Incidence rates of esophageal squamous cell carcinoma (ESCC) differ in various geographical areas and social and ethnic groups.1–3 Regions of enhanced risk are Henan, China,4 the Caspian littoral of Iran,5, 6 Normandy in France7 and some areas in South America.8–10 Local environmental conditions, individual genetic predisposition and life style may be involved in the development of esophageal cancer. A variety of factors is known to contribute to esophageal cancer, alcohol and smoking in particular.3N-nitroso-compounds are likely to be involved in the high risk areas of China.11 Other factors are also suspected to contribute, e.g., HPV,12–14 opium,15–17 deficiency of micronutrients,18 spicy food19 or barbecued meat, caffeine, hot drunk mate tea and hyperthermia.9, 10, 20 Chronic inflammation due to some of these factors may take part in the development of esophageal cancer. Incidence rates have been reported to be lower when protective substances are present like green tea,21 plant polyphenols and isothiocyanates,22 aspirin,23 zinc, vitamin E and selenium.18 Genetic predisposition may be based on individual types of carcinogen activation or detoxification and DNA repair in a population at risk.24–26 Because of the large number of involved factors it is difficult to analyze their specific contribution to esophageal carcinogenesis in a high risk area.
In recent years, the mutation spectrum of TP53 has been used as an indirect tool in disclosing the role of carcinogenic factors in defined types of cancer.27, 28TP53 is frequently mutated in human tumors.29 In 40–50% of all esophageal cancers TP53 mutations have been found,30–32 most frequently missense mutations, but also nonsense mutations, deletions and insertions. They are clustered in exons 5–8 that contain the DNA binding domain of TP53. Most mutations alter the p53 protein structure and lead to loss of its tumor suppressor function.28, 31
In some instances the TP53 mutation pattern in tumors has been shown to be related to the type of a putatively involved carcinogen,27e.g., a typical G > T transversion mutation at codon 249 of TP53 in hepatocellular carcinomas in aflatoxin B1 contaminated areas33–35 or a high prevalence of TP53 transition mutations in ESCC in central China compatible with nitrosamine contamination of food in this area.11
In southern Brazil the situation is more complex as a variety of different carcinogenic factors might interact in causation of ESCC. Besides alcohol and tobacco, the habit of drinking very hot mate tea has been assumed to be a major factor in the development of ESCC in this area.8–10, 20, 36TP53 mutation data of ESCC of southern Brazil are not yet available. The present study is aimed to provide a data set of TP53 mutations in ESCC of patients living in the high risk area of Rio Grande do Sul and to test a possible relation between documented anamnestic life style parameters and the TP53 mutation pattern. Comparison with the worldwide IARC TP53 database for esophageal cancer36 reveals major differences.
MATERIAL AND METHODS
Patients and tumor samples
Paraffin-embedded specimens of esophageal carcinomas were available from 238 patients. All patients lived in the high risk region of Rio Grande do Sul, Southern Brazil and underwent surgery at the Clinical and Surgical Gastroenterology Unit, Santa Casa Hospital, Porto Alegre, RS, Brazil, during the years 1988–95 (Cases ESP 1–106, first series) and 1995–96 (Cases ES 1–34 and ESN 1–75, second series). Patients‘ data were available for age, gender, profession, residence, smoking, alcohol and mate tea consumption and for tumor size, localization, differentiation and TNM. In the majority of the cases from the first series (until 1995) we were unable to extract DNA and to perform PCR and TP53 mutation analysis because of an inappropriate fixation for maintaining the DNA in an adequate condition. This analysis was possible in a total of 135 cases that had been fixed in 10% buffered formalin. These samples were included in the study reported here. After fixation, the samples were embedded in paraffin and stored at room temperature until time of analysis. Serial sections were used for histopathological evaluation, p53 immunohistochemistry and DNA isolation.
