Frequency of Defective DNA Mismatch Repair in Colorectal Cancer among the Alaska Native People

Colorectal cancer (CRC) is the second leading cause of cancer-related deaths and the fourth most frequently diagnosed cancer in the United States. Rates for incidence and mortality are not uniform across ethnic groups within the United States (1-3). Incidence rates for CRC among the Alaska Native men and women are the highest of all ethnic groups, and death rates are the highest or second highest in women and men, respectively (3). On the other hand, reported rates for American Indian people in the contiguous United States are among the lowest. The term Alaska Native refers to the indigenous population whose ancestors originally occupied the area that is now the state of Alaska. These include the Athabascan, Eyak, Tlingit, Haida, Tsimshian, Yupik, Cupik and Inupiat, and Aleut people, all of whom have diverse languages and cultural practices.

When the USPHS assumed responsibility for the delivery of health care services in Alaska in 1955 (4), these groups were categorized as Indian, Eskimo, or Aleut. One sixth of the current population of Alaska is Alaska Native. Based on census data, ∼52% are Eskimo, 36% are Indian, and 12% are Aleut. It is believed that the migration of the Indian groups into Alaska preceded that of the Eskimo/Aleut.

In national U.S. data, Alaska Native and American Indians in the contiguous United States are collectively reported as American Indian/Alaska Native. Nationally, CRC rates reported for American Indian/Alaska Native combined are the lowest in the contiguous United States. A recent report using Surveillance, Epidemiology, and End Results data comparing only the Indian of Alaska with Indian of New Mexico documented that CRC rates of Indian people in Alaska exceeded those of New Mexico nearly 2-fold (5).

Brown et al. (6) compared CRC among Alaska Native people with those of the Caucasian population in the Seattle/Puget Sound area of the northwestern United States. Among Alaska Native people, the Aleut and Eskimo, and particularly women, had the highest relative risk for CRC, twice that of Alaska Indian. The reasons for the varied CRC risks among populations are unknown.

Based on the presence or absence of DNA mismatch repair (MMR), CRC can be divided into two broad categories (7, 8). Tumors with defective MMR are characterized by the presence of a particular tumor phenotype, termed microsatellite instability (MSI). In sporadic CRC, two distinct MSI phenotypes have been described: MSI-L (MSI at <30% of the loci examined) and MSI-H (MSI at ≥30% of the loci examined; ref. 9). MSI-H has been identified in tumors from patients with hereditary nonpolyposis colon cancer and in ∼20% of sporadic CRC (10). The MSI-H phenotype is also associated with distinct clinicopathologic features (e.g., proximal tumor site, high grade, early stage, diploidy, and favorable survival; refs. 11-15). Among sporadic CRC, the majority of MSI-H cases are due to inactivation of hMLH1 (∼95%), with hMSH2 and hMSH6 accounting for a smaller percentage, ∼5% and <1%, respectively (16). Germ line mutations in these same MMR genes account for the majority of hereditary nonpolyposis colon cancer cases demonstrating MSI-H (17, 18).

Among all cases involving hMSH2 and hMSH6 (sporadic and inherited), the presence of a germ line mutation seems to be the most common mechanism of gene inactivation. For hMLH1, however, current data suggests that the most common mechanism of gene inactivation among unselected cases is promoter hypermethylation and less frequently by mutations in the gene itself (16, 19, 20). Additionally, although the etiology is still ill-defined, a specific mutation within BRAF (V600E) has been shown to be present in the majority (∼70%) of tumors with hypermethylation of the hMLH1 promoter but, importantly, not in cases with germ line hMLH1 mutations (21-23).

Certain characteristics of CRC reported in the Alaska Native population suggest that DNA MMR defects could contribute to the development of CRC in the Alaska Native population. These characteristics include increased rates in women, larger proportion of tumors in the proximal colon, and tumors of higher grade. We conducted this study among Alaska Native CRC patients to explore the frequency and potential cause(s) of defective MMR in this group of individuals.

