c-Ki-ras mutations in colorectal adenocarcinomas from a country with a rapidly changing colorectal cancer incidence

We have examined the incidence of mutation of the c-Ki-ras proto-oncogene in colorectal adenocarcinomas from two different time periods, namely 1962–1966 and 1994–1996. The first cohort of samples consisted of formalin-fixed, archival paraffin block and represent the oldest colorectal cancer samples for which ras mutation has been examined, while the second cohort of tumours were fresh, flash-frozen samples representative of genetic events occurring in contemporary times. Analysis of mutation status was undertaken by a mismatch-specific oligonucleotide hybridization analysis of exon 1 of the c-Ki-ras proto-oncogene after amplification by the polymerase chain reaction. Mutations in codon 12 or 13 of c-Ki-ras were detected in 28% (14/50) of contemporary samples, a figure consistent with locally established mutation rates. In contrast no mutation was detected in any of the 18 samples from the earlier period, a result that is statistically significant (P = 0.007). Age-standardized rates of colorectal cancer in Singapore have seen a marked increase over the last 30 years, and for the first time we have shown that such an increase in colorectal cancer is associated, at least in part with an increase in incidence of a specific mutagenic change. © 1999 Cancer Research Campaign

Colorectal cancer is believed to arise by the accumulation of genetic damage, in a process that may take years, or potentially decades (Fearon and Vogelstein, 1990;Fearon and Jones, 1992). Many of these changes have been well documented, and can include activation of oncogenes as well as the inactivation of tumour suppressor genes and genes of the mismatch repair family of genes (Baba, 1997).
Approximately 40% of colorectal adenocarcinomas have been shown to contain activated copies of the c-Ki-ras proto-oncogene (Breivik et al, 1994;Kiaris and Spandidos, 1995;Elnatan et al, 1996) This gene belongs to the ras family of proto-oncogenes, three closely related genes that code for proteins 188 or 189 amino acids long (Barbacid, 1987;Bos, 1989;Grand and Owen, 1991), which exhïbit intrinsic GTPase activity (Grand and Owen, 1991). All members of the ras family are converted to active oncogenes in naturally occurring tumours by point mutations in codons 12, 13 or 61 (Barbacid, 1987), although there is a strong preference for mutations in codons 12 or 13 (Bos, 1989;Yanez et al, 1987). These mutations abolish GTPase activity, and result in constitutive stimulation of autonomous growth and contribute to neoplastic c-Ki-ras mutations in colorectal adenocarcinomas from a country with a rapidly changing colorectal cancer incidence Age-standardised rates for colorectal cancer by sex and race for time periods 1968-1972-1992(adapted from Chia et al, 1996 development (Barbacid, 1987;Bos, 1989;Grand and Owen, 1991). The particular member of the ras gene family that is found to be activated is characteristic of a particular tumour type, and in colorectal adenocarcinomas the majority of ras activation is associated with c-Ki-ras (Bos et al, 1987). Activation of the c-Ki-ras gene has been found to be associated with a number of factors such as patient age, sex and tumour location (Breivik et al, 1994;Elnatan et al, 1996), and the presence of mutations has been variously correlated with neoplastic progression (Vogelstein et al, 1988;Fearon and Vogelstein, 1990;Fearon and Jones, 1992), pattern of disease dissemination (Finkelstein et al, 1993a(Finkelstein et al, , 1993b and patient survival Smith et al, 1996).
We were interested to discover whether the rate of c-Ki-ras mutation was constant in a single population that had experienced colorectal cancer as both a high incidence and a low incidence disease within the span of a single generation, and to possibly determine whether the spectrum of mutations in the c-Ki-ras gene had undergone significant alterations. To undertake this we analysed tumours harvested some 30-35 years ago in parallel with tumours collected from contemporary patients. Results showed that the mutation rates are significantly different between the two cohorts, and for the first time offering a potential link between the increase in the incidence of colorectal cancer seen in Singapore and a specific genetic change.

Patients and tumours
Tumours analysed were from patients undergoing colonic resection for colorectal adenocarcinoma either during the period 1962-1966 or during the period 1994-1996. Tumours from the contemporary period were collected along with matched normal mucosa from the same patient and stored at -80°C until analysed. Tumours from the period 1962-1966 were obtained as several contiguous 10-µm sections from paraffin blocks histologically verified to contain colorectal adenocarcinoma.