A TP53 mutation usually leads to an increased half life of the protein and an elevated p53 protein concentration that can be analyzed by immunohistochemistry. Paraffin sections (3 μm) were deparaffinized in xylene and rehydrated in decreasing concentrations of ethanol. After microwave treatment (700 W, 600 W and 2× 400 W, 5 min each) in citrate buffer (0.01 M; pH 6), repeated rinsing in TBS (5 mM Tris, 150 mM NaCl, pH 7.5) sections were exposed to monoclonal mouse anti-human p53 antibody AB-6 (1:35) (Calbiochem, Schwalbach, Germany) in TBS supplemented with 0.1% BSA (30 min), incubated in rabbit anti-mouse immunoglobulin (1:25) (DAKO, Hamburg, Germany) for 30 min and treated with APAAP-mouse-complex (1:50) (DAKO), for 30 min and incubated in Fast Red TR/Naphthol-AS-MX (20 min) including 1 mM Levamisole. Counterstaining was with Mayers Hemalum. DNA of positively stained sections was isolated and further analyzed.
p53-Positive areas (1–10 mm2 in size) were scraped from 8 μm sections and DNA was isolated with a QIAamp Tissue Kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. In brief, paraffin was removed by rinsing in 1,200 μl xylene and absolute ethanol twice. Sedimented material was dried. Tissue was digested in 200 μl proteinase K containing ATL buffer over night at 36°C. After addition of fresh proteinase K, incubation was continued at 55°C for about 2 hr. Samples were incubated in 200 μl AL buffer at 70°C for 10 min. After addition of 210 μl ethanol, samples were loaded on spin columns. After rinsing with buffer, DNA was eluted with 10 mM Tris (pH 9). Isolated DNA was divided in aliquots, quantified and quality was checked by running an aliquot on a 5% PAA gel. DNA was stored at −20°C until use.
Amplification of TP53 exons
TP53 exons 5, 6, 7 and 8 where most mutations of the TP53 are found in human tumors and exon 9 were amplified. PCR was performed with AmpliTaq Gold (Perkin-Elmer, Weiterstadt, Germany) according to the manufacturer's recommendations. About 50 ng of genomic DNA was incubated with 1.5 u enzyme in 50 μl sample volume. Primer amount was 20 pmol (exon 6: 40 pmol). The concentration of dNTP was 200 μM each. To the provided PCR-buffer II 1.75 mM MgCl2 were added when amplifying exon 5, 7, 9 and 2.75 mM or 3.25 mM MgCl2 was added for exon 6 and exon 8. In some cases reamplification of the first PCR reaction was necessary. For all exons the identical cycle profile was used in the thermal cycler TC 2400 (Perkin-Elmer): PCR started with enzyme activation at 95°C for 10 min. The first 3 cycles consisted (i) of denaturation at 95°C for 1 min, (ii) annealing at 52°C for 1 min and (iii) extension at 70°C for 7 min. The following 42 cycles consisted (i) of denaturation at 95°C for 50 sec, (ii) annealing at 54°C for 50 sec and (iii) extension at 70°C for 2 min. PCR was finished by a final extension for 10 min at 70°C. Amplification of the single exons was performed with the following primers: first primer pair of exon 5 spanning the exon intron borders: TCT TCC TAC AGT ACT CCC CT and AGC TGC TCA CCA TCG CTA TC; and second pair annealing to adjacent intron sequences: CCG TGT TCC AGT TGC TTT AT and CAA CCA GCC CTG TCG TCT CT. One primer pair for exon 6: TGG TTG CCC AGG GTC CCC AG and CGG AGG GCC ACT GAC AAC CA. First primer pair of exon 7 spanning the exon intron borders: TTA TCT CCT AGG TTG GCT CT and GCT CCT GAC CTG GAG TCT TC; and second pair annealing adjacent intron sequences: AGG TCT CCC CAA GGC GCA CT and GGG GTC AGC GGC AAG CAG AG. First primer pair of exon 8 spanning the exon intron borders: TCC TGA GTA GTG GTA ATC TA and GCT TGC TTA CCT CGC TTA GT; and second pair annealing adjacent intron sequences: TGA TTT CCT TAC TGC CTC TT and CCA CCG CTT CTT GTC CTG CT. One primer pair for exon 9: TTG CCT CTT TCC TAG CA and CCC AAG ACT TAG TAC CTG.37
For better results in SSCP and sequencing analysis, PCR products were purified in agarose gel (1.5% GST agarose (FMC), 1× TAE) using the QIAquick Gel Extraction Kit (Qiagen).
