5 years or more. An oral
R
ET
Oral precancer; Premalignancy; Oral carcinoma; Risk markers; Cyclooxygenase-2; Chemoprevention; Coxibs; DNA content; Aneuploidy
R
Received 13 January 2003; accepted 20 January 2003
§
None of the listed authors or acknowledged persons have any conflicts of interest, financially or personally, in relation to the data presented in the article. Presented in part at The Premier International Meeting on Cancer Prevention Research—Frontiers in Cancer Prevention Research, 14—18 October, 2002, Boston, and at the 14th EORTC-NCI-AACR Symposium on ‘‘Molecular Targets and Cancer Therapeutics’’, 19—22 November, 2002, Frankfurt, Germany. * Corresponding author. E-mail address: [email protected] (J. Sudbø). 1368-8375/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S1368-8375(03)00012-5
498
J. Sudbø et al. carcinoma developed in six of them (85%; P=0.02). These findings emphasize the need to determine whether coxibs can reduce the risk of oral cancer in patients with highrisk precancerous lesions. # 2003 Elsevier Science Ltd. All rights reserved.
TE
Patients with oral cancer have a poor prognosis because many have metastatic disease when a diagnosis is reached,13 and they are prone to second primary tumors of the upper aero-digestive tract.48 In developing countries, with limited resources in public health systems, head and neck cancer may account for as much as 50% of all cancers.9 Treating patients having oral precancers with cancer preventive agents is a possible lowcost approach towards an important global health problem that in western countries has proven difficult to control by cost-intensive surgery, radiotherapy, and chemotherapy.1,2,1015 Efficient chemopreventive intervention requires the identification of high-risk individuals and a plausible chemopreventive agent.16 Individuals with aberrant DNA content in precancerous oral white patches represent a subgroup of patients at a considerably increased risk of cancer.17,18 Recently, much attention has been paid to the role of cyxlooxygenase-2 (COX-2) in carcinogenesis.19,20 A large number of published epidemiologic studies provide data on the hypothesis that NSAIDs prevent colorectal cancer.21 COX-2 influences several processes important to cancer development, such as apoptosis, angiogenesis and invasiveness, although the relative importance of each of these effects in tumorigenesis is uncertain. The inducibility of COX-2 is explained, at least in part, by the presence of numerous cis-acting elements in the 50 -flanking region of the COX-gene, such as at the NF-B, NFIL6, and CRE sites.22 Increased stability of COX-2 mRNA also contributes to the induction of COX-2.23 When COX-2 is over-expressed in precancerous lesions, the levels of etheno-adducts are also increased, possibly as a result of increased oxidative stress.2426 Etheno-adduct may bind to DNA and give rise to permanent mutations in tissues if DNA repair mechanisms do not make complete DNA repairs during cell cycles.24,25,27 Permanent mutations by this mechanism is known to occur in headand neck cancers.28 DNA damage restricted to certain human genes may lead to aneuploidy,29 which has been found to be a driving force in several human malignancies.24,30 Previously published data indicate that up-regulation of COX-2 occurs in oral carcinogenesis.31 However, COX-2 expression has
previously not been correlated to other molecular prognostic markers in normal, dysplastic and cancerous oral mucosa. With the main purpose to see if individuals harbouring lesions at high risk of developing into oral cancer are candidates for chemopreventive treatment with coxibs, we investigated COX-2 expression in normal, dysplastic and cancerous oral mucosa, and then correlated these findings to DNA content as a risk molecular marker of oral cancer.
D
1. Introduction
2. Patients and methods
R
ET
R
AC
2.1. Tissue specimens
Tissue specimens from healthy oral mucosa were obtained by surgical resection from 25 healthy patients undergoing dental implant surgery. All patients gave informed consent to use the tissue specimens for research purposes. Archival material was obtained from 27 patients with who originally had received a diagnosis of epithelial dysplasias from white patches of the oral cavity, and from 29 patients with early stage oral squamous cell carcinomas. None of the patients with carcinoma had nodal or distal metastases (clinical staging T12N0M0). All histologic sections were evaluated by four separate pathologists according to guidelines given by WHO.32 Only cases in which complete agreement upon the diagnosis (normal, dysplastic, cancerous) was achieved among all four pathologist were included in the study. Clinical characteristics and follow-up data of the patients were obtained by chart review. All specimens from healthy oral mucosa were obtained from edentulous alveolar ridge, without any sign of mucosal inflammatory conditions, which would have precluded the insertion of a dental implant. This was done to avoid confounding staining from inflammatory conditions such as periodontitis. All specimens of dysplasias were obtained from leukoplakias from the floor of the mouth away from the alveolar ridge. Hence, COX-2 positivity is not likely to stem from inflammatory conditions such as periodontitis. For the same reason, all investigated carcinomas were taken from the lateral border of the tongue, which is the most prominent location of oral carcinomas.