Approval to conduct this study was granted by the institutional review board from several institutions, including the Mayo Clinic (Rochester, MN), the Indian Health Service Alaska Area Institutional Review Board, the Alaska Native Medical Center (Anchorage, AK), Mt. Edgecumbe Hospital (Sitka, AK), and Fairbanks Memorial Hospital (Fairbanks, AK).

Potential study cases were identified from the Alaska Native Tumor Registry. The Alaska Native Tumor Registry is a population-based registry that has identified all new cases at diagnosis among Alaska Native residents of Alaska who are eligible for Indian Health Service health care (24). For each patient identified with cancer, the registry abstracts and codes demographic, diagnostic and treatment information, pathology, stage of disease, and follow-up status. At the time of registration for health care, patients were asked to identify themselves as Eskimo, Aleut, or Indian. From the Alaska Native Tumor Registry, we identified those Alaska Native patients who were diagnosed in the calendar years 1984 through 2001, treated at one of three Alaska hospitals, and for whom there was an available pathology report and formalin-fixed tissue specimen. Selection was made to assure adequate numbers of each of the three ethnic groups. During this time period, ∼700 Alaska Native patients were diagnosed with invasive CRC and registered in the Alaska Native Tumor Registry. CRC from 329 Alaska Natives (165 Eskimo, 111 Indian, and 53 Aleut) met the criteria for inclusion in the study. Slides of all potential study cases were reviewed by a pathologist (L.J.B.) to confirm the diagnosis of invasive CRC and to select formalin-fixed, paraffin-embedded blocks that contained both tumor and normal tissue. From these blocks, 10-μm-thick sections were cut and placed on supercharged glass slides. Samples were deidentified and sent to the Mayo Clinic for laboratory studies. For analysis, select variables from the Alaska Native Tumor Registry were linked with the laboratory results. These included the year of diagnosis, site of tumor site within the colon/rectum, stage at diagnosis, and tumor histopathology grade. Age at diagnosis was provided within a specific decade (<50, 50-59, 60-69, 70-79, and 80+ years) to maintain patient anonymity.

Immunohistochemical staining for proteins produced by the hMLH1, hMSH2, and hMSH6 genes was done on slides of the paraffin-fixed sections as previously described (15). Normal colonic epithelium and lymphocytes exhibit strong nuclear staining for hMLH1, hMSH2, and hMSH6 and thus served as a positive internal control for staining of these proteins. All 329 study cases were evaluated for the presence of hMLH1 and hMSH2. Additionally, 146 of the 329 were evaluated for hMSH6.

From 10-μm-thick sections, areas containing >70% tumor nuclei, and areas containing only normal cells were identified for microdissection. DNA was extracted from microdissected sections using the Qiamp Tissue DNA extraction kit according to the manufacturer's instructions (Qiagen). Paired normal and tumor DNA were analyzed for MSI with the use of 10 microsatellite markers: four mononucleotide markers (BAT 25, BAT 26, BAT 40, and BAT 34c4), five dinucleotide markers (D5S346, D17S250, ACTC, D18S55, and D10S197), and one penta-mono-tetra compound marker (MYCL). Tumors were classified as MSH-H if ≥30% markers showed instability, MSH-L if <30% showed MSI, and MSS if no marker exhibited MSI.

A multiplex allele-specific PCR assay was created to directly detect both the BRAF wild-type allele and V600E mutation. The PCR reaction included Applied Biosystem buffer, deoxynucleotide triphosphates and TaqGold, and 2 μL of DNA extracted from paraffin-embedded tissue in a total volume of 25 μL. Cycling variables included 95°C for 2 min, 94°C denaturation for 30 s, 56°C annealing for 40 s, 72°C extension for 30 s, 72°C final extension for 10 min, and a 5°C hold. Primers for the PCR assay include the following: wild-type forward, 5′-NED-TGATTTTGGTCTAGCTACAGT-3′; V600E forward, 5′-56-FAM-CAGTGATTTTGGTCTAGCTTCAGA-3′; and reverse, 5′-GTTTCTTTCTAGTAACTCAGCAGC-3′. PCR amplicons were diluted 1:13 using the 400 Rox size standard and formamide. Samples were then run on an ABI 3100 Analyzer and the data were analyzed using Genotyper software from Applied Biosystems.

