Unusual Profile and High Prevalence of p53 Mutations in Esophageal Squamous Cell Carcinomas from Northern Iran¹

p53 tumor mutations have been linked to specific carcinogen exposures, suggesting a molecular epidemiology approach to investigation of cancers for which the causes have remained elusive. Although it is well-established that tobacco and alcohol, particularly in combination, are the major causes of SCC in the United States and Europe, neither is thought to be important in the etiology of this disease in Iran, where both men and women are at elevated risk (1). Age-adjusted incidence figures of up to 171/100,000 have been reported in rural areas of northeastern Iran for ESCC,³ the major histological subtype (2). The rates are >100-fold higher than in lowest incidence areas of the world, pointing to strong environmental and perhaps genetic influences. The precise risk factors responsible for the high prevalence of ESCC in northern Iran have remained a matter of conjecture. Cultural or environmental factors suggested to play a role include dietary vitamin and mineral deficiencies, mutagenic by-products of opiates, and consumption of regional specialties or scalding beverages, which are strong irritants to the esophageal mucosa (2, 3, 4).

We initiated a study to examine the p53 status in tumors of 74 ESCC patients from northern Iran, 40 from the capital city Tehran, and 34 from the Caspian Littoral, because no p53 mutation data were available in the literature on tumors from this region. Additional tumor tissue from paraffin blocks was available for 23 of the patients from Tehran, allowing immunohistochemical assessment of expression of COX-2 and iNOS, two enzymes known to be elevated in chronically inflamed tissues and in gastrointestinal tumors from other geographical areas.

Formalin-fixed, paraffinembedded esophageal tissue (surgically resected material or biopsy) from cancer patients diagnosed in Iran with ESCCs were collected for analysis. Informed consent was obtained from patients by participating scientists from Iran, and the study was approved by the Medical Ethics Committee, Ministry of Health, Iran as conforming to the ethical guidelines of the 1975 Declaration of Helsinki. Hospital records were used to verify age, permanent residence, smoking history, and ethnicity of individuals. Archived histology sections were examined by the collaborating pathologist in Tehran (P. M.), and serial sections of 10-μm thickness were prepared for DNA extraction. Diagnosis of SCC was confirmed by the pathologist at the German Cancer Research Center, Heidelberg (H-J. G.), who designated an area of tissue material with >50% neoplastic cellularity for each specimen on H&E-stained slides, used to guide dissection for DNA extractions. Forty tumors were examined from patients who were long-term permanent residents of Tehran (most of whom were also born there) and for whom diagnosis of ESCC could be confirmed. Fixed specimens from 34 patients residing in the Caspian Littoral and who had undergone surgery in regional hospitals also were examined in the present study and subjected to the same scrutiny by the pathologists as the material from Tehran. Tissue areas with high (>50%) neoplastic cellularity were dissected from dewaxed slides, and the material was digested by proteinase K in SDS-containing buffer for 3–5 days at 50°C to release DNA suitable for PCR.

Dewaxing, microdissection, DNA extraction, and PCR set-up were all performed in a special laboratory free of contamination from PCR products, with reagents and equipment reserved for these purposes as we have described previously (5). Each exon (exons 5–8) of the p53 gene was individually amplified by a single 40-cycle PCR, using intron-specific 20-mer primers as described by Lehman et al. (6), and 27–30-mer primers described by the Affymetrix GeneChip protocol (Affymetrix, Inc., Santa Clara, CA). The longer primers efficiently generated PCR product from template DNA of formalin-fixed specimens that had proved difficult to amplify with 20-mer primers. PCR products were purified with Microcon 100 (Millipore) filters and sequenced directly by BigDye fluorescent dye dideoxy sequencing and microcapillary electrophoresis with an ABI 310 Genetic Analyzer according to the supplier’s instructions (Applied Biosystems International). All samples with mutations were verified by two independent cycle sequencing PCR reactions and analysis of both DNA strands. In addition, of the 48 mutation-bearing tumor samples, 45 (94%) were reanalyzed by retrieving genomic DNA stocks, performing new PCR amplifications, and resequencing of new PCR amplification products. (No material was left for the remaining three samples.) Finally, 9 tumors, in which a common (hotspot) mutation had been identified, were independently examined a third time, by preparing new tissue sections for microdissection, extraction of genomic DNA, amplification by PCR, and a third repeat of the mutation analysis. All confirmation experiments performed produced the same result as the first analysis.