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
D
Western blotting procedures were performed on frozen biopsy tissues from the same patients as the paraffin embedded tissue samples had been obtained from. From each biopsy, a small part was removed and instantly frozen. The presence of the COX-2 gene product was verified by immunoblotting according to a western blot procedure. Briefly, Cells were lysed in ice-cold NP-40 lysis buffer (1% NP40, 10% glycerol, 20 mM Tris—HCl pH 7.5, 137 mM NaCl, 100 mM sodium vanadate, 1 mM phenylmethy sulphonyl fluoride (PMSF) and 0.02 mg/ml each of aprotinin, leupeptin and pepstatin). Lysates were sonicated and clarified by centrifugation. Protein quantitation was done by Bradford analysis and 30 mg protein/lane was resolved by SDS—polyacrylamide gel electrophoresis (SDS— PAGE), and transferred to acetate film according to standard procedures. The presence of COX-2 protein was verified by immunostaining with a human monoclonal antibody towards COX-2 protein (160112, Cayman Chemicals Co., Ann Arbor, MI). Carcinomas with verified massive up-regulation of COX-2 were used as positive controls.
R
ET
R
AC
Paraffin sections (5 mm) were dried for 30 min at 60 C and incubated overnight at 37 C. The sections were dewaxed in xylene and hydrated through graded alcohol to distilled water before quenching in 0.3% H2O2 in methanol for 30 min. For antigen unmasking, sections were immersed in DAKO1 Target Retrieval Solution (DAKO A/S, Glostrup, Denmark) and micro-waved (MW) at approximately 650 W for 45 min with 3 min rest in between each MW-treatment. The specimens were cooled at room temperature for 20 min before equilibrating in phosphate-buffered saline (PBS), followed by application of 5% normal horse serum in PBS for 20 min. Immunostaining was performed with mouse monoclonal anti human COX-2 in a concentration of 1 mg/ml (160112; Cayman Chemical Company, Ann Arbor, MI, USA) in PBS containing 1% bovine serum albumin and 0.1% sodium azide at 4 C overnight. The following day, sections were covered with biotinylated secondary antibody, 1:200 horse anti mouse IgG (Vector Laboratories Inc., Burlingame, CA, USA), and the antibody— antigen reaction was finally developed by avidin— biotin peroxidase complex (ABComplex; Vectastain; Vector Laboratories Inc., Burlingame, CA, USA) and 3,30 diaminobenzidine (Sigma-Aldrich Corp., Stylus, MS, USA). The specificity of the antibody was determined by pre-adsorption of the primary antibody with human COX-2-control-peptide (1—10mg/ml, Cayman Chemical Company, Ann Arbor, MI, USA) for 1 h at room temperature prior to the staining procedure. a-Smooth muscle cell actin peptide (50 mg/ ml; DAKO A/S, Glostrup, Denmark) was used as a non-COX-2-peptide. Negative controls were either PBS or an irrelevant antibody with the same isotype as the primary antibody. For evaluation of the sections we used a Nikon Eclipse 800 microscope (Nikon Corporation, Tokyo, Japan) with a magnification of 200. Sections with staining in more than 1% of the cells were regarded as COX-2 positive, all other sections were regarded as negative. Only intracytoplasmatic staining within epithelial cells was regarded as COX-2 positivity. Care was taken to exclude the possibility of inflammatory cells mistakenly being identified as COX-2 positive epithelial cells. The sections were blindly scored by four independent observers (two of the authors; T.S. and M.Br., in addition to B.D. and B.R., see Acknowledgements), and the final score was reached as a consensus among these four observers.