A 1-cm² 10-μm-thick section of paraffin-embedded tumor was scraped from a slide, placed in a microfuge tube along with 40 μL of the extraction solution (4 μL 100× TE, 2 μL of 20 mg/mL Proteinase K, 20 μL of 0.1 mg/mL tRNA, and 14 μL water), and incubated overnight at 50°C. The total reaction volume was then used to bisulfite modify tumor DNA with the use of the EZ DNA Methylation Kit according to the manufacturer's instructions. PCR primers designed specifically to detect differences between methylated and unmethylated DNA for the hMLH1 promoter were then used essentially as described by Grady et al. (23). Primers for the methylation-specific PCR assay included the following: methylated reaction (5′-FAM-AACGAATTAATAGGAAGAGCGGATAGCG-3′; 5′-CGTCCCTCCCTAAAACGACTACTACCC-3′), unmethylated reaction (5′-NED-taaaaatgaattaataggaagagtggatagtg-3′; 5′-AATCTCTTCATCCCTCCCTAAAACA-3′).

Cycling variables for the unmethylated hMLH1 primers included 95°C for 2 min, 94°C denaturation for 30 s, 61°C annealing for 30 s, 72°C extension for 30 s, 72°C final extension for 10 min, and a 5°C hold. Cycling variables for the methylated hMLH1 primers included 95°C for 2 min, 94°C denaturation for 30 s, 68°C annealing for 30 s, 72°C extension for 30 s, 72°C final extension for 10 min, and a 5°C hold. PCR products from the unmethylated and methylated reactions were pooled together (1:1 ratio) and diluted 1:13 using the 400 Rox size standard and formamide. Samples were then run on an ABI 3100 Analyzer and the data were analyzed using Genotyper software from Applied Biosystems.

The presence of defective MMR was assessed for associations with the following clinical and pathologic variables: patient age at diagnosis, gender, site of the tumor within the colon, stage at diagnosis, and tumor grade. The frequencies of defective MMR, as well as patient and tumor characteristics, were also assessed for associations with the three ethnic groups. Associations with categorical variables were assessed using either contingency table χ² statistics or Fisher's exact test when sample sizes were insufficient for χ² analysis. All analyses were done using the SAS software (25).

Invasive CRC cases included in this study were selected from cases diagnosed from 1984 through 2001. Testing was completed on 329 patients: 165 (50%) Eskimo, 53 (16%) Aleut, and 111 (34%) Indian. Patient and tumor characteristics of the three groups tested are presented in Table 1. Of the 329 patients studied, 174 were women and 155 were men. Although the numbers of women and men were more similar among the Indian and Aleut, there were more women than men tested in the Eskimo group. In addition to ethnicity and gender, data linked to the specimens included: year of diagnosis; age group (<50, 50-59, 60-69, 70-79, and 80+ years); site of tumor site within the colon/rectum (proximal defined as splenic flexure and proximal, and distal defined as descending colon through rectum); stage at diagnosis (local, regional, distant); and tumor histopathology grade (well differentiated, moderately differentiated, poorly differentiated, undifferentiated). No statistically significant differences were noted for any of the tumor-related variables examined among the three ethnic groups (Table 1).

Tumors were examined for the presence of MSI and for the absence of protein expression for hMLH1 and hMSH2 in all 329 patients and also for hMSH6 in a subset of 146 patients. Of the 329 patients examined, 46 (14%) tumors were MSI-H (Table 2). In all MSI-H cases, immunostaining showed absence of expression of one of the DNA MMR proteins. Forty-two tumors (91%) had an absence of protein expression for hMLH1; three tumors (7%) had an absence of expression for hMSH2; and one tumor (2%) showed an atypical pattern of loss of hMSH6 in combination with loss of hMLH1 and hPMS2 (Table 2). All of the MSS and MSI-L cases had normal protein expression patterns. Tumors were classified as having a defective DNA MMR if they showed a MSI-H phenotype and an absence of protein expression for either hMLH1, hMSH2, or hMSH6. On the other hand, tumors that had normal protein expression and were MSS/MSI-L were classified as having proficient MMR. The frequency of defective MMR was then examined among the three ethnic groups (Table 2), with all three groups showing similar results: 14% of the Eskimo CRC patients, 17% of the Aleut patients, and 13% of the Indian patients (P = 0.75).