Of the 74 ESCCs examined in this study for mutations, paraffin blocks of 23 tumors (all from residents of Tehran) were available for preparation of additional histology sections suitable for immunohistochemistry. Tissue sections (3 μm) were applied to precoated glass slides, and after dewaxing, were subjected to microwave antigen retrieval in citrate buffer. To block nonspecific reactivity and staining from endogenous peroxidase, sections were incubated in hydrogen peroxide (1%) for 15 min and in serum-free Protein Block (Dako, Inc.) for 5 min. After rinsing, slides were incubated overnight at 4°C with primary antibody, either polyclonal anti-COX-2 antiserum (Santa Cruz, dilution 1:200) or anti-iNOS monoclonal antibody (N32020; Transduction Laboratories; dilution, 1:100). Incubation with appropriate biotinylated secondary antibody was for 30 min, followed by incubation in avidin-biotin-peroxidase reagents (Vectastain ABC kit; Vector Laboratories), and visualization with 3,3′-diaminobenzidine (DAB kit; Vector Laboratories). Esophageal tumor cell lines TE-1 and TE-11 (7) served as positive and negative controls, respectively, for COX-2 expression in immunohistochemistry studies. The specificity of the COX-2 primary antibody was tested by incubating sections with primary antibody preabsorbed with the appropriate blocking peptide (C-20 P; Santa Cruz Biotechnology), which effectively blocked immunostaining of the COX-2-expressing cell line TE-11. Specificity also has been shown by Western blot analysis (8). Negative controls in which the respective primary antibody was omitted were also performed. Assessment of relative expression levels (0–3) was achieved according to a composite scoring system of staining intensity (0, negative; 1, weak; 2, moderate; 3, strong) and the percentage of positive cells (0, 0–5%; 1, 6–25%; 2, 26–50%; 3, 51–100%) When the two components of the score were in distinct categories, the composite score was rounded off to the lower value. Statistical analysis of p53 tumor mutation status and COX-2 or iNOS immunostaining was performed with Fisher’s exact test.

Of the 74 ESCC specimens analyzed from Iranian patients, 40 from Tehran and 34 from townships near the Caspian Sea, a total of 54 mutations were detected in exons 5–8 of the p53 gene. Forty-eight of the 74 cases (65%) harbored one or more p53 gene point mutations, including 5 tumors with two and one with three base changes. Patient list and mutation data from analysis of cases from Tehran are given in Table 1 and from the Caspian Littoral in Table 2 and Fig. 1. Because the average age of patients, sex ratio, mutation prevalence, and mutation patterns were highly similar in the two patient groups from Iran, we combined these two sets of data in Table 3 for comparisons with data from the IARC database (9), which comprises data from many areas around the world but which still has no data from Iran. Most mutations we identified in Iranian patients were base substitutions (81% substitutions and 19% frameshifts), among which transitions at CpG dinucleotides were the most common (20 mutations of 54 total, or 37%), followed by transitions at non-CpG sequences (27%), whereas transversions were relatively infrequent (19%; Table 3). p53 sequence sites, where a mutation was found in >3 tumors, were two CpG dinucleotides identified as mutation hotspots in analyses of other cancer types with a high CpG transition frequency: codons 248 (7 tumor mutations), and 213 (5 mutations). The Iranian ESCC mutation profile is thus clearly significantly different from the esophageal SCC mutation spectrum in the IARC database, both with respect to the high prevalence of CpG transitions (37% versus 16%; χ² statistic, P < 0.0002) and the low prevalence of transversions (19% versus 41%; P < 0.002; Table 3).

There were more women (n = 41) than men (n = 33) among the 74 cases we had available for this study, reflecting the particularly high ESCC incidence rates for women in Iran, in marked contrast to many other high-risk areas of the world, where men are at vastly greater risk for this cancer (1). No significant differences between men and women in overall mutation prevalence or pattern were observed in our study, although because of the number of cases at our disposal, only very sharp differences would be expected to appear.

IHC was performed with 23 tumor specimens from Tehran to assess levels of COX-2 and iNOS (Table 4; Fig. 2; and data not shown), two enzymes associated with inflammatory reactions and recently found to be elevated in various types of gastrointestinal tumors (10, 11, 12, 13). Tumor paraffin blocks for preparation of 3-μm sections on precoated slides for IHC were not available for the remaining patients. Seventy-four % of tumors examined by IHC (17 of 23) were immunoreactive in the COX-2 assay, and 91% were positive for iNOS expression. All of the p53 mutation-bearing specimens (15 of the 23 tumors examined for COX-2 by IHC) revealed moderate to strong cytoplasmic immunoreactivity against anti-COX-2 antibody in tumor cells, whereas neoplastic cells in most of the samples in which no p53 mutations were detected were scored negative with COX-2 IHC (6 of 8; Table 4; P < 0.0005, Fisher’s exact test). Reaction of inflammatory mononuclear infiltrating cells, known to express the enzyme, was also observed with COX-2 antibody. Although the sample size is small, this preliminary correlation is consistent with reports on inhibition of COX-2 expression by functional, nonmutated p53 (14) and on the association of p53 immunoreactivity with high levels of COX-2 expression in colorectal cancers (15).