Detection of mutations
SSCP analysis of the amplified exons was performed as a screening step before sequencing. An aliquot of 2–5 μl of purified PCR fragment was mixed with 10 μl formamide dye mixture. Denaturation for 5 min at 95°C was followed by rapid cooling on dry ice. Samples were applied to 10% nondenaturating SSCP gel (AA/Bis: 49/1), containing 5% glycerol and 0.75× TBE.38 Electrophoresis was performed at 15 V/cm in 0.75× TBE for about 16 hr at 20°C. DNA was stained using the Silver Stain Plus Kit (Bio-Rad, Munich, Germany) and PCR products showing a band shift or additional bands were sequenced. Sequence analysis of PCR fragments was carried out with the dRhodamine Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Weiterstadt, Germany) based on fluorescence labeled dideoxy nucleotides as chain terminators. Extension products were analyzed on ABI Prism 310 Genetic Analyzer (Applied Biosystems). PCR amplification primers served also as sequencing primers. To confirm the results, all detected mutations were checked twice. First, a second PCR product of the originally isolated DNA was sequenced. When confirming the detected mutation, an additional new PCR product was synthesized from a second DNA isolation and sequenced. Only the mutations that were confirmed in these 3 subsequent independent steps were scored as valid.
Fisher's exact test was used to determine correlations between different variables. The overall probability of obtaining a difference between groups was calculated. Only p-values smaller than 0.05 were considered statistically significant.
All known factors relevant to tumor development (age, gender, profession, residence, tumor stage and differentiation, consumption of alcohol, tobacco and mate tea) were compared to the observed mutation profile, i.e., mutation type, exon and hot spot localization and CpG distribution. Finally our data were compared to a TP53 mutation database published by the International Agency for Research on Cancer (IARC).36
Immunohistochemical proof of p53 accumulation in tumor cells suggests the presence of a missense mutation. To select tumors suitable for being analyzed for TP53 mutations, immunohistochemical staining for p53 was performed. PCR products of exon 5–9 of the patients with immunohistochemically p53-positive tumors were screened for mutations by SSCP (Fig. 1). Exons with a band shift in SSCP were sequenced. An example is given in Figure 2. The mutations that were found are listed in Table I. In a total of 135 ESCC 50 TP53 mutations (including 1 silent polymorphism) were detected, among them 39 missense mutations, 1 amber mutation, 2 silent mutations, 6 frameshift mutations and 2 mutations at intron borders leading to splicing defects. Three ESCC contained more than 1 TP53 mutation: Case 3 (ES 29) a silent and an amber mutation; Case 45 (ESN 2) a missense and a silent mutation that is obviously a polymorphism39 that could also be detected in noncancerous tissue and will not be considered further; Case 14 (ESN 17) a 26 nt deletion causing a frameshift and, in another part of the same tumor, a missense mutation.
Table I. Site and Type of TP53 Mutations Found in Squamous Cell Carcinomas of the Esophagus of Male and Female Patients From Southern Brazil of Different Age, Tumor Stage and Differentiation, With Various Life Style Factors1
Mutations are unequally distributed over exons 5–8. Eighteen of the 49 mutations (36.7%) are located in exon 5, 5 in exon 6 (10.2%), 8 in exon 7 (16.3%) and 16 are located in exon 8 (32.7%). Two mutations are intron mutations (4.1%). Mutations could not be detected in exon 9 (Table II).
Table II. Summary of 49 TP53 Mutations in 47 ESCC Patients in Southern Brazil1
35 males, 12 females.
Mutations at CpG
The type of point mutations differs widely, but a higher prevalence of transition mutations (53.1%) is found than transversion mutations (34.7%) and frameshifts (12.2%) (Table II).