2.3. Western blotting
TE
2.2. Quantitation of cyclooxygenase-2 expression
499
2.4. Measurement of DNA content as a risk marker of oral cancer
The methods for measuring DNA content have been presented elsewhere.17 Briefly, paraffinembedded tissue samples fixed in 4% buffered formaldehyde were sectioned and dysplastic areas were identified and retrieved with a microdissective technique. Two 50-mm sections were cut and enzymatically digested (type XXIV protease, Sigma Chemical, St. Louis) to yield isolated nuclei and subsequently a monolayer.33 Adjacent sections stained with hematoxylin and eosin were analyzed in order to verify the dysplastic content of each tissue sample. The DNA content of nuclei stained with Feulgen’s stain and periodic acid—Schiff stain was measured and analyzed with use of a specifically designed system for image analysis (Fairfield Imaging, Kent, United Kingdom), coupled to a Zeiss Axioplan II microscope (Zeiss, Oberkochen, Germany) with 40 lens, and 0.65 objective, and a final magnification of 1600. The number of homologous chromosomes was determined as a function of light absorption in the stained nuclear DNA, according to an established protocol.34
2.5. Statistical analysis Comparison of cyclooxygenase-2 expression in healthy, dysplastic and cancerous tissues and its
500
J. Sudbø et al.
3. Results 3.1. Patient characteristics
of 29 carcinomas (88%; P=0.04) (Table 2). There was no correlation between cyclooxygenase-2 expression levels and the histological grading of epithelial dysplasia (P=0.34) (Table 2). There was consensus between observers in all nine aneuploid cases with respect to quantification of cyclooxygenase-2 staining. Thus, the difference in negative versus moderate to strong cyclooxygenase-2 staining was unequivocal.
D
3.3. Cyclooxygenase-2 expression and its relation to DNA-content as a risk marker of cancer There was a signficant difference in the distribution DNA content between the the cyclooxygenase2 expressing and non-expressing lesions (Table 3). Of 30 cases with biopsies from healthy mucosa, all had a normal (diploid) DNA content, and in one specimen cyclooxygenase-2 expression was found. This case was, however, diagnosed as hyperplastic (Table 4). In 22 cases with dysplasia, nine patients had an aberrant (aneuploid), 12 had a normal (diploid) and one case was classified with an intermediate (tetraploid) DNA content (Table 4). Cyclooxygenase-2 expression was observed in lesions from 9 of the 22 patients with premalignant lesions, and exclusively in a subgroup of lesions with an aberrant (aneuploid) DNA content (Table 4). By contrast, none the 13 cases with diploid or tetraploid dysplasias showed cyclooxygenase-2 expression (P=0.02; Table 4). Of the 26 cyclooxygenase-2 positive carcinomas, 24 (92%) were classified as aneuploid and 2 (8%) were DNA tetraploid (Table 4).
TE
association with the DNA content was done by Fisher’s exact test, Student’s t-test, or the Wilcoxon—Mann—Whitney test, as appropriate. Disease-free and overall survival rates were computed by the Kaplan—Meier method, and the results compared with the use of the log-rank test. The logistic-regression analyses reflected the variables age, sex, DNA content, cyclooxygenase-2 expression levels, use of tobacco and alcohol, and use of non-steroidal anti-inflammatory drugs (NSAIDs) and coxibs. In a subanalysis, this model was applied to the patients with premalignant lesions. All P values were two-sided, and the threshold for significance was set at 0.05. All statistical analyses were performed with SPSS statistical software package (SPSS, Chicago).
R
ET
R
AC
Table 1 summarizes the main characteristics of the study subjects. None of the persons included in the study were under current treatment with a selective cyclooxygenase-2 inhibitor at the time of biopsy, or had been so within the previous 2 months (Table 1). None of the subjects had used a non-selective inhibitor of cyclooxygenase continuously for more than 3 weeks during the 3 years previous to the time when biopsies were obtained. There were no significant differences between the five groups of patients (with or without cyclooxygenase-2 expressing lesions, and with aneuploid, tetraploid or diploid lesions) with respect to clinical characteristics (Table 1). The age distribution among the persons with healthy oral mucosa (64.9 7.8 years) was not significantly different from the age distribution among patients with dysplasias or carcinomas (66.8 6.9 and 67.0 6.3 years, respectively; P=0.47). The median observation time of the patients with premalignant and malignant lesions was 43 months. After the histological re-evaluation, 5 of the 11 cases initially classified as mild dysplasia were typed as healthy or at least non-dysplastic. Hence, 30 cases were grouped as having healthy oral mucosa.