DNA preparation, PCR and DNA sequencing
Genomic DNA was extracted from tumour samples containing at least 70% neoplastic cells, as assessed histochemically, along with genomic DNA from matched normal mucosa from the same patient as described previously (Khine et al, 1994). Genomic DNA was prepared from archival paraffin sections using the Prep-A-Gene (BioRad, Hercules, CA, USA) protocol of Wang et al (1994). A 176-bp fragment of exon 1 of the c-Ki-ras gene was amplified from all DNA samples using custom synthesized oligonucleotides 5′-CATGTTCTAATATAGTCACA-3′ (forward); and 5′-AACAA-GAATTACCTCTATTG-3′ (reverse) essentially as described previously Smith et al, 1996). Polymerase chain reactions (PCR) were as previously described Smith et al, 1996), with the modification that to equalize yields of amplicon from the two source of starting material, archival samples were subjected to 40 cycles while fresh, flashfrozen materials (tumour and mucosa) were subjected to 30 cycles.
For DNA sequencing the PCR products were purified by agarose gel electrophoresis and excised from the gel matrix after which templates were purified using the Qiaex PCR Purification Kit (Qiagen, Hilden, Germany). DNA sequencing reactions were performed using the Big Dye Termination reaction kit (PE Applied Biosystems, Foster City, CA, USA) on approximately 60-90 ng of template from above. Analysis was undertaken on an automated ABI 377-18 DNA Sequencer (PE Applied Biosystems).

Mismatch-specific oligonucleotide hybridization
Mismatch-specific oligonucleotide hybridization was performed essentially as described by Verlaan de Vries et al (1986) with some modifications as described previously Smith et al, 1996) using a panel of primers specific for all possible mutations of the c-Ki-ras proto-oncogene in codons 12 and 13 (Human ras Muta-Lyzer Probe Panels, Clontech Laboratories Inc., Palo Alto, CA, USA).

RESULTS
A total of 68 colorectal adenocarcinomas were examined for mutation of the c-Ki-ras proto-oncogene. Eighteen of these samples were archival paraffin sections obtained from formalin-treated, paraffin-embedded blocks of tumour biopsy samples collected from patients operated on between 1962 and 1966. The remaining 50 samples were from patients operated on during the period 1992-1994 and kept as fresh, flash-frozen specimens, along with their matched normal mucosa until required for analysis. No statistically significant difference was found between the two cohorts with respect to patient age, sex, race, tumour location and stage ( Table 2). All tumour samples were histologically confirmed to contain at least 70% tumour cells.
Genomic DNA was extracted from contemporary tumour samples in parallel with genomic DNA from matched normal mucosa of the same patient by standard methodologies (Khine et al, 1994). Several methodologies were attempted for extraction of genomic DNA by archival samples, with consistent results being Table 2 Clinical data for archival (1962)(1963)(1964)(1965)(1966) and contemporary (1994)(1995)(1996)  obtained by the Prep-A-Gene methodology of Wang et al (1994; data not shown). Exon 1 of the c-Ki-ras gene was amplified by PCR, and samples dotted onto nylon support membranes. All samples were analysed for mutations occurring in codon 12 and 13 of c-Ki-ras by mismatch-specific oligonucleotide hybridization (Verlaan de Vries et al, 1986). Control hybridizations for both codon 12 and 13 wild-type sequences were also carried out (Figures 1 and 2). Sixteen mutants were detected in 14 samples (two tumours contained two mutant codons) from the contemporary cohort. Hence 28% of samples were mutated (14/50) and 8% (16/200) of codons. Ninety-three per cent (15/16) of the mutations occurred in codon 12 and 7% (1/16) in codon 13 (Figure 2). Both the frequency of the c-Ki-ras mutation and the distribution of mutants (Table 3) were consistent with our previously published results on c-Ki-ras gene mutations in the local Singaporean population .
In contrast, archival samples from 1962 to 1966 showed no evidence of mutation in any sample, a rate statistically different from that found in the contemporary cohort both with respect to the number of tumours bearing ras mutations (Samples 1992(Samples -1994 (Figure 2). To further confirm the identity of the PCR product from the archival samples, samples were also analysed by DNA sequencing using an automated DNA sequencer. Sequencing analysis confirmed the identity of the product as being exon 1 of the c-Ki-ras gene (Figure 3).