Within each exon, codons are differently affected, location of mutations differed, but clusters were found. Codon 151 and 248 contain 3 different mutations each, codons 179 and 279 2 different mutations each. Three identical mutations have been found in codon 282 and 2 identical mutations in each of the codons 157, 220, 266 (with different site in the codon) and 273. Ten mutations affect codons that are known as TP53 mutations hot spots,28, 40 most of them located at CpG sites. Only 1 mutation is a CpG mutation not located at a hot spot (Case 7, ESN 23, codon 152). No mutation was detected in codon 176, a TP53 mutation hot spot of esophageal carcinoma (Table III).36
Demographic data as well as information on known risk factors for esophageal cancer were collected. Thirty-five of the 47 patients with detected mutations were male (74.5%), 12 were female. This is identical with the gender distribution of all patients where 74.3% were male. Differences in the mutation pattern of men and women were evident. No deletions or G > T transversions were found in ESCC of women, whereas tumors in men showed 8 G > T transversions (21.6%) and 6 deletions (16.2%). G > A transitions were found more frequently in females than in males (p = 0.046). The average age of developing ESCC did not differ between patients with and without TP53 mutations. It was 58.3 years for all and 59.1 years for patients with detected mutations. Thirty-seven tumors with p53 mutations were moderately differentiated (78.7%), 5 were well (10.6%) and 5 were poorly differentiated (10.6%), a pattern that does not differ from the distribution in the whole patient group.
Main risk factors of interest are smoking, drinking of alcohol and hot mate tea consumption. For 108 patients (80%) smoking habits were documented. Thirty-seven of 38 patients with a TP53 mutation and with known smoking habits (97.4%) were smokers and 1 was a non-smoker. In contrast, only 82.9% patients without a TP53 mutation (58 of 70) were smokers and 12 (17.1%) did not smoke (p = 0.031). This correlation was specially strong for women (p = 0.027). Thus, smoking does apparently not only elevate the risk of esophageal cancer, but effects also the mutation frequency of TP53. Moreover, smoking seems to affect the TP53 mutation pattern. Because of the low number of nonsmokers, the patients with mutations who consumed 20 cigarettes or more per day were defined as heavy smokers [n = 15 (16 mutations)]. The mutation pattern of this group was compared to the pattern of weak (less than 20 cigarettes per day) or non-smokers (n = 13). Among the ESCC of heavy smokers with mutations, only 2 mutations were detected at a CpG site (12.5%), but 4 among the weak and nonsmokers (30.8%). Correlations between smoking and specific types of mutation are stronger and near the threshold of significance. Tumors of heavy smokers contain more frequently transversion mutations than tumors of weak or non-smokers who show a high prevalence of transition mutations (Table IV).
Table III. Clusters (‘Hot Spots’) of TP53 Mutations in ESCC From Southern Brazil1
Another documented risk factor is alcohol. Alcohol drinkers are more frequent in the TP53 mutation group (26 drinkers, 76.5% and 8 non-drinkers, 23.5%) than in the group without mutation (40 drinkers, 66.7% and 20 non-drinkers, 33.3%). A comparison of the 26 mutations in ESCC of known alcohol drinkers with the mutations of the 8 nondrinkers shows in tumors of drinkers a small accumulation of G > T and C > A transversions (5/26, 19.2%, vs. 0/8), deletions (5/26, 19.2%, vs. 0/8) and also of G > A transitions (6/26, 23.1% vs. 1/8, 12.5%) whereas the C > T transition mutations were more prevalent in non-drinkers (3/8, 37.5%, vs. 2/26, 7.7%) (p = 0.101) (Table IV).
Documentation about mate tea consumption is scarce. Among 135 patients only 32 (23.7%) are known to have continuously drunk hot mate tea and 3 (2.2%) did not. Mate tea consumption of 100 patients remained unknown. Twenty of the 47 patients with mutations in ESCC (42.6%) were verified tea drinkers and 1 patient was not.
To disclose whether there is a peculiar TP53 mutation profile in ESCC of Brazilian patients it was compared to the data of the IARC TP53 mutation library that comprises 389 cases of ESCC collected from various parts of the world. For ESCC from Brazilian patients an accumulation of TP53 mutations in exon 8 was evident [16/49 (32.7%), vs. 81/389 (20.8%) in IARC cases] (Table V). An obvious difference was also seen in the prevalence of hot spot mutations. Among 49 detected TP53 mutations of 135 Brazilian ESCC no mutation was found in codon 176, a mutation site that is strongly represented in the 389 listed ESCC of the IARC p53 mutation library.