3.2. Cyclooxygenase-2 expression in healthy, premalignant and cancerous oral mucosa Weak staining was observed in one case of healthy oral mucosa (3%), as compared with strong staining in 9 of 22 dysplastic lesions (41%) and in 26
3.4. Cyclooxygenase-2 expression in oral dysplasias and its relation to grading of the lesions We found no significant correlation between the histological grading of the dysplasias and COX-2 expression, when the dysplasias were graded as mild, moderate and severe, according to guidelines given by WHO. The interobserver variation as calculated by kappa statistics was 0.53 (range 0.39— 0.63)
3.5. Disease-free survival in patients with dysplasia Of nine patients with cyclooxygenase-2 expressing precancers, seven patients had a follow-up time for more than 60 months. Six of these seven patients (85%) developed a carcinoma after a median observation time of 67 months (range 61—
Median age—yearc Male sex—no. (%) Tobacco use—no. (%) No history of use Former use Current use Cigarettesd Smokeless tobaccod No information available
Subjects with COX-2 negative biopsies (N=45)
Subjects with aneuploid biopsies (N=34)
65.9 7.8 28 (67)
66.6 6.8 27 (60)
66.8 7.0 22 (65)
23 (51) 8 (18) 8 (18) 7 (16) 3 (7) 6 (13)
16 (47) 6 (18) 8 (24) 6 (18) 4 (12) 4 (12)
1 1 1 1
(25) (25) (25) (25) 0 1 (25)
23 (53) 8 (19) 8 (19) 6 (14) 5 (12) 4 (9)
3 (7) 13 (29) 21 (47) 8 (18)
2 (6) 8 (24) 20 (58) 4 (12)
0 0 2 (50) 2 (50)
2 (5) 16 (37) 17 (40) 8 (19)
18 (40) 16 (36) 11 (24)
4 (12) 17 (50) 13 (38)
3 (75) 1 (25) 0
15 (35) 15 (35) 13 (30)
26 (76) 8 (24) 0
4 (100) 0 0
37 (86) 6 (14) 0
17 (47) 7 (19) 9 (25) 6 (17) 6 (17) 3 (8)
1 (3) 11 (30) 18 (50) 6 (17)
Use of NSAID-no. (%)f No history of use Previous useg Current use
4 (11) 17 (47) 15 (42)
b c d e f
26 (72) 10 (28) 0
Subjects with tetraploid biopsies (N=4)
Subjects with diploid biopsies (N=43)
65.8 6.3 4 (100)
65.8 6.0 29 (68)
41 (91) 4 (9) 0
Because of rounding, percentages may not total 100. COX-2 indicates cyclooxygenase-2. Plus—minus values are S.D. No differences between groups were statistically significant. Because of overlap in tobacco-habits, percentages total more than 100. Not quantified. NSAID denotes non steroidal anti-inflammatory drugs.
501
ET
a
R
Alcohol use—no. (%) No history of use Former use Current usee No information available
Use of coxibs—no. (%) No history of use Previous useg Current use
D
Subjects with COX-2 positive biopsiesb (N=36)
AC TE
Characteristics
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
Table 1 Characteristics of the 81 study subjectsa
(26) (6) (12) (9) (74)
R
9 2 4 3 25
(64) (29) (9) (9) (11) (7)
b
a
29 13 4 4 5 3
ET (3)d (25) (6) (11) (8) (72) 1 9 2 4 3 26 Healthy oral mucosac Dysplasia—no. (%) Mild Moderate Severe Carcinoma—no. (%)
Patients with COX-2 positive biopsiesb (N=36) Characteristics
R
D
70 months) (P=0.02). The remaining two patients have been observed for 17 and 23 months without developing a carcinoma. The 13 patients with diploid or tetraploid lesion that did not express cyclooxygenase-2 have not developed a carcinoma during a median follow-up time of 63 months (range 49—72 months). The specificity and sensitivity of DNA content as a prognostic marker in oral premalignancies was 85 and 97%, respectively. None of the patients diagnosed with oral premalignancies have developed a carcinoma in other sites of the upper aero-digestive tract during follow-up.
TE
3.6. Overall survival in patients with carcinoma A minimum of 5-years of follow-up was available for all the 29 patients with stage I—II carcinomas (median observation time 88 months, range 108—217 months). Most of the patients were treated with radical surgery and postoperative radiotherapy up to 50 Gy, with fractions of 2 Gy. Of the 26 patients that had cyclooxygenase-2 positive aneuploid carcinomas, 13 had a relapse during the observation time. The patient with the cyclooxygenase-2 negative and aneuploid carcinoma died from progression of disease after an observation time of 96 months. The other two cases have not had a relapse after initial therapy, and the patients with diploid and tetraploid lesions have been observed for 128 and 203 months, respectively.
AC
Because of rounding, percentages may not total 100. COX-2 indicates cyclooxygenase-2. c In addition to the 25 specimens from patients undergoing dental implant surgery, five cases were added after the histological classification was revised in 5 of the 11 cases with mild dysplasia. In these five cases, the typing of dysplasia was altered to a non-dysplastic, non-premalignant diagnosis. d This case was classified as hyperplasia and belonged to the group of five reclassified specimens, initially classified and graded as mild dysplasia.