DISCUSSION
In this study we have reported a statistically different rate of c-Kiras mutation in samples of colorectal adenocarcinomas harvested 30-35 years ago to the rate of mutation detected in contemporary samples. Indeed, we were unable to detect any cases of c-Ki-ras mutation in the archival samples. This result was somewhat surprising. The c-Ki-ras proto-oncogene is currently found mutated in 20-40% of colorectal adenocarcinomas, dependent upon factors including the age and sex of the patient as well as the location of the tumour (Breivik et al, 1994;Elnatan et al, 1996), and c-Ki-ras activation is believed to play a significant role in tumorigenesis, possibly being involved in the transition from adenoma to carcinoma (Fearon and Vogelstein, 1990;Fearon and Jones, 1992). Our own data have shown a significant association between c-Ki-ras activation and tumour progression in tumours originating on the left of the colorectum and an association with a poorer prognosis .
However, recently several lines of evidence have suggested that ras activation may not be causally involved in tumour progression. In particular the findings that ras activation can be detected in phenotypically normal tissues Zhu et al, 1997), and in small non-dysplastic lesions of low tumorigenic potential (Jen et al, 1994) as well as the observation that activated ras genes are present in tumour cells at levels approximately 100fold less than that required to induce transformation in cell lines (Hua et al, 1997). These findings, coupled with those that show that ras activation is actually higher in aberrant crypt foci (Otori et al, 1995;Yamashita et al, 1995aYamashita et al, , 1995b) -the purported precursor to adenomas -than in developed adenocarcinomas suggest that ras activation may simply reflect the mutational burden of a tumour, rather that being specifically involved in the tumorigenic process.
We have previously shown that both p53 point mutation (Goh et al, 1995) and c-Ki-ras point mutation  are significantly associated with a poorer patient prognosis in tumours originating in tumours on the left of the colorectum. Perhaps more  The specific c-Ki-ras mutants in the present cohorts are listed, alongside previously established c-Ki-ras data for colorectal adenocarcinomas from Singapore (* Elnatan et al, 1996). interestingly we have also shown that when considering these two changes together the effect on patient prognosis is additive with patients showing both p53 and ras mutation showing a poorer prognosis than patients showing either one of these changes, and these patients in turn showing a poorer prognosis than patients who evinced neither of these changes . In a Cox regression analysis we noted that the terms for both p53 mutation and c-Ki-ras mutation were excluded in favour of a term that simply described the number of changes seen in a tumour (0, 1 or 2). As such, this analysis would seem to indicate that a better idea of patient prognosis could be determined by obtaining some idea of mutational burden. The cause of mutations in the c-Ki-ras proto-oncogene remain unclear, although some authors have suggested that bile acids may be implicated in the tumorigenic process (Cohen et al, 1989;Chaplin, 1998). Other studies have shown that cooked-meat heterocyclic amines such as 2-amino-3-methylimidazo[4,5f]quinoline can induce aberrant crypt foci with a high rate of ras mutation in animals (Tachino et al, 1995). The observation that the rate of mutation in tumours originating on the right of the colorectum is significantly higher than that found in tumours originating on the left of the colorectum while the spectrum of mutations detected is similar  is certainly consistent with a decreasing gradient of mutagen along the length of the human colorectum. As such c-Ki-ras activation is potentially a 'fingerprint' marker of exogenous mutagens.
Many studies have tried to elucidate a relationship between colorectal cancer and diet and despite the difficulties of these studies a common consensus would suggest a possible relationship between consumption of certain foodstuffs and an increased susceptibility to, or protection from colorectal cancer (World Cancer Research Fund, 1997). In particular, several studies have indicated a protective role for cruciferous vegetables such as cabbage and broccoli (Graham et al, 1978;Haenszel et al, 1980;Colditz et al, 1985), while other studies have implicated an increased risk of developing colorectal cancer based on either high meat consumption (Haenszel et al, 1973;La Vecchia et al, 1988) or a high meat to vegetable ratio (Manousos et al, 1983;Lee et al, 1989). In Singapore, while there are several studies on recent food consumption trends (Gourley et al, 1988;Lee et al, 1989), data on diet 30 years ago are somewhat more scanty. However, what data there are (Lee and Gourley, 1987) clearly point to a significant increase in the consumption of meat/offal and oily foods over the last 30 years. As such the data presented here indicate a somewhat tenuous link between changing diet, increasing incidence of colorectal cancer and increasing incidence of c-Ki-ras mutation suggesting that further studies using old (> 30 years) paraffin blocks may provide valuable insights into the tumorigenic process.