Approximately two-thirds of the ESCC patients in our study harbored a p53 mutation, a prevalence that is comparable with patients from other high-risk areas such as Normandy (16) and in contrast to the modest prevalence (less than one-fourth of tumors with p53 mutations) from ESCC patients not identified as belonging to a high-risk group (17). Unlike the ESCCs from northern Iran that we examined, ESCCs from Northern American and European patients (the majority of whom smoke tobacco and consume alcoholic beverages) are characterized by comparatively low numbers of p53 gene CpG transitions and a relatively high proportion of specific, strand-biased transversions (Ref. 9 and Table 3). This profile may reflect the mutagenicity of tobacco-derived compounds and the DNA repair-inhibiting activity of ethanol metabolites (18, 19). In keeping with this proposal, ESCCs of smoking and drinking patients in Europe and the United States are far more likely to harbor a p53 mutation than tumors of nonsmokers and nondrinkers, corroborating the notion that the named risk factors contribute importantly to mutation load in these populations (17), an observation that has been extended recently to lung cancer (20). In contrast, adenocarcinoma of the esophagus, a cancer thus far not closely linked to tobacco and alcohol consumption, is characterized by a high frequency of transitions at CpG sites (>50%) and low numbers of transversions (9, 21).

The high prevalence of p53 CpG transitions in Iranian SCCs of the esophagus we report here was thus unexpected and challenges the generality that CpG transition mutation prevalence is typical of adenocarcinoma but not SCC. Transitions at CpG sites can arise spontaneously by deamination of 5-methylcytosine to thymine but also may be provoked by various mechanisms including factors that cause chronic inflammation or high nitric oxide levels (22, 23). The frequent iNOS and COX-2 immunoreactivity among Iranian ESCCs is in keeping with observations in gastrointestinal cancers from other geographical areas and risk groups, however, and is thus not per se a distinguishing feature of the patients in this study. The antiapoptotic, growth-promoting role of COX-2 has been attributed to its capacity to eliminate free cellular arachidonic acid and its capacity to generate growth- and angiogenesis-promoting prostaglandins. Future studies may address the issue of whether COX-2 and iNOS levels are consistently higher or deregulation of enzymatic activity more precocious in ESCC carcinogenesis in Iranian patients and thus more effective in generating transitions at 5-methylcytosine-guanine dinucleotides. Altered cytosine methyltransferase or G:T mismatch glycosylase activities (19, 24) and adduction at 5-meC:G bp are also factors that theoretically could increase mutation load at CpG sites (25); therefore, it would be interesting to consider what dietary factors and cultural practices, such as consumption of beverages at high temperatures, could influence CpG transition rate by one of these mechanisms. Synergistic interaction of these biochemical processes with deregulation of cell cycle control by dietary zinc deficiency in generating transition mutations would be another avenue to investigate experimentally (26, 27, 28), because this mineral deficiency has been implicated in the etiology of esophageal cancer in Iran.

The ethnic composition of the Iranian population includes people of both Persian ethnicity and the far less numerous Turkomans of the northeastern townships near the Caspian Sea, such as Gonbad. All 40 patients in our study from Tehran were of Persian ethnicity as well as most patients from the Caspial Littoral participating in this study; only 4 patients, all residents of Gonbad, were of Turkoman (Mongolian; Table 2) origin. We thus were unable to address the important question of p53 mutation frequency/patterns in relation to genetic susceptibility/background. The public health problem that ESCC poses in Iran is particularly acute among the Turkomans of Gonbad, where incidence figures are among the highest recorded in the world (2, 29). In this and other rural areas of the eastern Caspian Littoral, dietary supplements to food staples low in vitamin and trace elements might attenuate the cancer-promoting effects of these deficiencies.

Investigations on the endogenous/environmental factors that can be linked to the presence of p53 mutations, especially CpG transitions, may point to preventive measures that could be undertaken. Our initial study of 74 Iranian ESCC patients calls for further work in this population, including design and implementation of a molecular epidemiology-oriented program of tumor analysis.

We thank M. Rezaei Manesh, Dr. Naghshvar (Sari, Iran), and Dr. Hazrati (Gorgan, Iran) for their valuable participation in the collection of material, Dr. M. Ashari for examination of tumor tissues, and K-R. Muehlbauer for expert assistance with sequence analysis.