0 1 (25) 0 1 (25) 0 3 (75) 0
30 12 4 3 5 1
Patients with tetraploid biopsies (N=4) Patients with aneuploid biopsies (N=34) Patients with COX-2 negative biopsies (N=45)
Table 2 Cyclooxygenase-2 expression and DNA content as related to histological typing and grading of oral mucosaa
(70) (28) (9) (7) (12) (2)
J. Sudbø et al.
Patients with diploid biopsies (N=43)
502
3.7. Multivariate analysis In univariate analysis, current use of tobacco, DNA content and cyclooxygenase-2 expression levels were significant prognostic factors (P=0.05, P=0.01, and P=0.03, respectively). However, when these three factors were fitted in a multiple-regression model applied to the complete set of data, only DNA content remained a significant prognostic factor (P=0.01), while cyclooxygenase-2 expression and the use of tobacco and were not (P=0.10 and P=0.07, respectively). In a corresponding univariate analysis, which only included the patients with premalignant lesions, current use of tobacco, DNA content an cyclooxygenase-2 expression levels were found to be significant prognostic factors (P=0.05, P=0.005, and P=0.01, respectively). In the following multivariate analysis, DNA content and cyclooxygenase-2 expression levels remained significant prognostic factors (P=0.01 and P=0.03, respectively).
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
503
Table 3 The DNA content and status of cyclooxygenase-2 expression in obtained tissue specimensa Characteristics
Patients with COX-2 positive biopsiesb (N=36)
Patients with COX-2 negative biopsies (N=45)
Diploid biopsies—no. (%) Tetraploid biopsies—no. (%) Aneuploid biopsies—no. (%)
1 (3) 2 (6) 33 (92)
42 (93) 2 (4) 1 (2)
a
Because of rounding, percentages may not total 100. In addition to the 25 specimens from patients undergoing dental implant surgery, five cases were added after the histological classification was revised in 5 of the 11 cases with mild dysplasia. In these five cases, the typing of dysplasia was altered to a non-dysplastic, non-premalignant diagnosis.
D
b
COX-2-positive biopsiesa
a
COX-2-negative bopsies
Aneuploid
Tetraploid
Diploid
Aneuploid
Tetraploid
Diploid
0 9 24
0 0 2
1c 0 0
0 0 1
0 1 1
29 12 1
AC
Healthy oral mucosab Dysplasia Carcinoma
TE
Table 4 Distribution of cyclooxygenase-2-expression and DNA content among individuals with healthy, premalignant and cancerous oral mucosa
4. Discussion
R
COX-2 indicates cyclooxygenase-2. In addition to the 25 specimens from patients undergoing dental implant surgery, five cases were added after the histological classification was revised in 5 of the 11 cases with mild dysplasia. In these five cases, the typing of dysplasia was altered to a non-dysplastic, non-premalignant diagnosis. c This case was classified as hyperplasia and belonged to the group of five reclassified specimens, initially classified and graded as mild dysplasia. b
R
ET
The novel finding of the present work is the link between overexpression of COX-2 and DNA aneuploidy in dysplastic oral lesion. We found that cyclooxygenase-2 was frequently expressed in oral premalignant and malignant lesions, but not in healthy mucosa. In premalignant lesions, cyclooxygenase-2 was exclusively expressed in lesions identified to be at considerably increased risk of carcinoma by their aberrant DNA content. It is known that COX-2 is up-regulated in oral cancers, but there is a lack of publications including data on the upregulation in oral precancers that have been prospectively followed over a longer period of time. Cyclooxygenase-2 is generally not found in healthy epithelial surfaces, but increases in response to mitogens, proinflammatory cytokines, and growth factors and has been linked to carcinogenesis.3537 The cancer-promoting effect of cyclooxygenase-2 may involve the formation of endogenous mutagens, immortalization of cells, promotion of local invasion and metastasis, or enhanced microcirculation.38 If the cancer preventive effect of coxibs were independent of epithelial cyclooxygenase-2
activity, elevated levels of cyclooxygenase-2 would not call for the use of coxibs. However, we found that cyclooxygenase-2 was only up-regulated in premalignant lesions with a particular high risk of malignant transformation, as given by their aberrant DNA content. This indicates that up-regulation of cyclooxygenase-2 is related to the early development of oral carcinomas. Accordingly, inhibition of cyclooxygenase-2 activity may revert the process leading to these cancers. Even if the primary up-regulation of cyclooxygenase-2 occurs in stromal cells, systemic treatment with a coxib would target these cell clones. The value of DNA content as a prognostic marker may lie in the fact that it is a harbinger of a multitude of early and significant events in cancer development,39 and its specificity and sensitivity as a prognostic marker of oral cancer found in this study (85 and 97%, respectively) confirms previous findings by us in a different group of patients.17 The fact that cyclooxygenase-2 expression was a significant prognostic marker only in patients with premalignant lesions, could reflect a cancer promoting effect of cyclooxygenase-2 in early stage of malignant transformation. Subsequent clonal
504
J. Sudbø et al. Acknowledgements
References
D
We are indebted to Inger Liv Nordli for preparation of tissue slides; Trond Storesund, Sadia Kahn, Saleh Ibrahimi, and Olav Schreurs for performing cyclooxygenase-2 staining of the tissue specimens; Ruth Punthervold and Signe Eastgate for preparation of monolayers for measuring DNA content; Professor Eva Skovlund for statistical support; and to Drs. Gary Kelloff and Steinar Aamdahl for helpful comments on the manuscript.