Table V. Comparison of the TP53 Mutation Pattern Between ESCC in Southern Brazil and the Worldwide ESCC Mutation Database at IARC1
C > T and G > A transitions are more frequent in Brazilian cases (16.3% and 26.5%, respectively) than in IARC cases (11.8% and 21.6%, respectively). In addition, a higher number of C > G transversions is observed in Brazilian cases (8.2% vs. 1.5% in IARC cases). Deletions are slightly more frequent in ESCC from Rio Grande do Sul (12.2% vs. 8.7% in IARC cases). G > T transversions typical for smokers are more frequent in IARC tumors (22.6%) than in Brazil (16.3%) although significantly more smokers in Brazil show a mutation (37/38 patients, 97.4%) than in the IARC database (91 of 116 patients with mutations, 78.5%, p = 0.005).
Mutations of the TP53 gene belong to the most common structural alterations in human cancer and may serve as a marker in studies on molecular cancer epidemiology.7, 41 We analyzed the TP53 mutation pattern of a series of esophageal squamous cell carcinomas (ESCC) of patients living in the high risk area of Rio Grande do Sul in Southern Brazil.42 Comparison of these data with those of the worldwide TP53 mutations database collected by IARC36 revealed several differences in the spectrum of type and localization of p53 mutations.
In 34.8% of the 135 analyzed ESCC we found mutations in TP53 exons 5–9 including 2 intron mutations. This percentage is lower than the reported value of 45.8% in 240 published cases, but in the same range as in the low-risk area of Western Europe43 and in Thailand.44 It cannot be excluded that immunohistochemically negative cases with TP53 mutations were missed in our series because only TP53 immunohistochemically-positive ESCC were selected for SSCP and DNA sequence analysis.
We did not find any codon 176 hot spot mutations, in contrast to the IARC collection with 12.3%, mainly G > T transversions. IARC database, however, might have been influenced or even distorted with respect to this mutation by results of a single large study45 with no further confirmation by any other report.
A peculiarity of TP53 mutations in Brazilian ESCC is a lack of insertions (0% vs. 5.9% in the IARC database) and a higher prevalence of deletions (12.2% vs. 8.7%). Five of 6 deletions are at or adjacent to repetitive sequences or monotonic runs, as typical for small deletions.46
G > T transversion is a common type of mutation in southern Brazil. It is present in tumors of smokers and may be generated by benzo(a)pyrene 7,8-diol-9,10-epoxide adduct formation at guanosine.24 Brazilian ESCC patients (97.4%) with a TP53 mutation were smokers.
A high prevalence of G > A (26.5%) and C > T (16.3%) transition mutations was found. The comparable data for the IARC collection are considerably lower (21.6% and 11.8%, respectively). C > T transitions could in part be due to spontaneous cytosine deamination at CpG dinucleotides. On the other side, the bulk of G > A transitions (and the corresponding C > T exchanges, if G is located on the opposite strand) are typical for alkylating agents like alkylnitrosamines that form O6-alkyl-guanine adducts that may lead to base mispairing during DNA replication.47 For the esophagus a mutation pattern characteristic of alkyl-N-nitrosamines has been demonstrated for nitrosomethylbenzylamine in rat esophageal papillomas that showed 90% G > A transitions and 10% deletions.48 The assumption that nitrosamines might be responsible for G > A and C > T transitions, especially those at non-CpG sites, is corroborated by observations in China where a high prevalence of G > A and C > T transitions was found, obviously related to nitrosamine contamination of food4, 49, 50 or to individual deficiencies of O6-methylguanine-DNA methyltransferase.51
If exogenous nitrosamines were involved in Brazil, their origin is obscure. In principle, they could be found in alcoholic beverages and foods like pickled meat, barbecued meat as well as tobacco smoke, but southern Brazil is not known to be an area where nitrosamines play a major role in food. Nitroso compounds, however, can also be generated by endogenous processes, e.g., nitrosation in the stomach at low pH or during inflammation.52 In Brazil, enhanced incidence of ESCC has been claimed to be related to consumption of very hot mate tea,9, 10, 53 with hyperthermia being the main cancer inducing factor.20 Cold mate tea commonly drunk in parts of Paraguay does not seem to enhance the incidence of ESCC in this area.20 Chronic irritation of the esophageal mucosa by hyperthermia and concomitant inflammatory processes might be involved in formation of nitrosamines via NO release and reactions of NO generated nitrogen oxides with amines.52 A positive relationship between a high prevalence of G > A transition mutations and nitric oxide synthase activity has been described in human cancers.54 Exposure of Salmonella typhimurium, plasmid DNA and the TP53 complementary DNA to NO induced G > A transition mutations in the p53 DNA.55 Endogenous formation of N-nitrosamines by induction of nitric oxide synthase has been found in rats with acute hepatic inflammation.56 Similar processes might be operative in generating the high prevalence of G > A transition mutations in ESCC after chronic thermal irritation. Although we cannot confirm this hypothesis statistically with the data available now, the enhanced G > A and C > T transition ratios supports the notion that in Brazil hyperthermia-induced inflammation caused by very hot mate tea might in part be responsible for ESCC in this high risk area.