Because MSI has previously been shown to be associated with a number of clinical and pathologic variables, similar variables were examined for their association with the MMR status in this group of patients. As with other patient samples, CRC characterized by defective MMR was more frequently seen in the proximal colon (85% of CRC with defective MMR versus 37% with proficient MMR, P < 0.0001) and tended to be more poorly differentiated (P < 0.0001; Table 3). Patients with defective MMR also were more likely to be female (P = 0.0007) and to present at a young age (<50 years old) than their counterparts with proficient MMR tumors (P = 0.047). In this study, tumor stage did not correlate with MMR status, in contrast to previous studies in which MSH-H tumors were significantly more likely to be early-stage disease.

Given that this patient sample was not selected for family history of cancer or young age of onset, the finding that patients with defective MMR were more likely to present at a young age was unexpected. Loss of hMSH2 protein expression is most likely due to the presence of a germ line mutation, but this is not the case for hMLH1. In an effort to assess the most likely mechanism of hMLH1 gene inactivation, those tumors exhibiting absence of hMLH1 protein expression were assessed for two somatic changes, namely hypermethylation of the hMLH1 promotor and the presence of the BRAF V600E mutation (Table 4). DNA from 36 of the 42 tumors was available for study. From Eskimo patients, all 19 tumors with absence of hMLH1 that were tested for promoter hypermethylation showed this epigenetic phenomenon. However, this was the case for only 3 of 7 of the Aleut and 5 of the 10 Indian CRC that had an absence of hMLH1 in their tumor (P < 0.0001). The majority of the hypermethylated cases (23 of 27) tested positive for the V600E alteration. Of the nine tumors from the Aleut and Alaska Indian patients that did not exhibit hMLH1 hypermethylation, eight tested negative for V600E. The age distribution (by decade) for both the hMLH1 hypermethylated and nonhypermethylated cases among each of the three ethnic groups is shown in Table 4.

Across the three ethnic groups, we did not detect any differences in clinical and pathologic characteristics of CRC. Overall, evidence of defective DNA MMR was detected in 14% of the CRC among all individuals tested, a frequency similar to that reported in other patient samples (12-24%; refs. 14-16, 26-28). Similarly, the relative involvement of hMLH1 and hMSH2 in defective MMR was essentially the same as that found in a Mayo Clinic series of 257 patients: 48 (94%) of 51 MSI-H had involvement of hMLH1 and 3 of 51 (6%) had an abnormality in the hMSH2 gene (16). Finally, the three ethnic groups had similar frequencies of CRC with defective DNA MMR (P = 0.75).

Several of the clinical and histologic features typically associated with defective MMR were also observed in the Alaska Native study. Tumors tended to be more poorly differentiated, were more likely to arise in women, and occurred more frequently in the proximal colon; these associations have been previously reported in CRC patients (11-15). Early-stage disease has been associated with the MSI phenotype, but we did not find this association in the Alaska Native study.

Of interest, an earlier age of onset for CRC with defective MMR was noted in the Alaska Native group. In cases with defective MMR, an earlier age of onset is typically associated with the presence of germ line mutations whereas an older age of onset is most often found in tumors caused by epigenetic inactivation of hMLH1 (29). To further explore the possible role of these two mechanisms, we evaluated all tumors with absence of hMLH1 expression for both promoter hypermethylation and for somatic BRAF mutations. Unlike hMSH2 and hMSH6, which most often involve germ line mutations, the current data suggest that the most common mechanism for hMLH1 gene inactivation among unselected cases is promoter hypermethylation and less frequently by the presence of either a somatic or germ line mutation the gene itself (16, 19, 20). Because we were not able to test directly for the presence of a germ line mutation, we used the methylation status of the hMLH1 promoter and the mutation status of BRAF as surrogate markers. Approximately 20% of sporadic colon cancers are the result of defective MMR, and the vast majority of these are due to hypermethylation of the hMLH1 promoter. Furthermore, the BRAF-V600E alteration has been shown to be present in the majority of tumors with hypermethylation of the hMLH1 promoter but not in cases with germ line hMLH1 mutations (21-23). Thus, the presence of a germ line mutation or of epigenetic inactivation of hMLH1 can be distinguished by direct assessment of the hMLH1 promoter methylation status and testing for the BRAF V600E mutation. Tumors that have the BRAF V600E mutation and show hMLH1 promoter hypermethylation are most likely sporadic, whereas tumors that show neither are most likely germ line.