TE
expansions, which may be reflected in gross genomic aberrations, could render pathways other than cyclooxygenase-2 activity equally important for the propagation of disease.40 The finding that histological grading did not correlate to DNA ploidy status or COX-2 expression confirms previous studies by our group and others, where it is demonstrated that histological grading of oral dysplasias may have limited prognostic value.17,4143 Nevertheless, quantitative methods for histological evaluation of tissue architecture might reveal some correlation between the degree of dysplasia, COX-2 expression and DNA ploidy status.41,44 It is not known whether coxibs are most effective as cancer preventive drugs when used alone or in combination with other agents. Phase 3 trials have documented that retinoids can reverse malignant development in the oral mucosa,45 but these agents have adverse effects that limit their longterm use for cancer prevention in asymptomatic individuals.4547 One effect of retinoids is to inhibit cyclooxygenase-2. Thus, a recent study48 demonstrated epidermal growth factor (EGF)-mediated induction of cyclooxygenase-2 in oral squamous cell carcinoma cells, and the suppression of this upregulation by retinoids. A synergistic effect of retinoids and cyclooxygenase-2 inhibitors therefore seems reasonable.49,50 Whether a combination of coxibs, retinoids or other inhibitors of the EGF pathway could be used as cancer preventive drugs in doses that would not give significant side effects must be settled from a dose-ranging study. The follow-up time for the patients with precancerous lesions was limited to a median of less than 4 years, and only for 15 out of 22 cases did we have a follow-up time of more than 5 years. However, available data indicate that a majority of carcinomas arise between 2 and 3 years after initial diagnosis of premalignancy.17,40,51 All patients with premalignant lesions included in this study were observed for more than 2 years. Therefore, the data presented may be an accurate estimate of the rate of malignant transformation in patients with such lesions. Since cyclooxygenase-2 is not a specific marker of malignancy, treating all patients having cyclooxygenase-2-containing lesions with a coxib could lead to overtreatment. By contrast, our data indicate that including only patients with concomitant aneuploid and cyclooxygenase-2 expressing lesions would not lead to undertreatment. It may be that patients with aneuploid lesions without cyclooxygenase-2 activity are not candidates for treatment with a coxib. Nevertheless, our findings emphasize the need to determine whether treating patients at high risk of oral carcinomas with coxibs can reduce the risk of cancer.