Recent results indicate that esophagus is highly susceptible to the tumor-initiating effect of nitrosamines. It was demonstrated57 that N-nitrosonornicotine is more effectively activated to mutagenic metabolites by esophageal microsomes than by liver microsomes. In a non-human primate model (patas monkey) it was shown that esophageal tissue contains a low amount of O6-alkylguanine alkyltransferase and correspondingly, high amounts of O6-methylguanine after treatment with N-nitrosodimethylamine.58 Coexposure to N-nitrosodimethylamine and alcohol resulted in the largest enhancement of O6-methylguanine content of esophageal tissue of all tested tissues.58 This agrees with our finding that alcohol consumption is connected with an elevated proportion of G > A transitions (23.1% in drinkers vs. 12.5% in nondrinkers). It is thus possible that alcohol increases the effect of nitrosamines that might be formed in small amounts during inflammation.
Chronic inflammation may generate, besides nitrosamines, other mutagenic factors like nitric oxide and nitrogen oxides, hydroxyl radicals or H2O2. N2O3 is able to deaminate amino residues of DNA bases. Deamination of 5-methyl cytosine results in formation of thymine to end up in a C > T transition. This might also contribute to a mutation pattern resembling a nitrosamine effect with high G > A and C > T transition ratios. Moreover, N2O3 may deaminate adenine to hypoxanthine. In case of nonrepair this results in A > G transitions. This type is more frequently found in ESCC of Southern Brazil than in ESCC from the IARC collection corroborating the hypothesis of inflammation processes being involved in ESCC. Other mutagenic substances generated during chronic inflammation, however, may induce a divergent mutation pattern. Hydroxyl radicals generate G > T transversions via 8-hydroxyguanine production, but they may be responsible for other mutations, too, e.g., the higher prevalence of C > G transversions in Brazil (8.2%) than in the IARC library database (1.5%) (Fig. 3).
Epidemiological data indicate that the increased incidence of ESCC in Southern Brazil might particularly be related, in addition to alcohol and smoking, to the consumption of very hot mate tea in this area.9, 10, 20 In other parts of the world, mainly the consumption of alcohol and tobacco smoking contributes to ESCC induction. The spectrum of TP53 mutations obtained in the present study indicates that all 3 factors are by all probability involved in ESCC development in the high risk region of Southern Brazil. The type and pattern of TP53 mutations are different from the average world wide values of the IARC database. Deviations could in part be explained by chronic inflammatory processes in the esophagus to be expected during irritation by very hot mate tea. Formation of nitrosamines and free radicals during inflammation might be responsible for the relatively frequent G > A and C > T transitions mutations found in Southern Brazil. The fact that almost all of the ESCC patients with a TP53 mutation were smokers, but the prevalence of G > A transitions was higher than in the worldwide database, argues for the hypothesis that consumption of hot mate tea might be an important cofactor in the increased ESCC incidence in the high risk area of Rio Grande do Sul. The implications for prevention of ESCC are evident.
The technical assistance of Ms. R. Koch, Ms. S. Madsen, Ms. A. Eberl and Mr. M. Ruiter is gratefully acknowledged.