Among the hMLH1-related CRC in the Eskimo tested, all 19 exhibited promoter hypermethylation, with 15 of these having somatic BRAF mutations. Among these cases, somatic and not germ line changes seem to be the primary cause of the defective MMR. However, in the Aleut and Indian cases, nearly half of the tumors had patterns more consistent with the presence of germ line hMLH1 mutations (i.e., lack of promoter hypermethylation and lack of the BRAF mutation). The higher apparent frequency of this pattern among the Aleut and Indian cases may help to explain the finding that some of the patients with defective MMR were more likely to present at a young age than those cases with proficient MMR. Thus, the findings in the Aleut and Indian groups suggest a greater role for germ line mutations in CRC in these two groups. However, to assure anonymity, patient identifiers were removed from our study sample. As a result, one potential problem is that we may have a high frequency of related individuals within our study group. Unfortunately, we have no information on the number of related subjects among the cases collected over this 18-year period.

The Alaska Indian population includes the Athabascan, from the interior of Alaska, and the Tlingit, Haida, Eyak, and Tsimshian from Southeastern Alaska (30). The Navajo and Apache tribes are related linguistically to the Athabascan. American Indian people in the contiguous United States have among the lowest rate of CRC of any population in North America (3). Although the CRC rate in the American Indian population in the southwest United States has risen slightly in the past several decades (31), this low rate has persisted in spite of the shift from plants and leaner lower fat protein food sources to higher fat and lower grain and vegetable intakes. The frequency of defective MMR in CRC arising in American Indians is unknown; however, one Navajo kindred has been reported to meet the Amsterdam criteria, with a causative germ line hMLH1 mutation detected within this kindred (32). In the current study, patients were characterized as recorded on the medical records (Eskimo, Indian, and Aleut), and tribal affiliation was not included in the research data file.

A variety of dietary factors, as well as exposure to environmental toxicants, have been associated with the development of CRC (33-35). The traditional diet of the Alaska Native population included predominantly marine mammals, fish, other seafood, wild game, berries, and wild greens. The diet was relatively high in protein, low in carbohydrates, low in fiber, and a large proportion of the fats included polyunsaturated (omega-3) fatty acids. There is no ongoing surveillance of the Alaska Native diet, and there have been limited dietary studies. A study done in 1956 to 1961 showed that, during those years, calorie sources were derived equally from both local (subsistence) and imported foods (36). At the time of this study, the source caloric intake was one-third protein, one-third carbohydrate, and one-third fat. In a more recent study (1987-1988), the nontraditional (imported) foods were shown to comprise an increasing portion of the diet (37). Of potential importance to the development of CRC were the findings that the diet contained a relatively low intake of calcium, fruits, and vegetables compared with the U.S. diet and recommended intake due to multiple factors, including the high cost and limited availability. Folate was not reported but the intake of six other vitamins and nutrients (vitamins A and C, iron, thiamine, riboflavin, and niacin) met the Recommended Daily Allowance. The most unique aspect of the Alaska Native diet continues to be the high intake of fish and seafood, and thus the high proportion of fat from ω-3 fatty acids.

In summary, although the incidence of CRC is high in the Alaska Native population, the relative proportion of defective MMR and the relative involvement of hMLH1 and hMSH2 is similar to that reported for other groups. However, the relative frequency of potential germ line versus epigenetic inactivation of the hMLH1 gene seems to differ among these three ethnic groups. Detailed collection of family history data combined with further testing of tumor tissue will be required to understand the role of genetics and environment in CRC in this population.