R
ET
R
AC
1. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics 1999. CA Cancer J Clin 1999;49(1):8—31. 2. Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer statistics 2000. CA Cancer J Clin 2000;50:7—33. 3. Silverman SJ, Gorsky M. Epidemiologic and demographic update in oral cancer: California and national data—1973 to 1985. J Am Dent Assoc 1990;120:495—499. 4. Slaughter DP, Southwick HW, Smejkal W. Field cancerization in oral stratified squamous epithelium: clinical implications of multicentric origin. Cancer 1953;6:963—968. 5. Lippman SM, Hong WK. Second malignant tumors in head and neck squamous cell carcinoma: the overshadowing threat for patients with early-stage disease. Int J Radiat Oncol Biol Phys 1989;17:691—694. 6. Scully C, Porter S. ABC of oral health. Swellings and red, white, and pigmented lesions. BMJ 2000;321:225—228. 7. Scully C, Porter S. ABC of oral health. Oral cancer. BMJ 2000;321:97—100. 8. Boysen M, Loven JO. Second malignant neoplasms in patients with head and neck squamous cell carcinomas. Acta Oncol 1993;32:283—288. 9. Macfarlane GJ, Boyle P, Evstifeeva TV, Robertson C, Scully C. Rising trends of oral cancer mortality among males worldwide: the return of an old public health problem. Cancer Causes Control 1994;5:259—265. 10. Callery CD, Spiro RH, Strong EW. Changing trends in the management of squamous carcinoma of the tongue. Am J Surg 1984;148:449—454. 11. de Vries N, van Zandwijk N, Pastorino U. Chemoprevention in the management of oral cancer: EUROSCAN and other studies. Eur J Cancer B Oral Oncol 1992;28B:153—157. 12. Kelloff GJ, Boone CW, Steele VK, Perloff M, Crowell J, Doody LA. Development of chemopreventive agents for lung and upper aerodigestive tract cancers. J Cell Biochem Suppl 1993;17F:2—17. 13. Schantz SP, Ostroff JS. Novel approaches to the prevention of head and neck cancer. Proc Soc Exp Biol Med 1997;216: 275—282. 14. Kelloff GJ, Sigman CC, Greenwald P. Cancer chemoprevention: progress and promise. Eur J Cancer 1999;35:1755— 1762. 15. Kelloff GJ, Crowell JA, Steele VE, Lubet RA, Boone CW, Malone WA, et al. Progress in cancer chemoprevention. Ann N Y Acad Sci 1999;889:1—13. 16. Lee JJ, Hong WK, Hittelman WN, Mao L, Lotan R, Shin DM, et al. Predicting cancer development in oral leukoplakia: ten years of translational research. Clin Cancer Res 2000;6: 1702—1710. 17. Sudbø J, Kildal W, Risberg B, Koppang HS, Danielsen HE,
Cyclooxygenase-2 (COX-2) expression in high-risk premalignant oral lesions
23.
24.
25.
26.
27.
28.
29. 30. 31.
R
32.
33.
34.
D
22.
TE
21.
35. DuBois RN, Abramson SB, Crofford L, Gupta RA, Simon LS, Van De Putte LB, et al. Cyclooxygenase in biology and disease. FASEB J 1998;12:1063—1073. 36. Herschman HR. Prostaglandin synthase 2. Biochim Biophys Acta 1996;1299:125—140. 37. Taketo MM. Cyclooxygenase-2 inhibitors in tumorigenesis (Part II). J Natl Cancer Inst 1998;90:1609—1620. 38. Li M, Wu X, Xu XC. Induction of apoptosis by cyclo-oxygenase-2 inhibitor NS398 through a cytochrome C-dependent pathway in esophageal cancer cells. Int J Cancer 2001; 93:218—223. 39. Ghadimi BM, Sackett DL, Difilippantonio MJ, Schrock E, Neumann T, Jauho A, et al. Centrosome amplification and instability occurs exclusively in aneuploid, but not in diploid colorectal cancer cell lines, and correlates with numerical chromosomal aberrations. Genes Chromosomes Cancer 2000;27:183—190. 40. Sudbø J, Kildal W, Johannessen AC, Koppang HS, Sudbø A, Danielsen HE, et al. Gross genomic aberrations in precancers: clinical implications of a long-term follow-up study in oral erythroplakias. J Clin Oncol 2002;20:456—462. 41. Karabulut A, Reibel J, Therkildsen MH, Praetorius F, Nielsen HW, Dabelsteen E. Observer variability in the histologic assessment of oral premalignant lesions. J Oral Pathol Med 1995;24:198—200. 42. Sudbø J, Bryne M, Johannessen AC, Kildal W, Danielsen HE, Reith A. Comparison of histological grading and large scale genomic status (DNA ploidy) as prognostic tools in oral dysplasia. Journal of Pathology 2001;194:303—310. 43. Sudbø J. Pathology in Disgrace? Journal of Pathology 2002; 196:244—245. 44. Sudbø J, Bankfalvi A, Bryne M, Marcelpoil R, Boysen M, Piffko J, et al. Prognostic value of graph theory-based tissue architecture analysis in carcinomas of the tongue. Lab Invest 2000;80:1881—1889. 45. Lotan R, Xu XC, Lippman SM, Ro JY, Lee JS, Lee JJ, et al. Suppression of retinoic acid receptor-beta in premalignant oral lesions and its up-regulation by isotretinoin. N Engl J Med 1995;332:1405—1410. 46. Lippman SM, Heyman RA, Kurie JM, Benner SE, Hong WK. Retinoids and chemoprevention: clinical and basic studies. J Cell Biochem Suppl 1995;22:1—10. 47. Lippman SM, Spitz MR, Huber MH, Hong WK. Strategies for chemoprevention study of premalignancy and second primary tumors in the head and neck. Curr Opin Oncol 1995;7: 234—241. 48. Mestre JR, Subbaramaiah K, Sacks PG, Schantz SP, Tanabe T, Inoue H, et al. Retinoids suppress epidermal growth factorinduced transcription of cyclooxygenase-2 in human oral squamous carcinoma cells. Cancer Res 1997;57:2890—2895. 49. Torrance CJ, Jackson PE, Montgomery E, Kinzler KW, Vogelstein B, Wissner A, et al. Combinatorial chemoprevention of intestinal neoplasia. Nat Med 2000;6:1024— 1028. 50. Gupta RA, DuBois RN. Combinations for cancer prevention. Nat Med 2000;6:974—975. 51. Sudbø J, Ried T, Bryne M, Kildal W, Danielsen HE, Reith A. Abnormal DNA content predicts the occurence of carcinomas in non-dysplastic oral white patches. Oral Oncol 2001; 37:558—565.
AC
20.
R
19.
ET
18.
Reith A. DNA content as a prognostic marker in patients with oral leukoplakias. N Engl J Med 2001;344: 1270—1278. Lippman SM, Hong WK. Molecular markers of the risk of oral cancer editorial. N Engl J Med 2001;344:1323—1326. Wong NA, Pignatelli M. Beta-catenin—a linchpin in colorectal carcinogenesis? Am J Pathol 2002;160:389—401. Dempke W, Rie C, Grothey A, Schmoll HJ. Cyclooxygenase2: a novel target for cancer chemotherapy? J Cancer Res Clin Oncol 2001;127:411—417. Vainio H. Is COX-2 inhibition a panacea for cancer prevention? Int J Cancer 2001;94:613—614. Inoue H, Yokoyama C, Hara S, Tone Y, Tanabe T. Transcriptional regulation of human prostaglandin-endoperoxide synthase-2 gene by lipopolysaccharide and phorbol ester in vascular endothelial cells. Involvement of both nuclear factor for interleukin-6 expression site and cAMP response element. J Biol Chem 1995;270:24965—24971. Sheng H, Shao J, Dixon DA, Williams CS, Prescott SM, DuBois RN, et al. Transforming growth factor-beta1 enhances Haras-induced expression of cyclooxygenase-2 in intestinal epithelial cells via stabilization of mRNA. J Biol Chem 2000; 275:6628—6635. Bartsch H. Studies on biomarkers in cancer etiology and prevention: a summary and challenge of 20 years of interdisciplinary research. Mutat Res 2000;462:25579. Nair J, Barbin A, Velic I, Bartsch H. Etheno DNA-base adducts from endogenous reactive species. Mutat Res 1999; 424:59—69. Bartsch H, Nair J. Ultrasensitive and specific detection methods for exocylic DNA adducts: markers for lipid peroxidation and oxidative stress. Toxicology 2000;153:105— 114. Bartsch H, Nair J. Exocyclic DNA adducts as secondary markers for oxidative stress: applications in human cancer etiology and risk assessment. Adv Exp Med Biol 2001;500: 675—686. Schmezer P, Rupprecht T, Tisch M, Maier H, Bartsch H. Laryngeal mucosa of head and neck cancer patients shows increased DNA damage as detected by single cell microgel electrophoresis. Toxicology 2000;144:149—154. Oberley TD. Oxidative damage and cancer. Am J Pathol 2002;160:403—408. Sen S. Aneuploidy and cancer. Curr Opin Oncol 2000;12: 82—88. Chan G, Boyle JO, Yang EK, Zhang F, Sacks PG, Shah JP, et al. Cyclooxygenase-2 expression is up-regulated in squamous cell carcinoma of the head and neck. Cancer Res 1999;59:991—994. Pindborg JJ, Reichart PA, Smith CJ, Waal Jvan der. Histological typing of cancer and precancer of the oral mucosa. 2nd ed. London: Springer; 1997. Hedley DW. DNA analysis from paraffin-embedded blocks. Methods Cell Biol 1994;41:231—240. Haroske G Giroud F, Reith A Bo ¨cking A. 1997 ESACP consensus report on diagnostic DNA image cytometry. Part I: basic considerations and recommendations for preparation, measurement and interpretation. European Society for Analytical Cellular Pathology Anal Cell Pathol 1998;17: 189—200.
505